# FRONTIERS IN PHYTOLITH RESEARCH

EDITED BY : Martin John Hodson, Terry B. Ball, Rivka Elbaum, Zhaoliang Song and Eric Struyf PUBLISHED IN : Frontiers in Plant Science, Frontiers in Earth Science and Frontiers in Ecology and Evolution

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-774-4 DOI 10.3389/978-2-88963-774-4

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# FRONTIERS IN PHYTOLITH RESEARCH

Topic Editors:

Martin John Hodson, Oxford Brookes University, United Kingdom Terry B. Ball, Brigham Young University, United States Rivka Elbaum, The Hebrew University of Jerusalem, Israel Zhaoliang Song, Tianjin University, China Eric Struyf, University of Antwerp, Belgium

Citation: Hodson, M. J., Ball, T. B., Elbaum, R., Song, Z., Struyf, E., eds. (2020). Frontiers in Phytolith Research. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-774-4

# Table of Contents

*05 Editorial: Frontiers in Phytolith Research* Martin J. Hodson, Zhaoliang Song, Terry B. Ball, Rivka Elbaum and Eric Struyf *08 Silicon Uptake and Localisation in Date Palm (*Phoenix dactylifera*) – A Unique Association With Sclerenchyma* Boris Bokor, Milan Soukup, Marek Vaculík, Peter Vd'ačný, Marieluise Weidinger, Irene Lichtscheidl, Silvia Vávrová, Katarína Šoltys, Humira Sonah, Rupesh Deshmukh, Richard R. Bélanger, Philip J. White, Hamed A. El-Serehy and Alexander Lux *25 Silicon Fertilizer Application Promotes Phytolith Accumulation in Rice Plants* Xing Sun, Qin Liu, Tongtong Tang, Xiang Chen and Xia Luo *32 Combined Silicon-Phosphorus Fertilization Affects the Biomass and Phytolith Stock of Rice Plants* Zimin Li, Fengshan Guo, Jean-Thomas Cornelis, Zhaoliang Song, Xudong Wang and Bruno Delvaux *43 Distributions of Silica and Biopolymer Structural Components in the Spore Elater of* Equisetum arvense*, an Ancient Silicifying Plant* Victor V. Volkov, Graham J. Hickman, Anna Sola-Rabada and Carole C. Perry *58 Spectroscopic Discrimination of Sorghum Silica Phytoliths* Victor M. R. Zancajo, Sabrina Diehn, Nurit Filiba, Gil Goobes, Janina Kneipp and Rivka Elbaum *70 Phytoliths in Inflorescence Bracts: Preliminary Results of an Investigation on Common Panicoideae Plants in China* Yong Ge, Houyuan Lu, Jianping Zhang, Can Wang and Xing Gao *90 Taxonomic Demarcation of* Setaria pumila *(Poir.) Roem. & Schult.,* S. verticillata *(L.) P. Beauv., and* S. viridis *(L.) P. Beauv. (Cenchrinae, Paniceae, Panicoideae, Poaceae) From Phytolith Signatures* Mudassir A. Bhat, Sheikh A. Shakoor, Priya Badgal and Amarjit S. Soodan *122 Palm Phytoliths of Mid-Elevation Andean Forests* Seringe N. Huisman, M. F. Raczka and Crystal N. H. McMichael *130 Bulliform Phytolith Size of Rice and its Correlation With Hydrothermal Environment: A Preliminary Morphological Study on Species in Southern China* Can Wang, Houyuan Lu, Jianping Zhang, Limi Mao and Yong Ge *145 Silicon Supplementation of Rescuegrass Reduces Herbivory by a Grasshopper* Showkat Hamid Mir, Irfan Rashid, Barkat Hussain, Zafar A. Reshi, Rezwana Assad and Irshad A. Sofi *153 Translocation of Phytoliths Within Natural Soil Profiles in Northeast China* Lidan Liu, Dehui Li, Dongmei Jie, Hongyan Liu, Guizai Gao and Nannan Li *168 Influence of Moisture and Temperature Regimes on the Phytolith Assemblage Composition of Mountain Ecosystems of the Mid Latitudes: A Case Study From the Altay Mountains* Marina Y. Solomonova, Mikhail S. Blinnikov, Marina M. Silantyeva and

Natalya Y. Speranskaja

*190 Soil Warming Accelerates Biogeochemical Silica Cycling in a Temperate Forest*

Jonathan Gewirtzman, Jianwu Tang, Jerry M. Melillo, William J. Werner, Andrew C. Kurtz, Robinson W. Fulweiler and Joanna C. Carey

*205 The Role of Macrophytes in Biogenic Silica Storage in Ivory Coast Lagoons*

Yefanlan Jose-Mathieu Koné, Bart Van de Vijver and Jonas Schoelynck


# Editorial: Frontiers in Phytolith Research

#### Martin J. Hodson<sup>1</sup> \*, Zhaoliang Song<sup>2</sup> , Terry B. Ball <sup>3</sup> , Rivka Elbaum<sup>4</sup> and Eric Struyf <sup>5</sup>

*<sup>1</sup> Department of Biological and Medical Sciences, Faculty of Health and Life Sciences, Oxford Brookes University, Oxford, United Kingdom, <sup>2</sup> Institute of the Surface-Earth System Science, Tianjin University, Tianjin, China, <sup>3</sup> Department of Ancient Scripture, Brigham Young University, Provo, UT, United States, <sup>4</sup> R.H. Smith Institute of Plant Sciences and Genetics in Agriculture, The Hebrew University of Jerusalem, Rehovot, Israel, <sup>5</sup> Department of Biology, Global Change Ecology Centre, University of Antwerp, Wilrijk, Belgium*

Keywords: phytolith, silica, silicon, biomineralisation, biogeochemistry, carbon sequestration

#### **Editorial on the Research Topic**

#### **Frontiers in Phytolith Research**

Interest in phytoliths has grown significantly in recent years. The Research Topic is unusual in its highly interdisciplinary nature, and in the huge range of scales covered: from cellular and molecular studies of phytolith formation to investigations focussing on the role of phytoliths in biogeochemical cycling. This Frontiers in Phytolith Research Topic includes high quality work across this whole range of phytolith research.

For phytoliths to form, plants need to absorb silicon (Si) from their environment. Since Ma et al. (2006) first described a Si transporter in the rice root, considerable interest was raised in establishing the molecular basis of plant Si uptake. The Bokor et al. paper in our issue reports a Si transporter in date palm for the first time. Two papers in our collection (Sun et al.; Li et al.) investigate the effects of Si fertilization on phytolith accumulation in rice, and in both cases showed significant increases in deposition. Whether Si transporters are directly involved in the formation of phytoliths remains to be studied.

Plant internal processes and structure also impact on phytoliths. Phytoliths not only vary in shape and size, but also in their chemistry, and this is influenced by the environment in which they form (Hodson, 2016). Carole Perry has worked on the chemistry of silica deposition in plants for many years, and we were pleased to include a paper from her group (Volkov et al.) in our collection. The authors investigate silica and its carbohydrate matrix in the elaters of Equisetum arvense, using Raman and scanning electron microscopy, assisted by density functional theory. Phytolith chemistry has usually been analyzed in bulk samples, but Zancajo et al. investigate individual phytoliths in the leaves of Sorghum bicolor using Raman and synchrotron FTIR microspectroscopies. They show that bilobate silica cells have a different silica molecular structure and type of occluded organic matter compared with prickles and long cells.

One of the areas of phytolith research where we have seen major advances in the last 20 years is morphometrics. This work was further advanced by the publication of the International Code for Phytolith Nomenclature (ICPN) 2.0, while we were in the midst of compiling our collection [International Committee for Phytolith Taxonomy (ICPT), 2019]. This will allow phytolith researchers to accurately describe the morphotypes they find in their work, and to compare their results with scientists around the world. Not surprisingly, anatomical and morphometric research feature strongly in five of the papers we received. Both Ge et al. and Bhat et al. worked on members of the Panicoideae. Ge et al. consider morphological variation in the phytoliths from the inflorescence bracts of 38 weed and crop species in China, while Bhat et al. work on the leaf and synflorescence phytoliths of three Setaria species. In both cases the authors report that it is

Edited and reviewed by: *Sebastian Leuzinger, Auckland University of Technology, New Zealand*

> \*Correspondence: *Martin J. Hodson mjhodson@brookes.ac.uk*

#### Specialty section:

*This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science*

Received: *26 February 2020* Accepted: *27 March 2020* Published: *17 April 2020*

#### Citation:

*Hodson MJ, Song Z, Ball TB, Elbaum R and Struyf E (2020) Editorial: Frontiers in Phytolith Research. Front. Plant Sci. 11:454. doi: 10.3389/fpls.2020.00454* possible to distinguish fairly closely related taxa using phytolith morphological and morphometric traits. Two papers take very different approaches to the study of palm phytoliths (Bokor et al.; Huisman et al.). Bokor et al. work on the phytoliths found in stegmata cells present in roots, stems and leaves. The stegmata are located on the outer surface of sclerenchyma bundles or associated with the vascular bundles. Huisman et al. study the phytoliths of 12 palm species from mid-elevation Andean forests and identify a number of distinctive morphotypes that are characteristic of a particular species. But phytoliths do not always distinguish between related taxa. Wang et al. show that bulliform phytolith size could not be reliably used to distinguish between cultivated rice and three wild rice species, and moreover hydrothermal factors (higher temperature, precipitation and water level) led to increased size.

Plants deposit phytoliths for many reasons. It has been known for many years that plant silica acts as a physical defense against grazing and pathogens. Mir et al. added to this body of literature, showing that Si fertilization of rescuegrass decreases herbivory by a grasshopper. They went on to show that the increased silica content of the plants caused greater mandibular wear of the grasshoppers.

When plant organs die and drop to the ground, they then rot and release phytoliths into the soil. Once released two key processes are important: migration of the phytoliths within the soil profile; and breakdown and dissolution of phytoliths. Liu et al. study the translocation of phytoliths in soil profiles in Northeast China. They find that 22% of phytoliths are translocated beneath the surface, and that translocation depends on phytolith size and aspect ratio. The authors suggest that phytolith translocation should be considered in investigations concerning palaeoclimate and palaeovegetation reconstructions. Strömberg et al. (2018) assessed translocation processes within the soil, but then went on to consider the dissolution and breakdown of phytoliths. The major factor in increasing phytolith solubility was geometric surface to bulk ratio. One area they did not cover was the chemical makeup of the phytoliths, and particularly any differences in the breakdown of cell wall and lumen phytoliths. Hodson reviews this topic and concludes that there is no evidence in the literature that cell wall phytoliths were either more or less soluble.

Phytoliths have found applications in many aspects of ecological work, and Solomonova et al. included in our collection is one example. These authors consider the influence of moisture and temperature on the phytolith assemblages of ecosystems in the Altay Mountains. They are able to distinguish between seven of 13 regionally important plant communities by using aggregated and more detailed phytolith morphotypes. For six communities there is too much overlap in their phytolith morphotypes. This kind of work on modern systems is needed before attempting to reconstruct past ecosystems using phytolith assemblages. Although many papers in our collection will be of use to those working in palaeoecology, palaeoclimatology, and archaeology, unfortunately the major gap in Frontiers in Phytolith Research are studies looking at using phytoliths to reconstruct past ecology, climates or human activities. Readers are referred to Ball et al. (2016) and Strömberg et al. (2018) for recent reviews of work in these areas.

It was Conley (2002) who first emphasized the importance of phytoliths as a sizable pool of Si in the terrestrial biogeochemical cycle. Because biogenic silica is more soluble than other mineral components of the soil (e.g., aluminosilicates) in most subsequent investigations it has been shown to be a significant source of Si for plant uptake. Our collection includes two papers concerning the role of phytoliths in Si cycling. Gewirtzman et al. investigate the effects of soil warming on cycling of Si in a temperate forest. They find that warming increases Si uptake by vegetation and accelerates the internal cycling of silica. In contrast Koné et al. find that biogenic silica storage in the sediments of Ivory Coast lagoons is dominated by diatom frustules and sponge spicules rather than the phytoliths produced by the abundant macrophytes. They conclude that the macrophytes contribute little to biogenic Si storage in sediments but speculate that fragile phytogenic silica structures may affect local silica cycling.

Parr and Sullivan (2005) first suggested that the carbon occluded within phytoliths (so-called PhytOC) might be significant in the global carbon cycle, and that sequestration within phytoliths might have some potential for tackling climate change. Their work created a whole new sub-discipline in phytolith research and it was not surprising that five of the papers submitted to Frontiers in Phytolith Research touched on this area. Two papers (Li et al.; Sun et al.) investigate the effects of fertilization on carbon sequestration in phytoliths from rice. In both cases fertilization has no effect on the carbon content of phytoliths, but it did increase the mass of phytoliths in the plants, and hence the total amounts of carbon sequestered. A further two papers (Chen et al.; Zhang et al.) emphasize the importance of bamboo in carbon sequestration. Chen et al. work on the belowground biomass of monopodial bamboo species in China, and find that this represents an important and overlooked PhytOC stock. Zhang et al. carry out a wider scale investigation of carbon sequestration in phytoliths in the forests of China. They find that sequestration is particularly high in bamboo, and that the litter layer beneath bamboo plants is very high in PhytOC. This could make a very significant contribution to the long term global biogeochemical carbon sink.

In recent years the whole topic of carbon sequestration in phytoliths has become mired in controversy. Some (e.g., Song et al., 2016) are convinced that the original hypothesis of Parr and Sullivan (2005) is correct, and that PhytOC is a highly important store of carbon on a global scale. Others (e.g., Reyerson et al., 2016) consider that carbon sequestration is not significant. The key issue is the extraction procedure used to prepare phytoliths for analysis. Strong extraction may remove carbon from within phytoliths giving low values for PhytOC, and then apparently poor sequestration on a global scale. Weak extraction may leave contaminants on the surface of phytoliths and lead to overestimation of sequestration. Hodson assesses this whole controversy, and attempts to find a way forward. He suggests that cell wall phytoliths are much richer in PhytOC than lumen phytoliths, as demonstrated by Zancajo et al., and that they may be highly significant in global carbon sequestration. Two hypotheses are advanced, one to explain what happens to phytoliths when they are prepared in the laboratory for analysis, and the other what happens in the soil. Hodson concludes that phytoliths probably are an important global carbon store.

The carbon dating of phytoliths has become another controversial area in phytolith research. Discrepancies in dating have been suggested to indicate that it is not a reliable technique, and some workers have suggested the "old carbon hypothesis" to explain these problems (Reyerson et al., 2016). Essentially this involves carbon being taken up from the soil and then selectively deposited in phytoliths. As this carbon will have an older date than that coming from the atmosphere it is postulated to cause problems with dating. However, others (e.g., Piperno, 2016) are critical of this idea and believe the dating problems are due to methodological issues. Zuo and Lu provide a comprehensive review of this topic. They are critical of the "old carbon hypothesis" and suggest that dating of phytoliths often gives consistent results.

Phytolith research is multidisciplinary and undertaken at many different scales. Often work in one area of research throws light on a topic at a different scale. So it is quite possible that the work of Zancajo et al. which suggests that bilobate silica cells in sorghum leaves have a different type of occluded organic

#### REFERENCES


matter compared with prickles and long cells may yet prove important when we consider carbon sequestration and dating. Therefore, phytolith researchers need to be aware of work that is some way from their immediate field of research. If this does not happen then we will all miss out. In his opinion article, Katz suggests that we need to break down the disciplinary barriers within phytolith research to produce a superdiscipline. He ends by stating, "Hence, embedding superdisciplinary thinking in plant silicon and phytolith research can not only advance our field, but increase its impact in the merger of Earth and life sciences into a single superdiscipline. Working toward this goal is a true new frontier for plant silicon and phytolith research, for Earth-life sciences and for science in general." There is much to be said in favor of this idea. We hope that Frontiers in Phytolith Research has, in some way, contributed to advancing the superdiscipline.

#### AUTHOR CONTRIBUTIONS

MH wrote the first draft of the editorial. ZS, TB, RE, and ES all commented on the draft. All authors agreed with the final draft.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Hodson, Song, Ball, Elbaum and Struyf. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Silicon Uptake and Localisation in Date Palm (Phoenix dactylifera) – A Unique Association With Sclerenchyma

Boris Bokor1,2, Milan Soukup1,3, Marek Vaculík1,4, Peter Vd'acný ˇ 5 , Marieluise Weidinger<sup>6</sup> , Irene Lichtscheidl<sup>6</sup> , Silvia Vávrová<sup>7</sup> , Katarína Šoltys2,7, Humira Sonah<sup>8</sup> , Rupesh Deshmukh<sup>8</sup> , Richard R. Bélanger<sup>8</sup> , Philip J. White9,10, Hamed A. El-Serehy<sup>11</sup> and Alexander Lux1,2 \*

<sup>1</sup> Department of Plant Physiology, Faculty of Natural Sciences, Comenius University in Bratislava, Bratislava, Slovakia, <sup>2</sup> Comenius University Science Park, Bratislava, Slovakia, <sup>3</sup> Institute of Chemistry, Slovak Academy of Sciences, Bratislava, Slovakia, <sup>4</sup> Institute of Botany, Plant Science and Biodiversity Centre, Slovak Academy of Sciences, Bratislava, Slovakia, <sup>5</sup> Department of Zoology, Faculty of Natural Sciences, Comenius University in Bratislava, Bratislava, Slovakia, <sup>6</sup> Core Facility of Cell Imaging and Ultrastructure Research, University of Vienna, Vienna, Austria, <sup>7</sup> Department of Molecular Biology, Faculty of Natural Sciences, Comenius University in Bratislava, Bratislava, Slovakia, <sup>8</sup> Department of Plant Science, Université Laval, Quebec, QC, Canada, <sup>9</sup> The James Hutton Institute, Dundee, United Kingdom, <sup>10</sup> Distinguished Scientist Fellowship Program, King Saud University, Riyadh, Saudi Arabia, <sup>11</sup> Zoology Department, College of Science, King Saud University, Riyadh, Saudi Arabia

Date palm (Phoenix dactylifera) can accumulate as much as 1% silicon (Si), but not much is known about the mechanisms inherent to this process. Here, we investigated in detail the uptake, accumulation and distribution of Si in date palms, and the phylogeny of Si transporter genes in plants. We characterized the PdNIP2 transporter following heterologous expression in Xenopus oocytes and used qPCR to determine the relative expression of Si transporter genes. Silicon accumulation and distribution was investigated by light microscopy, scanning electron microscopy coupled with X-ray microanalysis and Raman microspectroscopy. We proved that PdNIP2-1 codes for a functional Si-permeable protein and demonstrated that PdNIP2 transporter genes were constitutively expressed in date palm. Silicon aggregates/phytoliths were found in specific stegmata cells present in roots, stems and leaves and their surfaces were composed of pure silica. Stegmata were organized on the outer surface of the sclerenchyma bundles or associated with the sclerenchyma of the vascular bundles. Phylogenetic analysis clustered NIP2 transporters of the Arecaceae in a sister position to those of the Poaceae. It is suggested, that Si uptake in date palm is mediated by a constitutively expressed Si influx transporter and accumulated as Si aggregates in stegmata cells abundant in the outer surface of the sclerenchyma bundles (fibers).

Keywords: Arecaceae, cell wall composition, date palm (Phoenix dactylifera), phylogenetic analysis, phytoliths, plant anatomy, silicon (Si) transporters, stegmata

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Mohamed M. Hanafi, Putra Malaysia University, Malaysia Philippe Etienne, University of Caen Normandy, France Xinxin Zuo, Fujian Normal University, China

> \*Correspondence: Alexander Lux alexander.lux@uniba.sk

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 16 April 2019 Accepted: 12 July 2019 Published: 13 August 2019

#### Citation:

Bokor B, Soukup M, Vaculík M, Vd'acný P, Weidinger M, ˇ Lichtscheidl I, Vávrová S, Šoltys K, Sonah H, Deshmukh R, Bélanger RR, White PJ, El-Serehy HA and Lux A (2019) Silicon Uptake and Localisation in Date Palm (Phoenix dactylifera) – A Unique Association With Sclerenchyma. Front. Plant Sci. 10:988. doi: 10.3389/fpls.2019.00988

**8**

## INTRODUCTION

fpls-10-00988 August 13, 2019 Time: 15:51 # 2

Silicon (Si) is not considered to be an essential element for plants, but its tissue concentration can exceed that of many essential elements in some plant species (Hodson et al., 2005; White and Brown, 2010). The roles of Si as a beneficial element for plants, protecting them from a variety of abiotic stresses and biotic challenges, have been discussed in the literature for a long time (Epstein, 1999; Coskun et al., 2018).

In most circumstances, plant roots take up Si from the soil solution and it is then transported to the aboveground organs via the xylem (Casey et al., 2003; Mitani et al., 2005). The accumulation of Si varies greatly among plant species and those belonging to the commelinid monocot orders Poales (e.g., cereals, grasses, bromeliads, and sedges) and Arecales (e.g., palms) generally accumulate more Si than other plants (Hodson et al., 2005). The identification of genes encoding proteins responsible for Si transport have shown that Si accumulation is the result of an efficient symplastic pathway mediated by Si influx and efflux transport mechanisms in the plasma membrane of root cells (Ma and Yamaji, 2015). Silicon influx proteins, termed Lsi1, are members of the NIP III (nodulin 26-like intrinsic protein III) group of aquaporin-like proteins belonging to the large MIP (major intrinsic protein) superfamily that contains various classes of integral membrane proteins functioning as diffusion facilitators of water and small uncharged solutes (Wallace and Roberts, 2005; Ma and Yamaji, 2015; Pommerrenig et al., 2015; Deshmukh et al., 2016). Aquaporins in the NIP III group contain two hallmark domains: a unique selectivity filter (ar/R filter, also known as GSGR filter) formed by glycine (G), serine (S), glycine (G) and arginine (R) and two NPA motifs (also referred to as NPA boxes) consisting of asparagine (N), proline (P) and alanine (A) separated by 108 amino acids (Deshmukh et al., 2015; Ma and Yamaji, 2015).

Following its uptake by roots, Si can be deposited in plant tissues in various forms, most frequently in silica cells or silica bodies distributed within the leaf epidermis or as a dense layer beneath the cuticle (Datnoff et al., 2001; Coskun et al., 2018). Other common sites of Si deposition are specialized cells termed stegmata that form a sheath around sclerenchyma fibers attached to vascular bundles or individual fiber bundles in species such as palms in the commelinid monocot order Arecales and orchids in noncommelinid monocot order Asparagales (Møller and Rasmussen, 1984). Root tissues are also sites of Si accumulation in some plant species, with the endodermis being the dominant deposition site, especially in monocots (Sangster and Hodson, 1992; Lux et al., 2003).

Silicon deposition in palms is a well-known, but poorly understood phenomenon. This study is focused on a detailed description of date palm anatomy as it relates to the unique Si distribution in this species and presents novel observations on Si uptake mechanisms in date palms and the phylogenetic relationships between the Si transport proteins of date palms and other Si-accumulating species.

## MATERIALS AND METHODS

#### Plant Cultivation

In our studies, we compared three developmental stages of date palms: young seedlings (ca 1-month-old) grown in hydroponics, 1-year-old plants grown in perlite, and 10-year-old plants grown in soil. For the oocyte experiments, RNA was extracted from roots of 1-week-old date palm and rice plants grown in hydroponics.

Prior to cultivation in hydroponics and perlite, date palm seeds were surface sterilized in 2.5% NaClO solution for 10 min and washed several times with dH2O. After such treatment, germination took about 2.5 weeks. In hydroponics, two different treatments were imposed: a Si− control treatment with Hoagland solution (Hoagland and Arnon, 1950) and without silicon supplementation, and an Si+ treatment with Hoagland solution and Si addition as sodium silicate [Na2O(SiO2)x.xH2O, or given also as Na2O7Si<sup>3</sup> by Sigma-Aldrich] to a final concentration of 1 mM (this compound is referred as Si in the text), or 0.084 g/L of elemental silicon. This Si concentration was chosen because it is similar to the Si concentrations in soil solutions and is recommended for laboratory studies (Epstein, 1994; Liang et al., 2015). Plants were grown in a growth chamber with 12 h light/12 h dark, a light intensity of 200 µmol PAR m−<sup>2</sup> s −1 , relative humidity of approximately 75% and day/night temperatures of 28/24◦C. During the first 5 days, germinated plants were acclimatized to hydroponics by growing them in a half-strength Hoagland solution without Si addition. Subsequently, the two different treatments (Si− and Si+) were initiated and five plants were cultivated in 3 L pots for 21 days. Hoagland solution was renewed every third day and pH was adjusted to a value of 6.2.

Cultivation in 1L pots filled with perlite (68–73% SiO2, 7.5– 15.0% Al2O3, 1.0–2.0% Fe2O3, 0.5–2.0% CaO, 0.2–1.0% MgO, 2.0–5.5% K2O, 2.5–5.0% Na2O, max. 1.0% TiO2, max. 0.2% P2O5, max. 0.3% MnO) lasted about 12 months. Plants (one per pot) were watered once a week with half strength Hoagland solution (200 mL). Plants were grown in a growth chamber with conditions identical to hydroponically cultivated plants.

In addition, 10-year-old plants grown in soil in the greenhouse at the Department of Plant Physiology, Faculty of Natural Sciences, Comenius University in Bratislava, were studied. Plants were watered regularly with tap water and every second year they were transferred to a bigger pot containing fresh sandy-loam soil with a bioavailable Si concentration of 113 ± 15 mg kg−<sup>1</sup> as described by Bokor et al. (2017). The final volume of the pot at the end of the cultivation was 60 L.

#### Light Microscopy

Hand sections were prepared as described by Lux et al. (2015). Cross and longitudinal sections of all organs studied, primary, lateral and adventitious roots, stem, shoot apex, leaf petioles, leaf sheaths, and leaf blades, were examined under a microscope (Axioskop 2 plus, Carl Zeiss, Germany) and documented using a digital camera DP72 (Olympus, Japan).

For general anatomy, both unstained sections and sections cleared with lactic acid and stained in an aqueous 0.05%

(w/v) solution of toluidine blue were used. The Wiesner phloroglucinol-HCl reaction was used to identify lignification of cell walls in individual tissues. Suberin was visualized in sections cleared and stained with a 0.01% (w/v) solution of Fluorol Yellow 088 (FY088; Sigma-Aldrich) in lactic acid at 70◦C for 1 h (Lux et al., 2015) and examined under an epifluorescence microscope (Axioskop 2 plus, Carl Zeiss, Germany; filter set Carl Zeiss N. 25: excitation filter TBP 400 nm + 495 nm + 570 nm, chromatic beam splitter TFT 410 nm + 505 nm + 585 nm, and emission filter TBP 460 nm + 530 nm + 610 nm).

Serial cross and longitudinal sections, of fixed, paraffin embedded and stained sections were used for additional studies of all organs. Briefly, the samples of individual organs were fixed in formalin–acetic acid–alcohol (FAA), dehydrated in a graded ethanol series, transferred to xylene and embedded in paraffin (Johansen, 1940). Sections, 15–20 µm thick, were deparaffinised in xylene and stained with alcian blue/safranin and mounted in Canada balsam. Observation and documentation were performed as described above.

## Scanning Electron Microscopy (SEM) Coupled With X-Ray Microanalysis

Transversely and longitudinally sectioned and air-dried root, stem and leaf tissues were fixed on aluminum stubs covered with a carbon sticker. Surface conductivity was increased by carbon coating, which in turn also resulted in a uniform, approximately 60 nm thick, carbon layer on the tissue surface. The distribution of Si was analyzed with a Jeol JSM-IT300 scanning electron microscope (SEM) equipped with an energy dispersive X-ray (EDX) analyser (EDAX, Octane Plus, Ametek, United States).

Plant phytoliths were examined at several different spots on each of the three plant tissues studied (root, stem, and leaf). Raw data were processed with the TEAM Enhanced ver. 4.3 (EDAX-Ametek, United States) software and all values were expressed as weight % of the total analyzed Si element.

#### Total Si Concentrations in Plant Tissues

At the end of cultivation, the total Si concentration was measured in roots and second fully developed leaves of plants cultivated in perlite; and in roots, shoot apexes, leaf petioles, and leaf blades of plants cultivated in soil. The concentration of Si in the dry biomass of plant samples was determined using atomic absorption spectroscopy (AAS). Plant samples were dried at room temperature and ground to small pieces (<1 mm) with a mortar and a pestle. Digestions of plant samples were carried out in stainless steel coated PTFE pressure vessels ZA-1 (Czechia) in an electric oven at 160◦C for 6 h. Each vessel contained between 0.1 and 0.5 g dried plant sample, 5 ml of concentrated HNO3, 0.25 ml of concentrated HF and 2 ml of 30% H2O2. After digestion, 2 ml of a saturated solution of H3BO<sup>3</sup> was added and the resulting mixture was diluted to 25 ml with redistilled water and stored in a 100-ml polyethylene bottle. Silicon concentrations were determined by a flame atomic absorption spectrometry (AAS Perkin Elmer Model 5000, wavelength 251.6 nm, flame: acetylene-N2O). The concentration of bioavailable Si from the perlite (70 ± 5 mg kg−<sup>1</sup> ) was analyzed according to Rodrigues et al. (2003) with appropriate modifications. After extraction by 0.5 M acetic acid, Si was measured by ICP-MS in place of colorimetric determination using blue silicomolybdous acid procedure as used in the original procedure, and as a quality control certified reference material for Si was analyzed, too. Analyses were performed at a certified laboratory of the Institute of Laboratory Research on Geomaterials (Faculty of Natural Sciences, Comenius University in Bratislava).

## Isolation of Silica Phytoliths

Hand cross-sections from the basal part of the leaf sheath were placed on a microscope slide and a drop of 96% sulfuric acid was added. After 5 min, several drops of distilled water were added, the sample covered with a cover slip and gently pressed to break the digested tissues. The isolated phytolith samples were then used either for dark field light microscopy or for Raman analyses.

#### Raman Microspectroscopy

For Raman analyses, 15 µm thick microtome sections of paraffin embedded samples were prepared, dewaxed with 100% xylene for 30 min (2×) and gradually rehydrated in 20-min steps. A gradual series of mixtures of ethanol and distilled water was used (1:0; 1:0; 0.7:0.3; 0.5:0.5; 0.3:0.7; 0:1; 0:1). Sections were placed on microscope slides, mounted in distilled water, covered with coverslips and sealed with nail polish to avoid water evaporation. Hydrated silica gel was prepared as aqueous suspension of chromatography grade silica gel. Raman spectra were collected with a DXR Raman Microscope (Thermo Fisher Scientific, United States), equipped with a 532 nm laser, using 900 lines mm−<sup>1</sup> grating. Spectra were recorded using 9 mW laser power, 12 s photobleaching time, with 10–30 s acquisition time per collection and eight collections per measurement. At least five spectra per structure were collected and analyzed. Omnic Atlas software (Thermo Fisher Scientific, United States) was used to collect the spectra. Spectral processing was performed using Spectragryph 1.0.7 (F. Menges "Spectragryph – optical spectroscopy software," Version 1.0.7, 2017<sup>1</sup> ). Spectra were baseline-corrected, smoothed (Sawitzky-Golay, 9 points, polynomial order 4) and normalized against a peak at 2895 cm−<sup>1</sup> if not stated otherwise. Spectra are presented as means of all spectra collected from the object analyzed. The reference table used for peak assignments for these spectra are shown in **Supplementary Tables S1**, **S2**. The estimation of S/G-lignin ratio was based on the ratio of peak intensities 1334/1273 cm−<sup>1</sup> (Lupoi and Smith, 2012). The estimation of cellulose crystallinity was based on the ratio of peak intensities 380/1096 cm−<sup>1</sup> (Agarwal et al., 2010).

#### RNA Extraction and cDNA Synthesis

On the third day of hydroponic cultivation and for the next 5 days, root tissues were sampled from plants growing in both Si− and Si+ treatments to evaluate gene expression. Samples (up to 150 mg) were stored at –80◦C before RNA extraction. Total RNA was extracted and treated with DNase I using a Spectrum Plant Total RNA kit (Sigma–Aldrich, United States)

<sup>1</sup>http://www.effemm2.de/spectragryph/

according to the manufacturer's instructions, except for the duration of DNase I treatment which was extended to 60 min. The RNA concentration and sample purity were measured using a NanoDropTM 1000 spectrophotometer (Thermo Fisher Scientific, Germany) and RNA integrity was checked by agarose (1%) gel electrophoresis. The synthesis of the first strand of cDNA was performed using an ImProm-II Reverse Transcription System (Promega, United States), using Oligo(dT)15 primers according to the manufacturer's instructions. A control without RT was performed for each sample to determine whether there were any traces of genomic DNA. Samples containing only cDNA (10-times diluted) were used for qPCR analysis.

#### Plasmid Constructions for Heterologous Expression in Xenopus Oocytes

The cDNA prepared from rice and date palm was used to amplify the open reading frames (ORF) of OsLsi1 and PdNIP2-1. The ORFs amplified using Phusion Taq polymerase (New England Biolabs, Whitby, ON, Canada) were first cloned in a pUC18 plasmid vector and sequenced to confirm the accuracy of the ORFs. For heterologous expression in Xenopus laevis oocytes, the ORFs were further cloned using EcoRI/XbaI restriction sites into the Pol1 vector (PdNIP2- 1EcoR1F: CCGAATTCATGGCTTCCTTTCCGAGAC, PdNIP2- 1Xba1R: GTTCAATTGGAAAATGTTTGATCTAGAGC), a X. laevis oocyte expression vector derived from pGEM and comprising the T7 promoter, the Xenopus globin untranslated regions and a poly(A) tract (Caron et al., 2000). Both the plasmid constructs, OsLsi1-Pol1 and PdNIP2-1-Pol1, were transformed into Escherichia coli TOP10 strain and stored at −80◦C. Correctness of the constructs was checked by sequencing (T7P: TAATACGACTCACTATAGG, Xeno3UTR: GACTCCATTCGGGTGTTCTTG) prior to in vitro translation.

#### Si Transport Assays Using Heterologous Expression in Xenopus Oocytes

Plasmids containing either the OsLsi1 or PdNIP2-1 ORF were recovered from a fresh bacterial culture using a QIAprep Spin Miniprep kit (Qiagen<sup>2</sup> ). Five micrograms of each plasmid was linearized using NheI (Roche<sup>3</sup> ). Digested products were columnpurified using a PCR purification kit (Qiagen), and 1 µg of plasmid DNA was transcribed in vitro using the mMessage mMachine T7 Ultra kit (Ambion<sup>4</sup> ). Complementary RNAs (cRNAs) were purified using the lithium chloride precipitation method as described by the manufacturer and suspended in ultra-pure water.

The oocyte assays were performed as described by Deshmukh et al. (2013) with some minor changes. Oocytes at stage 5 or 6 were injected with 25 nl of 1 ng/nl cRNA or an equal volume of H2O as a negative control. Then oocytes were incubated for 1 day at 18◦C in Barth's (MBS) medium [88 mM NaCl, 1 mM KCI, 2.4 mM NaHCO3, 0.82 mM MgSO4, 0.33 mM Ca(NO3)2·4H20, 0.41 mM CaCl2, 15 mM HEPES, pH 7.6] supplemented with 100 µM each of penicillin and streptomycin. Then, 10 sets of 10 oocytes for each condition were exposed to MBS solution containing 1.7 mM Si for 30 or 60 min. After exposure, oocytes were rinsed in solution containing 0.32 M sucrose and 5.0 mM HEPES (pH 7.4). Si quantification was performed with a Zeeman atomic spectrometer AA240Z (Varian, Palo Alto, CA, United States) equipped with a GTA120 Zeeman graphite tube atomizer. Data from the spectrometer were analyzed using JMP 9.0.2 (SAS Institute Inc.). Three replicates were used for this assay.

#### Primer Design and RT-qPCR

In the NCBI database, two PdNIP2 transcripts (mRNA sequences) with the following accession numbers XM\_008804384.2 for PdNIP2-1 and XM\_008785804.2 for PdNIP2-2 were available for date palm. The primers for the reference gene actin (XM\_008778129.2) and NIP2 genes (**Supplementary Table S3**) were designed using the Primer3plus tool<sup>5</sup> . Gradient PCR was performed to determine annealing temperature of primers. After that, PCR products were checked by agarose (2%) gel electrophoresis and sequenced by the Sanger method to verify product specificity at the Department of Molecular Biology, Faculty of Natural Sciences, Comenius University in Bratislava. Before qPCR analysis, the stability of the reference gene and efficiency of gene amplification was assessed (Livak and Schmittgen, 2001; Pfaffl et al., 2004). The reference gene, PdNIP2-1 and PdNIP2-2 genes were amplified by the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific, Germany) in 96-well plates using a Light Cycler II 480 (Roche, Switzerland). Melt curve analysis of amplification products was included at the end of each run of the qPCR reaction. The main purpose of the melt curve analysis was to check PCR product specificity; i.e., to confirm that only specific amplification and no non-specific PCR products or primer dimers were formed. The relative change in gene expression was estimated according to the Pfaffl method, including the amplification efficiency of the selected genes (Pfaffl, 2001).

#### Bioinformatics and Statistics

Amino acid sequences were aligned using the MAFFT algorithm with one hundred bootstrap repeats on the GUIDANCE2Server<sup>6</sup> (Sela et al., 2015). The confidence level of the resulting base multi sequence alignment (MSA) was estimated by comparing bootstrap trees as guide-trees to the alignment algorithm. Unreliably aligned columns were removed from the MSA at a cutoff value of 0.93. To analyze the effect of masking on tree inferences, all phylogenetic analyses were conducted also on the unmasked MSA.

Phylogenetic trees were constructed using both the Bayesian and the maximum likelihood techniques. Bayesian inference was performed using the computer program MrBayes ver. 3.2.6 (Ronquist et al., 2012) on the CIPRES Portal ver. 3.1<sup>7</sup> , using the

<sup>2</sup>http://www.qiagen.com/

<sup>3</sup>http://www.roche.com

<sup>4</sup>http://www.invitrogen.com/site/us/en/home/brands/ambion.html

<sup>5</sup>http://primer3plus.com/web\_3.0.0/primer3web\_input.htm

<sup>6</sup>http://guidance.tau.ac.il/ver2/

<sup>7</sup>http://www.phylo.org

WAG amino acid substitution model, four independent chains, one million generations and a sample frequency of one hundred. The first 25% of sampled trees were considered as burn-in and discarded. A 50% majority-rule consensus of the remaining trees was computed, and posterior probabilities of its branching pattern were estimated. Maximum likelihood analyses were performed using the computer program PhyML ver. 3.0 on the South of France bioinformatics platform<sup>8</sup> (Guindon et al., 2010), with the SPR tree-rearrangement and 1000 non-parametric bootstrap replicates. The best amino acid substitution model for maximum likelihood analyses was selected automatically, using the Akaike Information Criterion as implemented in PhyML. Bayesian and maximum likelihood trees were computed as unrooted and were rooted a posteriori in FigTree ver. 1.2.3 (Andrew Rambaut<sup>9</sup> ) with the midpoint method.

The 3D structure of proteins was constructed using the Phyre<sup>2</sup> server<sup>10</sup> (Kelley et al., 2015). Profiling of transmembrane domains was done using the TMHMM tool<sup>11</sup> and functional annotation of NIP2-1 like proteins was performed using the Conserved Domain Database<sup>12</sup>. Amino acids were aligned in CLC Sequence Viewer (version 7.7.1) for visualization of NPA motifs and ar/R selectivity filters in PdNIP2 proteins.

The Statgraphics Centurion (version 15.2.05) and Microsoft Excel 365 software were used for statistical evaluation. The differences among group means were assessed by ANOVA (analysis of variance) and LSD (least significant difference) served as a post hoc test. Data from qPCR were evaluated by Student's t test (Microsoft Excel). Statistical significance was attributed at the 0.05 probability level.

#### RESULTS

#### Silicon Accumulation

Silicon accumulated in all organs of the date palm plants studied (**Figure 1**). The concentration of Si in plant tissues varied according to the developmental stage of plants and the cultivation method. The largest Si concentration was found in leaf blades of plants, whether cultivated in soil or perlite (**Figure 1**). The average concentration of Si in leaf blades of 10-year-old palm plants reached ca. 13 g kg−<sup>1</sup> dry weight (1.3% dry weight). The shoot apexes, leaf petioles and adventitious roots of 10-yearold plants had significantly lower Si concentrations than the leaf blades (**Figure 1**). The Si concentration in leaf blades of 1 year-old plants grown in perlite was significantly less than that in leaf blades of 10-year-old plants grown in soil, whereas the Si concentration in roots of 1-year-old plants grown in perlite was significantly larger than that of 10-year-old plants grown in soil (**Figure 1**). The Si concentration in primary roots of plants grown hydroponically was not significantly different from the Si concentration in roots of plants cultivated in perlite or soil.

### Anatomy of Vegetative Organs and Si Deposits

The structural organization of date palm, with the focus on Si deposition, is summarized in **Figures 2**, **3** and **Supplementary Figure S1**. Silicon deposits are present in the form of silica aggregates, termed phytoliths, in specialized small cells, termed stegmata. Stegmata in roots are exclusively attached to the sclerified bundles of fibers in the cortex. These bundles occur rarely in primary seminal roots (**Figure 2A**) but can be numerous in lateral roots (**Figures 2B,C**) and adventitious roots (**Figures 2D–J**). In the thinnest laterals (≤1 mm diameter), only individual bundles formed by 2–4 fibers are developed (**Figure 2C**). In thicker laterals (≥1 mm) one circle of fiber bundles is present formed by ∼10 fibers (**Figure 2B**). In the thickest adventitious roots, the number of fiber bundles can exceed 100 and they are scattered within the whole mid cortical region (**Figures 2D,F**).

Here, we have studied relatively young date palm plants and focus on the presence of sclerifying sheaths of vascular bundles and leaf traces occurring close to the shoot apex (**Figure 3A**). Already these sheaths are accompanied by stegmata accumulating Si.

The anatomy of the simple leaves of young plants is similar to the leaflets of the compound leaves of adult plants (**Figures 3B– E**). Stegmata with phytoliths are present in leaves in two anatomically distinct locations. One location is around the isolated bundles of sclerenchyma fibers occurring immediately subepidermally or deep in the mesophyll covered with axially arranged rows of stegmata (**Figures 3F,G**). The second location is around the sheath of sclerenchyma fibers surrounding the veins with a collateral arrangement of vascular tissues. The stegmata occurring in the petioles and leaf sheaths are of the same type and distribution as in the leaves and leaflets (3 H–L).

#### SEM/EDX and Raman Analysis of Si Phytoliths

A detailed investigation of various date palm tissues was performed to detect the pattern of Si distribution using SEM coupled with X-ray analysis of element distribution (EDX). In roots, stegmata cells containing Si aggregates were positioned on the outer surface of the sclerenchyma bundles (**Figures 4A,B**), organized in rows of cells with an average distance between the individual phytoliths of about 10–12 µm and an average size of Si phytoliths of between 6 and 8 µm (**Figures 4C,D**). Silicon is also present in the shoot apex, mostly in the form of individual Si phytoliths associated with the sclerenchyma of the vascular bundle. In leaves, the X-ray analysis showed that Si was localized in leaf tissues at two sites: as a part of sclerenchyma around the vascular bundles, and as a part of individual sclerenchyma bundles in the leaf mesophyll. Silicon was not detected in the epidermis, nor in association with the cuticle (**Figures 4E,F**). A very dense net of Si aggregates was observed in the leaf sheaths. The size of Si phytoliths varied between 5 and 10 µm, and they were associated with the surface cell layers of sclerenchyma bundles (**Figures 4G,H**).

<sup>8</sup>http://www.atgc-montpellier.fr/phyml/

<sup>9</sup>http://tree.bio.ed.ac.uk/software/figtree/

<sup>10</sup>http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index

<sup>11</sup>www.cbs.dtu.dk/services/TMHMM/

<sup>12</sup>www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml

FIGURE 1 | Silicon concentration in vegetative organs of Phoenix dactylifera cultivated in hydroponics, perlite or soil. The plants cultivated in perlite were 1-year-old in comparison to the well-developed, 10-year-old plants grown in a soil. Different letters indicate significant differences between the treatments at 0.05 level. Values are means (n = 4) ± standard deviation.

In general, stegmata were almost entirely filled by Si phytoliths (**Figures 5A–C**). X-ray analysis of the surface elemental composition of phytoliths revealed two major elements, Si and oxygen (**Figure 5D**). The presence of carbon was attributed to the surface carbon coating of samples prior the analysis. No other elements were detected in phytoliths.

Representative Raman spectra of isolated silica phytoliths and reference spectra of hydrated silica gel and opal were compared (**Figure 5E**). All three spectra were dominated by a broad band in the region 400–490 cm−<sup>1</sup> assigned to Si–O–Si bond-rocking vibration, underlining the amorphous nature of the silicas. In contrast to spectra from opal, spectra from both phytoliths and silica gel exhibited a well-resolved peak near 482 cm−<sup>1</sup> . The broad and asymmetrical band around 800 cm−<sup>1</sup> visible in all three spectra was assigned to symmetric Si–O–Si stretching vibrations arising from the heterogeneities in the geometry of SiO<sup>2</sup> subunits. The Si–O vibrations of non-bridging oxygen within the region 950–1000 cm−<sup>1</sup> reflects the abundance of Si– OH groups (985 cm−<sup>1</sup> ) and the presence of chemical impurities. Whereas the opal spectrum showed relatively low abundance of Si–OH groups, illustrating its compact inner structure, both phytoliths and silica gel exhibited a relatively high abundance of Si–OH groups, indicating a large surface area. A band assigned to asymmetric Si–O–Si stretching vibrations is located between 1050 and 1200 cm−<sup>1</sup> . Here, the phytolith spectra exhibit a peak around 1053 cm−<sup>1</sup> , indicating that some other elements or contaminants might be present.

## Raman Analyses of the Cell Wall Composition

Raman microspectroscopy was used to investigate the cell wall composition of root and leaf tissues (**Supplementary** **Figure S2**). In roots, the relatively thin hypodermal cell walls exhibited signals indicative of suberization and intense lignification with balanced S/G-lignin ratio, relatively high H-lignin content and ferulic/p-coumaric acids. The outer cortical layer displayed similar cell wall composition to the cortical fiber bundles, characterized by high cellulose crystallinity, relatively weak lignification of the cell wall, but intense lignification in the compound middle lamellae. The thin-walled cell strands separating the aerenchyma lacunae in the mid cortex showed high abundance of both aromatic and aliphatic esters. The cell walls of the inner cortex displayed a low abundance of phenolic compounds (1600–1660 cm−<sup>1</sup> ), but their ester-rich constitution was indicated by a broad band between 1660 and 1750 cm−<sup>1</sup> (C=C and C=O stretching). These cells probably represented an early developmental stage of the thin-walled cells of the mid cortex. The endodermis has developed a thick U-shaped cell wall with relatively high content of phenolic compounds (including H-lignin) in comparison to the thin-walled cells of the mid cortex as well as to the fiber walls. In addition, multiple signals associated with lipidic substances indicated suberin deposits and a relatively large amount of ferulic/pcoumaric acid.

The cell walls of the pith sclerenchyma exhibited a qualitatively similar composition to the outer cortex but with a slightly higher degree of wall lignification. Early metaxylem walls were heavily lignified with a high S/G-lignin ratio. The spectra from late metaxylem walls exhibited a very similar profile, but with less wall lignification. The phloem cell walls exhibited a profile associated with simple primary cell walls, displaying a relatively high pectin signal (817 cm−<sup>1</sup> ), a very low signal from phenolic compounds, low cellulose crystallinity and a relatively high abundance of hemicelluloses (region 470– 515 cm−<sup>1</sup> , 1462 cm−<sup>1</sup> ).

## Phylogenetic Placement of Si Transporters From Date Palm

Two putative Si transporters, PdNIP2-1 (XP 008802606.1) and PdNIP2-1 (XP\_008784026.1) share an 87% identity based on a BLAST alignment and both show the hallmark features required for Si transport (**Figure 6A**). The 3D model of both proteins showed an hourglass-like structure (**Figures 6B,C**). The TMHMM tool for prediction of transmembrane domains showed six transmembrane helices for both proteins, identical to the known Si transporters of other plant species (**Figures 6D,E**). Both proteins were classified functionally as membrane channels that are members of the MIP superfamily using this tool.

Phylogenetic analyses showed that the PdNIP2-1 and PdNIP2- 2 transporters from date palm belong to the well-defined group of Si influx transporters previously identified in various plant species (**Figure 7**). Transporter sequences from Arecaceae were clustered together with strong statistical support in Bayesian and maximum likelihood trees. In both phylogenetic analyses, sequences from the family Poaceae were classified in a sister position to those from the Arecaceae, supporting a common phylogenetic ancestry of Si transporters in the monocotyledonous cluster. The NIP2 transporters from dicotyledons formed a distinct, statistically fully supported group (**Figure 7**).

#### Silicon Permeability of PdNIP2-1

To prove the functionality of PdNIP2, X. laevis oocytes expressing PdNIP2-1 were assayed for their ability to accumulate Si (**Figure 8A**). Oocytes expressing either

FIGURE 4 | Scanning electron microscopy images of various P. dactylifera tissues with corresponding maps showing the distribution of Si (violet color). (A,B) Cross section of an adventitious root showing detail of a fiber band (white arrowhead) with adjacent stegmata cells containing Si phytoliths (red arrowheads). Multiple phytoliths are not visible in (A), though detected by EDX (B). (C,D) Longitudinal section through the fiber band (white arrowhead) in an adventitious root. Cell walls of several stegmata cells are disrupted, uncovering Si phytoliths (red arrowheads). (E,F) Cross section of a leaf showing the presence of Si phytoliths in stegamata cells associated with vascular bundles (vb) and fiber bands (white arrowheads). A detail on a stegma (red arrowhead) associated with the vascular bundle sclerenchyma (scl). (G,H) A surface view on a fiber band (white arrowhead) with a dense net of adjacent stegmata. Cell walls of multiple stegmata are disrupted, uncovering Si phytoliths (red arrowheads).

PdNIP2-1 or rice OsLsi1 accumulated significantly more Si than oocytes injected with water, and the same amount after 60-min incubation, confirming the function of PdNIP2-1 as a Si transporter as predicted from in silico analyses (**Figure 8A**).

### Expression of PdNIP2 Si Transporters in Roots of Date Palm Plants

The expression of PdNIP2 genes in roots of date palm plants was constant. The relative amount of the PdNIP2-1 transcripts in roots showed only slight daily variation (**Figure 8B**), varying between 0.62–1.29 and 0.54–1.12 for mRNA in the Si− and Si+ treatments, respectively. The second transcript PdNIP2-2 showed a general increase in expression with length of cultivation in both Si− and Si+ treatments (**Figure 8C**). However, the fold change of this transcript ranged only between 1.0 and 2.35 in the Si− treatment and 1.0–1.92 in Si+ treatment. Because this variation of both transcripts is rather low, we also used the BestKeeper tool to determine the stability of expression of the transcripts in the Si− and Si+ conditions, based on the correlation coefficient of all possible pairs of the candidate reference genes (**Supplementary Table S4**). Both transcripts showed a low (<1) standard deviation of the threshold cycle values (SD CT) and a low SD (<2) of the fold change of gene expression (x-fold), with a strong correlation for all transcripts (**Supplementary Table S4**).

## DISCUSSION

There is little knowledge of the role of Si in date palm, with limited data being available (Fathi, 2014). The present study might stimulate research on this important element

in this economically (FAOSTAT, 2019) and medicinally (Zhang et al., 2017) important species. This study provides conclusive evidence of the presence and functionality of Si influx transporters in date palm and highlights a unique pattern of Si deposition in stegmata cells. Stegmata containing Si phytoliths are present in all organs of the date palm, attached to the surface of sclerenchyma bundles in roots, leaves and stem and to the surface of sclerenchyma sheaths of vascular bundles in stems and leaves (**Figures 2**–**4**).

## Morphology of Phytoliths

The phytoliths of Phoenix dactylifera are classified as spherical, with surface appearance ranging between warty and echinate/spiculate (**Figure 5**) (Prychid et al., 2003; Tomlinson et al., 2011). Such morphology is recognized as typical for palm species and provides a reliable taxonomical identifier (Piperno, 2006; Tomlinson et al., 2011). The hat-shaped/conical phytoliths are the only other morphotype found in palms and can be found, for example, in Caryota, Sclerosperma, and Reinhardtia

(Tomlinson et al., 2011). In contrast, grass phytoliths seem to exhibit much greater morphological variability, where it is possible to identify several morphotypes within the leaf epidermis alone (Kumar et al., 2017).

Mature stegmata possess thick inner tangential and radial cell walls and thin primary outer tangential walls (**Figures 2**, **3**). In the majority of cases, each stegmata contains a single phytolith that occupies almost the entire cell volume. The size of stegmata varies between 10 and 12 µm and the size of phytoliths varies between 6 and 8 µm.

## Phytolith Structure

Raman microspectroscopy confirmed the amorphous nature of the silica framework (a broad band in region 400–490 cm−<sup>1</sup> ), which is a well-known attribute of silica phytoliths in general (Currie and Perry, 2007). A well-resolved peak near 482 cm−<sup>1</sup> and a relatively strong signal near 985 cm−<sup>1</sup> further indicated a large surface area of the silica and suggested that the phytoliths have a microporous structure (Iqbal and Vepˇrek, 1982; Gailliez-Degremont et al., 1997). This is consistent with the study by Lins et al. (2002), revealing the porous structure of phytoliths in the palm Syagrus coronata. A high abundance of superficial – OH groups might favor the adsorption of new silica species via hydrogen bonding (Coradin and Lopez, 2003) and enable the growth of the phytolith. According to Lins et al. (2002) the phytoliths of S. coronata were composed of granules of varying size and morphology. This feature is reflected in the Raman spectra by a broad band around 1200 cm−<sup>1</sup> , indicating that multiple degrees of silicate unit polymerization are present in the phytoliths of date palm (McMillan and Remmele, 1986). This might have resulted from contaminants disrupting the silica framework during polymerization (McMillan, 1984; Marsich et al., 2009).

## Phytolith Association With Cell Walls

The phytoliths from date palm do not seem to contain any organic backbone (**Figure 5**), which was also reported for the palm S. coronata (Lins et al., 2002). In contrast to palms, the phytoliths of grasses are typically associated with the cell walls, particularly, within lignified tissues (Guerriero et al., 2016; Kumar et al., 2017). Raman signals from their scaffolding organic materials can be detected even if harsh procedures are used to isolate phytoliths (Gallagher et al., 2015), usually indicating the presence of phenolic compounds and hemicelluloses (Guerriero et al., 2016; Soukup et al., 2017). Recent studies suggest that lignification might be required to initiate silica deposition (Zhang et al., 2013; Soukup et al., 2017). The association of phytoliths with lignified cell walls has also been reported in dicots, despite the fact that they have low tissue Si concentrations (Scurfield et al., 1974; Hodson et al., 2005). It is speculated that a tradeoff between the accumulation of silica and lignin might occur in plants (Schoelynck et al., 2010; Yamamoto et al., 2012; Klotzbücher et al., 2018). Such a phenomenon is often considered beneficial, with the cost of silicification being estimated to be only 3.7% that of lignification (Raven, 1983). However, although the stiffness provided by these two components might be comparable,

FIGURE 8 | (A) Silicon influx transport activity of PdNIP2-1 from date palm evaluated at two different time points in Xenopus oocyte assays. Oocytes injected with OsLsi1 from rice, or water were used as positive and negative controls, respectively. Values are means ± standard deviation. Different letters indicate significant differences in the same time point. The relative transcript level of PdNIP2-1 (B) and PdNIP2-2 (C) genes in roots of hydroponically grown date palm seedlings in the Si– treatment (orange line) and the Si+ treatment (blue line) from the third to the seventh day of cultivation. Gene expression for the control was set as 1.0. Statistically significant differences between control and treated plants were analyzed by Student's t test and are denoted as <sup>∗</sup>P < 0.05. Values are means ± standard deviation. The mean values are based on three technical and three biological replicates.

they are not entirely interchangeable due to the much lower density of lignin and its water repelling properties (Raven, 1983; Soukup et al., 2017). However, unlike grasses, Si phytoliths in palms are probably formed intracellularly in the vacuole and seemingly without an organic backbone. Schmitt et al. (1995) performed a detailed TEM study of stegmata ontogenesis in the rattan palm species Calamus axillaris. They concluded that the "silica-body" grows within the vacuole. The growth of the "silica body" is probably controlled via active Si accumulation progressively supersaturating the vacuole, and by additional modulation of its physico-chemical environment.

Anatomical observations of date palm show that stegmata that are almost completely filled with Si phytoliths are very abundant near lignified tissues, principally in the outer surface of the sclerenchyma bundles (fibers) in roots, stem and shoots of date palm (**Figures 2**, **3**). Therefore, we also performed Raman spectra analysis of cell walls in lignified tissues. Despite Raman spectra from the fiber cell walls indicating relatively weak lignification, they exhibited good responsiveness to Wiesner reaction (phloroglucinol-HCl). This can be associated with a relatively high abundance of sinapyl/coniferyl aldehydes, which are the key cell wall reagents in this reaction. Furthermore, high content of phenolic aldehydes in the lignin polymer indicate an early stage of lignification (Pomar et al., 2002). In older tissues, additional H-lignin signals appeared in the spectra and the S/G-lignin ratios declined, suggesting that in later stages of the cell wall development predominantly G- and H-lignin were deposited. Relatively weak lignification of the cell wall, high cellulose crystallinity and strong lignification of the compound middle lamellae indicate the gelatinous character of these fibers (Mellerowicz and Gorshkova, 2012). As such, these fibers might provide adjustable mechanical support, that is gradually stabilized by the deposition of lignin as the tissue matures and the organ achieves its optimal position in the environment. This anatomical trait might have substituted for secondary growth, allowing palms to achieve a stable erect posture of the trunk with much lower metabolic costs invested into rigid mechanical tissues.

#### The Role of Silica Phytoliths

Silica phytoliths are traditionally perceived as structures supporting the mechanical properties of plant tissues (Currie and Perry, 2007; Yamanaka et al., 2009). The abrasive nature of silica also deters grazing animals and phytophagous insects (Massey and Hartley, 2009). Moreover, leaf phytoliths might facilitate the transmittance of light to the mesophyll and improve the efficiency of photosynthesis (Sato et al., 2016). Despite these benefits, demands driving the evolution of silica phytolith formation are still unclear (Strömberg et al., 2016). A contrasting evolutionary perspective views silicic acid as a potentially toxic substance and controlled silicification as a mechanism for its detoxification (Exley, 2015). In concentrations exceeding 2 mM, silicic acid is prone to polymerize and might lead to silica scaling on the surfaces of membranes or enzymes and impair their functionality. On the other hand, it offers protection against fungi and insects and might stabilize the membrane against harmful effects (Coskun et al., 2018). So far, the roles of silica phytoliths in palms have not been assessed experimentally. Besides possible prophylactic roles, intracellular formation of Si phytoliths might indicate a role in harnessing excess Si accumulated by the plant. This might be crucial for the longevity of palm tissues and/or slow progression of fiber lignification. For instance, up to 20–30% less lignin

was recorded in rice straws due to the silica-lignin trade-off (Klotzbücher et al., 2018), and in aged bamboo leaves, epidermal silicification was reported to extend to chlorenchyma and reduce the leaf photosynthetic efficiency (Motomura et al., 2008). Curiously, a negative correlation between leaf longevity and silicon content was found across various plant groups (Cooke and Leishmann, 2011).

#### Molecular Aspects of Si Transport

The Lsi1 transporter, which mediates Si influx to roots, was first discovered in rice plants (Ma et al., 2006). Since then, the list of plant Si transporters has been extended to include those of many other species (Yamaji et al., 2008; Chiba et al., 2009; Mitani et al., 2009, Mitani-Ueno et al., 2011; Montpetit et al., 2012; Vivancos et al., 2016; Ouellette et al., 2017). The NIP2 transporters, especially the well characterized NIP2-1 (Lsi1), have a role in Si uptake from soil to root cells and are, therefore, intimately involved in Si accumulation by flowering plants. In addition, a more efficient NP3,1 aquaporin has been identified in horsetail (Equisetum arvense) that contains a STAR pore in contrast to the GSGR pore in monocots including date palm (Grégoire et al., 2012). However, in our study, we take only NIP2 proteins into consideration. For this reason, we focused the molecular study on the properties of PdNIP2 in date palm roots. In our study, transcriptomic data, bioinformatic analyses and oocyte assays revealed the presence and functionality of PdNIP2 transporters in P. dactylifera similar to those known in other plants. These proteins share the hallmark features, such as the ar/R pore and the 108 amino acid sequence between the NPA loops required for Si transport across the plasma membrane (Deshmukh et al., 2015). The permeability of PdNIP2-1 to Si was proven to be comparable to that of rice Lsi1 using a Xenopus oocyte bioassay, a heterologous expression system that has proven reliable for testing the functionality of Si transporters. Using Si as a substrate rather than germanium, our data have also eliminated any possible complication associated with a surrogate substrate (Garneau et al., 2018). A phylogenetic analysis clustered the sequences of NIP2 transporters from the Arecaceae separately, but in a sister position to the Poaceae.

The expression of PdNIP2 genes in roots of hydroponically grown date palm plants was relatively unaffected by the presence or absence of Si in the growth medium (**Figure 8**). Plant species appear to differ in the effects of rhizosphere Si supply on the expression of NIP2 genes and their expression can be up-regulated, down-regulated or unaffected by Si addition to cultivation media (for a detailed review, see Ma and Yamaji, 2015). Analysis using the BestKeeper tool suggests that both PdNIP2-1 and PdNIP2-2 have the transcriptional attributes of a reference gene, although PdNIP2-1 is a better reference gene than PdNIP2-2. It is possible that the constitutively large Si accumulation in P. dactylifera plants might be a consequence of the relatively high stable expression of the PdNIP2 genes that are most probably responsible for Si uptake. In contrast, plants that do not accumulate Si, especially dicots, have a constitutively low expression of NIP2 genes that is even supressed by the presence of Si in cultivation media, as for example NIP2-1 (XM\_013836541) in Brassica napus (Haddad et al., 2019), which might explain smaller accumulation of Si by dicots than monocots. We found homologous sequences to OsLsi2 and OsLsi6 transcripts in the sequence of P. dactylifera. It is, therefore, suggested that the uptake of Si from soil into root epidermal cells is mediated by PdNIP2. Silicon is subsequently transported from cortical cells to the xylem by a Lsi2-like protein and translocated and distributed in leaves by a Lsi6 like proteins.

## CONCLUSION

In conclusion, Si is accumulated in all tissues of P. dactylifera plants, where Si aggregates are present in stegmata. In contrast to grasses, in which Si is generally associated with epidermal tissues, the stegmata of palms are abundant in the outer surface of the sclerenchyma bundles (fibers) present in roots, shoot apex, leaf petioles and blades with the diameter of Si aggregates/phytoliths ranging from 6 to 8 µm. The surface of phytoliths is composed of only silicon and oxygen, without any organic constituents. The analysis of the fiber cell walls suggests they possess a gelatinous character and together with Si phytoliths might provide strong mechanical support for the plant. Again, in contrast to grasses, in which Si phytoliths are mostly associated with cell walls, those of P. dactylifera appear to be formed intracellularly. As P. dactylifera is a Si accumulator homologous sequences of Lsi genes typical for grasses, which are also Si accumulators, were predicted from its genome and found to be functional. Phylogenetic analysis of those transporters within Arecaceae, suggested that they occupied a sister clade to those of the Poaceae and both were distinct from those of dicots. It is likely that, as the palms and grasses diverged, different patterns of Si accumulation became established in each clade.

## DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

## AUTHOR CONTRIBUTIONS

BB, KŠ, and SV carried out the gene expression study, bioinformatic analysis and other molecular biology experiments. PV carried out the phylogenetic analyses. RB, RD, and HS carried out oocyte assays and functional annotation of NIP2-1 gene. MS carried out the cell wall analysis. AL, MV, MW, and IL carried out the scanning electron microscopy coupled with X-ray microanalysis. AL carried out the anatomical study, designed the research together with PW and HE-S, and supervised the project. All authors discussed the results and commented on the manuscript.

#### FUNDING

fpls-10-00988 August 13, 2019 Time: 15:51 # 15

This work was the result of the project implementation: Comenius University in Bratislava Science Park supported by the Research and Development Operational Programme funded by the ERDF (Grant No. ITMS 26240220086). This work was also supported by the Slovak Grant Agency VEGA by grant VEGA 1/0605/17; partially supported by the Slovak Research and Development Agency under the Contract No. APVV-17-0164; and the Distinguished Scientist Fellowship Program of King Saud University (PW and HE-S). The work at The James Hutton Institute was supported by the Scottish Government Rural and Environment Research and Analysis Directorate.

#### REFERENCES


Datnoff, L. E., Snyder, G. H., and Korndörfer, G. H. (2001). Silicon in Agriculture. Studies in Plant Science, 8. Amsterdam: Elsevier.

Deshmukh, R. K., Sonah, H., and Bélanger, R. R. (2016). Plant aquaporins: genomewide identification, transcriptomics, proteomics, and advanced analytical tools. Front. Plant Sci. 7:1896. doi: 10.3389/fpls.2016.01896

Deshmukh, R. K., Vivancos, J., Guérin, V., Sonah, H., Labbé, C., Belzile, F., et al. (2013). Identification and functional characterization of silicon transporters in soybean using comparative genomics of major intrinsic proteins in arabidopsis and rice. Plant Mol. Biol. 83, 303–315. doi: 10.1007/s11103-013-0087-3


Epstein, E. (1999). Silicon. Annu. Rev. Plant Physiol. Plant Mol. Biol. 50, 641–664.

Exley, C. (2015). A possible mechanism of biological silicification in plants. Front. Plant Sci. 6:853. doi: 10.3389/fpls.2015.00853

FAOSTAT (2019). Available at: http://faostat.fao.org (accessed April 16, 2019).

#### ACKNOWLEDGMENTS

Technical assistance of Mrs. Zuzana Šulavíková was gratefully appreciated. HS, RD, and RB would like to thank the Natural Sciences and Engineering Research Council of Canada (NSERC), the Fonds de recherche du Québec – Nature et technologies (FRQNT), and the Canada Research Chairs Program.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00988/ full#supplementary-material



(phloroglucinol-HCl) reaction. Protoplasma 220, 17–28. doi: 10.1007/s00709- 002-0030-y



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Bokor, Soukup, Vaculík, Vd'aˇcný, Weidinger, Lichtscheidl, Vávrová, Šoltys, Sonah, Deshmukh, Bélanger, White, El-Serehy and Lux. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Silicon Fertilizer Application Promotes Phytolith Accumulation in Rice Plants

#### Xing Sun1,2, Qin Liu<sup>1</sup> \*, Tongtong Tang<sup>1</sup> , Xiang Chen<sup>1</sup> and Xia Luo<sup>2</sup>

1 Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China, <sup>2</sup> School of Biological Science and Food Engineering, Chuzhou University, Chuzhou, China

In this study, a pot experiment was designed to elucidate the effect of varying dosages of silicon (Si) fertilizer application in Si-deficient and enriched paddy soils on rice phytolith and carbon (C) bio-sequestration within phytoliths (PhytOC). The maximum Si fertilizer dosage treatment (XG3) in the Si-deficit paddy soil resulted in an increase in the rice phytolith content by 100.77% in the stem, 29.46% in the sheath and 36.84% in the leaf compared to treatment without Si fertilizer treatment (CK). However, the maximum Si fertilizer dosage treatment (WG3) in the Si -enriched soil increased the rice phytolith content by only 32.83% in the stem, 27.01% in the sheath and 32.06% in the leaf. Overall, Si fertilizer application significantly (p < 0.05) increased the content of the rice phytoliths in the stem, leaf and sheath in both the Si-deficient and enriched paddy soils, and the statistical results showed a positive correlation between the amount of Si fertilizer applied and the rice phytolith content, with correlation coefficients of 0.998 (p < 0.01) in the Si-deficient soil and 0.952 (p < 0.05) in the Si-enriched soil. In addition, the existence of phytoliths in the stem, leaf, and sheath of rice and its content in the Si-enriched soil were markedly higher than that in the Si-deficient soil. Therefore, Si fertilizer application helped to improve the phytolith content of the rice plant.

#### Edited by:

Zhaoliang Song, Tianjin University, China

#### Reviewed by:

Xinxin Zuo, Fujian Normal University, China Yong Ge, Institute of Vertebrate Paleontology and Paleoanthropology (CAS), China

> \*Correspondence: Qin Liu qliu@issas.ac.cn

#### Specialty section:

This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science

Received: 12 December 2018 Accepted: 21 March 2019 Published: 16 April 2019

#### Citation:

Sun X, Liu Q, Tang T, Chen X and Luo X (2019) Silicon Fertilizer Application Promotes Phytolith Accumulation in Rice Plants. Front. Plant Sci. 10:425. doi: 10.3389/fpls.2019.00425 Keywords: Si fertilizer, phytolith accumulation, Si-deficient paddy soils, PhytOC, rice organs

## INTRODUCTION

Phytoliths derive from bio-mineralization in plants and usually take the shape of the plant cell or cell spatium where Si is deposited. The phytolith content of plants ranges from less than 50 g kg−<sup>1</sup> to as high as 150 g kg−<sup>1</sup> (Epstein, 1994; Parr et al., 2010; Song et al., 2013, 2017; Ji et al., 2017), mainly due to phylogenetic differences in Si requirements of most dicotyledons and some Gramineae (Hodson et al., 2005), as well as the amount of available silica in the environment (Seyfferth et al., 2013; Guo et al., 2015; Si et al., 2018; Wen et al., 2018).

Rice is a staple crop, with a global planting area of approximately 1.64 × 10<sup>8</sup> ha as of 2014 (Prajapati et al., 2016). When rice is harvested, the rice straw and husks are removed from the paddy field and used for other purposes, including animal feeding and firewood, or simply incinerated (Savant et al., 1996). Thus, most of the Si taken up by rice is removed from a field when the rice straw is removed, and the loss of SiO<sup>2</sup> is from 75 to 130 kg hm−<sup>2</sup> every production season (Zhang et al., 2014). Such large losses of Si make it difficult to maintain the balance of Si in soils from natural weathering alone. Currently, most paddy soils in China are Si-deficient. For example, 73% of paddy soils in Zhejiang Province and approximately 60% in Henan Province are Si-deficient (Cai, 2015). Some research has shown that Si fertilizer application can significantly increase the biomass of rice (Wu et al., 2014; Zhang et al., 2014).

In plants, monosilicic acid is taken up from the soil by a specific transporter (Ma et al., 2006; Song et al., 2014a) and deposited throughout the cellular structures, thereby forming amorphous Si particles known as "phytoliths" (Piperno, 1988; Pearsall, 1989). There is a significant correlation between the Si content and the phytolith content of crop materials, including the leaves, stems and sheaths, and the Si concentration of the plant phytoliths is approximately 90% (Song et al., 2014a).

Phytoliths can occlude small amounts of many elements, such as C, N, S, and so on (Kameník et al., 2013; Li et al., 2014; Anala and Nambisan, 2015). The C-occluded content of phytoliths ranges from less than 1 g kg−<sup>1</sup> to as high as 100 g kg−<sup>1</sup> (Clarke, 2003). This PhytOC can be stored in the soil for thousands of years (Parr and Sullivan, 2005). Thus, it plays a vital role in global carbon (C) pools (Song et al., 2014a). The Si cycle is tightly coupled to the C cycle, and this interaction is relevant for research on climate change (Chadwick et al., 1994). The formation of phytoliths in rice plants depends not only on the crops (Li et al., 2013a; Guo et al., 2015) but also on the plant cultivars (Hodson et al., 2005; Henriet et al., 2008; Yang et al., 2015), the soil's Si availability (Henriet et al., 2008; Klotzbücher et al., 2018) and so on.

The application of Si fertilizer in soils with different available Si contents needs further study regarding the accumulation of phytoliths in rice. Thus, in this work, we designed a pot experiment to elucidate the effect of varying dosages of Si fertilizer application on the rice phytolith and PhytOC contents of plants grown in Si-deficient and enriched paddy soils.

#### MATERIALS AND METHODS

#### Experimental Soils and Rice Cultivar

The Si-deficient paddy soil (red paddy soil) was obtained from Yangliu Town, Xuanchen City, Anhui Province, China. The Si-enriched paddy soil (Wushan soils) was obtained from the Changshu Agroecological Experimental Station, Chinese Academy of Sciences. The base is located in Xinzhuang County, South Changshu, Suzhou, Jiangsu Province, China. The physicochemical properties of the two soils are shown in **Table 1**.

The rice cultivar (Oryza sativa) Nanjing 46 was obtained from the Changshu Agroecological Experimental Station, Chinese Academy of Sciences.

#### Pot Experiment

Two soils (Si-deficient and enriched paddy soils) were selected from Xuanchen City and the Changshu Agroecological Experimental Station, Chinese Academy of Sciences, respectively. Four available Si dosages were designed in the pot experiments: (1) CK (Si fertilizer not applied); (2) low slag Si fertilizer I (SiO2150 kg ha−<sup>1</sup> ); (3) high slag Si fertilizer II (300 kg ha−<sup>1</sup> ); and (4) high slag Si fertilizer III (600 kg ha−<sup>1</sup> ). Thus, this experiment comprised 8 treatments repeated 3 times. Two soils were placed in the pot bowl for a total volume of 0.0175 m<sup>3</sup> ; each pot contained N 46%, P2O<sup>5</sup> 13.5%, and K2O 60%, Si fertilizer was applied as the base fertilizer and three rice plants were planted in every pot. Pots were placed in the greenhouse of the Changshu Agroecological Experimental Station, Chinese Academy of Sciences in June 2014, and the whole rice growth period was maintained using conventional management.

#### Sample Preparation

After the rice cultivar harvest, each rice plant was separated into five different organs: sheath, leaf, root, stem, and grain. All rice samples were rinsed twice in distilled water, placed in an ultrasonic bath for 20 min and subsequently dried in oven at 70◦C for 24 h. After hulling, the rice organ samples were stored for phytolith extraction and PhytOC determination.

#### Phytolith Extraction From Rice Organs and PhytOC Analysis

The phytolith extraction was used for a revised wet digestion measurement previously described by Zuo and Lü (2011); Sun et al. (2016). Phytolith extraction sample assemblages were installed on glass slides in Balsam Canada mounting medium. The slides were viewed at 400 × magnification using a microscope (Jiangnan XP-213, China) fitted with a polarizing filter and a 5.0 MP color CCD camera to ensure the absence of organic material residue as shown by Parr et al. (2010; **Figure 1**). The PhytOC was measured using an Elemental Analyzer 3000 (GmbH Company, Germany).

#### Statistical Analyses

The mean values of all parameters were calculated from the determination of three replicates, and the standard errors of the means were determined. A one-way ANOVA was used to measure the significance of the results between different varieties, and Tukey's multiple range tests (p < 0.05) were subsequently performed. All the statistical analyses were performed using SPSS v.17 for Windows.

#### RESULTS

#### Phytolith and C Contents of the Phytoliths in the Rice Organs

With an increase in the application of the Si fertilizer dosages, the content of the phytoliths in the rice organs was increased in the Si-deficient red paddy soil (**Table 2**). For example, the content of the phytoliths in the XG3 (26.10 g kg−<sup>1</sup> ) and XG2 (18.50 g kg−<sup>1</sup> ) stems was significantly (p < 0.05) higher than that of the control (13.00 g kg−<sup>1</sup> ), and the rate increased by 100.7 and 42.3%, respectively. In addition, the content of the phytoliths of XG1 in the stem was not significantly (p > 0.05) different than that of the control. However, the content of the phytoliths in the rice sheath and leaf could be significantly (p < 0.05) increased by the application of all the Si fertilizer dosages. The content of the phytoliths in the XG3 treatment rice grains could only be increased by a high dose of Si fertilizer application. However, the content of the phytoliths in all the root treatments was not significantly (p > 0.05) different from that of the control.

TABLE 1 | Basic chemical properties of the two soils.

fpls-10-00425 April 14, 2019 Time: 11:9 # 3


FIGURE 1 | Optical microscope images of phytoliths extracted from the rice samples using the wet ashing method according to Zuo and Lü (2011) and Sun et al. (2016); magnification 400×, scale bar 30 µm.


W, Wushan soil; X, Red paddy soil; G, Leg silicon fertilizer. Different lowercase letters after the data mean that the difference between different types of Si-fertilizer dosage treatments are significant (p < 0.05).

With the increase in the application of the Si fertilizer doses, the content of the phytolith in the rice organs could be increased in the Si-enriched Wushan paddy soil (**Table 2**). For example, the content of the phytoliths of the WG3 (100.60 g kg−<sup>1</sup> ) and WG2 (93.13 g kg−<sup>1</sup> ) leaves was significantly (p < 0.05) higher than that of the control (76.18 g kg−<sup>1</sup> ), and the increase in the rate was 32.06 and 22.25%, respectively. In addition, the content of the phytoliths of WG1 in the stem was not significantly (p > 0.05) different from that of the control. However, the content of the phytoliths in the other rice organs could be significantly (p < 0.05) increased by the application of high Si fertilizer dosages.

Thus, different Si fertilizer doses might increase the content of the phytoliths in the rice organs in either Si-deficient red paddy soil or Si-enriched Wushan paddy soil. The C content in the phytoliths in the organs was not affected by the increase in the Si fertilizer dose. However, the content of the C in the phytoliths was different in all the organs. Generally the content of the C of the leaf phytoliths was higher than that of the other organs (**Table 3**).

## Phytolith Content and the Estimated PhytOC Fluxes in Whole Rice Plants

Compared with the control treatment, the content of phytoliths in the whole rice plant was significantly (p < 0.05) increased by the use of a high Si fertilizer dose in the two types of soils (**Table 4**). The C content of the phytoliths and the PhytOC content of the dry organ weights were not significantly (p > 0.05) different in the rice plant. In Si-deficient red paddy soil, the estimated PhytOC fluxes were calculated by the content and proportion of the phytoliths and the C content of the phytoliths in each part of the rice plant. The results showed that the application of Si fertilizer could significantly (p < 0.05) increase the content of the estimated PhytOC fluxes in the whole plant with the increase in the Si fertilizer dosage. The estimated PhytOC fluxes of the XG2 (11.36 kg-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> ) and XG3 (12.93 kg-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> ) treatments were 43.04 and 49.70%, respectively, and were significantly (p < 0.05) higher than those of the control treatment (8.41 kg-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> ). In the Si-enriched soil, the phytolith content of all the Si fertilizer


W, Wushan soil; X, Red paddy soil; G, Leg silicon fertilizer. Different lowercase letters after the data mean that the difference between different types of Si-fertilizer dosage treatments is significant (p < 0.05).

TABLE 4 | Different effects of silicon fertilizers on rice plant content of phytoliths, C content of phytoliths, PhytOC content of dry organ weight, and the estimated PhytOC fluxes per ha in kg of CO<sup>2</sup> equivalents (kg∼e∼ CO2) for rice.


W, Wushan soil; X, Red paddy soil; G, Leg silicon fertilizer. Different lowercase letters after the data mean that the difference between different types of Si-fertilizer dosage treatments are significant (p < 0.05).

treatments in the rice plants was higher than that of the control treatment, but it was not significantly (p > 0.05) different in all the Si fertilizer treatments compared with the control treatment. The estimated PhytOC fluxes of WG1 were 1.3% lower than those of the control.

#### The Correlation Coefficients Between the Six Variables of the Red Paddy Soil

As shown in **Table 5**, the coefficient of variation in the different factors in the Si-deficient red paddy soils was high, illustrating considerable variation among these different Si fertilizer dosages. The results demonstrated that there was a significant correlation (R = 0.998 and p < 0.01) between the phytolith content and the Si fertilizer dose. The C contents of the phytoliths were not correlated (R = −0.177 and p > 0.05) with the phytolith content in the rice plants treated with different fertilizer doses. The correlation coefficient was 0.986, indicating a significant relationship (p < 0.05) between the phytolith content and the estimated PhytOC fluxes. The biomass of the rice was significantly related to the phytolith content (R = 0.972 and p < 0.05) and the estimated PhytOC fluxes (R = 0.994 and p < 0.01).

#### The Correlation Coefficients Between the Six Variables of the Wushan Soil

As shown in **Table 6**, the coefficient of variation in the different factors in the Si-deficient red paddy soils was high, illustrating considerable variation among the different Si fertilizer doses. The results demonstrated that there was a significant correlation (R = 0.952 and p < 0.05) between the phytolith content and the Si fertilizer dose. The C contents of the phytoliths were not correlated (R = −0.035 and p > 0.05) with the phytolith content in the rice plants of different fertilizer treatments. The correlation coefficient was 0.598 and there was significant correlation (p > 0.05) between the phytolith content and the estimated PhytOC fluxes. The biomass of the rice was significantly correlated with the phytolith content (R = −0.890 and p > 0.05) and the estimated PhytOC fluxes (R = 0.076 and p > 0.05).

#### DISCUSSION

Rice accumulates Si (Seyfferth et al., 2013), and the Si concentration is approximately 10–15% in the rice plant (Marschner, 1995), with approximately 90% of the Si present in the phytolith (Wang, 1998). There was a significant correlation between the Si content and the phytolith content of the crop materials, such as the phytolith contents of the rice leaves, stems and sheaths (Song et al., 2014a). The shape of the phytoliths in the different rice organs varied (e.g., double-peaked, bulliform, and parallel dumbbell phytoliths) (Prajapati et al., 2016). Prajapati et al. (2016) reported that the phytolith content in the different rice organs (stem, sheath, leaf, and grain) ranged from 0.14 to 26.4 g kg−<sup>1</sup> . Similar results and trends were reported by other

TABLE 5 | The correlation coefficients between the six variables of the red paddy soil.


<sup>∗</sup>Correlation is significant at the 0.05 level (2-tailed). ∗∗Correlation is significant at the 0.01 level (2-tailed).

TABLE 6 | The correlation coefficients between the six variables of the Wushan soil.


<sup>∗</sup>Correlation is significant at the 0.05 level (2-tailed). ∗∗Correlation is significant at the 0.01 level (2-tailed).

researchers (Li et al., 2013c; Guo et al., 2015). Our results showed that whether the paddy soil was Si-deficient or Si-enriched, the utilization of Si fertilizer could significantly (p < 0.05) improve the phytolith content of the rice organs (**Tables 2**, **3**) such as the stem, sheath, leaf, grain and root. According to the formation mechanism of phytoliths, the available Si in the soil is taken up by rice plants at the roots, usually taking the shape of the plant cell or cell spatium where Si is deposited (Piperno, 1988; Ma, 2003; Neumann, 2003; Song et al., 2016). Thus, the use of Si fertilizer increased the content of effective Si in the soil (Ma et al., 2004; Liu et al., 2006; Cai, 2015) and increased the absorption capacity of Si in the rice (Li et al., 2013c; Seyfferth et al., 2013; Guo et al., 2015; Zuo et al., 2016; Huan et al., 2018), thereby increasing the phytolith content of the rice plant (**Table 4**).

A substantial amount of research reported that the factors of the PhytOC content were as follows: different varieties (Parr et al., 2009, 2010; Parr and Sullivan, 2011; Li et al., 2013b; Song et al., 2017; Sun et al., 2017), pest and disease resistances (Ma et al., 2002), nitrogen utilization (Zhao et al., 2016), basalt powder (Guo et al., 2015), soil-effective Si content (Song et al., 2014b; Klotzbücher et al., 2018), and net production on the ground (Blecker et al., 2006). It has been shown that Si is an important element for rice growth and the deficiency of plantavailable Si may exert an adverse effect on the rice yield through biotic stresses, disease and pests, etc. (Ma, 2004; Ma et al., 2004). Our results also showed that the contents of phytolith in rice plants were different in Si-deficient and Si-enriched paddy soil. The content of Phytolith in rice plants with Si-enriched paddy soils was higher than that in rice plants with Si-deficient paddy soil (**Tables 2**, **4**). Moreover, whether in Si-deficient or in Sienriched paddy soils, there was a positive correlation (p < 0.05) between the phytolith content of rice plants and the Si fertilizer dosages (**Tables 5**, **6**). Previous studies have demonstrated that the content of the Si (phytoliths) in crops may be promoted through Si fertilizer application (Alvarez and Datnoff, 2001; Liang et al., 2010; Mecfel et al., 2010). Further, in the Si-deficient paddy soil, the estimated PhytOC fluxes were significantly related to the Si fertilizers (R = 0.973 and p < 0.05), the phytolith content (R = 0.986 and p < 0.05) and the biomass of the rice (R = 0.994 and p < 0.01) (**Table 5**). However, in the Si-enriched paddy soil, the estimated PhytOC fluxes were not correlated (P > 0.05) with these factors. Zhang et al. (2014) showed that the yield of rice was increased 14.5% by the use of 225 kg ha−<sup>1</sup> Si fertilizer; when the application of Si fertilizer was increased to 375 kg ha−<sup>1</sup> , the yield of the rice increased only by 10.1%. Similarly, Wu et al. also recommended the use of 225 kg ha−<sup>1</sup> Si fertilizer as the most economical measure (Wu et al., 2014). We also obtained the same results. The application of Si fertilizer to the Si-enriched paddy soil did not increase the biomass of the rice but reduced it. In particular, when the amount of the Si fertilizer reached 600 kg ha−<sup>1</sup> , the rice biomass decreased significantly by 29.10% compared with the control treatment (**Table 4**). Therefore, excessive Si fertilizer not only has no benefit to the accumulation of estimated PhytOC fluxes in rice plant, but also reduces the yield of rice. However, for Sideficient soils, the application of Si fertilizer can not only increase rice yield, but also increase the phytolith content of rice plants and the estimated PhytOC fluxes (**Table 4**). Thus, different Si fertilizer doses were one of the measures to improve the phytolith

content and the biomass of the rice plant. Thus, how to promote the phytolith content and C content of phytoliths will require further in-depth study.

The global rice cultivation area was approximately 1.64 × 10<sup>8</sup> ha in 2014 (Prajapati et al., 2016); when rice is harvested, the rice straw and husks are removed from the paddy field and used for other purposes, including animal feeding and firewood, or simply incinerated (Savant et al., 1996). Thus, most of the Si taken up by rice is removed from a field when the rice straw is removed, and the loss of SiO<sup>2</sup> is from 75 to 130 kg ha−<sup>1</sup> every production season (Zhang et al., 2014). Such large losses of Si make it difficult to maintain the balance of Si in soils from natural weathering alone. Appropriate dosages of Si fertilizer could solve the problem of Si deficiency in soil, and increase the biomass of rice and the content of phytolith in rice plants, and indeed result in the occlusion of increased CO<sup>2</sup> in the rice plants (Liang et al., 2010; Mecfel et al., 2010). The estimated PhytOC fluxes increased from 0.49 to 4.52 Kg-e-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> (**Table 4**). More than 8.04 × 10<sup>4</sup> to 7.41 × 10<sup>5</sup> Mg-e-CO<sup>2</sup> would have been occluded within the phytolith of the rice plants per year globally. Taking the largest estimated PhytOC flux (12.93 Kg-e-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> ) of the rice plants, 2.12 × 10<sup>6</sup> Mg-e-CO2, would have been occluded within the phytolith of rice plants every year. However, the annual CO<sup>2</sup> bio-sequestration within the rice phytoliths of the unit area is likely to be lower than that of other plants, such as bamboo leaf litter (1.56 × 10<sup>7</sup> Mg-e-CO<sup>2</sup> year−<sup>1</sup> ) (Parr, 2006), wetland plants (4.39 × 10<sup>7</sup> Mg-e-CO<sup>2</sup> year−<sup>1</sup> ) (Guo et al., 2015), grasslands (4.14 × 10<sup>7</sup> Mg-e-CO<sup>2</sup> year−<sup>1</sup> ) (Song et al., 2012), millet (2.37 × 10<sup>6</sup> Mg-e-CO<sup>2</sup> year−<sup>1</sup> ) (Pan et al., 2017) and sugarcane leaf (0.72 × 10<sup>7</sup> Mg-e-CO<sup>2</sup> year−<sup>1</sup> ) (Parr et al., 2009). In this study, we showed that Si fertilizer application could promote the phytolith content and biomass of rice plants and further improve the estimated PhytOC flux of rice plants. Thus, the measure provided a theoretical basis for the bio-carbon sequestration of the rice plant and laid a foundation for PhytOC fixation in paddy soil by the return of straw.

### REFERENCES


## CONCLUSION

The use of Si fertilizer could significantly increase the phytolith content of rice plants in Si-deficient red paddy soil or Si-enriched Wushan soil. The phytolith content of rice plants was positive correlation with the Si fertilizer dose in two types paddy soil. The estimated PhytOC fluxes in Si-deficient red paddy soil had a positive correlation with the phytolith content, the biomass of the rice and the Si fertilizer dose. In this study, we estimated that the PhytOC fluxes increased from 0.49 to 4.52 Kg-e-CO<sup>2</sup> ha−<sup>1</sup> year−<sup>1</sup> . More than 8.04 × 10<sup>4</sup> to 7.41 × 10<sup>5</sup> Mg-e-CO<sup>2</sup> would have been occluded within the phytoliths of the rice plants per year globally. Therefore, Si fertilizer application might provide a new approach to increase the atmospheric CO<sup>2</sup> occluded within the phytoliths, offering a potential method.

## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct, and intellectual contribution to the work. XS completed the experiments independently, carried out the data analysis, and finished the final writing of the article. QL made great contributions to guide the process of experiments. TT, XC, and XL helped in sampling, experimentation, and essay writing.

## FUNDING

This work was partially supported by the National Natural Science Foundation of China (Nos. 41271208 and 31400464), the Anhui Province University Natural Science Research Foundation (Nos. KJ2017A423 and KJ2018A0430), the Excellent Researcher Program of the Education Department of Anhui Province (No. gxyq2018097) and the Laboratory Opening Subject of the School of Biology and Food Engineering of Chuzhou University (No. SWSP201816KF).



Neumann, D. (2003). Silicon in Plants. Berlin: Springer.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Sun, Liu, Tang, Chen and Luo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Combined Silicon-Phosphorus Fertilization Affects the Biomass and Phytolith Stock of Rice Plants

Zimin Li <sup>1</sup>† , Fengshan Guo2† , Jean-Thomas Cornelis <sup>3</sup> , Zhaoliang Song4\*, Xudong Wang2 and Bruno Delvaux <sup>1</sup>

<sup>1</sup> Soil Science, Earth and Life Institute, Université catholique de Louvain (UCLouvain), Louvain-la-Neuve, Belgium, <sup>2</sup> School of Environment and Resources, Zhejiang Agricultural and Forestry University, Lin'an, China, <sup>3</sup> BIOSE Department, Gembloux Agro-Bio Tech, University of Liege, Gembloux, Belgium, <sup>4</sup> Institute of the Surface-Earth System Science, Tianjin University, Tianjin, China

#### Edited by:

Julia Cooke, The Open University (United Kingdom), United Kingdom

#### Reviewed by:

Jitender Giri, National Institute of Plant Genome Research (NIPGR), India Timothy J. Gallaher, Bernice P. Bishop Museum, United States

\*Correspondence:

Zhaoliang Song zhaoliang.song@tju.edu.cn † These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science

Received: 20 December 2018 Accepted: 17 January 2020 Published: 18 February 2020

#### Citation:

Li Z, Guo F, Cornelis J-T, Song Z, Wang X and Delvaux B (2020) Combined Silicon-Phosphorus Fertilization Affects the Biomass and Phytolith Stock of Rice Plants. Front. Plant Sci. 11:67. doi: 10.3389/fpls.2020.00067 Phytoliths are silica bodies formed in living plant tissues. Once deposited in soils through plant debris, they can readily dissolve and then increase the fluxes of silicon (Si) toward plants and/or watersheds. These fluxes enhance Si ecological services in agricultural and marine ecosystems through their impact on plant health and carbon fixation by diatoms, respectively. Fertilization increases crop biomass through the supply of plant nutrients, and thus may enhance Si accumulation in plant biomass. Si and phosphorus (P) fertilization enhance rice crop biomass, but their combined impact on Si accumulation in plants is poorly known. Here, we study the impact of combined Si-P fertilization on the production of phytoliths in rice plants. The combination of the respective supplies of 0.52 g Si kg–<sup>1</sup> and 0.20 g P kg−<sup>1</sup> generated the largest increase in plant shoot biomass (leaf, flag leaf, stem, and sheath), resulting in a 1.3-fold increase compared the control group. Applying combined Si-P fertilizer did not affect the content of organic carbon (OC) in phytoliths. However, it increased plant available Si in soil, plant phytolith content and its total stock (mg phytolith pot−<sup>1</sup> ) in dry plant matter, leading to the increase of the total amount of OC within plants. In addition, P supply increased rice biomass and grain yield. Through these positive effects, combined Si-P fertilization may thus address agronomic (e.g., sustainable ecosystem development) and environmental (e.g., climate change) issues through the increase in crop yield and phytolith production as well as the promotion of Si ecological services and OC accumulation within phytoliths.

Keywords: phytolith, crop yield, silicon-phosphorus fertilization, rice, silicon cycle

#### INTRODUCTION

Amorphous biogenic silica (SiO2·nH2O) can accumulate in living plant tissues during their growth and development (Conley, 2002; Piperno, 2006). These silica bodies, named phytoliths, are released into the soil after the decomposition of litter and plant residues (Smithson, 1956; Alexandre et al., 1997; Fraysse et al., 2006). Depending on their chemical composition and structure, phytoliths can accumulate in soils and sediments over centuries or millennia, or dissolve and then contribute to the

**32**

pool of aqueous monosilicic acid (dissolved silicon: DSi), which is available for plant root uptake (Bartoli, 1985; Meunier et al., 1999; Fraysse et al., 2009; Struyf et al., 2010; Cornelis and Delvaux, 2016). The elemental composition of phytoliths is influenced by plant species and phytolith morphology (Bartoli and Wilding, 1980; Li et al., 2014; Nguyen et al., 2014). Organic carbon (OC) (0.2–6%) can be associated with phytoliths (Bartoli, 1985; Parr and Sullivan, 2005; Parr et al., 2010; Zuo and Lü, 2011; Li et al., 2013c; Alexandre et al., 2015). In particular, the occlusion of organic carbon (OC) within phytolith (PhytOC), which is formed in plant tissues, has been proposed as a mechanism which traps the photosynthesized molecules within silica bodies (Parr and Sullivan, 2005; Santos et al., 2012; Alexandre et al., 2015; Reyerson et al., 2016). The occurrence of PhytOC has been reported in various studies (Parr and Sullivan, 2005; Parr et al., 2010; Parr and Sullivan, 2011; Song et al., 2012; Song et al., 2013; Li et al., 2013a; Li et al., 2013b; Huang et al., 2014; Song et al., 2015; Guo et al., 2015; Sun et al., 2016; Pan et al., 2017; Qi et al., 2017; Li et al., 2018a). However, the biological processes leading to the occurrence of PhytOC has not been demonstrated. Therefore, OC content in phytoliths varies depending on the extraction procedure (Parr and Sullivan, 2014; Santos and Alexandre, 2017; Song et al., 2016). These variations led to a debate on the scale of OC occlusion within phytoliths, and on the significance of the PhytOC sink for the global C cycle and climate change mitigation (Parr and Sullivan, 2005; Song et al., 2012; Hodson, 2016; Reyerson et al., 2016; Lorenz and Lal, 2018; Crifò and Strömberg, 2019; Ramesh et al., 2019; Song et al., 2016; Santos and Alexandre, 2017). In addition, OC associated with phytoliths might have a non-photosynthetic origin attributed to the uptake of organic molecules from soil (Santos et al., 2012; Alexandre et al., 2015; Reyerson et al., 2016), which could lead to erroneous C dating using phytoliths (Hodson, 2016). Therefore, the accurate determination of the phytolith OC content must not only completely eliminate external OC, but also keep the phytolith structure intact and the oxidation of C in the phytolith to a minimum (Parr and Sullivan, 2014). Overoxidation may significantly underestimate phytolith C sequestration and should be avoided (Parr and Sullivan, 2014).

In any case, whether phytoliths sequester OC or not, the increase in silicon (Si) uptake undoubtedly enhances plant biomass, Si and phytolith content in plants [a.o. (Li et al., 2018b; Li et al., 2019)]. The amount of OC that could be associated with phytoliths would depend on plant Si accumulation and thus phytolith content (Li et al., 2013c); therefore, suggesting that regulating Si supply might increase phytolith-associated OC in croplands. In this respect, the combination of Si and phosphorus (P) fertilization may enhance the contents of plant phytolith and OC associated within phytoliths.

This study is how co-fertilization combining Si and phosphorus (P) can affect Si availability and plant uptake, as well as phytolith formation in rice. Si uptake improves the growth of Si-accumulator cereals such as rice (Savant et al., 1997; Ma et al., 2001; Ma et al., 2006; Liang et al., 2015). Si fertilization can enhance rice resistance to biotic and abiotic stresses (e.g., pests, water and heat stress, disease, etc.) (Liang et al., 2007; Cooke et al., 2016; Cooke and Leishman, 2016; Coskun et al., 2019), and thus promote rice crop yields and Si accumulation (Savant et al., 1997; Ma et al., 2001; Keller et al., 2012). However, P fertilization also plays an important role in improving yields and promoting plant precocity (George et al., 2001; Lambers et al., 2006; Hammond and White, 2008). In paddy soils, Si and P fertilization could alleviate P deficiency, increase P uptake by plants (Ma and Takahashi, 1990; Liang et al., 2007; Hu et al., 2018), and enhance plant available Si in soil, hence improving crop yields (Song et al., 2014; Klotzbücher et al., 2015; Carey and Fulweiler, 2016; Li et al., 2019). Furthermore, plant available Si content in soil may increase after P supply. Besides, Si supply can increase P bioavailability in soil through the competition between silicate and phosphate for sorption on Al and Fe oxide surfaces that bear positive charges (Parfitt, 1989; Su and Puls, 2003). Combined Si-P fertilization may thus substantially influence Si and P biocycling in the soil-plant system, as well as plant phytolith and chemical composition.

Through a pot experiment in controlled conditions, we aim to address three interconnected questions: 1) does Si-P fertilization increase rice biomass? 2) does increased biomass promote plant phytolith formation? and 3) does combined Si-P supply impact the amount of OC associated within phytoliths?

## MATERIALS AND METHODS

The pot experiment was carried out at Zhejiang Agricultural and Forestry University, Lin'an, Zhejiang Province, Eastern China (29°56'–30°27'N, 118°51'–119°52'E). This region is characterized by a mid-subtropical monsoon climate with a mean annual precipitation of 1,500 mm, a mean annual temperature of 15.8° C, 237 frost-free days, and an annual 1,939 h of sunshine.

#### Pot Experiment Design and Management

The soil used was a Cambisol, according to the World Reference Base (WRB) key (IUSS, 2014), sampled from the agricultural station at Zhejiang Agricultural and Forestry University. The soil was air-dried, sieved at 2 mm, and mixed with Si-P fertilizers. The soil physico-chemical properties were as follows (Lu, 2000): pHwater = 5.34 ± 0.02, soil organic matter = 30.26 ± 4.28 g kg−<sup>1</sup> , available Si = 155.59 ± 22.73 mg kg−<sup>1</sup> , available P = 113.87 ± 1.35 mg kg−<sup>1</sup> , available K = 10.33 ± 1.11 mg kg−<sup>1</sup> and available N = 87.15 ± 2.47 mg kg−<sup>1</sup> (Guo et al., 2015). The analytical methods were described by Lu (2000). Here, plant available Si was assessed using extracts of NaOAc and acetic acid. Jiayu 253 was selected as the experimental rice (Oryza sativa) cultivar because of its high yield and wide distribution in Zhejiang province.

The experiment was carried out using three fertilization levels, zero (0), medium (m), and high (h), for Si (Si0: 0, Sim: 0.26, Sih: 0.52 g SiO2 kg–<sup>1</sup> ) using Na2SiO3, and P (P0: 0, Pm: 0.2, Ph: 0.4 g kg–<sup>1</sup> ) using P2O5. Nine treatments (Si0P0, Si0Pm, Si0Ph, SimP0, SimPm, SimPh, SihP0, SihPm, and SihPh) and five replicates per treatment were set up (Table 1). N and K fertilizers were



applied in all treatments as, respectively, urea ammonium nitrate (0.20 g N kg−<sup>1</sup> ), and KCl (0.25 g K kg−<sup>1</sup> ). All fertilizers were added to soil before planting rice. Soil pH value and available Si and P contents under different levels of Si and P supply were

TABLE 2 | Soil pH value and available silicon (Si) and phosphorus (P) contents under different levels of Si and P supply\*.


\*The data were collected from Sun et al., 2015.

determined by Sun et al. (2015), as presented in Table 2. Each pot (0.24 m diameter, 0.28 m height) contained 8.5 kg air-dried soil and was regularly irrigated using tap water (Si: 0.36 mg L–<sup>1</sup> ) at the same level until rice grain harvesting. After a first irrigation of 500 ml, 1,000 ml of water were supplied per pot during the whole growing period, once every 2 days. Crop harvesting was done 4 months after planting. The rice plant parts were sampled separately: sheath, leaf, flag leaf, and stem. Plant samples were thoroughly washed with deionized water, and then oven dried at 75°C until a constant weight was attained, as equal to dry shoot biomass. Rice grains, including rice husk, were also dried at 75°C and weighed.

#### Plant Chemical Analysis

Dried plant samples were cut into small pieces by stainless steel scissors for the analysis of Si and phytolith content. Plant samples were fused with Li-metaborate at 950°C and dissolved in nitric acid (HNO3 4%), prior to molybdenum blues colorimetry to determine Si content (Lu, 2000).

Microwave digestion in combination with Walkley–Black digestion was used to isolate the phytoliths from plant material (Walkley and Black, 1934; Parr et al., 2001), in order to remove extraneous organic materials thoroughly (Li et al., 2013c). We first checked the presence of phytoliths by optical microscopy to ensure that all extraneous organic materials had been removed (Li et al., 2013c). Then, we further assess the purity of phytolith extract using the scanning electron microscope (SEM) images and energy-dispersive spectroscopy (EDS) (Figure 1). The phytoliths were then oven dried at 75°C for 24 h, cooled and weighed. Phytolith particles were dissolved in HF 1 M at 45°C during 100 min, so that associated OC could be released in the acidic solution (Li et al., 2013c). Associated OC content was determined using the potassium dichromate procedure and the soil standard reference GBW07405, ensuring a relative precision below 5% (Li et al., 2013c). Using plant dry matter, OC and phytolith contents, we computed OCphyt and OCpdm, as the OC

FIGURE 1 | (A) Scanning electron microscope (SEM) image of rice leaf phytolith. (B) Semi-quantitative element concentration (wt. %, n = 5) measured by SEMenergy-dispersive spectroscopy (EDS) of the selected area.

contents per mass unit of, respectively, phytolith and plant dry matter.

### Data Treatment

Phytolith stock (mg pot−<sup>1</sup> ) = phytolith content (mg g−<sup>1</sup> ) × biomass of dry plant tissue (g pot−<sup>1</sup> ) where phytolith stock is used to refer to the mass of phytoliths per pot (mg pot−<sup>1</sup> ); phytolith content is used to refer to the mass of phytoliths per gram of dry plant tissue (mg g−<sup>1</sup> ); biomass of dry plant tissue is used to refer to the mass of dry plant tissue per pot (g pot−<sup>1</sup> ).

A two-way analysis of variance of was performed to assess the effects of combined Si-P fertilization levels using SPSS (24.0). Fisher's least significant difference (LSD) test was used to compare the average values of the contents of SiO2, phytolith, OCphyt, OCpdm in the different plant parts (leaf, flag leaf, sheath, and stem) (at P < 0.05 level, n = 5). Origin 8.0 software was used to plot the figures.

## RESULTS

#### Rice Shoot Biomass and Grain Yield

The rice shoot biomass (g pot−<sup>1</sup> ) significantly varied from 168 in Si0P0 to 213 in SimPm or SimPh (Table 3). Among the Si0 treatments, there was a significant increase in shoot biomass between Si0P0 and Si0Ph whereas Si0Pm was intermediate between and not significantly different from the other two treatment levels (Table 3). At the given level Pm = 0.2 <sup>g</sup> kg−<sup>1</sup> , increasing Si application rate from Si0 to Sim increased the leaf and shoot biomass (Table 3). At the same Pm level, rice grain yield increased from Si0 to Sim and from Si0 to Sih (Table 3).

#### Content and Stock of Phytoliths Formed in Rice Plants

Considering all plant parts, phytolith content significantly varied (<sup>p</sup> < 0.05) from 4.73 to 59.12 mg g–<sup>1</sup> (Tables 4–6). At all given levels of Si0, Sim, and Sih, the increase in P application rate did

TABLE 3 | Effect of silicon-phosphorus (Si-P) levels on biomass in different plant parts and rice dry shoot.


The data of rice organ (leaf, flag leaf, stem and sheath) collected from Sun et al., 2015; Different lowercase letters represent significant differences of rice shoot biomass (Duncan's multiple range test; at p < 0.05 level, n = 5). \*Grains including rice husk.

TABLE 4 | Contents of SiO2, phytolith, organic carbon (OC) associated with phytolith as expressed per unit mass of phytolith (OCphyt) and of plant dry matter (OCpdm) in different plant parts (leaf, flag leaf, sheath, and stem).


#### TABLE 4 | Continued


Different lowercase letters indicate significant differences among the treatments in different P treatments and rice plant parts at a given Si level, respectively [least significant difference (LSD) test; p < 0.05, n = 5]. Different uppercase letters indicate significant differences among the treatments in different Si treatments and rice plant parts at a given P level, respectively (LSD test; p < 0.05, n = 5). Uppercase letters of bolded texts indicate significant differences among different plant parts (leaf, flag leaf, sheath, and stem).

TABLE 6 | Two-way analysis of variance (ANOVA) of silicon-phosphorus (Si-P) levels on the rice shoot biomass, stock of phytolith, and OCpdm in rice shoot.


not significantly increase phytolith content regardless of plant part, while this effect was not true for sheath with a significant increase from Si0P0 to Si0Pm (Table 5). Yet at given levels P0, Pm, and Ph, the increase in Si application rate significantly increased phytolith content in all plant parts (Table 5). Phytolith content in leaves was the highest, and varied from 28.36 to 59.12 mg g–<sup>1</sup> , with an average of 39.82 mg g–<sup>1</sup> (Table 4). As compared to the other plant parts, stem phytolith content was the lowest, with an average value of 7.11 mg g–<sup>1</sup> . Considering all plant parts, the stock of phytolith formed during the experimental period varied significantly from 152.6 to 1,002.7 mg pot−<sup>1</sup> (Figure 2). Si-P fertilization increased the stock of phytoliths formed in all plant parts, including plant shoot biomass (Figure 2). At all given levels of Si0, Sim, and Sih, the increase in P application rate did not significantly increase phytolith stock regardless of plant part, including in plant shoot biomass (Figure 2; Table 5). Yet at given levels P0, Pm, and Ph, the increase in Si application rate significantly increased phytolith stock in all plant parts, including plant shoot biomass. The mean phytolith stock was the highest in sheath (758.3 mg pot−<sup>1</sup> ), followed by leaf (621.0 mg

TABLE 5 | Two-way analysis of variance (ANOVA) of silicon-phosphorus (Si-P) levels on the contents of SiO2, phytolith, organic carbon (OC) associated with phytolith as expressed per unit mass of phytolith (OCphyt) and of plant dry matter (OCpdm), as well as the stock of phytolith and OCpdm in different plant parts (leaf, flag leaf, sheath, and stem).


pot–<sup>1</sup> ), flag leaf (374.3 mg pot–<sup>1</sup> ), and stem (average 289.1 mg pot–<sup>1</sup> ). Considering shoot biomass and including rice grains, the stock of phytolith significantly varied from 1,296.6 to 2,778.6 mg pot−<sup>1</sup> , the latter and maximal value being measured at SihPm level (Figure 2).

## Organic Carbon Content Associated With Phytoliths Formed in Rice Plants

Considering all plant parts, OCphyt ranged from 11.16 to 18.17 mg g–<sup>1</sup> , but did not differ between Si-P treatments and plant parts (Tables 4–6). OCphyt content did not vary following P application irrespective of the Si supply (Si0, Sim, and Sih), while this effect was not true for stem and flag leaves with a significant increase from SimP0 to SimPm and SimP0 to SimPh, respectively (Table 5). At a given level Sih in leaf, and a given level Sim in stem, the increase in P application rate significantly decreased their OCpdm content (Table 5). At a given level Si0, the increase in P application rate significantly increased the OCpdm stock in all plant parts as well as plant shoot biomass except leaves, while at a given level Sih level, the increase in P application rate significantly decreased the OCpdm stock in all plant parts as well as plant shoot biomass except sheath (Table 6 and Figure 3). However, OCpdm content and its stock significantly increased with increasing Si application rate due to the increased phytolith content and phytolith stock in all plant parts, respectively (Table 4 and Figure 3).

## DISCUSSION

#### Effects of Silicon-Phosphorus Supply on Rice Shoot Biomass and Yield

Our experimental data show that the addition of P alone increased biomass and grain yield (a significant increase from Si0P0 Si0Pm Si0Ph); but when a combined Si-P fertilization were applied there was no significant increase in biomass and yield except that at SimPm and SimPh (Table 3). This supports the results of previous experiments carried out either in the field (Liu et al., 2014; Liang et al., 2015; Song et al., 2015) or in pots (Agostinho et al., 2017; Liang et al., 1994; Ma and Takahashi, 1990). Si fertilizer supply increased the stock of bioavailable Si that is crucial for sustainable paddy rice yield production (Klotzbücher et al., 2015). Furthermore, once available Si is taken up by plant roots, the accumulation of phytoliths in plant tissues can enhance the efficiency of plant photosynthesis and water use (Meunier et al., 2017), as well as their tolerance to biotic stresses (Epstein, 1994; Cooke and Leishman, 2016; Coskun et al., 2019). On the other hand, P supply likely

increased plant growth and fecundity as well as root growth (Lambers et al., 2006; Brown et al., 2012). Indeed, low P levels (i.e., SimP0 or SihP0; Table 3) did not significantly increase rice biomass regardless of plant part (Tables 2 and 3), confirming that rice growth was clearly limited at low P supply (Ma and Takahashi, 2002; Ma, 2004; Cooke and Leishman, 2016; Agostinho et al., 2017; Hu et al., 2018) even with increasing the addition of Si fertilizer. Excessive inorganic P within rice plant inhibits enzyme reactions, induces abnormal osmotic pressure in plant cell, which further decreases rice growth (Ma and Takahashi, 1990). As reported by Ma and Takahashi (1990), the levels of bioavailable P and Si in soil influence plant P content. At Si0 level, the increase in P supply did not result in a change of stem, sheath and flag leaf biomass (Table 3) likely because the positive side-effects of P nutrition were limited at a high P supply, as mentioned here above. However, these sideeffects may have been enhanced by low Si level. Yet once available P content increases up to 17.8–20.3 mg kg−<sup>1</sup> at Ph level (Table 2), the increase in bioavailable Si is beneficial to rice plants by decreasing P uptake (data not shown; Ma and Takahashi, 1989; Owino-Gerroh and Gascho, 2005; Greger et al., 2018), which, in turn, decreases plant P content (Ma and Takahashi, 1990). This Si-induced decrease in plant P uptake can also result from the molecular mechanism of down-regulating the expression of P transporter gene, OsPT6 in rice (Hu et al., 2018). The Si-P interaction thus contributes to increase rice

biomass at SimPm, SimPh, and SihPh levels (Table 5), suggesting Si supply may alleviate excessive P application.

#### Effects of Silicon-Phosphorus Supply on the Production of Phytoliths

At a given P level, Si2O content significantly increased with increasing Si application rate compared to control (Si0), regardless of plant part. Thus, the addition of Si fertilizer as monosilicic acid (H4SiO4) taken up by roots resulted in silica accumulation in plant tissues through the formation of phytoliths (Figure 4A). This significant increase was due to the addition of Si fertilizer that can improve the well-observed increase in plant available Si in soils (Table 2). The DSi release from highly soluble Na2SiO3, wollastonite and other Si fertilizers (Haynes et al., 2013; Haynes, 2014; Keeping, 2017; Li et al., 2018b; Li et al., 2019) largely contributed to the pool of bioavailable Si, from which it was taken up by plant roots to accumulate around plant transpiration termini. As expected, P fertilizer supply did not change the concentration of available Si in Si0 level (Table 2), and thus of phytolith content, regardless of plant part (Table 4). Interestingly, our data further show that, at given levels Sim and Sih, the increase in P application rate decreased the formation of phytoliths, but not always significantly, and regardless of plant part, except in flag leaf at Sih treatment (Table 4). This trend is in accordance with Ma and Takahashi (1990) who reported that Si content of rice shoots

FIGURE 4 | Plot of: (A) phytolith content of plant parts against SiO2 content considering all silicon-phosphorus (Si-P) treatments (leaf: y = 0.9151x−1.2668, R<sup>2</sup> = 0.9254 P < 0.01; flag leaf: y = 0.8248x + 1.3865, R<sup>2</sup> = 0.8035 P < 0.01; Sheath: y = 0.5457x + 10.337, R<sup>2</sup> = 0.6938 P < 0.01; Stem: y = 1.0171x−1.6823, R<sup>2</sup> = 0.8929 P < 0.01). (B) OCpdm content of plant parts against phytolith content considering all Si-P treatments (leaf; y = 0.0134x + 0.0233, R<sup>2</sup> = 0.8557 P < 0.01; flag leaf; y = 0.011x + 0.1038, R<sup>2</sup> = 0.8097 P < 0.01; sheath; y = 0.008x + 0.1541; R<sup>2</sup> = 0.6845 P < 0.01; stem; y = 0.0121x + 0.0166; R<sup>2</sup> = 0.7924 P < 0.01). (C) OCpdm content of plant parts against C content of phytoliths (OCphyt) considering all Si-P treatments (leaf; y = 0.019x + 0.29, R<sup>2</sup> = 0.0273 P > 0.05; flag leaf; y = −0.0079x + 0.5291; R<sup>2</sup> = 0.0329 P > 0.05; sheath; y = −0.003x + 0.4191; stem; y = 0.0027x + 0.0629; R<sup>2</sup> = 0.0491 P > 0.05).

decreased with increasing P availability in soil (Tables 2 and 3). As here discussed above, this trend of decreasing Si deposition in plant tissues resulted from dilution caused by increased plant growth following P application and the molecular mechanism of down-regulating the expression of P transporter gene, OsPT6 in rice (Hu et al., 2018). Since shoot biomass significantly increased following P addition, our data thus suggest that combined Si-P fertilization contributes to increased Si bioavailability in soil, Si root uptake, phytolith formation, and rice plant biomass, which, in turn, increases the stock of phytolith production in plants, while this effect is limited at the high P levels.

#### Effects of Silicon-Phosphorus Fertilization on Carbon Associated With Rice Phytoliths

Considering all plant parts (Figure 4), our data suggest that OCpdm may be controlled by phytolith accumulation in plant tissues (Figures 4A, B), during which the incorporation of OC seems to be constant (Figure 4C) and therefore does not influence the OC content of phytoliths, OCphyt, in line with previous hypotheses (Li et al, 2013c). Evidently, the increase in phytolith stock increases the stock of OCpdm, i.e., the quantity of OC associated with phytolith in living plant tissues.

Si-P fertilization does not affect OCphyt content, regardless of plant part and biomass whereas it affects OCpdm (Table 4). SEMenergy dispersive X-ray spectroscopy (EDX) image (Figure 1) illustrates that OC can be associated with the extracted phytoliths. However, the associated OC levels, irrespective of its source, do not change with the fertilizer treatments. SEM-EDX is semi quantitative, and thus, we used this technique not to quantify but to check the OC content as determined chemically. Therefore, we may not conclude about the possible entrapment of OC during polymerization of biogenic amorphous silica as previously proposed (Hodson et al., 1985; Parr and Sullivan, 2005; Zuo and Lü, 2011; Parr and Sullivan, 2014; Alexandre et al., 2015; Alexandre et al., 2016; Reyerson et al., 2016; Hodson, 2016; Song et al., 2016). Similarly, the hypothetical ability of plant phytoliths to occlude OC does not vary depending on the application rate (this study) and type of Si supply: basalt powder (Guo et al., 2015) or slag-based silicate (Song et al., 2015). According to Zhao et al. (2016), increased N supply in degraded grasslands decreased the phytolith content in grass shoots, while significantly increased OC content of their phytoliths. These authors hypothesized that the increase in OCphyt was probably caused by improved cell growth, partly enlarged cell volume and decrease in the specific surface area of phytoliths. Similarly, Gallagher et al. (2015) reported, that growing conditions impact the OC content of phytoliths in Sorghum bicolor irrespective of the type and rate of application of inorganic fertilizers. These growth conditions, referring to different nutritive regimes of N, P, K, and microelements, affected the plant transpiration stream, and thus Si accumulation (Gallagher et al., 2015), which in turn, affect the OC content of phytolith (Blackman, 1969; Hodson et al., 1985). In addition to the growth conditions, the nature of plant part or organ might influence the phytolithic OC content through its impact on phytolith morphology and specific surface area (Li et al., 2013c and Li et al., 2014; Table 4).

Although Si-P fertilization did not increase OCphyt, the application of Si and P fertilizer can substantially improve the OCpdm content in rice plant through increasing phytolith accumulation (Figures 2 and 3; p < 0.001). Our data further show that the content of phytolithic OC in rice plants mainly depends on Si supply. Indeed, phytolith accumulation in rice plant tissues significantly increased with increasing supply of Si fertilizer. Thus, regulating Si supply promoted the OC content associated within phytolith by increasing phytolith accumulation in plant notably through the increase in biomass production. Consequently, increasing crop productivity could play a crucial role in increasing the stock of phytolithic OC, while the processes explaining OC associated within phytoliths are still debated. Here the largest rice biomass was obtained at SihPm level (Si = 0.52 g kg–<sup>1</sup> ; P = 0.2 g kg–<sup>1</sup> ), regardless of plant part (Table 3). The level SihPm largely contributed to double the stock of phytolithic OC (mg pot−<sup>1</sup> ) from 18.9 at Si0P0 to 36.8 at SihPm (Figure 3E). Another lesson is that P should not be neglected if rice productivity is to be improved as discussed above. Thus, regulating Si-nutrient supply combined with optimal P supply is promising to enhance both phytolith formation and associated organic carbon in Siaccumulating plants, as well as crop productivity.

## CONCLUSION

Our experimental results show that i) phytolith concentration increases with increasing Si fertilization, ii) phytolithic OC concentration does not depend on Si or P fertilization, iii) as the biomass increases with Si fertilization, the stocks of phytolith and phytolithic OC increase, iv) P fertilization has no clear impact either on phytolith or phytolithic OC concentration, but increases plant biomass and grain yield. Despite the occurrence of OC associated within phytoliths, we cannot be sure of OC occlusion within phytoliths. We conclude that the combined Si-P fertilization increases the phytolith stock by increasing the biomass and phytolith content of rice plants. Through these positive effects, combined Si-P fertilization may thus address agronomic (e.g., sustainable ecosystem development) and environmental (e.g., climate change) issues through the increase in crop yield and phytolith production as well as the promotion of Si ecological services and OC accumulation within phytoliths.

## AUTHOR CONTRIBUTIONS

We thank Mrs. Linan Liu and Mr. Xiaomin Yang for laboratory assistance (Tianjin University). ZL and FG carried out the experiment, analyzed all data and prepared the draft. XW and ZS guided the experiment and revised manuscript. J-TC and BD reworked and revised the manuscript. All authors played a significant role in the development of the study and in writing of the manuscript. The submitted version of the manuscript has been read and accepted by all co-authors.

## FUNDING

The work was supported by National Natural Science Foundation of China (41571130042, 41930862 and 41522207) and the State's Key Project of Research and Development Plan of China (2016YFA0601002, 2017YFC0212703). ZL is supported by ASP (aspirant)-FNRS of Belgium in 2015–2019 and was also supported by Fonds spécial de recherché of UCLouvain (UCLouvain-FSR) in 2014–2015. The authors declare no competing financial interests.

## REFERENCES


## ACKNOWLEDGMENTS

We thank Mrs. Linan Liu and Mr. Xiaomin Yang for laboratory assistance (Tianjin University).


Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Li, Guo, Cornelis, Song, Wang and Delvaux. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Distributions of Silica and Biopolymer Structural Components in the Spore Elater of Equisetum arvense, an Ancient Silicifying Plant

Victor V. Volkov, Graham J. Hickman, Anna Sola-Rabada and Carole C. Perry\*

Interdisciplinary Biomedical Research Centre, School of Science and Technology, Nottingham Trent University, Nottingham, United Kingdom

Equisetum species are primitive vascular plants that benefit from the biogenesis of silica bio-organic inclusions in their tissues and participate in the annual biosilica turnover in local eco-systems. As means of Equisetum reproduction and propagation, spores are expected to reflect the evolutionary adaptation of the plants to the climatic conditions at different times of the year. Combining methods of Raman and scanning electron microscopy and assisted with density functional theory, we conducted material spatialspectral correlations to characterize the distribution of biopolymers and silica based structural elements that contribute to the bio-mineral content of the elater. The elater tip has underlying skeletal-like structural elements where cellulose fibers provide strength and flexibility, both of which are necessary for locomotion. The surface of the elater tips is rich with less ordered pectin like polysaccharide and shows a ridged, folded character. At the surface we observe silica of amorphous, colloidal form in nearly spherical structures where the silica is only a few layers thick. We propose the observed expansion of elater tips upon germination and the form of silica including encapsulated biopolymers are designed for ready dispersion, release of the polysaccharide-arginine rich content and to facilitate silica uptake to the developing plant. This behavior would help to condition local soil chemistry to facilitate competitive rooting potential and stem propagation.

Keywords: Equisetum, spore, microscopy, Raman, silica, DFT

## INTRODUCTION

The occurrence of silica in algae (i.e., diatoms) (Brunner et al., 2009), simple animals (i.e., sponges) (Mann et al., 1983) and in plants (Sachs, 1862; Lewin and Reimann, 1969; Page, 1972; Hodson et al., 2005) are important examples of bio-mineralization in evolution. Silica may accumulate in pith (stem), cortex (stem or root), mesophyll (leaves) and vascular tissues of plants (Lewin and Reimann, 1969). Equisetum spp. (commonly known as horsetail) are ancient examples of living vascular plants (Page, 1972). It has been reported that Equisetum spp. take up mono-silicic acid from the soil to accumulate silica in their tissues (Timell, 1964; Grégoire et al., 2012). The genus benefits from a broad global distribution and yet can also be considered as an invasive and persistent weed.

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Christopher Exley, Keele University, United Kingdom Richard Belanger, Laval University, Canada

> \*Correspondence: Carole C. Perry carole.perry@ntu.ac.uk

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 13 December 2018 Accepted: 07 February 2019 Published: 05 March 2019

#### Citation:

Volkov VV, Hickman GJ, Sola-Rabada A and Perry CC (2019) Distributions of Silica and Biopolymer Structural Components in the Spore Elater of Equisetum arvense, an Ancient Silicifying Plant. Front. Plant Sci. 10:210. doi: 10.3389/fpls.2019.00210

**43**

Electron microscopy of silica depositions in epidermal cells of an Equisetum sp. (Kaufman et al., 1971) provided strong support for the earlier expectations (Lewin and Reimann, 1969) that silica may serve to (1) provide mechanical strength and rigidity of cellular wall and/or of tissue (Perry and Fraser, 1991; Hodson et al., 2005; Grégoire et al., 2012), (2) prevent excessive water loss through the epidermis (Gao et al., 2004), and (3) protect against pathogens and predators (Stahl, 1888; Guerriero et al., 2018; Coskun et al., 2019). Indeed, recently, correlation between the localization of silica and callose (a type of polysaccharide in plants) was reported which allowed the suggestion of a unique relationship between uptake of silicic acid and depositions of biogenic silica and callose, which were considered to provide resistance against fungal infection in horsetail (Guerriero et al., 2018) though alternative hypotheses as to how resistance is achieved have recently been published (Coskun et al., 2019).

Silica deposited in living organisms, often referred to as biogenic silica or biosilica, is generally accepted to be in amorphous forms (Perry et al., 1984; Brunner et al., 2009; Neethirajan et al., 2009). Silicification of Equisetum and its spores is a complex but potentially useful model system with insights (i.e., optical properties) that could aid commercial applications (Neethirajan et al., 2009). The formation of amorphous biosilica through the biosilicification process in Equisetum spp. is notable in its divergence from crystalline inorganic silica, with characteristics that have been described as a xerogel (Holzhüter et al., 2003). The chemical and morphological characteristics of the biosilica influence the biocompatibility of the material, being remarkably less harmful than crystalline polymorphs both natural and man-made, though silica toxicity is an active area of research (Fruijtier-Pölloth, 2012; Murugadoss et al., 2017). Silica produced by plants presents ordered hierarchical porous structures giving this material interesting properties with possible applications both in industry and in medicine (Davis, 2002; Sola-Rabada et al., 2018).

To understand the role of silica in biology and survival strategies of Equisetum, it is necessary to correlate both the distribution of biosilica inclusions and chemical properties of the inorganic component (at the junction with bio-tissue) with plant biology and biochemistry. In the early 70s (Kaufman et al., 1971), it was reported that in E. hyemale, silica is uniformly distributed over and within the outer epidermal cell walls, while in E. arvense, silica is concentrated in discrete structures (knobs and rosettes) projecting from the outer epidermal walls. Even though both rosette and knob structures were reported to contain inorganic silica, the character of silica distribution and the chemical nature of such inorganic inclusions in such structures, as well as in particulate structures on the outside of elaters, may be very different (Duckett, 1970; Kaufman et al., 1971; Perry and Fraser, 1991). In our previous study on the subject, we characterized the distribution of silica in stem nodes, internodes, basal branches, distal and leaves of E. arvense as fibrillary, globular and sheetlike silica ultra-structures (Perry and Fraser, 1991). According to the results, we anticipated the inorganic component would assert mechanical strength and rigidity and discussed the possible role of the cellulose micro-fibrillary network to template some of the considered silica depositions.

As a lower vascular plant species Equisetum reproduces with the aid of spores (Sadebeck, 1878; Newcombe, 1888; Beer, 1909; Erdtman, 1952), see **Figure 1**. Spores are notable for their complex morphology and motile nature. Beer (1909), Rudolf Beer reported that a "ripe spore contains a very considerable quantity of chlorophyll in its protoplast," and that when "spores are heated with concentrated sulphuric acid on a cover-glass, very pretty siliceous skeletons are left behind." Later spore protoplasm was reported to differentiate into a peripheral one, where storage substances are dominant, a middle zone containing chloroplasts, and an internal one surrounding the nucleus (Gullvag, 1968). The central body of a spore is approximately of 30–50 µm in diameter (Duckett, 1970). Each spore has four elaters, which respond to humidity variations (Newcombe, 1888).

Under dry air (depending on humidity) elaters can demonstrate periodic opening, and upon drying from a fully hydrated state, elaters may open rapidly to pitch a spore 1.5 mm above the surface (Marmottant et al., 2013). The described locomotion helps dispersal strategies using both hydrodynamics and aerodynamics because of local changes in environmental conditions. The structural mechanism of elater mobility is likely due to their bilayer structure: the inner layer consists of longitudinal microfibrils, the matrix of the outer elater formed because of structural granulations is expected to be less dense and porous (Uehara and Kurita, 1989). The differential volume change of one layer with respect to the other (Elbaum et al., 2007; Reyssat and Mahadevan, 2009) is suggested to be responsible for the curvature changes of the elaters (Marmottant et al., 2013). Besides a role in dispersion and spore motility, one may anticipate that due to the terminating paddle structure, elaters will have other purposes either when folded or when unfolded. The morphology of the paddles, structuring and elemental distributions at their surface (Perry, 1989) may suggest that in a dry environment paddles may assist the physical fixation of a spore at a site and conditioning of local biochemistry prior to germination.

In one of our earlier studies, we pioneered the characterization of silica distribution in elaters of E. arvense spores (Perry, 1989). On the surfaces of the spores and elaters, we identified small rounded objects ca. 500 nm in diameter protected by a coating of small silica particles, that contained proteinaceous material rich in arginine, low in aromatic amino acids and with a small amount of glucose containing polysaccharide material (Perry, 1989). In a relatively recent study, fluorescence labeling was used to observe "punctate" deposits of silica on the spore surfaces (Law and Exley, 2011). To understand the role of silica deposits on the surface of the spore/elater germination machinery we need to provide chemical/biological spatial-functional correlations between the bio-organic and bio-inorganic components. Correlations in chemical composition for different extractions (Currie and Perry, 2009) lent support to the idea that the observed spatial codistributions of the organic and inorganic components are according to the genetically programmed biochemistry of the plant. However, a deeper insight would require characterization of the structural states of the silica and the organic moieties at the junctions. This is where, due to resolution and chemical selectivity, the methods of Raman microscopy sampling are

particularly valuable. For example, the technique has been applied to the analysis of the chemical composition of Equisetum hyemale, where Gierlinger et al. (2008) described the nonuniform distribution of silica below a cuticular wax layer.

The challenge of what is being attempted becomes obvious if we compare the results by Raman microscopy with the conclusions of Kaufman et al. (1971) who used electron microscopy to report that silica is uniformly distributed over and within the epidermal surface in Equisetum hyemale. The differences, however, may be accounted for by variations in sample preparation (if the plane of cuts were explored or surfaces) and due to differences in penetration depth for different frequencies of radiation: being either a few or several hundreds of nanometers for X and Y, respectively. From this perspective, the results of the two techniques are both relevant and should be discussed comparatively if possible.

In this article we combine Raman microscopy with scanning electron microscopy (SEM) and elemental analysis to explore (and correlate) structural, vibrational and elemental properties on the paddle structure of a selected elater of a spore of Equisetum arvense. Further, we use computational studies of structures representing the major classes of materials anticipated to confirm identification of the materials present. The structure of the results section of the article is as follows: after describing the spore elater complex we present, (a) experimental spectra obtained from different regions of the system studied, (b) theoretical vibrational spectra calculated for representative (bio)chemical markers necessary for the reconstruction of Raman microscopy images, and (c) Raman difference microscopy images reconstructed for the selected spectral markers where for this task we adopt an ansatz for renormalization of Raman intensities. The described details on the distribution of bio-inorganic structural components allow us to discuss the bio-functionality of these structures and to hypothesize on survival strategies.

#### MATERIALS AND METHODS

Spore heads of a field horsetail, Equisetum arvense, were collected in Nottinghamshire (May 2018), see details in the **Supporting Information**.

Silica nanoparticles were synthesized by a modified Stöber method (Stober et al., 1968). The synthesized particles were rehydrated several times in the presence of deuterium oxide at high-temperature and vacuum avoiding annealing to allow for solvent exchange with deuterium oxide. For the optical studies 200 ± 12 nm diameter particles were used, the sizes of which were determined by employing dynamic light scattering (Zetasizer Nano-ZS, Malvern Instruments, Malvern, United Kingdom).

Material and elemental analysis: sample imaging and energydispersive X-ray spectroscopy (EDS) were conducted using a JEOL 7100FEG SEM equipped with an Oxford Instruments X-MaxN 80 mm<sup>2</sup> EDS. Samples were mounted on an aluminum stub with carbon tape (TAAB, Aldermaston, United Kingdom). The instrument was operated in secondary electron mode with a 1.0 kV accelerating voltage for imaging with a beam of about 2 nm diameter. For elemental analysis (EDS), the accelerating voltage was set to 10.0 kV. Micrographs were collected and exported using PC-SEM v. 5.1.0.6 and EDS spectra were collected, processed and exported using Aztec 3.3 SP1.

Raman spectral studies were conducted using a DXR microscope from Thermo Fisher Scientific, Madison, WI, United States equipped with 50× and 100× microscopy objectives. The spectral resolution in the Raman experiment was down to 2 cm−<sup>1</sup> according to the instrumental limit of the microscope operated with a 25 micron confocal slit or pinhole. The former was used for Raman spectral measurements when spatial resolution was not considered, while the latter was used for sampling of Raman maps. Raman measurements were made using 532 nm excitation radiation of 2 mW.

To correlate results of SEM and Raman microscopy studies, in the latter we used a spore sample on the same aluminum stub fixed with double-sided carbon adhesive tape as prepared for SEM. However, since Raman studies cannot be conducted on samples deposited on carbon tape (due to immediate thermal degradation of carbon under laser 532 nm radiation even when working at minimal power), it was necessary to search for an elater that would be free and hanging from the edge of the carbon tape. Hence, upon Raman mapping with a short focal length 100× microscope objective, the elater would not be physically disturbed by the objective. Hereafter, every time a suitable elater was found, numerous pre-tests were conducted by taking Raman samples to monitor the mobility and structural stability of the elater. In particular, it was determined by periodic focusing of field radiation of different powers that the minimal possible power of 2 mW] could be used as under such the elater would stop changing the bending of its stem. After sampling Raman maps of different regions of a selected elater (and before the elater would demonstrate structural degradation and decomposition), we conducted SEM microscopy and elemental analysis for carbon, oxygen and silicon atoms on the same elater (as shown in **Figure 2**).

Raman activities at different sites of the selected elater were sampled with a spatial resolution of 1 micron in both directions of the imaging plane. As a Raman microscopy scan collects a set of spectra specific for defined positions in the image plane, in order to present Raman microscopy images specific to spectral signatures of interest, we must process each spectrum from the detected set to extract amplitudes of spectral components of interest. Being dependent on the nature of the vibration (normal) modes, abundance of contributing structural species, orientation and degree of orientational variance at the sampling sites, the extracted amplitudes therefore are informative on molecular structural distributions at interfaces. To approach this, we conducted reconstructions of Raman activity microscopy images (RAM) specific to selected vibrations according to:

$$\operatorname{RAM}(X,\,\,Y,\,\alpha)\,\,=$$

$$\sum\_{\mathbf{i}} A\_{\alpha,i} \frac{1}{2\pi\sigma^2} \text{Exp}\left[ -\frac{(X - X\_{\mathbf{i}})^2}{2\sigma\_{\mathbf{x}}^2} - \frac{(Y - Y\_{\mathbf{i}})^2}{2\sigma\_{\mathbf{y}}^2} \right] \tag{1}$$

where i, is the index of a site where a spectrum is taken, Aω,<sup>i</sup> , is the amplitude of the Raman resonance of interest in the detected

spectrum, and ω, is the frequency of the resonance. Furthermore, the equation shows that we sum image projections of twodimensional Gaussian source functions over all the defined sites i. Setting σ 2 <sup>x</sup> = σ 2 rmy = 0.5 µm<sup>2</sup> provides the spatial full width of a source function. X<sup>i</sup> and Y<sup>i</sup> describe the position of the projection of the site i into the image plane. X and Y variables are sample distances from the site i in terms of the dimensions of detector pixels or displacements of a pinhole.

To utilize Raman microscopy imaging properly, we need to (1) understand "which" spectral resonance can be used to describe a particular structural species. In the last two decades, multivariate algorithms, such as principal component analysis (Pearson, 1901; Dieing and Ibach, 2011), independent component analysis (Ans et al., 1985; Hyvarinen et al., 2001) and methods of cluster analysis (Hedegaard et al., 2011), have gained popularity as powerful tools in the processing of microscopic images and Raman microscopy images of Equisetum tissues have previously been processed using principal component analysis (Gierlinger et al., 2008). In our studies, however, we approach Raman image reconstruction using (i) spectral markers obtained by exploring and comparing our experimental results, (ii) peak assignments previously reported in literature and, (iii) our predictions of quantum chemistry for the model molecular systems. The adopted approach is more computationally demanding than extraction of orthogonal (independent) spectral signatures upon application of linear algebra on detected spectral sets. However, this allows us, first, to avoid possible artifacts due to non-linear variances of Raman amplitudes as the surface of elaters in relation to molecular orientations are not trivial; and, second, trying to understand better the nature of the observed Raman responses.

As we have mentioned, we adopted a computationally demanding approach to select responses suitable for molecular structural analysis using Raman image reconstructions. To manage the task, we conducted quantum chemical calculations for selected model systems using the 6–31 g<sup>∗</sup> basis set and the restricted b3lyp functional (Becke, 1988; Lee et al., 1988) within the Gaussian 09 program package (Frisch et al., 2010). To model vibrational properties of inorganic structural components which may resemble that of the elater's surface, we adopted (see **Figure 3**): (i) Silica-10, four hexagonal cycles merged in a cage system; (ii) Silica-16, four hexagonal cycles merged in a single layer system; (iii) Silica-48, single layer spherical structure with two pores on opposite sides; and (iv) Silica-60, single layer spherical structure, with two pores on opposite sides, fused with a cage to model a small span of a double layered system. To model vibrational properties of bio-organic components as expected in Equisetum tissues, we adopted structural segments of cellulose, glucomannan, pectin, lignin and a methylated dipeptide with arginine side group and arginyl-n-acetyl-di-glucosamine (NDGA) at a Silica-6 hexagonal cycle. Here, we used the following structural definitions, as described in the literature: cellulose is a linear polysaccharide consisting of thousands of β(1-4) linked D-glucose units (Updegraff, 1969), glucomannan is a hemicellulose polymer, with linearly linked β(1-4)-linked Dmannose and D-glucose in a ratio of 1.6:1 (Katsuraya et al., 2003), pectins are hetero-polysaccharides with chains and branches of α(1-4)-linked D-galacturonic acid (Ridley et al., 2001), lignins are bio-organic polymers composed of phenylpropanoids p-hydroxyphenyl, guaiacyl, and syringyl (Boerjan et al., 2003; Martone et al., 2009). The scaling factor for the calculated Raman dispersions was 0.97.

Using the results of quantum chemistry and experimental Raman spectroscopy to correlate the presence of different chemical species at the elater surface, we show differences between Raman microscopy images reconstructed for selected vibrations, while the intensities of the Raman images are scaled to be equal. This intensity renormalization allows qualitative (relative) characterization of spatial co-distributions of chemical species of interest at very complex interfaces and surfaces.

In the following Results section we describe: Raman spectral properties of a spore central body and of its elaters; review theoretically predicted spectral responses for the model molecular systems; select spectral markers and demonstrate the results of Raman difference microscopy images constructed for the selected spectral markers.

## RESULTS

## Bright Field Optical Microscopy

The structural changes of spores of E. arvense upon humidity are shown in **Figure 1**. Under low humidity elaters are unfolded or partially unbent as shown in **Figure 1A**, whereas, in an aqueous environment (or in humid air), elaters wrap around the main spore body (see **Figure 1B**). The diameter of the central body of the spores corresponds well to the typical reported for this species, approximately 30–50 µm (Duckett, 1970). It is also observed, **Figure 1C**, that at the surfaces of the paddle structures and at the sides of the stems of the elaters there are small rounded objects, **Figure 1D** which have been previously reported to contain polysaccharide and proteinaceous material rich in arginine and reinforced with porous silica layers (Perry, 1989). Let us now describe the Raman spectral properties of a spore central body and of its elaters.

## Raman Responses From the Spore Central Body

Raman spectra sampled at the spore body under air and from the squeezed content are shown in **Figure 4A**. Accordingly, **Figures 4A1,A2** show the corresponding optical microscopy images of the spore body under air and its content in water squeezed by application of a 300 µm glass cover slip. As expected, under 532 nm excitation, the detected spectra are dominated by spectral responses of carotenoid molecules which are generally present in light-harvesting proteins (Ruban et al., 1995, 2001). The spectra show the ν<sup>1</sup> mode specific to C = C- stretching vibrations; ν<sup>2</sup> mode of C-C stretches coupled either to C-H inplane bending or C-CH<sup>3</sup> stretching, the ν<sup>3</sup> mode characteristic to CH<sup>3</sup> in-plane rocking vibrations; and the ν<sup>4</sup> mode specific to C-H out-of-plane bending (Rimai et al., 1973). The peak at 1520 cm−<sup>1</sup> was fitted with two components, the major at 1523 cm−<sup>1</sup> and the minor centered at 1515 cm−<sup>1</sup> . Accounting for the reported sensitivity of Raman on excitation wavelength (Ruban et al., 2001) and the possible linear regression (Gall et al., 2015) of the

wavelengths of electronic transitions for zeaxanthin (Landrum and Bone, 2001), β-carotene (Britton, 1995), lutein (Takaichi and Shimada, 1992), and lutein epoxide (Melendez-Martinez et al., 2005), on the frequencies of the ν<sup>1</sup> of C = C- stretching vibrations, we may anticipate that the main component fitted at 1523 cm−<sup>1</sup> is due to the contribution of β-carotene. The smaller component is likely due to Fermi resonance with possible combinations and overtones. At the same time, it cannot be completely ruled out that the lower frequency component at 1515 cm−<sup>1</sup> may be a signature of another carotenoid, like, for example, rhodoxanthin as was reported to be present in significant amounts in sporiferous shoots of E. arvense (Czeczuga, 1985). Accounting for dependencies on the wavelength of Raman excitations, the reported electronic and vibrational properties for rhodoxanthin (Chabera et al., 2009; Berg et al., 2013) may fit approximately the correlation electronic transitions on ν<sup>1</sup> of the C = C- stretching frequency, as suggested in Boerjan et al. (2003).

#### Raman Responses From Spore Elaters

Next, let us review the results of Raman spectroscopy of elaters, which, as organelles, have been described as four narrow spiral bands, formed upon division of the external coat of a spore at maturity (Sadebeck, 1878). **Figure 4B** shows several Raman spectra sampled at various sites on elater paddle structures. The spectra show two rich and complex subsets of Raman activity: between 250 and 750 cm−<sup>1</sup> and between 870 and 1750 cm−<sup>1</sup> which is expected to be due to both, inorganic and organic structural components (Sapei et al., 2007; Gierlinger et al., 2008). Also, in the spectral region covering 2600 to 3100 cm−<sup>1</sup> , resonances which are typical for -CH stretching modes of bioorganic molecules are observed (Atalla and Agarwal, 1985; Wiley and Atalla, 1987; Edwards et al., 1997; Agarwal and Ralph, 1997; Agarwal et al., 2001; Jahn et al., 2002; Synytsya et al., 2003). The spectra shown are site specific and are selected to demonstrate spectral diversity.

To analyze the observed spectral responses, we use the results of computational quantum-chemical density functional theory (DFT) calculations to account for the spectral contributions of bio-organic and bio-mineral species we expect to be present: (1) silica structures, (2) carbohydrates, and (3) lignin(s), amino acids and polypeptides; as spectral markers to discuss the spatial distribution of the expected molecular species. In **Figure 5** we present the calculated isotropic Raman responses (DFT predictions) for silica and hydro-carbon molecular structures, as shown in **Figure 3**, which we may consider as representatives of the main structural moieties present at the surface of the elater paddle structure.

#### Deduction of Spectral Markers Specific to Silica Structures

Raman responses of inorganic amorphous silica have been previously reported in literature. According to the results of early studies on Raman responses in silica gels (Gailliez-Degremont et al., 1997), attribution of (a) strong activities at 430–440 and 490–495 cm−<sup>1</sup> to in-plane Si-O-Si vibrations and to the modes associated with SiO<sup>4</sup> tetrahedra with a non-bridging oxygen atom; (b) medium and weak intensity signals at 800 and 1070 cm−<sup>1</sup> to Si-O-Si symmetric and asymmetric stretching, respectively; and, (c) weak and medium responses at 910–920 and 970–980 cm−<sup>1</sup> to surface and internal silanol stretching, respectively. These assignments were adopted to discuss the experimental results of Raman microscopy studies on the distribution of silica in a knob structure of E. hyemale (Gierlinger et al., 2008). In particular, Gierlinger et al. suggested that (i) an "overall remarkably high intensity" below 580 cm−<sup>1</sup> should be a signature of amorphous silica in Equisetum tissue; (ii) Raman resonance at 802 cm−<sup>1</sup> is due to Si–O–Si and Si–C stretching (Gailliez-Degremont et al., 1997); and (iii) the strong band at 973 cm−<sup>1</sup> is specific to surface and internal silanol stretching. In a separate report (Sapei et al., 2007), the authors discussed the role of silica hydration in contact with polysaccharides to explain the presence of the peak at 973 cm−<sup>1</sup> . Note that this resonance was detected at the knob tip but not when the signal was sampled from a silica rich layer adjacent to epidermal cells.

To understand better the possible contributions of silica inclusions into the Raman spectra of elaters (**Figure 4B**), in **Figure 4C** we show Raman spectra from silica rich rosette structures at the side of an aged dry branch (the corresponding microscope images are also shown in **Figures 4C**1**,C**2); and in

**Figure 4D** we show the Raman response from synthesized silica nanoparticles after different thermal treatments (TEM image is shown in **Figure 4D**1). The response from the latter systems is particularly helpful to verify the spectral contributions specific to Si-OH groups, which are expected at the particle surface (Perry and Keeling-Tucker, 2000). As it emerges from IR (Bunker et al., 1989; Morrow and McFarlan, 1992), Raman (Brinker et al., 1982; Humbert, 1995), and NMR (Chuang and Maciel, 1996)

FIGURE 4 | Raman spectra of Equisetum structures and silica nanoparticles. (A) Raman spectrum of the central body of an Equisetum spore in water (green line) and its optical microscopic image, (A1) (scale bar 2000 µm); and the Raman spectrum of the squeezed-out content of the central body of an Equisetum spore (blue line) and its optical microscopic image, (A2) (scale bar 10 µm). (B) Raman spectra sampled at different sites on the Equisetum spore elater. Numbers indicate spectral regions used for reconstructions of Raman microscopy images specific to these frequencies. (C) Raman spectra of silica rich rosette structures at the surface of the dried branch and its microscopic images, (C1,C2) (scale bars 10 and 20 µm, respectively). (D) Raman spectra of ca. 200 nm diameter amorphous silica nanoparticles (gray line) and the same sample after several hours of high temperature treatment at 1200 C (red line) and a TEM image (D1).

FIGURE 5 | Raman responses calculated with DFT for a series of model systems designed to represent structural components of cellulose, glucomannan, pectin. arginine side group, and silica structures as indicated. The structures of the model systems are presented in Figure 3. The dotted marks denote spectral contributions of NH<sup>2</sup> bending modes specific to the arginine side group in the considered systems.

experimental studies, amorphous silica surfaces may undergo dehydroxylation via condensation of vicinal silanols upon heating. Consistently, Raman spectra in **Figure 4D** shows a significant decrease of Raman activity at 976 cm−<sup>1</sup> for thermally treated amorphous silica particles compared to samples maintained at room temperature.

Inspection of the nature of the normal modes by DFT reveals that admixture of delocalized out-of-plane Si-OH bending and in-plane Si-O-Si symmetric vibrations of SiO<sup>4</sup> tetrahedra contribute to the Raman activities anticipated at 500 cm−<sup>1</sup> , **Figure 5**. The variations in this spectral region are due to: (a) relative weights of the two types of vibrations, (b) geometry of hydrogen Si-OH. . .O-H bonding, which may not be optimal, (c) extent of delocalization, and (d) effects of coupling that contribute to excitonic splits. There is a noticeable tendency for the normal modes, where the contributions of out-of-plane Si-OH bending mode dominate, to manifest at the lower frequency side. The delocalized in-plane Si-OH bending modes are shown in the spectral region between 700 and 1200 cm−<sup>1</sup> . The diversity in this region is due to different degrees of admixing of such bending activities with O-Si-O symmetric stretching at about 800 cm−<sup>1</sup> , or with Si-O stretching at about 900 cm−<sup>1</sup> , or with Si-O-Si antisymmetric vibrations at 1000 cm−<sup>1</sup> and above. The results of our calculations partially agree with assignments for silica gels (Gailliez-Degremont et al., 1997). However, DFT theory suggests a significant (if not a leading) role for various outof-plane and in-plane Si-OH bending modes in the considered structures. This is certainly due to the enhanced contribution of the surface in small and single-double layered systems and this is what we expect at the surface of Equisetum elaters.

The calculated Raman spectral activities in the region between 700 and 1200 cm−<sup>1</sup> for the silica structures (**Figure 5**) agree better with the experimental spectral responses from plant tissues, as shown in **Figures 4A–C**, rather than with spectral responses from the amorphous silica nanoparticles, as described in **Figure 4D**. Therefore, considering previous peak assignments, the spectral properties of silica nanoparticles and the suggestions of theoretical studies on silica cage systems, we may adopt the broad scattering intensity between 500–580 cm−<sup>1</sup> as a signature of biosilica at the surface of an elater (marker #1 in **Figure 4B**). Also, we ascribe Raman activities in the spectral region 920– 1000 cm−<sup>1</sup> , see marker #3 in **Figure 4B**, to biosilica. However, due to the weak intensity, we do not use such responses in our discussion based on Raman microscopy image reconstructions (see next section of the manuscript, Raman microscopy). Here, we wish to emphasize that we do not observe resonance at 805 cm−<sup>1</sup> in the spectral response from elaters, while such a signal is strong in the response from the nanoparticles. It is interesting that both Raman signatures at 813 and 976 cm−<sup>1</sup> are present in spectra detected form the rosette structures, though they are quite weak. The observed variances in the spectral responses of silica at the elater paddle structure suggest that structural composition of this mineral component at the surface of elaters may have a peculiar, distinct character, that likely differs from that found in amorphous inorganic nanostructures or in bio-organic deposits developed by plants for the purposes of defense and mechanical strength (Timell, 1964; Kaufman et al., 1971; Perry and Fraser, 1991; Hodson et al., 2005; Grégoire et al., 2012). This observation may have justifications from the perspective of spore biology – we will address this in our general discussion.

#### Deduction of Spectral Markers Specific to Carbohydrates

Polysaccharides are the main bio-organic structural components for plants and for Equisetum spp. tissues (Sapei et al., 2007; Gierlinger et al., 2008). To discuss our experimental observations, we adopt previous assignments for polysaccharide biopolymers such as cellulose (Wiley and Atalla, 1987; Edwards et al., 1997), glucomannan (Agarwal and Ralph, 1997), and pectin (Synytsya et al., 2003); as well as our predictions of Raman responses for the model systems, **Figure 5**. In these structural cases, theory predicts a set of spectrally narrow resonances in the frequency range 300–700 cm−<sup>1</sup> . C-OH out-of-plane bending contributes in this range but dominates at lower frequency. Symmetric and antisymmetric ring deformations with possible admixing of out-of-plane wagging of C-O-C bridges and C-OH outof-plane bending contribute in the central region and at the high frequency side of this spectral range. In earlier studies, normal modes in this spectral region were mainly assigned to skeletal-bending modes involving the CCC, COC, OCC, and OCO internal coordinates of glucose-like moieties (Wiley and Atalla, 1987; Edwards et al., 1997). It is interesting to notice that for more regular (polycrystalline like) cellulose and glucomannan structures, normal modes in the frequency range of 300–700 cm−<sup>1</sup> tend to group into higher and lower frequencies subsets with a window of relative transparency at 500 cm−<sup>1</sup> , which is where theory predicts that Raman resonances of silica shell systems would contribute the most, see the previous section. In the case of a more distorted pectin-like system, the out-ofplane C-OH bending modes admixed with the ring vibrations; fill the spectral range 300–700 cm−<sup>1</sup> more uniformly.

Theoretical prediction (**Figure 5**) shows an absence of Raman activities for the more regular (polycrystalline) cellulose-like system in the spectral range between 700 and 1000 cm−<sup>1</sup> . For cellulose-like molecules, there are weak resonances at 870– 900 cm−<sup>1</sup> due to stretching/bending localized on C4-C<sup>5</sup> and C5-C<sup>6</sup> of glucopyranose rings. Besides this, the delocalized ring deformations which involve C4-C5-O and C4-O-C<sup>1</sup> bending and C1-O stretching experience strong splitting at 723 and 989 cm−<sup>1</sup> . Calculations for glucomannan- and pectin-like systems provide similar results for this spectral range, but the normal modes are less delocalized due to the more deformed and less crystalline character of the materials. As a result, the C2-C<sup>3</sup> and C1-O stretching, C1-C2-C<sup>3</sup> symmetric and antisymmetric stretching, C1-O-C<sup>4</sup> and C1-O-C<sup>5</sup> symmetric stretching, and C1-O-C<sup>4</sup> bending start to contribute in the spectral region between 700 and 1000 cm−<sup>1</sup> . In early studies, Raman activities in this spectral region were assigned to ν(COC) in plane symmetric mode at 897 cm−<sup>1</sup> for glucomannans (Agarwal and Ralph, 1997; Gierlinger et al., 2008), to γ(COH)ring for pectins at 817 and 832 cm−<sup>1</sup> , to a glycosidic asymmetric (COC) skeletal mode of α-anomers of pectins at 855 cm−<sup>1</sup> , and to α-glycosidic bonds of acidic pectin at 859 cm−<sup>1</sup> (Synytsya et al., 2003). Considering the

results of our theoretical studies we ascribe the experimentally detected Raman responses between 860–900 cm−<sup>1</sup> (see marker #2 in **Figure 4B**) to local C1-O and C2-C<sup>3</sup> stretching and C1-O-C<sup>4</sup> bending modes of less regular and more deformed glucomannanand pectin-like polysaccharides.

Comparing the experimental data with the results of DFT studies on polysaccharide model systems, we can state that the broad band between 1000 and 1150 cm−<sup>1</sup> (**Figure 4B**) is dominated by delocalized C-O-C symmetric, C-O-C antisymmetric and C-O stretching admixed with some contributions from C-C stretching of the pyranose rings. The degree of delocalization is smaller for more deformed structures, like pectin. At lower frequencies, there are also contributions of partially delocalized pyranose rings corresponding to C-C stretching admixed with COH bending. The peak assignments are consistent with those reported in early studies for cellulose (Wiley and Atalla, 1987; Edwards et al., 1997), glucomannan (Agarwal and Ralph, 1997), and pectins (Synytsya et al., 2003). From our Raman microscopy studies, we attribute the contribution of Raman intensities between 1000 and 1150 cm−<sup>1</sup> (marker #4 in **Figure 4B**) as a "generic" spectral signature of bio-organic components (mainly polysaccharides) for the normal modes, where interatomic displacements are mainly in the plane of pyranose rings. The next, higher energy spectral subsets of Raman activities centered at 1250 and 1375 cm−<sup>1</sup> are specific to (possibly) delocalized C-CH bending modes associated with pyranose rings with contributions of both, C-CH and C-OH bending of the side groups, consistent with the early assignments reported in literature (Wiley and Atalla, 1987; Agarwal and Ralph, 1997; Edwards et al., 1997; Synytsya et al., 2003). It is interesting that for the three considered polysaccharides, our theory predicts (with minor spectral deviation) the CH<sup>2</sup> scissor modes at around 1500 cm−<sup>1</sup> . In the experimental spectra this should correspond to spectral signatures at about 1464 cm−<sup>1</sup> , which we adopt as a spectral marker #5 in **Figure 4B**.

Finally, theoretical predictions for Raman activities specific to C-H stretching modes in the spectral range 2800–3010 cm−<sup>1</sup> were investigated. In the case of the more regular cellulose and glucomannan structures, DFT anticipates (a) CH stretching of pyranose rings should dominate at the lower frequency side from 2870 to 2915 cm−<sup>1</sup> ; (b) side group CH<sup>2</sup> symmetric stretching mixed with CH activities of pyranose rings contribute mainly in the spectral range 2929–2949 cm−<sup>1</sup> ; and, (c) side group CH<sup>2</sup> antisymmetric stretching mixed with CH activities of pyranose rings would dominate at the higher frequency side from 2980 to 3009 cm−<sup>1</sup> . In comparison, theory predicts that CH Raman activities for less ordered pectin-like systems are broader. Inspection of the normal modes in that region reveals that in such systems, the vibrations with contribution of CH<sup>2</sup> symmetric and antisymmetric stretching modes "explore" wider spectral ranges. From this perspective, the maximal Raman activity at 2882 cm−<sup>1</sup> , which we adopt as spectral marker #7 (**Figure 4B**) should be informative on the more ordered celluloselike structural components (see **Figure 3**) at the interface of the elaters where CH<sup>2</sup> antisymmetric stretching mixed with CH activities of pyranose rings dominate at the higher frequency side from 2980 to 3009 cm−<sup>1</sup> .

#### Deduction of Spectral Marker Specific to Lignin

Firstly, it is important to note that the intensities of the two Raman transitions at about 1607 and 1683 cm−<sup>1</sup> (see **Figure 4B**) vary depending on the sampling spot at the surface of the elater paddle structure. Further, considering the spectral response from the rosette structures at the side of aged Equisetum dry branch, **Figure 4C**, two analogous resonances are observed, though the lower frequency resonance becomes very narrow and is shown at 1620 cm−<sup>1</sup> . According to previously reported results (Atalla and Agarwal, 1985; Agarwal et al., 2001; Jahn et al., 2002; Sapei et al., 2007), these two spectral signatures can be attributed to vibrations of the lignin structural component. Indeed, quantum calculations for the normal modes of a lignin-like segment (see **Figure 5**), suggest that the lower frequency resonance may be assigned to symmetric stretching in the aromatic ring, analogous to the '8a' and '8b' modes in substituted benzenes as assigned by Varsanyi (1974). The higher frequency resonance may be a signature of carbonyls which are present in lignin (Kirk and Farrell, 1987; Christopher et al., 2014).

To distinguish the role of lignin at the surface of elaters, we identified another independent spectral marker specific for this bio-polymer. For example, **Figure 4C** shows a dominant resonance at 2930 cm−<sup>1</sup> , which was reported as a spectral signature specific to lignin (Atalla and Agarwal, 1985; Agarwal et al., 2001; Jahn et al., 2002; Sapei et al., 2007). Considering that in the spectra detected from elaters, the Raman response at 2930 cm−<sup>1</sup> is present as a smaller shoulder and it is not proportional to the spectral signatures at about 1607 and 1683 cm−<sup>1</sup> , we anticipate that lignin is a minor structural component at an elater surface, and, if present, its distribution is not uniform, and it is not well-ordered. According to the Raman response as shown in **Figure 4C**, this contrasts with the clear presence of lignin as a constituent structural component in a dried branch, where lignin fibers likely form well aligned depositions to contribute to mechanical stability. Accounting for both variances of experimental Raman responses (in **Figures 4B,C**) and DFT predictions for lignin in the spectral range between 1520 and 1750 cm−<sup>1</sup> , we adopt the experimentally observed Raman resonance at 1607 cm−<sup>1</sup> as a representative signature for the lignin contribution and indicate it with marker #6 in **Figure 4B**.

#### Deduction of Spectral Markers Specific to Amino Acids and Polypeptides

The results of our previous studies (Perry, 1989) suggest that in this spectral region may be expected possible spectral signatures due to arginine functional groups and other protein related structural motifs. Exploring Raman activities predicted for the arginine side group, we confirm that there are two relatively weak and two strong in-plane NH<sup>2</sup> bending modes for this moiety. These demonstrate variance in relative atomic displacements and in frequencies that depend on possible coordination of the side group with silica and orientation in respect to the backbone of a structural moiety: see the dotted markers next to the corresponding spectra in **Figure 5**. The results of theoretical studies suggest that the observed underlying spectrally broad background in the spectral range 1663–1754 cm−<sup>1</sup>

where the Raman activities of the subtracted Raman microscopy maps, reconstructed at frequency by marker #7 (see Figure 4), are larger than the Raman activities

of the Raman microscopy maps reconstructed at frequency by marker #2. Red areas indicate spatial regions where the situation is the opposite of this.

and observation at some sites of a relatively narrow spectral signature at 1675 cm−<sup>1</sup> (see the upper spectrum in **Figure 4B**), may indicate the presence of arginine and carbonyl moieties, respectively. However, due to the weak intensity, breadth and non-systematic character of the spectral responses measured, we do not use these spectral signatures in our further analysis.

#### Raman Difference Microscopy on the Bio-Inorganic Composition of an Elater Surface

Taking differences between equally scaled images (specific to the selected Raman activities) we gain a qualitative comparison of how one Raman active vibrational activity correlates or anticorrelates in respect to another in space. Accordingly, **Figure 6** shows three sets of Raman microscopy maps corresponding to the upper, the middle and the lower sections of the same elater paddle structure. Here, we recapitulate the chemical-structural aspects we wish to stress using Raman difference maps. The Raman difference signal mapping labeled as 1–4, 1–6, and 1–7 would contrast spatial distributions of biosilica versus spatially aligned structural elements of generic extended celluloselike polysaccharides mainly, lignin, and other generic organic contributions expressed through CH stretching, respectively. In contrast, the Raman microscope maps named 2–7, 5–7, and 6–7 would contrast spatial distribution of pectin-like polysaccharides (Synytsya et al., 2003; Gierlinger et al., 2008), δCH<sup>2</sup> scissor modes of hydrocarbons and lignin versus distribution of organic contributions expressed through CH stretching.

To gain deeper insight, in **Figures 7**, **8** we show Raman difference signals along selected directions of the lower and the upper sampling sections, as demonstrated in **Figure 6**. Note that, selected directions match those in **Figure 2** where X-ray spectroscopy in the electron microscope was attempted. Considering the moderate accelerating voltage applied (to allow for the fragility of the specimen) and accounting for the approximate density of carbon atoms of about 24 atoms at 120Å<sup>2</sup> , we anticipate a relatively good sampling efficiency along

line B1, but a loss of sampling efficiency when signals were detected along lines B<sup>2</sup> and B3, see **Figure 2**. This is likely due to alteration in the orientation of the sampling plane, the gun and the detector during successive line scans. X-ray spectroscopy (EDS) instructs on the average relative levels of contributions of carbon and silicon atoms, and on the relative local variability. The former characteristic helps us understand the character of silica deposition. We see that, on average, the presence of silicon (as silica) is about 20 times smaller than that of the bio-organic component. This suggests the presence of thin (likely one or rarely few layers) clusters of silica at the interface. This agrees with the results of secondary ion mass spectrometry on some silica depositions in Equisetum arvense (Guerriero et al., 2018).

We first consider the bottom section of the elater, which comprises the connection between the stem and the paddle structure (**Figure 6**). This structural region has a very complex three-dimensional character, with a steep slope of the stem at the lower side. The EDS responses in **Figures 7O1,C1,Si1** show the underlying trend - atomic abundances increase (from left to right). This is likely to be due to orientation effect – reflection toward detector is favorable at the slope. In the same spatial region, Raman difference signals 1–4, 1–6, and 1–7 in **Figure 7A1** anti-correlate with the differences 2–7, 5–7, and 6–7 in **Figure 7B1**. For the biosilica component signals are collected more efficiently than those from the bio-organic structures and

the extended cellulose-like contribution is prominent amongst the others. Comparatively, the atomic abundances (C3, O3, Si3) and the Raman difference signals maps (A3 and B3) as shown in **Figure 7** are sampled from a more uniformly spatial region (structural components located at the same height), thus avoiding the tilting of the sampling plane. The SEM image shows that at the very left side of the sampling line, there is the presence of a spherical structure. Since, here, the sampling plane is not tilted; we may attribute the positive Raman differences 1–7 and 1–4 (in **Figure 7A3**) as signatures of silica surplus, likely due to the material of the vertical side walls of the structure. This is consistent with the fact that the positive signatures are spatially broader and "tipped" from the top. It is also interesting, that, in contrast to the dependences in **Figure 7B1**, the Raman differences 5–7 and 6–7 in **Figure 7B3** in the spatial region of the spherical structure are positive and relatively narrower, comparing to differences 1–7 and 1–4 in **Figure 7A3**. This suggests that the deposition of pectin-glucomannan glycoside may be associated with that of silica, and that such polysaccharide components are likely imbedded in the bio-inorganic component/structure. Further, the lignin-like contributions are diminished at the center of the spherical structure (see orange line in **Figure 7A3**). Finally, the Raman difference signals maps shown in **Figures 7A2,B2**, reveal an intermediate character considering the above described cases. This can be explained as a spatial overlap of the areas where Raman signals are sampled. Note that, the physical limit of Raman microscopy resolution cannot be better than λ/2, and

experimentally will always be inferior comparing to SEM. In **Figure 8** are shown the atomic abundances and Raman difference signals in the upper section of the elater paddle structure. This section of the paddle structure is the most distant from the stem; however, similar tendencies to that of the lower section are observed when comparing Raman mapping (A1, B1 and A2, B2) with **Figures 7A3,B3**.

After exploring possible correlations of several Raman difference signals with mapping spatially parallelized with atomic abundances, it is time to review the two-dimensional Raman difference microscopy images (in **Figure 6**), to help unravel the morphogenetic plan of the Equisetum spore developed by millions of years of evolution. First, we consider the Raman difference map of the stem part, which is expected to provide mechanical stability and locomotion properties. Images specific to differences corresponding to Raman maps 1–4, 1–6, 1–7, and 5–7 indicate that silica deposition may have a spiral character, see arrows in the lower set of panels in **Figure 6**, and it may be coordinated with spatial distributions and orientation of CH moieties associated with a cellulose-like structural component.

Exploring the middle section Raman map, we may confirm that formation of spherical structures requires a morphogenetic plan requiring the embedding of polysaccharide and silica layers and a supply of material. To address this further, black and green stars have been placed on the images for the middle and the upper sections in **Figure 6**. For example, the lower black stars (of the middle section) indicate the region where silica is strongly associated with CH modes of cellulose-like component, but anti-correlates with a pectin-glucomannan-like fraction which is embedded inside the structure, more concretely in the center of a spherical structure. Similar trends are observed in the spatial region marked by green stars in the middle section.

Here, it is interesting to notice that SEM images of the upper part of the paddle structure indicate a more rigged and folded morphology of the surface at the tip. This is consistent with Raman microscopy maps. Raman maps reveal a more red-blue rapidly altering pattern toward the edge, and the very edge, apparently is reinforced with celluloselike polycrystalline terminations – see the blue ridge at the edge in differences 1–7, 2–7, 5–6, and 6–7 in the Raman images specific to the upper part (see blue and red arrows in **Figure 6**).

Finally, it is necessary to address that in contrast to large scale preferences for distributions for pectin-glucomannan- and cellulose-like components reported in cells of stems and branches of Equisetum spp. (Perry and Fraser, 1991; Speck et al., 1998; Perry and Keeling-Tucker, 2003; Sapei et al., 2007; Gierlinger et al., 2008), distributions of polysaccharides and lignin at the elater paddle structure exhibits structure on the submicron scale. Overall, this correlates with the ridged-folded morphology of the surface of the elater paddle. This ridged-folded morphology becomes more obvious toward the edges.

## DISCUSSION

Early TEM studies demonstrated that an elater can be found on the surface of the plasmodial plasma membrane as a thin beltlike structure spirally coiled around the middle layer (Uehara and Kurita, 1989). The structure consists of an inner granulofibrous zone and of an outer micro-fibrillar region which aligns parallel to the longitudinal axis. This structure provides structural heterogeneity of the inner and outer layers of the elaters in stems to demonstrate locomotion capacity by rapid differential volume change (Marmottant et al., 2013). In our studies, we address the complexity of bio-organic decoration of the paddle structure of the elater, which should be consistent with the needs of biological survival and propagation. Our SEM studies clearly indicate: (a) a ridged complex, of about 0.5 micron, at the elater's surface, suggesting a folding complex nature of structuring of the bioorganic matter under the surface; (b) the presence of sub-micron diameter spherical structures on the ridged surface of the paddle structure and on the surface of the connecting stems; (c) more or less uniform silica deposition at the surface of the paddle structures. Our Raman microscopy results suggest that; (d) silica is deposited in its amorphous form in thin layered structures; (e) there is a relative increase of silica at the sites of spherical containers; (f) pectin- and glucomannan-like glycosides may have some preference in the interior of the spherical containers; (g) the spherical containers are attached at the surface of paddle structures where cellulose-like planar bio-organic components dominate and where lignin contributions are diminished. The observed differences in distribution and character of inorganic and bio-organic structural elements in the elater of Equisetum

arvense arise from the restrictions of prior morphogenesis and structural and elemental capacities gained during evolution to answer the practical challenges of initial proliferation and survival at the early stages of vegetation. To understand this, in **Figure 9** are shown the SEM and optical microscopy images of spores on carbon tape 10 days after their deposition. These images show that the paddle structures undergo motion to interact with the carbon substrate with the consequence that some paddles and some spore central bodies demonstrate partial submerging into the carbon substrate. Further, some of the paddle structures became flattened and expanded and opening of spherical container to expose its contents to the environment is revealed (see **Figure 9**). The observed structural reorganization suggests that there exists an on-going morphogenesis and biochemistry which involves the elater paddles and spore central body to support effective germination. From these results, it can also be concluded that silica deposition at the elaters' surface and their paddles as layered structures is beneficial for fast break down into a chemical substrate suitable for both, conditioning of local microbiology and to become ready for re-consumption as expected upon following the vegetation cycle. Opening of the internal spherical structures is likely to occur for the same purpose, while flattening and spreading of elaters may hinder photosynthesis of algae and other plants in the spatial region next to the spore central body.

Colloidal silica dissolution is a challenging task when applied to deposits in process cooling systems. Earlier studies suggested that hydroxyl ions play a catalytic role in the process (Jendoubi et al., 1997). However, later experimental studies indicated that an increase in the number of -COOH groups in various agents (capable to dissolve colloidal silica) does not have an obvious effect on dissolution efficiency. Instead, the presence of -PO3H<sup>2</sup> and NH<sup>2</sup> groups would appear to be important, and, particularly, under acidic environments (Mavredaki et al., 2005). Consequently, we may suggest that since degrading plant tissues typically generate an acidic environment (due to humid acid and other products), the side-group of arginine containing polysaccharide matrix (Perry, 1989) may facilitate the anticipated fast uptake of silica provided by the elater surface. This silica is particularly conditioned to: (i) not hinder the necessary flexibility and morphology of elaters; and, (ii) partition fast into the substrate in a colloidal form to be easily re-absorbed by young vegetation, facilitated by substrate chemistry pre-conditioned by decay and prior release of the content of the containers. Further, there are a range of opinions concerning whether silica may (Guerriero et al., 2018) or may not (Coskun et al., 2019) play a role in anti-fungal and anti-bacterial resistance mechanisms though it is not unreasonable to propose that fast re-absorption of silica would help improve the strength of structural elements within a living organism. If this is a skeletal component of a "fewcell" developing organism this may be along the biogenetic law stated by Haeckel of "ontogeneous recapitulation of phylogeny" (Haeckel, 1866). At the same time, we cannot exclude the opinion of the editor that fast silica re-absorption could be an evolutionary memory of mechanisms which are not obvious now but were possibly helpful in the past. It is possible that morphological and (bio)chemical studies of early Equisetum embryogenesis may help us understand if silica uptake was among the evolutionary advances of organisms (keeping in mind diatoms and sponges). This, however, is far beyond the scope of the current study.

## CONCLUSION

The use of sensitive Raman microscopy assisted with density functional theory is an attractive approach to explore the biomineral composition of Equisetum spore elaters spatially, both at the surface and within the biological structure. The spatialspectral optical sampling correlates with structural properties detected using scanning electron microscopy. The approach suggests that silica is deposited in an amorphous, nearly colloidal form within structures that are up to a few layers thick that are readily dissolved and dispersed on germination. Silica and pectin-glucomannan-like glycosides may have some preference in the internal content of the spherical containers which are attached at the surface of paddle structures where a celluloselike planar bio-organic component dominates and where ligninlike contributions are minimal. Spatial correlations of spectral signatures assist in addressing how structural properties and biochemical decoration of the elaters may support the physiology of the organelles and contribute to reproduction success.

## AUTHOR CONTRIBUTIONS

CP and VV conceived the study. CP collected the plant material, performed the early studies on germination, and supervised the study. VV conducted the Raman microscopy and DFT calculations. GH conducted the SEM microscopy. AS-R conducted the material studies. All authors contributed in writing and reviewing the manuscript.

## FUNDING

The authors gratefully acknowledge funding from AFOSR FA9550-16-1-0213 and thank Dr. Joanna Aizenberg, Harvard University, for continued access to the Odyssey Cluster at Harvard University. Fees for open access were provided by a research contingency fund allocated to CP by NTU.

## ACKNOWLEDGMENTS

Some of computations used in this paper were run on the Odyssey Cluster supported by the Faculty of Arts and Sciences Division, Research Computing Group at Harvard University.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00210/ full#supplementary-material

#### REFERENCES

fpls-10-00210 March 1, 2019 Time: 18:30 # 14



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Volkov, Hickman, Sola-Rabada and Perry. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Spectroscopic Discrimination of Sorghum Silica Phytoliths

*Victor M. R. Zancajo1,2,3\*, Sabrina Diehn2, Nurit Filiba4, Gil Goobes4, Janina Kneipp1,2,3\* and Rivka Elbaum5\**

1 School of Analytical Sciences Adlershof (SALSA), Humboldt-Universität zu Berlin, Berlin, Germany, 2 Chemistry Department, Humboldt-Universität zu Berlin, Berlin, Germany, 3 BAM Federal Institute for Materials Research and Testing, Berlin, Germany, 4 Department of Chemistry, Bar Ilan University, Ramat Gan, Israel, 5 R. H. Smith Institute of Plant Sciences and Genetics in Agriculture, The Hebrew University of Jerusalem, Rehovot, Israel

Grasses accumulate silicon in the form of silicic acid, which is precipitated as amorphous

silica in microscopic particles termed phytoliths. These particles comprise a variety of morphologies according to the cell type in which the silica was deposited. Despite the evident morphological differences, phytolith chemistry has mostly been analysed in bulk samples, neglecting differences between the varied types formed in the same species. In this work, we extracted leaf phytoliths from mature plants of Sorghum bicolor (L.) Moench. Using solid state NMR and thermogravimetric analysis, we show that the extraction methods alter greatly the silica molecular structure, its condensation degree and the trapped organic matter. Measurements of individual phytoliths by Raman and synchrotron FTIR microspectroscopies in combination with multivariate analysis separated bilobate silica cells from prickles and long cells, based on the silica molecular structures and the fraction and composition of occluded organic matter. The variations in structure and composition of sorghum phytoliths suggest that the biological pathways leading to silica deposition vary between these cell types.

#### Keywords: phytoliths, biosilicification, Raman, sorghum, solid state NMR, synchrotron FTIR

## INTRODUCTION

Grasses are silicon accumulators, concentrating silicic acid (herein Si) from the soil solution through the activity of Si transporters (Ma et al., 2006; Ma et al., 2007; Sakurai et al., 2015). Si moves with the water transpiration stream and deposits as hydrated amorphous silica (SiO2·nH2O) impregnating cell walls and filling cell lumens and intercellular spaces (Prychid et al., 2003). These microparticles are termed phytoliths. We can find phytoliths in root endodermis, leaf epidermis, inflorescence bracts, preferentially in highly transpiring organs (Jones et al., 1963). Phytoliths studies are relevant to geology and archaeology. This is because, similarly to pollen grains, under ambient conditions they are the more stable than other plant parts (Kelly et al., 1991; Shahack-Gross et al., 1996; Albert et al., 1999; Elbaum et al., 2003; Piperno et al., 2009; Ball et al., 2016). Organic molecules are trapped within phytoliths (Perry, 1985; Harrison, 1996; Elbaum et al., 2009; Parr and Sullivan, 2010; Gallagher et al., 2015; Asscher et al., 2017) and possibly reflect the chemical environment in which the silica formed (Perry and Keeling-Tucker, 2000). These organic entities can be studied by nuclear magnetic resonance (NMR) (Ravera et al., 2016). A seminal study of the hairs in the grass *Phalaris canariensis* demonstrates that plant silica has a significant fraction of surface silanol groups (Mann et al., 1983; Perry and Mann, 1989).

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Carole Celia Perry, Nottingham Trent University, United Kingdom Minh N. Nguyen, Vietnam National University, Vietnam

#### \*Correspondence:

Victor M. R. Zancajo rodriguez.zancajo@gmail.com Janina Kneipp janina.kneipp@chemie.hu-berlin.de Rivka Elbaum rivka.elbaum@mail.huji.ac.il

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 17 April 2019 Accepted: 11 November 2019 Published: 11 December 2019

#### Citation:

Zancajo VMR, Diehn S, Filiba N, Goobes G, Kneipp J and Elbaum R (2019) Spectroscopic Discrimination of Sorghum Silica Phytoliths. Front. Plant Sci. 10:1571. doi: 10.3389/fpls.2019.01571

1 **58**

In order to study phytoliths, the plant tissue around them is digested, many times by harsh chemistry, high temperature, or mild chemistry during very long time periods (archaeologic or geologic). These processes change the physical and chemical properties of phytoliths. These changes were monitored in phytolith assemblies (Jones and Milne, 1963; Cabanes et al., 2011; Watling et al., 2011; Cabanes and Shahack-Gross, 2015). Individual phytoliths were also characterized (Perry et al., 1984a; Perry et al., 1984b; Elbaum et al., 2003; Watling et al., 2011; Alexandre et al., 2015; Gallagher et al., 2015), and variation in the mineral structure was identified within one phytolith type (Perry et al., 1990). However, different phytolith morphotypes were not compared, and we do not know whether a specific morphotype has a unique chemical signature, which is different from other morphotypes.

Raman and fourier transformed infrared (FTIR) microspectroscopy enable the probing of individual phytoliths and assessing their mineral structure and occluded organic matter. In these vibrational microspectroscopy methods, information on chemical bonds and thereby structure and composition of a sample is obtained. By combining a microscope with FTIR or Raman spectrometer, the spectra are collected at a micrometre resolution. FTIR absorption spectroscopy gives fingerprint-like information that has been widely used to study cell wall constituents like proteins, aromatic phenols, cellulose, and to characterize biologically produced silica (also referred to as biogenic silica or biosilica, (e.g. Fröhlich, 1989; Kačuráková et al., 2000; Gendron-Badou et al., 2003; Kerr et al., 2013). Raman spectroscopy complements the information from FTIR spectroscopy and was used to analyse cell wall polymers, silica, phenolics, and lipids in varied plant tissues (e.g., Sapei et al., 2007; Chylińska et al., 2014; Prats Mateu et al., 2016). Spectral information is often encoded in very minute features. Principal component analysis (PCA) transforms the spectral dataset into a variance weighted vector-space, and provides us with a highly sensitive analysis for subtle spectral variations.

In this work, we extracted silica phytoliths from sorghum leaves, using two wet digestion methods, and compared the extracts using bulk and individual phytolith analyses. We used vibrational microspectroscopy, both Raman and FTIR, to characterize individual phytoliths and evaluate the differences between phytolith morphotypes. Our results indicate a significant influence of the extraction method on the structure and composition of phytoliths silica and occluded organic matter. Nonetheless, we could show that specific phytolith morphotypes contain characteristic organic molecules.

## MATERIAL AND METHODS

## Controlled Plant Growing Conditions

Seeds of *Sorghum bicolor* (L.) Moench (line BTx623) were sown in 1-L pots in universal potting soil (Bental 11, Tuff Merom Golan), and grown in a greenhouse at The Robert H. Smith Institute of Plant Sciences and Genetics greenhouse in Rehovot, Israel during September 20 2016 to January 1, 2017 under natural light and temperature of the Israeli autumn (21°C –33°C). The plants were irrigated automatically twice a day by water supplemented with N-P-K fertilizer (nitrogen (N), phosphorus (P2O5), and potassium (K2O)) at respective % weight ratio of 5-3-8. Leaves were harvested at flowering stage. Only fully developed green leaves were collected, and cut to exclude the main vein.

## Sample Preparation and Phytolith Extraction

Leaf pieces and cross sections were prepared manually using razor blades. Phytoliths were isolated from mature healthy leaves using two wet extraction methods: (a) H2SO4/H2O2/HNO3 extraction (herein SONE), (*Protocol 2* in Corbineau et al., 2013). Leaves were cut and rinsed with 10% HCl, immersed in 70% H2SO4 solution at 70°C for 2 hours, and left overnight at room temperature. The sample was heated to 70°C, 30% H2O2 was added slowly until the supernatant became clear, and then kept heated for 3 h. The sediment was collected, rinsed with DI water thrice, and reheated to 70°C in concentrated HNO3 for 2 h. About 50 mg of KClO3 was added and the sample was kept overnight at room temperature. The sediment was collected, rinsed with DI water, washed with 0.001 M KOH solution, rinsed three times with DI water, and dried at 70°C until its weight remained constant; (b) Microwaveassisted digestion (herein MAD) using a Discover SPD-80 sample digestion system (CEM, USA). Cut leaves were oxidized by 65% HNO3 for 30 min at room temperature in quartz vessels, afterwards the temperature was raised linearly to 200°C over 5 min and retained for 5 min at a pressure of 200 psi. The sample was rinsed three times with DI water and dried at 70°C. Phytolith samples from both extraction methods were stored in paraffine sealed Eppendorf tubes at ambient temperature until analysis.

#### Raman Microspectroscopy

Extracted phytolith samples were placed on a calcium fluoride slide without a cover slip. Raman spectra were collected from individual particles by a Jasco Raman spectrometer, using a 532-nm wavelength laser with a power of 5.6 mW for excitation, focused by a 100x objective to a spot size of ~1 µm2 . Spectra were obtained from 25 phytoliths of each morphology (bilobate silica cells, prickles or trichomes and long cells or plates), with 30 s acquisition time and 10 accumulations in the spectral range of 136 – 3977 cm-1. The spectra were calibrated using a spectrum of 4-acetamidophenol, and preprocessed with MATLAB, including background correction using asymmetric least squares method (AsLS), spectra interpolation yielding a spectral resolution of 1.8 cm-1, vector normalisation and selection of the spectral range of interest. PCA was performed on preprocessed spectra and on their first and second derivatives. By PCA, variations in the dataset were identified, which led to the formation of groups of similar spectra that were represented in scores plots. The loadings estimated how much each of the old coordinates, that is the wavenumbers, contributed to the PCs. Therefore, beyond differentiation and classification, PCA allowed us to highlight features in the collected dataset that are the basis for discrimination between the spectral groups, corresponding to each PC.

#### Synchrotron Fourier Transform Infrared (FTIR) Microspectroscopy

Extracted phytolith samples were placed on zinc selenide slides and FTIR transmission spectra were collected from individual particles in the range from 700 to 4,000 cm-1 using a FTIR microscope (ThermoNicolet) at the IRIS beamline of BESSY-HZB, Berlin. The spot size from which the spectrum was acquired, was approximately 60 µm2 (12 × 5 µm) but was adapted to the size of each phytolith to avoid contributions by Mie scattering and maximize the signal-to-noise ratio. We collected 35 spectra of bilobate silica phytoliths and 36 long cell phytoliths. Prickle phytoliths led to strong scattering contribution to the absorbance spectra due to their morphology, and thus their spectra were excluded from the analysis. Preprocessing of the spectra included selection of the spectral range of interest, interpolation of the data, baseline correction with asymmetric least square smoothing (AsLS), and vector normalization. Extended multiplicative signal correction (EMSC) was applied to the data to correct baseline variations, noise, and scattering effects that were caused by the micron range size of the samples. The window size and polynomial order of the fitting curve for the Savitzky-Golay (SG) numerical algorithm and EMSC were optimized following a procedure previously evaluated and described (Zimmermann and Kohler, 2013). We removed nine spectra outliers during the EMSC analysis.

## Nuclear Magnetic Resonance

Solid state nuclear magnetic resonance (SSNMR) measurements were performed under magic angle spinning (MAS). Approximately 40 mg of extracted phytoliths were placed in the NMR rotor and the samples were spun at 10 kHz in all experiments. Spectra of 29Si Direct polarization (DP) MAS SSNMR and cross polarization (CP) MAS SSNMR were acquired at room temperature on a Bruker 11.7T Avance ІІІ spectrometer equipped with a 4-mm VTN CPMAS probe employing 1 H decoupling at a field of 85.7 kHz. The 1 H-29Si cross polarization spectra were recorded using a CP contact time of 6 ms, recycle delay of 6 s and 2048 scans. The 29Si direct polarization spectra were taken with a 3 μs 90° pulse followed by acquisition of 2,048 points with 8 μs dwell time a recycle delay of 60 s and 137 scans. Time domain signals (2,048 points) were zero filled to 4,096 points and multiplied by exponential decaying function (with line broadening of 100 Hz) and then Fourier transformed, phase adjusted and baseline corrected using automatic 5th order polynomial function. Line deconvolutions in all 29Si NMR spectra shown were performed using the DMFIT program which minimizes the line shape generated by a set of simulated lines to the line shape of the convoluted spectrum (ref to https:// doi.org/10.1002/mrc.984). The Q4 line in **Figure 3** was best fit by adding three more Q4 peaks aside from the main Q4 signal at −111.4 ppm (see **Table S1**). These peaks represent Q4 species with minor populations having slightly different local environments resulting from etching of the silica surface by the harsh acidic treatment. These Q4 species contribute less than 2% to the total intensity and therefore were neglected in the Q4/ Q3+Q2 calculation. I.e. only the Q4 specie at −111.4 ppm was taken in calculating this ratio. The program assigns each line four parameters (position, amplitude, width, and Gaussian-to-Lorentzian ratio) which were varied until a minimum in the calculated least square function comparing the two line shapes

was found. It generated a standard deviation value as a score for the goodness of fit. It also calculated the intensity percentage that each line takes, out of 100% intensity of the spectrum based on the other peak parameters. An example for the fitting parameters of the 29Si CP spectrum of SONE is given in the supplementary information, **Table S1**.

## Thermogravimetric Analysis

Thermogravimetric analysis (TGA) of the phytolith samples were performed with a Bargal Q500 instrument (Bargal Analytical Instruments Ltd, Israel) following Tishler et al. (2015). Approximately 5 mg of phytoliths were placed in a platinum crucible, equilibrated at 25°C and the weight variation recorded in the range of 30°C to 900°C under nitrogen flow of 60 ml per min, using the high-resolution sensitivity mode and a ramp of 30°C per min. Data were processed using the Universal Analysis 2000 software from TA instruments (Waters).

#### Scanning Electron Microscopy - Energy Dispersive X-Ray Analysis

Leaf samples were imaged by a JCM-6000PLUS NeoScope scanning electron microscope (SEM, JEOL, Japan) at the backscattered electrons mode, under accelerating voltage of 15 kV using the low vacuum mode. Si elemental maps were obtained by energy-dispersive X-ray (EDX) with a dwell time of 2 ms, high probe current, and gain 1. Extracted phytoliths were imaged by a FEI/Philips XL-30 field emission with accelerating voltage 15 kV. Samples were mounted on a carbon tape and coated by a gold layer of 5 or 10 nm.

## RESULTS

#### Extraction Methods Affect the Structure and Chemistry of the Biosilica

Several types of phytoliths can be found in sorghum leaf epidermis (**Figure 1**), including bilobate silica cells, silicified long cells, prickles, and cross cells, similarly to other grasses (Prychid et al., 2003). We compared plant biogenic silica isolated by two very common extraction methods: (1) sulphuric acid-hydrogen peroxide-nitric acid extraction (SONE), and (2) microwavedassisted digestion (MAD). Both ways resulted in a similar assemblage of phytoliths, governed by long cells, bilobate silica cells and prickles (**Figures 2A**–**C**). Low magnification scanning electron microscopy (SEM) revealed no variation between the extraction methods. Higher magnifications of phytoliths extracted by MAD (**Figures 2D**–**G**) and SONE (**Figures 2H**–**K**) revealed spherical loosely aggregated particles in long cells only when extracted by SONE (**Figure 2J**). This finding suggested that the SONE damaged the structure of the silicon and the occluded organic matter.

#### Magic Angle Spinning - Solid State Nuclear Magnetic Resonance

Direct 29Si polarization (DP) spectra detected silicon atoms attached to oxygen atoms that were coordinated either to

FIGURE 1 | Back-scattered scanning electron micrographs (SEM) of sorghum leaves demonstrating typical silica deposition. Epidermal surface showing a row of bilobate silica cells (red arrows), cross cells (black arrows) randomly distributed between epidermal long cells, and prickles (blue arrows) (A), and Si EDX map (B). The white contrast in panel (A) matches the Si map in panel (B), showing heavily silicified bilobate and cross cells, in contrast to the prickles where silica accumulates at the tips. (C) Leaf cross section. (D) Close-up of the dashed rectangle in panel (C) showing a bilobate cell cut transversally (arrow), and (E) Si EDX map of the dashed rectangle in C. (F) Overlay of panels (D) and (E) localizing silica to the cell walls of epidermis cells and the volume of the bilobate cell.

FIGURE 2 | Scanning electron micrographs (SEM) of sorghum phytoliths extracted by sulphuric acid-hydrogen peroxide-nitric acid extraction (SONE) or microwavedassisted digestion (MAD). Under low magnification (panels A–C), we did not identify differences between the extractions. (A) Long cells creating a silica skeleton imaged without gold coating, scale bar 50 µm. (B) Lateral view of uncoated bilobate silica cell showing asymmetric shape, scale bar 5 µm. (C) Lateral view of a prickle, scale bar 10 µm. High magnification scans of phytoliths extracted by MAD, showing tightly packed silica in a prickle (D), bilobate (E), long (F), and cross cell (G). High magnification scans of phytoliths extracted by SONE, showing a prickle (H), bilobate (I), long (J), and cross cell (K). The scale bars in panels D–K are 2 µm.

another silicon atom, or to hydrogen that formed a terminal hydroxyl. We did not identify silicon covalently bound to atoms other than oxygen. Species of O3-Si(OH) (termed Q3) at a chemical shift of −101.6 ppm, and O4-Si (Q4) at −111.3 ppm were detected in phytoliths from both extraction methods. Q2 species (O2-Si(OH)2), shifted to −91.8 ppm, were found only in the MAD (**Figures 3A**, **B**). The bulk (Q4) to surface (Q3+Q2) ratio was 2.9 in MAD and 4.8 in SONE samples. Selective excitation of surface Si by measuring an 1H-29Si cross polarization (CP) spectrum showed that in the SONE the Si surface species intensity ratios Q2:Q3:Q4 is 2.2:46.6:51.2. The MAD phytoliths showed the typical Si surface species intensity ratios of 6:55:39 for Q2:Q3:Q4. The siloxane to silanol ratio on the surface, calculated as Q4 to Q3+Q2, was 1.05 for SONE (**Figures 3A**, **B**) and was 0.64 for MAD (**Figures 3C**, **D**). The higher ratio in the SONE indicated a more hydrophobic surface than the surface of the MAD phytoliths.

We examined the Q2, Q3, and Q4 line intensities in the DP and CP 29Si spectra and compared them to values reported before for plants i.e. equisetum (Bertermann and Tacke, 2014), rice (Park et al., 2006), and diatom cell walls, called frustules (Bertermann et al., 2003; Tesson et al., 2008; La Vars et al., 2013). The bulk/surface ratio in silica from the phytoliths extracted by the MAD was similar to the ratio in native and acid extracted silica hairs of *Phalaris canariensis* (Mann et al., 1983) and extracted *C. fusiform* frustules (Bertermann et al., 2003). The bulk/surface ratio in silica from the phytoliths

FIGURE 3 | Magic angle spinning - solid state nuclear magnetic resonance (MAS-SSNMR) of 29Si atoms in sulphuric acid-hydrogen peroxide-nitric acid extraction (SONE) and microwaved-assisted digestion (MAD) sorghum leaf silica. Measurements of 29Si direct polarization spectrum (blue) (A), and 1H-29Si cross polarization spectrum (blue) (B) of SONE silica. Optimal fit was achieved by adding minor Q4 peaks. See the fitting parameters in Table S1. 29Si direct polarization spectrum (blue) (C), and 1H-29Si cross polarization spectrum (blue) (D) of MAD silica. Spectral decomposition into 3 lines, Q2 (green), Q3 (purple), and Q4 (cyan) is shown with the total simulated spectrum (red). Q2 corresponds to a Si atom bound to 2 hydroxyl groups, Q3 to 1 hydroxyl group, and Q4 to Si surrounded by oxygen bridging atoms with no hydroxyl groups.

extracted by SONE was higher than any reported values for biosilica.

The relative intensities of the Qn lines in the CP spectrum are dependent on parameters of the experiment and sample properties. For example, the CP spectra measured on phytolith silica using a CP contact time of 6 ms and a recycle delay of 6 s are roughly comparable to diatom CP spectra measured with a CP contact time of 5 ms and a recycle delay of 4 s (Bertermann et al., 2003). The Q4/(Q3+Q2) ratio in CP of phytolith silica is, therefore, crudely compared to the ratios in other reported biosilica samples. This ratio in silica extracted by the MAD is similar to the ratio in *E. giganteum* (Bertermann and Tacke, 2014) and dried extracted frustules of several diatoms such as *C. fusiformis* (Bertermann et al., 2003) and *T. pseudonana* (Tesson et al., 2008). In silica extracted by SONE, this ratio is similar to the value reported for dried *C. muelleri* diatom grown in high salt concentrations (La Vars et al., 2013). The seminal early report by Perry does not contain details on cross polarization times (Perry, 1985).

#### Thermogravimetric Analysis

After pyrolysis, the SONE sample lost about 17% of its weight, while the MAD sample lost only about 12% (**Figure 4**). The weight loss is assumed to be composed mainly of bound water (up to 150°C) and organic matter (150°C – 800°C). We identified a peak in the weight loss rate at 120°C (**Figure 4**), associated to bound water, and representing 4.9% of the weight of MAD and 3.8% weight of the SONE sample. The TGA is consistent with our NMR analysis, showing a more hydrophilic character of the silica extracted by MAD. Differential thermal gravimetric (DTG) broad peaks at 250°C, 380°C, 450°C, and 700°C appear only in the SONE sample (**Figure 4**). The lack of peaks in the DTG of the MAD phytoliths indicates that much less organic matter remained after this extraction. The continuous weight loss between 150°C to 800°C in both extraction

FIGURE 4 | Thermogravimetric analyses of extracted phytoliths. Curves of percent weight loss (full line, left Y-axis) and derivative of weight loss by temperature (dashed line, right Y-axis) of the microwaved-assisted digestion (MAD) (red) and sulphuric acid-hydrogen peroxide-nitric acid extraction (SONE) (black) samples.

methods can be attributed to removal of chemically bound water (OH in the surface of silica powders), (Mueller et al., 2003), changing the surface chemistry from silanol to siloxane groups.

## Microspectroscopic Characterization of Individual Phytoliths

#### Raman Analysis

We measured Raman spectra of prickle, long, and bilobate phytolith cells (**Figures 5A**, **B**). Due to the noncrystalline and nonuniform molecular structure of the silica, the Raman bands were broad. The signal extending from 400 to 500 cm-1 with a maximum at 478 cm-1 (Si-O-Si bending modes) was assigned to five-, six-, and seven-membered SiO ring (Sharma et al., 1981). Other characteristic Raman silica bands appeared at 808 cm-1 (Si-O-Si symmetric stretching), 970 cm-1 (Si-OH stretching mode of nonbridging oxygen atoms) and 1,070 cm-1 (Si-O-Si asymmetric bond stretching), (Bertoluzza et al., 1982). To estimate the hydroxyl density in the different phytoliths we calculated the intensity ratio of the band at 970 cm-1 to that at 808 cm-1 after AsLS baseline correction (**Figures 5C**, **D**). The latter band was used for normalization because it is a lattice band characteristic to the silica network and remains unchanged in different silicas (Humbert, 1995). The ratios calculated for bilobate cells were significantly higher than those ratios calculated for both prickles and long cells under MAD and SONE (p < 0.05, T-test). No significant differences were detected between long cells and prickles. Our results suggest a larger surface to volume ratio and a lower degree of condensation of the silica in the bilobate cells in comparison to that in prickles and long cells.

All other bands in the spectrum were attributed to organic matter occluded within the silica: C-C twisting and rocking at 1,153 cm-1, CH2 deformation in alkane long chains at 1,298 cm1 , and CH2 deformation vibrations in n-alkanes at 1,440 cm-1 (Parker, 1983). A small band at 1,613 cm-1 was attributed to C = C stretching or aryl stretching vibrations was also identified. We associated it with the presence of modified lignin (Ram et al., 2003). Prickle cells presented two unique features: a band at 1,665 cm-1 that was assigned to the C = C stretching, C = O stretching, and amide I vibrations, and the absence of a band at 1,043 cm-1, which was assigned to ring vibrations of substituted benzenes and C-C stretches in n-alkanes (Parker, 1983), (**Figure 5B**). Raman spectra of prickles were the only place we could identify contributions that are typical to proteins, in peaks associated to amide I (1,600–1,690 cm-1) and amide II (1,480–1,580 cm-1), (Tuma, 2005).

Discrimination between the two extraction methods was achieved by PCA of the Raman spectra. The separation was particularly clear when the PCA was applied to the spectra of long cells (**Figure S1A**). In this case, the loading spectra that indicate the source of the variation, revealed differences in the silica structure and the amount of occluded organic matter (**Figure S1B**). Based on PCA, a separation between different phytolith types was possible regardless of the extraction method (**Figure 6A**). A clear separation between the bilobate and long cells was achieved when we analysed only the SONE phytoliths spectra (**Figure 6C**). The source of separation was

silica cells (S, red), long cells or plates (L, black) and prickles (P, blue) are shown in bright-field micrographs (A). Mean spectra ± standard deviation are plotted in the same respective colour and denoted with the same abbreviations (B). Averages of 25 spectra of phytoliths of each type extracted by microwaved-assisted digestion (MAD) are shown. The area of the peak at 970 cm-1, assigned to Si-OH surface groups, was normalized to the area of the 808 cm-1 band, assigned to Si-O-Si stretching. Ratios of band areas calculated in spectra of bilobate cells were significantly higher (p < 0.05) than both prickles and long cells under MAD (C) and sulphuric acid-hydrogen peroxide-nitric acid extraction (SONE) (D) methods.

studied based on the PCA loadings (**Figures 6B**, **D**). In both the full dataset as well as the SONE dataset the highest variation, which is represented by PC1, is explained by an increase in the bands at 475, 808 and 970 cm-1 and a decrease in the band at 1,435 cm-1. In the scores plot, long cells and prickles appeared at negative values of PC1, indicating a higher contribution of the 1,435 cm1 CH2 deformation band, associated with lipids (Parker, 1983). The other bands that contribute to the variance represented by PC1 corresponded to vibrational modes of Si-O-Si and Si-OH.

PCA was also applied to the derivatives of the Raman spectra (**Figure S2**). We found high variation within the group of the bilobate cells in comparison to the long cells and prickles that formed a compact distribution in the scores plot. The discrimination was based on differences in the shape of bands between 440 and 500 cm-1, indicating differences in the structure of the silica.

#### Synchrotron Infrared Microspectroscopy

We further characterized the long and bilobate cells extracted by SONE by FTIR microspectroscopy (**Figure 7**). The main spectral features were attributed to the silica: the band at 800 cm-1 are assigned to the deformation of Si-O-Si bonds bridging between two adjacent tetrahedral (Kirk, 1988), and the bands at 1,000– 1,250 cm1 are assigned to Si-O asymmetric stretching modes. The latter band has a maximum at 1,093 cm-1 in the phytoliths of long cells, and at 1,020 cm-1 in bilobate cells (**Figure 7A**). This variation indicates differences in the silica structure between the phytolith types. PCA analysis supported this observation (**Figure S3**), resulting in clear separation of the two cell types. Infrared bands in the 2,700–3,100 cm-1 region suggested that a considerable amount of organic matter remained linked to the extracted silica. The spectra of the long cells display bands at 2,854, 2,866, 2,925, and 2,959 cm-1 (**Figure 7B**), which are attributed to C-H stretching in -CH3 and CH2 groups (Silverstein

et al., 2005). These bands are expected in biological materials due to the presence of terminal -CH3 and of CH2 groups in cellular components like proteins, carbohydrates, and lipids. PCA of the spectra in the 2,700–3,100 cm-1 range separated the long cells and the bilobate cells along PC1 (**Figure 7C**). The loading of PC1 represents absorption bands of organic matter (**Figure 7D**). However, these spectral features are represented in negative values. Therefore, the negative values for PC1 coefficients, at which the long cell phytoliths spectra are found (**Figure 7C**), lead to the conclusion that more organic material must be occluded in the long cell phytoliths as compared to the bilobate cells.

(D) Corresponding loading spectra of PC1 and PC4.

## DISCUSSION

In this work, we aimed to discriminate between phytolith types extracted from the same sorghum leaf. Our hypothesis was that the phytoliths that are produced by varied cell types will vary in their occluded organic matter. We also assumed that the harsh extraction conditions alter and mask genuine variations between phytolith types (Alexandre et al., 2015). Our NMR and Raman data indicated that the extraction changes the silica structure. The number of silanol groups on the silica surface was lower in phytoliths extracted by SONE in comparison with the MAD (**Figure 3** and **Figure S1**), making SONE silica less polar and more hydrophobic. TGA supported this by showing lower percentage of water molecules released below 150°C in the SONE sample (**Figure 4**). Our TGA measurements further showed that the SONE was less aggressive than the MAD, which left hardly any organic matter in the phytoliths. SEM indicated that the long cells behaved differently under the two extraction methods, in accordance with the Raman PCA that could discriminate between the extractions based on long cells spectra. Our results clearly show that long cells react differently to the extractions. More research is needed to elucidate the native state of the silica and its occluded organic matter as synthesized in the plant.

Using single particle spectroscopy, we could show that under the same extraction, the silica structure is different between phytolith types. In general, IR vibrational spectra of microscale particles are masked by Mie scattering that depends on the particles' shapes. Even so, the spectra of bilobate phytoliths show a prominent shift to lower energies in the Si-O asymmetric stretching vibration as compared to long cells. This difference indicates variation in the atomic organization of the mineral. Our results thus conform with the hypothesis that silica organization is under biological control, as was suggested by Perry et al, showing that variation in the

extraction (SONE). (A) Representative spectra of long and bilobate cells. Yellow shade at 2,700–3,000 cm1 showing bands typical to hydrocarbons. (B) Average FTIR spectra ± standard error of long cell phytoliths in the range 2,700 to 3,100 cm-1. (C) Scores plot of a PCA at the spectral region 2700–3000 cm1, attributed to the organic matter occluded in long (L) and silica (S) cells, discriminating between the two phytolith types. (D) Loadings of the principal component analysis (PCA) correlate the discrimination with the terminal -CH3 and CH2 groups absorption bands.

mineral nanostructure in correlation to cell developmental stages (Perry and Mann, 1989) correlates to the silanol groups exposed on the silica surface (Perry et al., 1990). In agreement with FTIR, the PCA of the Raman spectra resulted in the formation of two groups: one includes spectra of bilobate silica cells and another of prickles and long cells. The Raman spectra indicated a larger ratio of surface to bulk Si atoms in the bilobate cells in comparison to prickles and long cells (**Figures 5C**, **D**). These differences may arise from higher number of silica nucleation sites in bilobate cells in comparison to long cells and prickles. Biogenic moieties that integrate to the bulk mineral or attach to its surface may also alter the mineral structure. The variation in the mineral structure was persistent within a phytolith type, suggesting that within the same cell type, similar plant factors interact with the mineral, and these materials may differ between cell types—specifically between bilobate and long cells.

From our results, it is not possible to determine the hydroxylation degree of the native biosilica before its extraction. However, the variations between phytolith types extracted similarly indicate either an initially distinct variation in hydroxylation or structure of the silica of different phytolith types. Regardless of the actual origin of the variation in hydroxylation degree, it most probably indicates that there is more than one pathway of silica deposition in sorghum leaves.

The SONE allowed us to analyse organic matter that was intimately associated with the silica (**Figure 4** and **Figure S1**). Our results indicated that the Si atoms are coordinated to oxygen, similarly to silica gel and opal, in agreement with analyses of *in planta* silica (Yoshida et al., 1959; Casey et al., 2004) and *in vitro* precipitation with lignin (Cabrera et al., 2016; Soukup et al., 2019). We cannot exclude the existence of Si-O-C bonds as detected by X-ray photoelectron spectroscopy in cell walls extracted from rice cell suspension (He et al., 2015). These bonds may be below the detection limit because obviously they are not abundant, and their vibrations are expected at very similar energies to Si-O-Si vibrations. In addition, they may wash out or decompose during extraction.

Si in cell walls of *Equisetum arvense* is associated with cell wall polymers, including polysaccharides, proteins, and phenolic acids, suggesting that silica may form in a range of chemical conditions independent of a charged matrix (Currie and Perry, 2009). Raman and Infrared bands associated to lipids were more intense in the spectra of long cell and prickle phytoliths, suggesting that the cuticle incorporated into the mineral (**Figure 7**). This is in agreement with the existence of a cuticle-silica double layer, observed first in the epidermis of rice by Yoshida et al. (1962). Cell wall polymers (possibly polysaccharides) are involved in the deposition of silica in hairs and epidermis, similarly to hairs and outer epidermis cells in lemmas of the grass *Phalaris canariensis*

(Hodson et al., 1984; Perry et al., 1987). In comparison to long cells and hairs, we found that the mineral in bilobate cells contained lower fraction of organic residues. In sorghum bilobate cells silica deposits between the cell membrane and wall, constricting the protoplast and creating a secondary wall made of silica (Kumar and Elbaum, 2018). Thus, the bilobate silica deposition pathway excludes cuticle materials and includes only small amounts of cell wall polymers in the mineral.

Acidic proteins and glycoproteins are found in association with mineral phases as components of the organic matrix encapsulated in phytoliths (Harrison, 1996; Elbaum et al., 2009). Specifically in bilobate, protein residues were identified embedded in their silica (Alexandre et al., 2015). A protein (Siliplant1) was identified inside sorghum bilobate cells that is active in *in planta* silica deposition (Kumar et al., 2019). Nonetheless, our results did not provide direct evidence of amino acids in bilobate cells, possibly because they degraded during the phytolith extraction. We suggest that other organic compounds such lipids and carbohydrates are much more abundant than proteins in the extracted sorghum phytoliths. The presence of more organic matter entangled within the silica of long cells and prickles in comparison to bilobate phytoliths may be explained by a slow co-deposition of silica and other cell wall components like lignin, cutin, hemicelluloses, and cellulose (Perry et al., 1987; Fry et al., 2008; Law and Exley, 2011; Soukup et al., 2017; Kulich et al., 2018). The observed differences in hydroxylation and amount of occluded organic matter between phytolith types are also expected to have an effect on the dissolution rate of phytoliths (Nguyen et al., 2019).

## CONCLUSIONS

Due to the strong influence of the method used to extract the phytoliths on the silica structure and occluded organic matter, it is important to study plant silicification *in situ* in the native tissues. Differences between phytolith types extracted similarly from the same leaf suggest that the mineral deposits through a cell type-dependent pathway. Two mechanisms are suggested by our data: one involves the mineral impregnation of a cuticlecellulose matrix (in long cells and prickles) and another suggests a low fraction of organic matrix (in bilobate silica cells) on which silica deposits.

## REFERENCES


## DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

## AUTHOR CONTRIBUTIONS

VZ, JK, and RE planned the research, designed the study, and wrote the manuscript. VZ collected the TGA data, Raman, and IR spectra, and SEM images; NF and GG performed the NMR experiments. VZ and SD analyzed the data. All authors commented, added, and revised the manuscript and approved for publication.

## FUNDING

This work was funded in part by the Excellence Initiative of the German Research Foundation (DFG) GSC 1013 (SALSA) and the Israel Ministry of Agriculture grant 12-01-0031. We thank BESSY-HZB for the allocation of synchrotron radiation beam time.

## ACKNOWLEDGMENTS

We thank V. Rosen (HUJI) for his help extracting the phytoliths by microwave digestion, M. Soukup for discussion on silicification mechanism, I. Feldmann for help with SEM images of phytoliths, I. Gardi (HUJI) for his help with the TGA analysis, L. Puskar (HZB, Berlin) for support at the BESSY IRIS beamline, and B. Zimmermann and A. Kohler (NMBU, Aas, Norway) for the EMSC scripts and discussion of IR spectra. We acknowledge the support by the German Research Foundation (DFG) and the Open Access Publication Fund of Humboldt-Universität zu Berlin.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01571/ full#supplementary-material

of 14C depleted carbon fraction and implications for radiocarbon dating. *J. Archaeol. Sci.* 78, 57–65. doi: 10.1016/j.jas.2016.11.005


vegetation and climate change. *Quat. Res.* 35, 222–233. doi: 10.1016/ 0033-5894(91)90069-H


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor is currently organizing a Research Topic with one of the authors RE, and confirms the absence of any other collaboration.

*Copyright © 2019 Zancajo, Diehn, Filiba, Goobes, Kneipp and Elbaum. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Phytoliths in Inflorescence Bracts: Preliminary Results of an Investigation on Common Panicoideae Plants in China

Yong Ge1,2\*, Houyuan Lu3,4,5\*, Jianping Zhang3,4, Can Wang6 and Xing Gao1,2,5

<sup>1</sup> Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China, <sup>2</sup> Center for Excellence in Life and Paleoenvironment, Chinese Academy of Sciences, Beijing, China, <sup>3</sup> Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China, <sup>4</sup> Center for Excellence in Tibetan Plateau Earth Science, Chinese Academy of Sciences, Beijing, China, <sup>5</sup> University of Chinese Academy of Sciences, Beijing, China, <sup>6</sup> Department of Archaeology, School of History and Culture, Shandong University, Jinan, China

#### Edited by:

Terry B. Ball, Brigham Young University, United States

#### Reviewed by:

Monica Tromp, University of Otago, New Zealand Luc Vrydaghs, Université libre de Bruxelles, Belgium

#### \*Correspondence:

Yong Ge geyong@ivpp.ac.cn Houyuan Lu houyuanlu@mail.iggcas.ac.cn

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 19 December 2018 Accepted: 10 December 2019 Published: 20 February 2020

#### Citation:

Ge Y, Lu H, Zhang J, Wang C and Gao X (2020) Phytoliths in Inflorescence Bracts: Preliminary Results of an Investigation on Common Panicoideae Plants in China. Front. Plant Sci. 10:1736. doi: 10.3389/fpls.2019.01736 Phytoliths in the inflorescence of Poaceae plants can be of high taxonomic value in some archaeological contexts and provide insight into plant taxonomy and crop domestication processes. In this study, phytoliths in every inflorescence bract of 38 common Panicoideae weeds and minor crops in China were studied. Based on dissection of the inflorescence into different bracts using a treatment that retained the phytoliths anatomical position, observations of inflorescence phytoliths types and distribution were described in detail. We found that INTERDIGITATING, Blocky amoeboid, Rectangular dentate, and Elongate dendritic with multi tent-like arch tops were of higher taxonomic value than the other types in our studied species. Both morphological and morphometric traits of the INTERDIGITATING were summarized and compared with previous studies; the findings suggested that genus level discrimination of some Paniceae species could be reliable, and tribe/species level discrimination might be feasible. The phytoliths in the involucre of domesticated and wild type Coix lacryma-jobi provided insight into the domestication process of this plant. Our data also indicated that phytolith production in the inflorescence bracts might be under the genetic and molecular control of inflorescence development. Thus, the findings of this study could assist future studies in plant taxonomy and archaeobotany.

Keywords: phytolith morphology, Poaceae taxonomy, inflorescence phytolith, seed protection strategy, archaeobotanical implication

#### INTRODUCTION

Phytoliths are plant-produced micro silica bodies which can, in some cases, have a diagnostic morphology that can be distinguished among taxa; in particular, phytoliths in the Poaceae family can have taxonomic value in archaeological contexts and natural sediments (Twiss et al., 1969; Wang and Lu, 1993; Piperno, 2006). Based on phytolith taxonomy and morphology, phytolith analysis has been proven to be a reliable tool in understanding the taxonomy and evolution of plants

**70**

(Prychid et al., 2003; Rudall et al., 2014; Dinda and Mondal, 2018), paleoecology (Blinnikov et al., 2002; Stromberg, 2005; Gu et al., 2008; Stromberg et al., 2013; Dunn et al., 2015), and paleoclimate (Prebble and Shulmeister, 2002; Lu et al., 2007; Zuo et al., 2016; Liu H. et al., 2018); in recent years, it has been extensively employed in investigating the origin, development, and spread of agriculture (Lu et al., 2009a; Piperno et al., 2009; Madella et al., 2014; Ball et al., 2016a; Hilbert et al., 2017; Deng et al., 2018; He et al., 2018). However, compared to studies on the leaf phytoliths of Poaceae plants, inflorescence phytoliths have not been extensively studied and have generally focused on crop species and their relatives (Ball et al., 2016a). Thus, a broad and systematic investigation of inflorescence phytoliths would provide an important advancement and enable multidisciplinary application of phytolith analysis.

The study of inflorescence phytoliths has a long history; as early as 1908, Schellenberg studied inflorescence phytoliths in archaeological contexts (Schellenberg, 1908), and in 1966, Parry and Smithson studied the inflorescence phytoliths of grasses and cereals in Britain using an acid treatment to extract the phytoliths and observing them under a light microscope (Parry and Smithson, 1966). Thereafter, scanning electron microscopy was introduced to study inflorescence phytoliths (Terrell and Wergin, 1981; Sangster et al., 1983; Rosen and Weiner, 1994). With the introduction of phytolith analysis among archaeologists, Poaceae inflorescence phytoliths were valued due to their relationship with human food gathering activities, and the inflorescence phytoliths of both major (Hodson and Sangster, 1988;Piperno and Pearsall, 1993;Tubb et al., 1993; Pearsall et al., 1995; Zhao et al., 1998; Ball et al., 1999; Rosen, 1999; Ball et al., 2001; Pearsall et al., 2003; Rosen, 2004; Hodson et al., 2008;Madella et al., 2014; Ball et al., 2017) and minor (Lu et al., 2009b; Radomski and Neumann, 2011; Zhang et al., 2011; Madella et al., 2013; Kealhofer et al., 2015; Novello and Barboni, 2015; Weisskopf and Lee, 2016; Ge et al., 2018; Zhang et al., 2018; Duncan et al., 2019) crops were extensively studied to investigate crop domestication. Nevertheless, many other wild relatives and minor crops have not been studied, which not only results in identification uncertainty, but also has stalled further application of phytolith analysis in the investigation of early plant resource exploitation.

Panicoideae plants such as foxtail millet (Setaria italica), common millet (Panicum miliaceum), and barnyard millet (Echinocloa sp.) are widely recognized as minor crops that could have been important plant resources in ancient times (Bellwood, 2004; Fuller, 2006; Crawford, 2017). These millets have been cultivated and harvested in many countries as food crops, especially in Asia and Africa (Anderson and Martin, 1949). Compared with the long domestication history of major crops that extends back to the early Holocene, the domestication or utilization history of other useful species is short or unclear (Zohary et al., 2012), and might be partially due to a lack of evidence. The development of new methods and proxies for the identification criteria of crop phytoliths has revealed the early domestication process of many species (Denham et al., 2003; Piperno and Stothert, 2003; Ezell et al., 2006; Horrocks and Rechtman, 2009; Piperno et al., 2009; Yang et al., 2013; Yang et al., 2015), and implies the possibility of using a similar method to investigate the early exploitation of Panicoideae species. Phytoliths are more stable under various preservation conditions and are generally abundant (Wang and Lu, 1993; Piperno, 2006). Moreover, inflorescences are the part of the plant generally collected for harvest; the occurrence of inflorescence phytoliths could reflect these activities. Thus, investigating inflorescence phytoliths in Panicoideae species could improve our understanding of the early process of plant resource exploitation.

In this study, we examined inflorescence phytoliths in every bract of a single specimen of the 38 most common Panicoideae species in China to provide a preliminary detailed phytolith morphology dataset. Further, we investigated the morphometric differences in a phytolith morphotype that we propose naming INTERDIGITATING (see the results for a detailed description) on the lemma and palea of Digitaria, Oplismenus, and Paspalum genera to determine to what level (at genus, section, or species) the morphological traits might be robust. We also report on novel phytolith types that may be of high taxonomic value. This study provides insight into inflorescence phytoliths and reinforces the importance of treatment that preserves the anatomical position of phytolith in different bracts. Our detailed description of phytoliths in every bract of the inflorescence could provide the baseline information for further archaeological and taxonomical studies.

## MATERIALS AND METHODS

#### Sample Collection and Pretreatment

A total of 38 species (one specimen per species) (Table 1) were collected to investigate the morphological differences of phytoliths in the inflorescence of common Panicoideae plants in China. These species included the most common weeds and several minor crops from across China. They were collected during several field trips over decades led by colleagues from the Institute of Geology and Geophysics, the Chinese Academy of Sciences, and China Agricultural University, and identified by colleagues from the Institute of Botany, Chinese Academy of Sciences.

Mature spikelets from the inflorescence of collected samples (more than three entire spikelets from the same specimen) were dissected into different parts according to plant anatomy and included the following five parts: (1) involucre, (2) glume, (3) lemma, (4) palea, and (5) seed. All samples were divided into four groups (Figure 1) according to the dissection results: type I, with a thin and soft involucre covering other bracts and the seed; type II, with a thick and hard involucre or glume covering other bracts and the seed; type III, with thin and soft bracts; and type IV, with thicker and harder lemma and palea covering the seed compared to those of type III. After dissection under a microscope, every part (except the seed) was ultrasonically cleaned and dried for further treatment.

## Phytolith Preparation for In Situ Analysis

Whereas traditional wet oxidation methods for phytolith extraction (Piperno, 1988) can easily break down and disarticulate phytoliths (Jenkins, 2009), we prepared our samples Ge et al. Inflorescence Phytolith in Panicoideae Plants



(Continued)

#### TABLE 1 | Continued


Spikelet types correspond to Figure 1.

1

2 Details of phytoliths types in the inflorescence bracts were described in the supplementary file and shown in the Supplementary Figures 1–9.

following the published methods of Lu et al., 2009b, with minor modifications, to ensure that the phytoliths in the whole bract structures remained articulated and undamaged. For the bracts that were thick and hard (e.g. involucre from type II, or lemma and palea from type IV), a saturated nitric acid (HNO3) treatment was used. A total of 5–10 ml HNO3 was added to each bract (to merge the bracts) in a 15 ml centrifuge tube, then the sample was placed in a water bath at 50–60 ºC. When the bract turned transparent, all contents in the tube were poured into a glass dish, and the bract was carefully moved onto a slide. Distilled water was used to wash the bract on the slide to remove the HNO3, and absolute ethanol was used to wash the bract to remove the water. After the bract was dry, a drop of xylol was added to it. Before the xylol was totally volatilized, a drop of Canada Balsam was added, and the bract was covered with a cover glass. All procedures were performed in a fume cupboard. Similar procedures were followed for bracts that were thin and soft, except saturated nitric acid was replaced with hydrogen peroxide (H2O2). This method increased the chances of keeping the whole bract structure undamaged and allowed for articulated or in situ observation of the phytoliths in the bracts. At least two replicates were prepared for each bract.

#### INTERDIGITATING Phytolith Measurement

The phytolith morphotype that we name INTERDIGITATING has been reported to be a useful tool in discriminating samples at the genus level for some taxa (Lu et al., 2009b; Madella et al., 2013; Weisskopf and Lee, 2016; Ge et al., 2018), however, discrimination at species level requires the assistance of

the glumes (left and right ones), c the lemma (left) and palea (right) of fertile floret, d the seed.

morphometric analysis (Zhang et al., 2011; Zhang et al., 2018). The INTERDIGITATING phytoliths from Digitaria, Paspalum, and Oplismenus genera were employed to aid in morphometric discrimination from each other in our samples. The measurement parameters are shown in Figure 2A, and are described as follows: h-total is the width of the whole undulation pattern; h-undulation is the mean value of the two individual undulation parts; h-body is the difference between htotal and h-undulation; w is the length of the protuberant ends, and was measured along one direction (either upward or downward) in the same sample; and L is the total length of the undulation patterns. Two additional parameters used were: (1) R (w/hu) = w/h-undulation; (2) R (hu/hb) = h-undulation/h-body. All parameters were measured in 150 individuals [the number has been tested by the suggested formula (Ball et al., 2016b)] of each species and were measured in different areas. Fifty measurements were taken near the base area, 50 from the center area, and 50 from the top area. Data parameters are shown in Table 2. All observations and measurements of inflorescence phytolith parameters were conducted under a Leica DM 750 microscope with 400× magnification. Statistical analysis (conical discriminate analysis) was performed using IBM SPSS Statistics 24 software.

#### RESULTS

In general, phytoliths were abundant in the inflorescence bracts of the studied species. The distribution of phytolith types in the inflorescence bracts are shown in Table 3, detailed descriptions can be found in the Supplementary file, and the details of phytolith morphology in different inflorescence bracts can be found in Supplementary Figures 1–9.

## Phytolith Nomenclature and Classification

Phytolith morphology nomenclature followed ICPN 2.0 rules (ICPT, 2019), and the description and classification of phytolith morphology followed those of previous studies:


epidermal long cells, but also to the silica layer between the epidermal cuticle layer and the epidermal cells (Rosen, 1992; Madella et al., 2013; Madella et al., 2014; Weisskopf and Lee, 2016; Ge et al., 2018). Thus, in this paper to clarify the different morphology and anatomical origins, the 'Elongate dendritic' only refers to the phytoliths derived from epidermal long cells, and the "Interdigitating" only refers to the phytoliths derived from the silica layer between the epidermal cuticle layer and the epidermal cells. Different types of Elongate phytoliths are shown in Figure 4.


#### TABLE 2 | Parameters of the INTERDIGITATING phytolith in Digitaria, Paspalum, and Oplismenus.


#### TABLE 2 | Continued


SD, standard deviation.

TABLE 3 | Phytoliths types in different bracts of the studied samples.


#### TABLE 3 | Continued


#### TABLE 3 | Continued


The abbreviations used in the table refer to 1) NB, no bract; NP, no phytolith observed; 2) BIL, BILOBATE; CRO, CROSS; POL, POLYLOBATE; conv, convex ends; conc, concave ends; var1, variant 1; var2, variant 2; var5/6, variant 5/6; var7, variant 7; vars, variant saddle-like; 3) ELO, ELONGATE; DET, dentate; CYL, cylindric; ENT, entire; DEN, dendritic; tent, multi tent-like arch top; PAR, PAPILLAR; 4) INT, INTERDIGITATING; nPAP, no PAPILLATE; mPAP, PAPILLATE attached with main body; sPAP, PAPILLATE separated with main body; sun, smooth-type undulation; nun, ntype undulation; oun, W-type undulation; scon, smooth connection; acon, articulated connection; obod, ovate main body; rbod, rectangular main body; 5) RON, RONDEL with small chamber; ACU, ACUTE; ACU\_BUL, ACUTE BULBOSUS; dPAP, disaggregated Papillate; BLO\_AMO, BLOCKY AMOEBOID; sREC\_DET, RECTANGULAR DENTATE with smooth short sides; pREC\_DET, RECTANGULAR DENTATE with protuberant short sides; PRI, PRISMATIC silicified hair base.

acon rbod

cells and the epidermal cuticle layer. This type was named after Parry and Hodson's first observation of the morphotype in S. italica, in which they described what they observed on inflorescence bracts as "interdigitating epidermal cells" (Parry and Hodson, 1982); however, these could be the silica layer covering the surface of the lemma and palea. Other names that have been used by other studies to describe this type of phytolith include "silica skeleton" (Rosen, 1992; Madella et al., 2013;

acon rbod

FIGURE 3 | Illustration on the morphology of the LOBATE phytoliths.

FIGURE 4 | Illustration on the morphology of the ELONGATE phytoliths.

Madella et al., 2014; Weisskopf and Lee, 2016), "dendriform" (Lu et al., 2005), "silicified epidermal long cells"(Lu et al., 2009a; Lu et al., 2009b; Zhang et al., 2011; Kealhofer et al., 2015), and "epidermal silica layer" (Ge et al., 2018). As the anatomical origin of this type of phytolith has been discussed by Ge (Ge et al., 2018), and associated with a study on rice husk (Yoshida et al., 1962), we propose INTERDIGITATING as the formal name for this phytolith type to show its different anatomical origin and morphology. Previous studies have shown that morphological trait combinations could be helpful in discriminating INTERDIGITATING. In the present study, we followed the description of Setaria, Panicum (Lu et al., 2009b), and Echinochloa (Ge et al., 2018) and used the morphological traits of PAPILLATE (present or not), undulation patterns (ntype, smooth-type, and W-type), ending structure (smooth connection or articulated connection), and main body (ovate or rectangular) to describe INTERDIGITATING morphology. All the traits that describe INTERDIGITATING morphology and are shown in Figure 5.

5. As the treatment retained the undamaged anatomical structure, the different types of phytoliths could combine to form a pattern of taxonomic value. In this study, one pattern, a BILOBATE- ELONGATE DENDRITIC/DENTATE pattern, and a special involucre phytolith layer were observed (Figure 6). The BILOBATE-ELONGATE DENDRITIC/DENTATE pattern was comprised of BILOBATE and ELONGATE DENDRITIC/DENTATE, one BILOBATE and one ELONGATE DENDRITIC/DENTATE alternating formed the common pattern. Their long axes were parallel to the veins, sometimes BILOBATE was replaced by PAPILLATE or ACUTE or just disappeared. A similar pattern has been reported in the inflorescence of cereals, the PAPILLATE- ELONGATE DENDRITIC/ DENTATE pattern (Parry and Smithson, 1966; Rosen, 1992; Tubb et al., 1993), which includes a combination of PAPILLATE and ELONGATE DENDRITIC/DENTATE. Sometimes PAPILLATE could be replaced by the RONDEL, and the PAPILLATE in this pattern has pits (radiating marks) on the base. Thus, the BILOBATE- ELONGATE DENDRITIC/DENTATE pattern could be a potential tool to distinguish between some Panicoideae grasses and Pooideae cereals. The involucre phytolith layer was found in Coix lacryma-jobi and C. lacryma-jobi var. ma-yuen on the surface of the involucre and was comprised of tightly connected phytoliths with various morphologies (could be stretched or condensed in morphology). This phytolith layer was reported for the first time in this study and recognized as the involucre phytolith layer. The involucre phytolith layer (Figure 6) differed among species in our samples: (1) In cultivated C. lacryma-jobi var. ma-yuen, the involucre phytolith layer was comprised of different types of phytoliths, including BILOBATE (some could be condensed or stretched in morphology) and ELONGATE DENDRITIC/DENTATE, each in a different column and with their long axis parallel to the veins; (2) In wild type C. lacryma-jobi, the involucre phytolith layer was only comprised of BLOCKY AMOEBOID, this type of phytolith was cubic or oblong with granules on the surface. They were tightly connected with each other and the column was parallel to the veins.

## Phytoliths in Different Inflorescence Bracts

Involucres were found in Cymbopogon goeringii, C. lacryma-jobi var. ma-yuen, C. lacryma-jobi, Themeda caudata, and Themeda japonica. In our samples phytoliths in the involucre could be divided into two groups, corresponding to spikelet type I and type II (Figure 1). In the involucres of C. goeringii (Supplementary Figure 1-V), T. caudata (Supplementary Figure 6-I), and T. japonica (Supplementary Figure 6-II), which were spikelet type I, phytolith types were mainly BILOBATE, ELONGATE, and ACUTE (Table 3), and the phytoliths were separated from each other and presented a scattered distribution in the involucre. However, in the involucres of C. lacryma-jobi var. ma-yuen (Supplementary Figure 2-I) and C. lacryma-jobi (Supplementary Figure 2-II),

Frontiers in Plant Science | www.frontiersin.org 11

which were spikelet type II, phytoliths were all tightly connected to form an involucre phytolith layer on the surface. The phytolith types were LOBATE, ELONGATE, RECTANGULAR DENTATE, and BLOCKY AMOEBOID.

Glumes were found in all studied species. Most of the glumes were thin and soft in the studied species; however, in Sorghum bicolor and Hackelochloa granularis (both are spikelet type II), glumes were thick and hard (H. granularis had a thin but hard glume). Thus, the phytolith type in the glumes appeared to differ: in thin and soft glumes, phytoliths were LOBATE, ELONGATE, ELONGATE DENDRITIC/DENTATE, and ACUTE, and scattered from each other; while in S. bicolor (Supplementary Figure 5-III), phytoliths were LOBATE, ELONGATE, ELONGATE DENDRITIC/DENTATE, and ACUTE, and the BILOBATE- ELONGATE DENDRITIC/DENTATE pattern covered most of the area of the glume surface. In the lower glume, ELONGATE DENDRITIC with multi tent-like arch tops (Figure 4) were observed, and this type was only observed in the lower glume of S. bicolor. In H. granularis (Supplementary Figure 3-I I), phytoliths in the glumes formed an INTERDIGITATING with PAPILLATE pattern separated from the main body, n-type undulation, articulated connection, and rectangle main body; this type of phytolith covered the entire surface of the glumes, and was the only type observed in H. granularis. An exception of the soft and thin glumes was Arthraxon hispidus, which produced weakly silicified INTERDIGITATING (Supplementary Figure 2-II) and BILOBATE phytoliths could be found among the INTERDIGITATING.

Lemmas were found in most of the studied species and could be divided into lemmas of the sterile floret and lemmas of the fertile floret. Sterile lemmas were remarkably similar to the thin and soft glumes in morphology, and the phytolith types were the same. The fertile lemmas could be divided into two groups according to spikelet type (III and IV): fertile spikelet type III lemmas were similar to the glumes and sterile lemmas in morphology and the phytolith types were the same. Fertile spikelet type IV lemmas were hard and glossy, INTERDIGITATING was the major phytolith type that covered the entire surface of the lemmas.

Paleas were similar to the lemmas; however, the paleas may stop growing or be absorbed during the growth of the inflorescence and finally disappear in some of the studied species. Phytoliths in the paleas could be the same as that in the lemmas. In the sterile paleas andfertile spikelet type III paleas, phytolithswere LOBATE, ELONGATE, and ACUTE, and far fewer occurred compared to the number in the lemmas of the same floret. In the fertile spikelet type IV paleas, phytoliths formed the INTERDIGITATING TYPE, the same as those in the lemmas from the same floret.

In all the inflorescence bracts, LOBATE, ELONGATE, ELONGATE DENDRITIC/DENTATE, and ACUTE phytoliths were the most commonly observed phytolith types. The morphology of these types could overlap among different species and were of relatively low taxonomic value in distinguishing among the studied species. However, phytolith types from spikelet type II and IV (which usually produce more phytoliths than spikelet type I and III), namely the BLOCKY AMOEBOID, RECTANGULAR DENTATE, ELONGATE DENDRITIC with multi tent-like arch top, and INTERDIGITATING, as well as the combination of phytolith types such as the BILOBATE- ELONGATE DENDRITIC/DENTATE pattern and the involucre phytolith layer showed higher taxonomic value in distinguishing among our samples of the studied species. The morphological traits are summarized in Table 4.

#### Morphological and Morphometric Approaches to the INTERDIGITATING Phytolith

Among the phytolith types found in inflorescences with potentially high taxonomic value, the INTERDIGITATING had more complex morphological traits than the others; it could be observed in A. hispidus, H. granularis, Digitaria chrysoblephara, Digitaria ciliaris, Digitaria sanguinalis, Setaria faberi, Setaria pallidifusca, Setaria plicata, Setaria pumila, Oplismenus compositus, Oplismenus undulatifolius, Paspalum dilatatum, and Paspalum orbiculare. By using a combination of morphological traits of the INTERDIGITATING we could distinguishing among the taxa at the genus level for our samples of the studied specimen.

A. hispidus (Supplementary Figure 2-II) produced the INTERDIGITATING with PAPILLATE separated from the main body, W-type undulation, articulated connection and rectangular main body. BILOBATE could sometimes be found among the INTERDIGITATING. H. granularis (Supplementary Figure 3-II) produced the INTERDIGITATING with PAPILLATE separated from the main body, n-type undulation, articulated connection, and rectangle main body. These two species could be distinguished from other species by the PAPILLATE separated from the main body, this distinct feature allows quick discrimination of A. hispidus and H. granularis. As these two species belongs to the tribe Andropogoneae, and all other species belong to the tribe Paniceae; this morphological trait might have the potential to be a discriminating feature at the tribe level.

Some of the studied Digitaria specimen (n = 3, Supplementary Figure 7 I-III) produced the INTERDIGITATING with PAPILLATE attached to the main body, smooth type undulation, articulated connection, and rectangle main body. These morphological traits allowed discrimination from other species in our samples. However, in Digitaria ischaemum and Digitaria violascens (Supplementary Figure 7-IV and V), only the PAPILLATE were silicified to form the disaggregated PAPILLATE, and the other parts of the INTERDIGITATING were very weakly or not silicified. The different phytolith types of Digitaria showed the differences within the genus level, which was consistent with the taxonomy: D. chrysoblephara, D. ciliaris, and D. sanguinalis belongs to the section Digitaria, while D. ischaemum and D. violascens belongs to the section Ischaemum (according to the Flora Reipublicae Popularis Sinicae, in Chinese, http://frps. iplant.cn/). The observed differences in our samples showed the potential of a subgenus (section) level discrimination.

All the studied Setaria specimen (n = 4, Supplementary Figure 8) produced the INTERDIGITATING with PAPILLATE attached to the main body, W-type undulation, smooth connection, and rectangle main body, which could be used for discrimination from other species in our samples. The W-type undulation observed in the Setaria species was mostly from the TABLE 4 | Morphological traits of the inflorescence-types of phytoliths and the corresponding species.


The numbers before the descriptions showed the categories of morphological traits, using the combination of morphological traits from different categories could conduct identification of certain species.

basic type to W-II type; only a few of W-III type were observed in the center area of the lemmas in S. faberi (Supplementary Figure <sup>8</sup>-I). The PAPILLATE usually grew very large (compared with its main body) in Setaria species, thereby affecting the connection part, hampering observation of the connection part, while PAPILLATE in other samples of our studied species did not have such a feature. Some minor morphological traits such as small nodes on the tip of the undulation (Supplementary Figure 8-IV-c) and the flat top of the undulation (Supplementary Figure 8-II-b and 8-IV-c) were not used as morphological traits to distinguish samples. As there were limited number of specimens, we could not confirm whether such morphological traits were individual variation or a common feature; however, these minor morphological traits also showed the potential to discriminate among the taxa at species level.

All studied Oplismenus specimens (n = 2, Supplementary Figure 9-<sup>I</sup> and II) produced the INTERDIGITATING with no PAPILLATE, smooth/W-type undulation, smooth/articulated connection, and rectangle main body, which could be used for discrimination from other species in our samples. The combination of undulation and connection: smooth undulation with smooth connection, and W-type (up to W-I type) undulation with articulated connection in the two studied species, suggested the potential of discrimination at the species level.

All studied Paspalum specimen (n = 2, Supplementary Figure 9-III and IV) produced the INTERDIGITATING with PAPILLATE attached to the main body, W-type undulation, articulated connection, and ovate/rectangle main body. In Paspalum, a larger main body than the undulation part was observed to be the identifying feature which was not present in other species. The shape of the main body in the two studied species of this genus, ovate, and rectangle, also suggested the potential of discrimination at the species level.

By using a combination of the morphological traits, we could achieve a reliable discrimination at the genus level among our samples. Although not enough species were studied, the variation in the morphology within the same genus also showed the possibility of discrimination at a more precise level (section or species level). A general distribution pattern of INTERDIGITATING is summarized in Figure 6-A, which shows that the undulation, connection, and main body all have a continuum variation from small to large along the gradient from the edge to the center.

As described above, A. hispidus and H. granularis belongs to the tribe Andropogoneae and could be easily discriminated from other species in our samples using the morphological traits of the INTERDIGITATING. In Setaria species, parameters w and L could not be measured due to the growth of PAPILLATE. In order to apply the same parameters as in previous studies (Lu et al., 2009b; Ge et al., 2018), morphometric analysis was only applied to Digitaria, Oplismenus, and Paspalum genera. Measurements of the parameters are shown in Table 4. The four basic parameters, w, L, h-undulation, and h-body were described as follows: the w value was highest (above 4 mm) in Paspalum, while it was low (below 4 mm) in Digitaria and Oplismenus; the L value was lowest (below 30 mm) in Digitaria, higher in Paspalum (40–60 mm), and highest in Oplismenus (60–90 mm); the h-undulation value showed a large overlap among the three genera and could not be distinguished; the h-body in Paspalum had the highest value (10–30 mm), while it was low (below 10 mm) in Digitaria and Oplismenus. The calculated parameters, h-total, R(w/hu) and R (hu/hb), also varied among the three genera due to the large main body: Paspalum had the highest h-total and R(w/hu) values, and the lowest R(hu/hb) value, while in Digitaria and Oplismenus, a large overlap occurred among h-total, R(w/hu), and R(hu/hb). In our samples it could be found that the parameters could vary from species to species, however, parameters were much similar within the same genus and greater differences could be found among different genera, especially when combining all the parameters.

For statistical analysis of the parameters, a discriminant analysis was applied to examine if morphometric parameters could aid in distinguishing between the samples. The parameters involved in the discriminant functions included w, L, hundulation, h-total, R (w/hu), and R (hu/hb) values. The hbody parameter was excluded as it had the largest absolute correlation between each variable and any discriminant functions. Two discriminant functions were generated (shown in Figure 7). The discriminant analysis showed that by using these parameters, genus level classification could be achieved among our samples; the genera Digitaria, Oplismenus, and Paspalum could be classified successfully. Furthermore, classification accuracy through cross validation reached 94.3%. However, only 53.9% of the original data could be correctly classified to the species level using the same dataset, suggesting that discrimination at the genus level was much more robust than that at the species level.

FIGURE 7 | Canonical discriminant analysis of genera Digitaria, Paspalum, and Oplismenus. The number of species refer to 1 Digitaria sanguinalis, 2 Digitaria chrysoblephara, 3 Digitaria ciliaris, 4 Paspalum orbiculare, 5 P. dilatatum, 6 Oplismenus compositus, 7 Oplismenus undulatifolius.

#### DISCUSSION

#### The INTERDIGITATING and the Silica Skeleton

Although it consists of individual phytoliths articulated together, we defined the INTERDIGITATING as a single type of phytolith in this study, as it has a different anatomical origin and morphology than other types of phytoliths. In the previous studies, the term silica skeleton has been used to describe the articulated ELONGATE DENDRITIC/DENTATE (Rosen, 1992), which was mostly found in wheat and barley inflorescences, and originates from silicification of epidermal long cells. Previous studies sometimes recognized the INTERDIGITATING and the silica skeleton as the same type of phytolith (Madella et al., 2013; Madella et al., 2014; Weisskopf and Lee, 2016), while we note two major differences: 1. silica skeletons originate from epidermal long cell silicification, while the INTERDIGITATING originate between the epidermal cells and the cuticle layer; 2. silica skeletons are silicified single cells, such as in the PAPILLATE- ELONGATE DENDRITIC/ DENTATE pattern, and are an assemblage of single phytoliths, while the INTERDIGITATING are an intact layer with interdigitating ornamentation. Thus, we propose that INTERDIGITATING should be defined as a single type of phytolith, a silicon layer with interdigitating ornamentation that covers the surface of a bract. According to this definition, the BILOBATE-ELONGATE DENDRITIC/ DENTATE pattern and the involucre phytoliths layer should belong to the silica skeleton type, as well as other types of phytoliths that originate from silicified epidermal cells.

### Factors Influencing Phytolith Production in Inflorescence Bracts

The Poaceae inflorescence structure is distinct from that of other plants (Berbel et al., 2007), with a more complex organization of bracts under both genetic and molecular control (Kellogg, 2007; Kellogg et al., 2013; Zhang and Yuan, 2014). Thus, the development of phytoliths might also be affected by both genetic and molecular control. Glumes are leaf-like structures that enclose the florets. In the glumes of our samples the phytoliths were generally similar to leaf type phytoliths: BILOBATE, ELONGATE, ACUTE, and ACUTE BULBOSUS (Table 2). In the lemmas and paleas of sterile florets, phytolith types tended to be the same as those in glumes; however, in the lemmas and paleas of fertile florets, phytolith types generally differed. The divergence of glumes, lemmas, and paleas is known to be under genetic control (Kellogg, 2007; Kellogg et al., 2013), while the divergence of sterile and fertile florets is mostly under molecular control (Zhang and Yuan, 2014). We found that the types of phytoliths produced in the inflorescence bracts differed among those bracts whose divergence is under genetic control and those under molecular control.

Seed setting requires more energy than flowering and is important for plant regeneration (Bazzaz et al., 2000), thus, seed protection is of great importance to plants. Silicon has been proven to be beneficial to plants (Guntzer et al., 2012) as it aids defense against insects (Massey et al., 2006) and fungi (Remus-Borel et al., 2005); it is presumed that phytoliths in the inflorescence bracts also provide similar effects (Ge et al., 2018). In the present study, we observed that phytoliths were most abundant in bracts from spikelet types II and IV, in which phytoliths cover the surface of the bracts that wrap the seed, and the silicification rate and phytolith quantity are positively correlated with seed size in the studied species. As shown in Figure 8, specimens with large, plump seeds, such as Job's tears (C. lacryma-jobi var. ma-yuen), produced numerous phytoliths to form the involucre phytolith layer on the involucre surface (spikelet type II); specimens with small, plump seeds, such as S. pallidifusca, produced an INTERDIGITATING type covering the lemma and palea (spikelet type IV); specimens with small, shriveled seeds, such as Eremopogon delavayi, produced BILOBATE and ELONGATE DENDRITIC/DENTATE phytoliths on the glume, and a very low number of phytoliths are found on the lemmas and paleas (spikelet type III). As a result, the silica layers (including the involucre phytolith layer and INTERDIGITATING) could provide seed protection, preventing biotic and abiotic harm. Phytolith formation consumes less energy (approximately 1/27) than that required for lignification (Raven, 1983); therefore, species with larger seeds tend to invest additional energy to protect the seeds. As these species require large amounts of energy for seed setting and lignification, phytolith formation could be a relatively economical and effective way to protect seeds.

Lignification also affects phytolith production, as revealed by the hard rind genetic locus (Hr) in the genus Cucurbita (Piperno et al., 2002). Similar phenomena were observed in the inflorescence bracts of the present study. Strongly lignified bracts, such as the glume of S. bicolor and the involucre of Coix sp., all produced many more phytoliths than other bracts in the same spikelet. The bracts that produced the INTERDIGITATING were also observed to be of stronger lignification compared to those of other bracts, and very weakly lignified bracts (mostly transparent) did not produce phytoliths at all. As discussed above, a combination of lignification and silicification might be an economical and effective way to provide additional seed protection. Further studies on the genetic control of inflorescence development could facilitate the identification of genes related to phytolith production.

In the current study, more inflorescence phytoliths were observed in species that produced edible seeds (with a larger size and higher seed production rate that would be worth collecting as a food resource), which belonged to the spikelet types II and IV than other species. This corroborated prior observations on other major and minor crops (Ball et al., 2016a). These phenomena suggest the possibility that more inflorescence-type phytoliths might be observed in other unstudied species that possess edible seeds and strongly lignified bracts than those that do not produce edible seeds.

#### Discrimination Among Taxa Based on INTERDIGITATING Differences

The INTERDIGITATING was an important phytolith for discriminating among some of the taxa in our samples. Of all

FIGURE 8 | Comparison of seed size. Big and plump seed: Coix lacryma-jobi var. ma-yuen. Small but plump seed: Setaria pallidifusca and Digitaria ciliaris. Small and shriveled seed: Eremopogon delavayi.

the subtypes, the INTERDIGITATING phytolith with n-type undulation was of low taxonomic value; the n-type undulation being the basic type of all undulation patterns (Ge et al., 2018) and was found in both in the inflorescence and leaves of many taxa (Wang and Lu, 1993). Based on the dataset of the present study, the INTERDIGITATING was mostly observed in Panicoideae species (Lu et al., 2009b; Radomski and Neumann, 2011; Zhang et al., 2011; Madella et al., 2013; Kealhofer et al., 2015; Weisskopf and Lee, 2016; Ge et al., 2018).

A. hispidus and H. granularis from the tribe Andropogoneae produced INTERDIGITATING phytoliths that presented two significant differences from other INTERDIGITATING producers in our samples: A. hispidus and H. granularis produced the INTERDIGITATING type on the glume and the PAPILLATE type were separated from the main body. Based on the PAPILLATE morphological traits, A. hispidus and H. granularis could be easily distinguished from other INTERDIGITATING producers, indicating that the PAPILLATE type separated from the main body might be a potential distinguishing morphological trait at the tribe level. Among the rest of the INTERDIGITATING producers in our study, P. dilatatum and P. orbiculare belong to the tribe Paspaleae. The main body of INTERDIGITATING in these two species are wider than the undulations along the margins, in contrast to other INTERDIGITATING producers where the undulations are larger than the main body. This morphological trait discriminates tribe Paspaleae from tribes Andropogoneae and Paniceae in our samples. Thus, the morphological traits of the INTERDIGITATING: PAPILLATE types separated from the main body and the size of the main body showed great potential for tribe level identification.

Because we only sampled a single specimen of each of the taxa we analyzed in this study, we recognize the need for further analyses of many specimens for each taxon in order to confirm, validate and/or refine our findings. We note that although the number of specimens in the present study was limited, our morphological trait findings were consistent with those reported in other studies. For example, the figures and description provided in a study on D. ciliaris (Madella et al., 2013) show that the INTERDIGITATING had PAPILLATE attached to the main body, smooth type undulations, and rectangular main bodies. Another study (Radomski and Neumann, 2011) reported that disaggregated PAPILLATE were abundant in Digitaria species (including D. ciliaris, D. exilis, and D. iburua). A study on Digitaria adscendans (Weisskopf and Lee, 2016) also reported "short, regular, and cone like papillae." All of these studies corroborate the present study findings that Digitaria species might produce two types of phytoliths in the lemmas and paleas. Further, figures and descriptions of S. pumila (Madella et al., 2013; Weisskopf and Lee, 2016), S. plicata (Lu et al., 2009b), and Setaria verticillate (Madella et al., 2013; Weisskopf and Lee, 2016), show that the INTERDIGITATING all have PAPILLATE attached to the main body, Wtype undulations (only level-I or II), smooth connections, and rectangular main bodies; all these morphological traits are likewise consistent with the findings of the present study. Similarly, a study on Paspalum conjugatum (Weisskopf and Lee, 2016) showed that the INTERDIGITATING had PAPILLATE attached to the main body, Wtype undulations, articulated connections, and a rectangular main bodies, which again are similar to our findings for P. dilatatum in the present study. These prior studies support our findings and confirm the possibility of genus level identification using these morphological traits.

Within the same genus, although some minor morphological commonalities were observed, these morphological traits might not be useful as identification features due to the limited number of studied specimens. The morphometric analysis showed a large overlap within genera (Table 2 and Figure 7), and morphological traits also overlapped among species (Zhang et al., 2011; Kealhofer et al., 2015; Ge et al., 2018; Zhang et al., 2018). However, some minor morphological traits also suggest the possibility of species level identification, but again this must be further evaluated by additional studies.

Morphological traits focusing on the presence of PAPILLATE, undulation pattern, connection shape, and main body shape of Paniceae species are shown in Table 5 and compared with data from related studies (Lu et al., 2009b; Ge et al., 2018). Thus the use of a combination of morphological traits for genus level discrimination, and species level discrimination is promising.

Kealhofer et al. (2015) studied the common Setaria species in China and doubted the stability of discrimination criteria with regard to discriminating Setaria from Panicum (Lu et al., 2009b) In their study, they did not apply a combination of morphological traits (apply all traits to the same INTERDIGITATING phytolith), but rather focused on single morphological traits (compare one trait for all INTERDIGITATING phytoliths). The basic n-type and the level-I of W-type, h-type, b-type, and smooth type undulation on the edge of the bracts can be highly similar, as highlighted in our previous study (Ge et al., 2018) and this study.As shown in the present study, single morphological traits overlapped at the genus level (Table 5), while a combination of morphological traits could provide more robust discrimination at the genus level. In their study (Kealhofer et al., 2015), they also noticed the overlapping occurrence of papillae, the morphology of undulations and connections among different species, thus suggesting that the basic n-type should not be used as a sole identification criterion; their data also revealed the insufficiency of single measurements for differentiating among Setaria species, which was consistent with the current study findings. Thus, the key to discrimination is to identify a diagnostic combination of the most common and representative traits, rather than single trait variables. In this study, we identified morphological traits that might be used to discriminate S. italica and S. viridisfrom their wild relatives (Table 5); however, based our limited sample size, further verification is required. Thus, our findings in this preliminary study support previous published identification criteria for distinguishing between S. italica and P. miliaceum (Lu et al., 2009b), and the idea that examining a combination of morphological traits has the potential to provide reliable discrimination at the genus level.

## Phytolith Types With High Taxonomic Value in C. Lacryma-Jobi

Job's tears (C. lacryma-jobi) is an important plant resource that was used approximately 24,000 years ago (Liu L. et al., 2018). Liu et al. conducted an experiment on starch and phytoliths for their


identification in archaeological remains (Liu et al., 2019). In their study, they emphasized cross shaped BILOBATE phytoliths found in the glumes, lemmas, paleas, and leaves, which could also be found in other Panicoideae species (their study did not compare other cross shaped BILOBATE phytolith producers). In the present study, we found that RECTANGULAR DENTATE and BLOCKY AMOEBOID phytoliths were unique to the inflorescence of this taxon, which has not been reported before. This differentiation has the potential to be the diagnostic criteria for Coix identification, even at the subspecies level.

The two specimens of Job's tears used in this study represented the two commonly used types: edible (with an easy to break involucre) and decorative (with a rigid involucre). The phytoliths differed between the two types (Supplementary Figure 2): The edible species (C. lacryma-jobi var. ma-yuen) produced BILOBATE and ELONGATE phytoliths on the involucre surface which had some height or thickness differences when compared with the flat phytoliths found in the glumes, lemmas, and paleas. The phytolith articulations were not very tight and resulted in an easy to break involucre. However, the decorative species (C. lacryma-jobi) produced BLOCKY AMOEBOID phytoliths that were tightly connected to form a more rigid phytolith layer on the involucre surface. The different phytolith types on the involucre surface created difficulty in the hulling process; the edible species could be hulled by hand, while the decorative species required the use a hammer. As hulling difficulty would have been an important selection trait (Arora, 1977) in the domestication of Job's tears, our preliminary results indicated that the BLOCKY AMOEBOID involucre phytoliths have a great potential for investigating the domestication process of Job's tears. As an important minor crop, the domestication of Job's tears has not been fully studied, partly due to lack of evidence. The different phytolith types on the involucre surface observed in cultivated and wild Job's tears in this study may provide insight into the domestication process. Again, we note that as these findings were limited to the studied specimens, further studies are needed to validate our findings.

#### CONCLUSIONS AND PERSPECTIVES

Phytoliths in every inflorescence bract of 38 common Panicoideae species were observed, various phytolith types were described, and the inflorescence-type of phytoliths were identified. We proposed a new phytolith morphotype, the INTERDIGITATING, and identified several other types of phytoliths, BLOCKY AMOEBOID, RECTANGULAR DENTATE, and ELONGATE DENDRITIC with multi tent-like arch top, that might be of high taxonomic value. Some of these types we report for the first time in the taxa analyzed in this study. From our observations we suggest that phytoliths in the inflorescence bracts may be positively related to inflorescence development, which might be under both genetic and molecular control. The inflorescence involucre phytolith layer and INTERDIGITATING phytoliths might also be related to a seed protection strategy; species with larger seeds might produce more phytoliths in the outermost bracts to protect the seeds. Thus, species with larger seeds and lignified bracts might have higher potential to produce more phytoliths of higher taxonomic value; more attention should be paid to such species in future studies.

The INTERDIGITATING was an important phytolith type in the inflorescence in our study, especially for millet identification. We summarized INTERDIGITATING morphological traits and found that reliable resultsfor genus level identification among our samples was possible using a combination of morphological traits (Table 5). We also found that morphological variation may have the potential for identification at the tribe level (INTERDIGITATING with PAPILLATE separated from the main body and the size of the main body) or species level (morphological variations within the same genus). Again, studies ofmore specimens are neededfor confirmation of the potential. The BLOCKY AMOEBOID and RECTANGULAR DENTATE types from the involucre of Job's tears have great potentialfor studying the domestication process of Job's tears, and the ELONGATE DENDRITIC with multi tent-like arch topfrom the glumes of S. bicolor might also assist in the identification of sorghum remains. In future studies, more species, more samples per species, and further efforts are needed to provide applicable and robust identification criteria at the tribe/species level.

## AUTHOR CONTRIBUTIONS

YG and HL designed the research. YG performed the experiment. YG, JZ and CW carried out the image process and

#### REFERENCES


data analysis. YG, HL and XG wrote the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This study was jointly supported by the National Natural Science Foundation of China (Grant Nos. 41802021, 41830322 and 41430103), the Strategic Priority Research Program of Chinese Academy of Sciences (Grant No. XDB26000000) and the China Postdoctoral Science Foundation (Grant No. 2018M641480).

#### ACKNOWLEDGMENTS

We thank Prof. Guoan Wang, Prof. Yongji Wang and Prof. Zhijian Feng for the providing of same samples, and Prof. Shuzhi Cheng for the identification of the samples.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01736/ full#supplementary-material


Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Ge, Lu, Zhang, Wang and Gao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Taxonomic Demarcation of *Setaria pumila* (Poir.) Roem. & Schult., *S. verticillata* (L.) P. Beauv., and *S. viridis* (L.) P. Beauv. (Cenchrinae, Paniceae, Panicoideae, Poaceae) From Phytolith Signatures

#### Mudassir A. Bhat, Sheikh A. Shakoor, Priya Badgal and Amarjit S. Soodan\*

*Plant Systematics and Biodiversity Laboratory, Department of Botanical and Environmental Sciences, Guru Nanak Dev University, Amritsar, India*

Background and Aims: The role and significance of phytoliths in taxonomic diagnosis of grass species has been well documented with a focus on the types found in foliar epidermis and the synflorescence. The present paper is an attempt to broaden the scope of phytoliths in species diagnosis of grasses by developing phytolith signatures of some species of the foxtail genus *Setaria* P. Beauv. through *in situ* location and physico-chemical analysis of various phytolith morphotypes in different parts of the plant body.

#### *Edited by:*

*Terry B. Ball, Brigham Young University, United States*

#### *Reviewed by:*

*Jennifer Bates, University of Cambridge, United Kingdom Lisa K. Kealhofer, Santa Clara University, United States*

> *\*Correspondence: Amarjit S. Soodan assoodan@gmail.com*

#### *Specialty section:*

*This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science*

*Received: 21 February 2018 Accepted: 04 June 2018 Published: 22 June 2018*

#### *Citation:*

*Bhat MA, Shakoor SA, Badgal P and Soodan AS (2018) Taxonomic Demarcation of Setaria pumila (Poir.) Roem. & Schult., S. verticillata (L.) P. Beauv., and S. viridis (L.) P. Beauv. (Cenchrinae, Paniceae, Panicoideae, Poaceae) From Phytolith Signatures. Front. Plant Sci. 9:864. doi: 10.3389/fpls.2018.00864* Methods: Clearing solution and dry ashing extraction methods were employed for *in situ* location and isolation of phytolith morphotypes respectively. Ultrastructural details were worked out by Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy. Morphometric and frequency data of phytolith morphotypes were also recorded. Biochemical architecture of various phytolith types was worked out through SEM-EDX, XRD, and FTIR analysis. Data were analyzed through Principal Component Analysis and Cluster Analysis.

Key Results: *In situ* location of phytoliths revealed species specific epidermal patterns. The presence of cystoliths (calcium oxalate crystals) in the costal regions of adaxial leaf surface of *S. verticillata* (L.) P. Beauv. is the first report for the genus *Setaria*. Our results revealed marked variations in epidermal ornamentation and undulation patterns with a novel "3" (Lamda) type of undulated ornamentation reported in *S. verticillata*. Dry ashing method revealed species specific clusters of phytolith morphotypes.

Conclusions: The study revealed that phytoliths can play a significant role in resolution of taxonomic identity of three species of *Setaria*. Each species was marked out by a unique assemblage of phytolith morphotypes from various parts of the plant body. Apart from *in situ* location and epidermal patterning, diagnostic shapes, frequency distribution, size dimensions, and biochemical architecture emerged as complementary traits that help in developing robust phytolith signatures for plant species.

Keywords: grasses, morphotypes, phytoliths, *Setaria* spp., silica, taxonomic demarcation

## INTRODUCTION

The foxtail genus, Setaria P. Beauv., so named by the presence of sterile bristles that subtend spikelets in a close panicle, belongs to the "bristle clade" (subtribe Cenchrinae, tribe Paniceae, subfamily Panicoideae) of the grass family Poaceae (Morrone et al., 2012). The genus has a labile morphology requiring additional characters for the resolution of phylogenetic relations among the 113 odd species of the genus (Clayton et al., 2016 onwards). One of the species, the foxtail millet Setaria italica (L.) P. Beauv. has been cultivated along with other millets in dryland farming system since prehistoric times (Madella et al., 2016; Weisskopf and Lee, 2016). Some other species of the genus also serve as significant sources of forage and fodder (Aliscioni et al., 2011; Marinoni et al., 2013). Several studies have attempted to resolve the infrageneric (Stapf and Hubbard, 1934; Webster, 1987; Pensiero, 1999) and intergeneric (Webster, 1993, 1995; Veldkamp, 1994; Morrone et al., 2014) relations of the genus. Molecular studies on the chloroplast gene ndhF have revealed polyphyletic nature of the genus with three well supported clades (Kellogg et al., 2009). Even though leaf blade anatomy has traditionally been employed for taxonomic characterization of grasses (Prat, 1936, 1948; Metcalfe, 1960; Ellis, 1979, 1984), the role of anatomical characters in grass taxonomy and phylogeny has been, so to say, rediscovered in the recent past (Ingram, 2010) with Setaria P. Beauv. as a model genus (Aliscioni et al., 2016). Apart from epidermal cell patterns and vasculature, phytoliths in leaf epidermis and other parts of the plant body have been utilized for species characterization and taxonomic analysis of grass taxa.

Phytolith studies have been utilized both for characterization of individual Setaria species (Rovner, 1971; Hodson et al., 1982) as also for taxonomic demarcation among species within the genus (Zhang et al., 2011; Layton and Kellogg, 2014; Wang et al., 2014; Madella et al., 2016) and from related genera (Hunt et al., 2008; Lu et al., 2009; Out et al., 2014; Wang et al., 2014; García-Granero et al., 2016; Madella et al., 2016). The ever increasing role of phytoliths in the resolution of intrageneric and intergeneric taxonomy of the genus can be ascribed to the simple fact that even among grasses, Setaria spp. show exceptional levels of silica accumulation in the form of phytoliths in all parts of the plant body. During the present investigations, an attempt has been made to supplement the information available on the phytolith profiles of three closely related species of the foxtail grass genus through a multiproxy approach and the development of phytolith signatures as additional evidence for their taxonomic demarcation. Analysis of several aspects of phytoliths from different parts of the plant body of the selected species was done through a battery of techniques employed in a logical sequence from in situ location of phytolith morphotypes in foliar epidermis to advanced level of physico-chemical analysis involving sophisticated instruments and methodology. In this context, the present study marks a significant advance toward developing a comprehensive and robust framework for the use of data on morphotype diversity, distribution in different parts of the plant body and their ultrastructural and biochemical characterization in identification and taxonomic demarcation of plant taxa.

## Silica and Phytolith Production in Plants

Plants absorb monosilicic acid (H4SiO4), which is released to the soil by weathering of siliceous minerals, by action of an aquaporin-like channel Low-silicon 1 (Ls1) and a proton antiporter Low-silicon 2 (Ls2) and polymerizes it into amorphous silica (SiO2.nH2O) in cell lumens (internal casts), intercellular spaces, and cell walls (external casts) of the parenchymatous tissue (Baker, 1959b; Jones and Handreck, 1967; Rovner, 1971; La Roche, 1977; Bombin, 1984; Piperno, 1988; Mulholland, 1989; Ma et al., 2011; Ma and Yamaji, 2015). A number of unknown silica transporters are believed to be involved in directing silica transfer to different silicification sites (Kumar et al., 2017). Being hard and resistant to dessication and disfiguration, these amorphous silica bodies are commonly called phytoliths [phyton (ϕυτ oν) = plant + lithos (λιθoς) = stone]. As casts (both internal and external) of plant cells, phytoliths vary in shape, size, frequency, surface ornamentation and other structural features (Ollendorf et al., 1988; Piperno, 1988, 2006; Lu and Liu, 2003; Lu et al., 2009; Zhang et al., 2011; Szabo et al., 2015; Ge et al., 2016). Genetic control of shape, size and frequency of phytoliths has been demonstrated in some monocots (e.g., Zea mays L.) and dicots (e.g., Cucurbita spp. L.) (Bozarth, 1987; Piperno et al., 2000).

Phytoliths have been implicated in several biological functions including that of providing an endoskeletal framework which prevents wilting (Parry and Smithson, 1958a) and offering resistance to herbivory (Rovner, 1971; Stebbins, 1972, 1981; Coughenour, 1985; Epstein, 1994, 1999), and alleviating biotic (Jones and Handreck, 1967; Gould and Shaw, 1983; Mazumdar, 2011) and abiotic (Hodson et al., 1985; Hodson and Evans, 1995; Lux et al., 2003; Hattori et al., 2005) stress. Phytoliths have also been reported to play a role in checking the rate of transpiration and at the same time reducing the heat load of plants growing in exposed habitats (Jones and Handreck, 1967; Sangster and Parry, 1971; Krishnan et al., 2000).

Ecological functions played by phytoliths include a role in biogeochemical and bio-cycling of silicon in terrestrial ecosystems (Conley, 2002; Gerard et al., 2008; Borrelli et al., 2010; Struyf and Conley, 2012) and sequestration of occluded carbon (Rajendiran et al., 2012; Parr and Sullivan, 2014; Alexandre et al., 2015; Ru et al., 2018; Yang et al., 2018). Isotopic dating of phytolith occluded carbon (PhytOC) has been employed to determine the age of sediments and that of elements of vegetation trapped in these sediments (Parr and Sullivan, 2014). The use of phytoliths in dating of plant fossils can be attributed to the fact that upon death and in situ decay of the plant body, phytoliths are released into the soil where they stay through the millenia resisting deformation and destruction by the vagries of geological and climatic conditions. Their long time persistence in the soil make them ideal plant microfossils which have been recovered from sediments as far back as 60 mya in the Cenozoic (Jones, 1964), including the glacials (Twiss et al., 1969; Fredlund et al., 1985) and the Holocene (Baker, 1959a; Crawford, 2009). Phytoliths have been recovered from diverse habitats including swamps (Baker, 1959a), arid zones (Pease and Anderson, 1969), humid areas (Jones and Beavers, 1964) and vegetation types including grasslands and forests (Wilding and Drees, 1973).

Owing to widespread production across several plant groups and excellent preservation as microfossils, phytoliths have found an ever increasing role as proxies in diverse fields of scientific enquiry including archeaobotany of the centers of civilization and cultivation (Schellenberg, 1908; Pearsall, 1978; Rovner, 1983; Piperno, 1984; Shillito, 2013; Gao et al., 2018), paleoecology and paleoclimatology (Rovner, 1971; Carbone, 1977; Fox et al., 1996; Piperno, 2006; Albert et al., 2007), the mapping of ancient land use patterns, and vegetation structure (Gross, 1973; Pearsall and Trimble, 1984; Fisher et al., 1995). Phytolith profiles of present day crop species and soil samples of ancient sites have been compared and calibrated for developing historical calendars for the origin of agriculture and routes of spread and diversification of crop species and calculating the crop ratios (Rovner, 1983; Piperno, 1998, 2009; Pearsall et al., 2003; Albert and Henry, 2004; Fuller et al., 2007; Itzstein-Davey et al., 2007; Tsartsidou et al., 2007; Hunt et al., 2008; Crawford, 2009; Lu et al., 2009; Zhang et al., 2010, 2012; Zhao, 2011; Chen et al., 2012; Madella et al., 2014, 2016; Weisskopf et al., 2014; Out and Madella, 2016; Weisskopf and Lee, 2016), the food and non-food uses of plants in crafts and building materials (Ryan, 2011), agricultural practices (e.g., irrigation, Rosen and Weiner, 1994; Slash-n-burn; Piperno, 1998), paleoagrostology (Piperno and Pearsall, 1998), taphonomy (Madella and Lancelotti, 2012) and colonization of islands and distant lands (Astudillo, 2017).

On account of the wide range of availability and ease of recovery from unused parts of cereals (and other crop species) and the purity of silica obtained, phytoliths have also found a role in nanotechnology (Neethirajan et al., 2009; Qadri et al., 2015). In the contemporary environmental context, phytoliths are being employed as models for assessment of the effects of global warming and climate change (Hongyan et al., 2018).

## Phytoliths in Grass Systematics

Notwithstanding the above mentioned applications, phytoliths have found the most significant role in taxonomic characterization and demarcation of plant taxa. At this juncture it would be quite instructive to review the landmarks in plant phytolith research that have provided the framework for the use of phytoliths in grass systematics as well. After the revisionary work of (Netolitsky, 1929), attempts were made to identify the marker morphotypes for plant taxa at different levels of taxonomic hierarchy. Within grasses, branched cells were typically associated with Nardus stricta L. (Parry and Smithson, 1958a,b). Twiss et al. (1969) expanded the scope of "marker morphotype" approach to major groups within the family through a study of 26 morphotypes of which eight were ascribed to festucoid group, two to chloridoid, and 11 to panicoid grasses and the rest (five) had no particular subfamilial affiliation. Soon afterwards, Rovner (1971) pointed out that a search for "marker" types for plant taxa would run into difficulty on account of "multiplicity" of types within a single species (more so for taxa at higher ranks) and "redundancy" of occurrence of same types "appearing in related as well as taxonomically unrelated species." Rovner (1971) suggested that assemblages or "type-sets" of phytoliths would provide better taxonomic demarcation among plant species and soil samples.

Apart from types, Mulholland (1989) presented data on frequencies of various types to characterize 19 wild grasses collected from their natural habitats. Piperno and Pearsall (1993) pointed out that phytoliths from reproductive parts proved more useful in separating maize (Zea mays L.) from teosinte. This work focused on an organ-specific approach in using phytoliths in taxonomic demarcation of grass species. Pearsall et al. (1995) further narrowed it down to "silicified glumes" as the most revealing in distinguishing cultivated rice (Oryza sativa L.) from its wild relatives. Piperno (1998) identified diagnostic morphotypes of phytoliths for the subfamilies Pooideae, Arundinoideae, Chloridoideae, and described the diagnostic and diverse types in the Bambusoideae in great detail. Several subsequent workers have utilized typology and frequency (abundance) approachs to phytolith analysis for taxonomic characterization and demarcation of cultivated and wild grasses (Piperno, 1985; Zhang et al., 2012; Tripathi et al., 2013).

Rudall et al. (2014) employed the shapes of costal phytolith morphotypes and their orientation to elucidate phylogenetic relationships among different grass subfamilies and supported the recognition of three clades within the family. The APP (Anomochloideae, Pharoideae, Puelioideae) clade was treated as the most primitive followed by BEP (Bambusoideae, Ehrhartoideae, Pooideae) and species rich PACCMAD (Panicoideae, Arundinoideae, Chloridoideae, Micrairoideae, Aristidoideae, Danthonioideae) clades. Kealhofer et al. (2015) carried out phytolith analysis of leaf and synflorescence of the foxtail millet [S. italica (L.) Beauv.]. In India, Jattisha and Sabu (2015) brought out the taxonomic significance of foliar phytoliths as diagnostic markers in some grasses of South India. More recently, Shakoor et al. (2016) employed phytoliths from underground (root) and aerial (culm, leaf & synflorescence) parts for taxonomic demarcation of two reed grasses, Arundo donax L. and Phragmites karka (Retz.) Trin. ex Stued.

Parry et al. (1984) marked the biochemical dimension in phytolith characterization by reporting a time dependent accumulation of some elements (K, Cl, P, and S) along with silicon in the silicified microhairs from the lemma of the canary grass, Phalaris canariensis L. and giving evidence of genetic control of silicification. In recent years, physico-chemical characterization of phytoliths has been extended to a study of the physical states (as amorphous vs. crystalline), the mineral composition and the study of functional groups and their bonding patterns through sophisticated methods of analysis (Chauhan et al., 2011; Shakoor et al., 2016).

The work reported in this paper is a part of the ongoing program of research on the role of phytoliths in the systematic analysis of grass flora in the area of study. Setaria species selected for the present investigations show morphological similarity with one another as well as the foxtail millet S. italica (L.) P. Beauv. and are placed closely in keys to species identification of the genus (Layton and Kellogg, 2014). Setaria viridis (L.) P. Beauv. had an Asian origin with phylogenetic relations with its domesticated derivative the foxtail millet, S. italica with which it remains interfertile (Shi et al., 2008). The second species of the present sample, S. verticillata is the polyploid derivative of S. viridis (L.) P. Beauv. (Layton and Kellogg, 2014). The third species, S. pumila (Poir.) Roem. & Schult. had an African origin (Rominger et al., 2003) but shares a wide distribution with S. viridis and growth in mixed populations and is included in the "S. viridis clade" of the genus. The foxtail millet, Setaria italica would have been a useful and desirable addition to the material but it is not cultivated in the Punjab plains and was thus unavailable for this work. Even though permanent herbarium sheets of this species were available in the departmental herbarium, sufficient material could not have been extracted from them for the present analysis.

#### MATERIALS AND METHODS

#### Area of Study

About twenty plant specimens of S. pumila and S. verticillata were collected from the campus of Guru Nanak Dev University, (32.64 ◦N and 74.82 ◦E) Amritsar, Punjab (**Figures 1a–c**). A similar number of plants of S. viridis were collected from the campus of Sher-i-Kashmir University of Agricultural Sciences and Technology, (32.65 ◦N and 74.81 ◦E) Srinagar, Jammu & Kashmir (**Figures 1d,e**). The specimens were collected at flowering and fruiting stage. Taxonomic descriptions and illustrations of the species were made from fresh material in the standard formats of grass description proposed by Grass Phylogeny Working Group (GPWG (Grass Phylogeny Working Group)., 2001) and GPWG (Grass Phylogeny Working Group II). (2011) systems and maintained by the online sources (Clayton et al., 2016; GrassBase—The Online World Grass Flora: The Board of Trustees, Royal Botanic Gardens [online]. Available at http://www.kew.org/data/grasses-db.html and 2. Tropicos (2011) http://www.tropicos.org. Name Search.aspx. 3.eflora of China: http://www.efloras.org. Missouri Botanical Garden, St. Louis, MO and Harvard University Herbaria, Cambridge, MA). The species identity of the specimens was established by comparison of the vegetative and reproductive morphology and micromorphometry with standard descriptions available in works of grass floristics of the world (Bor, 1960; Gould, 1968; Cope, 1982; Gould and Shaw, 1983; Clayton and Renvoize, 1986; Watson and Dallwitz, 1992; Kellogg, 2015; Soreng et al., 2017 and the region Sharma and Khosla, 1989; Kumar, 2014). For preparation of herbarium sheets, three dried specimens for each of the species have been deposited in the Herbarium of the Department of Botanical and Environmental Sciences, Guru Nanak Dev University, Amritsar (Voucher nos. 7373 to 7381).

### Phytolith Analysis

About five to ten plants remaining intact after taxonomic descriptions and dry preservation for hebarium specimens, were dismembered into underground (root) and above ground (culm, leaf and synflorescence) parts. The material was homogenized (part wise) into lots. Some of the material from each lot was preserved in 70% ethanol at 4◦C for in situ location of phytoliths. The rest of the material in each lot was washed to clear dust and adhering soil particles, sundried and stored in plastic jars for dry ashing and further analysis.

Methodology of the present study followed a logical and systematic sequence from in situ visualization of the phytoliths in the leaf epidermis to dry ashing of plant parts for disarticulation of individual morphotypes for recording qualitative (morphotypic) data and collection of quantitative (micromorphometric) data on phytoliths among the species and their parts. Quantitative assessment also included frequency distribution of various morphotypes. After data collection at the level of light microscopy (LM), scanning electron microscopy (SEM) of morphotypes was carried out to record their surface ornamentations and three dimensional structures. Transmission electron microscopy (TEM) was employed to study variations in texture, interplanar spacing, and crystallinity of various morphotypes. EDX analysis was employed to study elemental composition of phytolith morphotypes and the rhizospheric soil. XRD analysis revealed the physical phases of silica and other minerals in the phytoliths. Similarly, FTIR analysis was carried out to know the functional groups of phytoliths from different plant parts.

#### *In Situ* Location

A study of in situ location and epidermal patterning of phytoliths on both adaxial and abaxial surfaces of the leaf was carried out according to the method of Clarke (1959) with some modifications. The leaf segments from mature leaves were boiled in tubes for 5–10 min in distilled water. After cooling down the tubes, leaf segments were put in ethanol (absolute) and heated gently (80◦C) in a water bath till they were decolorized. Thereafter, the segments were immersed in a solution of lactic acid and chloral hydrate (3:1 v/v) and boiled again for 20– 30 min in a water bath. After clearing, they were placed on clean ceramic tiles with the adaxial surface upwards and the epidermis was peeled off the middle part of mature leaf blades. Similarly, peelings from abaxial surface of leaf segments were obtained. Epidermal peelings were stained in Gentian Violet and passed through a dehydration series of ethanol (30% through 50, 70, 90% and absolute ethanol) and mounted in DPX for light microscopy and microphotography with a Micro Image Projection System (MIPS-USB 0262) mounted on an Olympus Binocular and connected to a computer for imaging.

## Dry Ashing Method

The standard protocol of Carnelli et al. (2001) with some modifications was employed for dry ashing of the plant material. The dried and stored material of individual parts was taken from the plastic jars, further dried in an oven, weighed and transferred to porcelain crucibles. The material was incinerated at 550◦C for 4–6 h to ashes. The crucibles were taken out of the furnace, allowed to cool and ash contents were transferred to test tubes. A sufficient amount of hydrogen peroxide (30%) was added to submerge the ash and the test tubes were kept at 80◦C for 1 h in a water bath. The mixture was decanted and rinsed twice in distilled water. Hydrochloric acid (10%) was added to the pellet and incubated at 80◦C for 1 h. Thereafter, the mixture was washed in distilled water and centrifuged for 15 min at 7,500 rpm. The supernatant was decanted off and the pellet was washed twice in distilled water. The process was repeated till the pellet was clear. Finally, the pellet was oven dried for 24 h at 60◦C to a powder form and weighed. The silica content (%) was calculated by the formula: final ash content/dry mass × 100.

FIGURE 1 | Distribution of sampling sites in India (a–e): *Setaria pumila* (Poir.) Roem. & Schult. and *Setaria verticillata* (L.) P. Beauv. (b,c) and *Setaria viridis* (L.) P. Beauv. & Schult. (d,e).

A small amount (ca. 0.1 mg) of dried ash was dipped in 10 ml of Gentian Violet in a watch glass and stirred. A drop of this mixture was put on a glass slide and covered with a cover slip. The slides were heated gently and excess stain drained off with a filter paper. Ten slides were prepared for each sample. Morphotypes of phytoliths were photographed by Olympus Micro Image Projection System (MIPS-USB 0262) at a uniform magnification (40X). The phytoliths isolated by the dry ashing method from underground (root) and aerial (culm, leaf, and synflorescence) parts showed considerable diversity of phytolith morphotypes in terms of their shapes and were classified according to the International Code of Phytolith Nomenclature (ICPN 1.0; Madella et al., 2005). Some of the morphotypes whose description was not available in the ICPN nomenclature were classified as per the schemes presented in **Table 1**.

## Morphometry

Morphometric measurements of phytolith morphotypes were made using Image J software (version 1.46r.). A total of 5 morphometric parameters of size and shape descriptors were recorded on a minimum sample size calculated as per recommendations of the International Committee for Phytolith Morphometrics (ICPM, Ball et al., 2016) by the formula:

$$N\_{\rm min} = Z\_{\alpha/2}^2 \times \mathcal{S}^2 / (ME)^2$$

Where: **N**min = the minimum adequate sample size; **Z 2** <sup>α</sup>/<sup>2</sup> = 1.64, which is the square of the two tailed Z value for level of significance (α) = 0.10; **S <sup>2</sup>** = the variance, and **(ME)<sup>2</sup>** = the square of the permissible margin of error (in this case 0.05) × the sample mean. This calculation estimates the minimum adequate sample size required for 95% confidence that the sample means are within 5% deviation from actual population means.

#### Scanning Electron Microscopy (SEM)

For SEM, dry ash was spread evenly over the stubs with the help of double-sided adhesive tape under the stereoscope. Silver paint was applied on edges of the stub and the samples dried overnight at 40◦C. The next day, stubs were coated with graphite using a vacuum evaporator and later coated with gold by a sputter coater (QUORUM) and imaged under SEM (CARL ZEISS EVO 40) at an accelerating voltage of 40 KV.

#### Transmission Electron Microscopy

TEM micrographs were obtained using a JEOL JEM-2100 operating at 200 keV. Samples were prepared by suspending a


Frontiers in Plant Science | www.frontiersin.org

 and Distel,

2004

9.

 Bilobate class VIII BCVIII

10. Blocky crenate

11. Blocky irregular

 BIR

 Blocky

 Irregular

 ———— —

+ – + +

+ + + +

+ + + +

—–

 1.0 (0.91)

Barboni et al., 2010 *(Continued)*

 BC

 Blocky

 Crenate

 ———— —

– – – –

– + – –

– – – –

SVC

 0.33 (0.08)

Barboni et al., 2010

 Bilobate class

—————

 ———— —

– – – –

– – – –

– – + –

SV

 0.33 (0.08)

Blinnikov, 2005

> VIII

 —

VII


TABLE 1 |

Continued



TABLE 1 |

Continued


small quantity of powder (crushed in pestle and motar) in double distilled water (DDW). The samples were sonicated for 30 min and a drop of material was placed on a carbon coated copper grid. The grids were dried on filter paper using an electric lamp and were subsequently analyzed. Structural details as well as the chemistry of the samples were worked out by High Resolution Transmission Electron Microscopy (HRTEM) and Selected Area Electron Diffraction (SAED) of various phytolith types.

#### Biochemical Architecture

Elemental analysis of phytolith morphotypes and soil samples were carried out with Scanning Electron Microscope-Energy Dispersive X-ray analysis (SEM-EDX). Infrared spectra of silica powder were obtained on a Fourier Transform Infrared (FTIR) Spectrophotometer (System92035, Perkin-Elmer, England) at room temperature using the standard KBr method. The functional group spectra were recorded over a wavelength range of 500–4,000 cm−<sup>1</sup> . X-ray Diffraction (XRD) studies were performed on powder XRD system (Bruker D8 Advance) using Cu Kα radiation (k = 1.5418 Å) in the 2θ (Bragg's angle) at a range of 20–70. The data were analyzed for presence of different polymorphic structures of silica and other compounds using the origin pro 8 software and following the notation of the Joint Committee on Powder Diffraction.

Elemental composition of rhizospheric soil samples was carried out with SEM-EDX. Soil samples (ca. 5 g) from the rhizospheric region of the specimens taken for phytolith analysis were collected and ground into fine powder. Small bits of the powder were spread uniformly on the stubs and were scanned using Energy Dispersive X-ray analyzer coupled with the SEM through Inca software.

### Statistical Analysis

Descriptive statistics of morphometric and elemental composition data was carried out with the help of paleontological statistics (PAST) software (Hammer et al., 2001). Cluster analysis of presence/absence data of bilobate classes of phytoliths and Principal component analysis (PCA) of morphometric and elemental composition data was carried out using C2 data analysis software (Juggins, 2003). This software has also been used for plotting the stratigraphic diagram of the frequency data of phytoliths. Pearson's coefficient of association of phytolith morphotypes of the three species were also calculated employing computer programs developed for the purpose.

## RESULTS AND DISCUSSION

Taxonomic descriptions of grasses include several (vegetative and reproductive) characteristics that help to characterize and classify taxa from subfamily down to the species and infra specific levels. Morphological and morphometrical characters that diagnose Setaria pumila, S. verticillata, and S. viridis from one another are presented in Supplementary Table 1. Whereas qualitative characteristics provide a clear cut account of similarities and dissimilarities in paired comparisons, quantitative characteristics show overlapping ranges and cryptic distinctions requiring additional evidence for taxonomic resolution. In the present context, phytolith analysis was employed to supplement and substantiate taxonomic demarcation among the three species of the genus Setaria P. Beauv.

#### Epidermal Patterns

Leaf epidermal characteristics play an important role in taxonomic demarcation of grass taxa due to the unique arrangement of epidermal long and short cells in the costal and intercostal regions (Prat, 1936, 1948; Metcalfe, 1960; Ellis, 1979; Hilu, 1984; Rudall et al., 2014).

The present study has revealed two distinct distribution patterns of silica cells and associated epidermal cells. The first one comprises long-short cell alternation in the intercostal region of the epidermis and the second one includes axial rows of closely spaced short silica cells in the costal region. These costal rows of silica bodies are separated from each other by a single short intervening cell known as the cork cell and are considered highly diagnostic in grasses (Prasad et al., 2011). The intervening cells are relatively thin walled, but resemble silica bodies in size and shape.

The underlying causes for the concentration of the silica bodies over the veins remain unknown though there is an apparent positive correlation between silica deposition and proximity to lignified tissues of the vascular bundles. Indirect support for this association between lignin and silica deposition comes from studies indicating a correlation between silica deposition and lignin metabolism in grasses (Schoelynck et al., 2010).

Supplementary Table 2 summarizes epidermal patterning and the distribution of silica cells and other associated epidermal cells on the adaxial and abaxial leaf surfaces of grass species under investigation. S. pumila revealed one to three axially oriented rows of bilobate phytoliths with each bilobate phytolith flanked by silica cork cell in the coastal region on the adaxial surface of cleared leaf segments (**Figures 2Aa–e**). It also showed the presence of nodular bilobate phytoliths (**Figure 2Aa**). The costal rows of silica cells showed the presence of prickle hairs (**Figure 2Ae**)**.** The intercostal region on adaxial surfaces completely lacked silica cells except for occasional prickle hairs (**Figure 2Af**) with those on the margin having the length of base greater than the barb (**Figure 2Ag**). The abaxial surface of cleared leaf segments of S. pumila presented a different scenario. The costal region showed 1–3 bilobate to cross shaped silica cells with each bilobate/cross pair of silica cells separated by silica cork cells (**Figures 2Ah–j**). The intercostal region of the abaxial surface in S. pumila showed prickle hairs between each pair of epidermal long cells in alternating axial rows (**Figures 2Ak,l**). The margins on abaxial surfaces showed prickle hairs with a much higher base to barb length ratio than those on the margins of adaxial surfaces.

We have classified bilobate phytolith morphotypes into eight subtypes based upon the length of their shank (the interconnecting segment between two lobes) and the shape of the outer margin of their lobes as proposed by Lu and Liu (2003) (Supplementary Tables 2, 3 and **Figure 2B**). The bilobate shape is highly conserved and has been employed in identification of grass species (Lu and Liu, 2003; Gallego and Distel, 2004; Fahmy, 2008). The costal region on adaxial surface of S. pumila showed

three structural variants of the bilobate phytoliths, (III, V, and VI) out of a total eight variants of bilobates recorded from different parts of Setaria species (Supplementary Table 3 and **Figure 2B**). The bilobate and nodular bilobate type of phytoliths with each lobate pair separated by silica cork cells have been reported in the costal region of S. pumila (Sharma and Kaur, 1983; Ishtiaq et al., 2001; Shaheen et al., 2011). However, these authors did not report structural variations within the bilobates as recorded in the present investigations.

S. verticillata showed an axial row of phytoliths comprising of 3–6 bilobates, a cross and a nodular bilobate flanked by prickle hairs, with each phytolith pair separated by silica cork cells in the costal region (**Figures 2Ca–d**). The costal region on adaxial surfaces of S. verticillata had only two structural variants of bilobate phytoliths (VII and IV as compared to three variants in S. pumila (Supplementary Table 3). Shaheen et al. (2011)reported bilobates on adaxial surfaces of the costal region of S. verticillata. However, this work made no mention of the presence of the nodular bilobate type of phytolith in the costal region on adaxial surfaces. The intercostal regions lacked silica cells and prickle hairs but showed the presence of long hairs (**Figures 2Ce,h**) in contrast to S. pumila in which prickle hairs were present and long hairs were completely absent. The presence of cystoliths (calcium oxalate crystals) on the adaxial epidermal surface of S. verticillata

FIGURE 3 | (A) Undulated patterns and ornamentations on the epidermal long cells of *Setaria pumila* (Poir.) Roem. & Schult. synflorescence. Columellate extensions (a–c); η-I type (d–g); granulate (h) and n-I type (i) type of epidermal undulation patterns. (B) Undulated patterns and ornamentations on the epidermal long cells of *Setaria verticillata* (L.) P. Beauv. synflorescence. η-I (a–c); -I (d), 3-I (e,f) 3-II (g,h), 3-III (i) and n-I (j), and n-II (k) type of epidermal undulation patterns. (C) Undulated patterns and ornamentations on the epidermal long cells of *Setaria viridis* (L.) P. Beauv. synflorescence. -I (a,b,g), –II with papillate structures (encircled) (c–e) and granulate epidermal extensions (f).

leaf as quadrihedrons and hexahedrons has emerged as the most important diagnostic feature of the species. The cystoliths showed greater concentration in costal rather than the intercostal regions (**Figures 2Cf,g**). Even though cystoliths have been reported from leaf epidermis in several other grass species (Benecke, 1903; Sato, 1968; Dayanandan et al., 1977; Sato and Shibata, 1981; Lerseten, 1983; Prasad et al., 2005), the present study is the first report of cystoliths for the genus Setaria. The abaxial surface in the costal regions showed a single axial row of bilobate and nodular bilobate types of phytoliths (**Figure 2Cj**). The bilobate class revealed two structural variants (III and IV). The intercostal region showed 1–2 stomatal files of high domed stomata (**Figure 2Ck**). The margins on abaxial surface showed the presence of prickle hairs with base lengths greater than the barb (**Figure 2Ci**).

S. viridis showed, on the adaxial surfaces 1-4 axial rows of bilobate and nodular bilobate type of phytoliths in the costal region with each phytolith pair flanked by silica cork cells (**Figures 2Da,b**). Three variants of bilobate phytoliths were present in S. viridis (II, IV, and V). These phytoliths are flanked by a pair of prickle hairs in the costal region. The intercostal region showed prickle hairs between each epidermal long cell pair. In contrast to S. pumila and S. verticillata, the intercostal regions of S. viridis occasionally showed a single row of phytoliths. S. viridis abaxial leaf surface had 1–3 rows of bilobate phytoliths with occasional nodular bilobate types in the costal region (**Figures 2Dc,d**). The bilobate class included two structural variants (V and VI). The species had small one celled prickle hairs in the intercostal regions in addition to prickle hairs with bases smaller than the barb on the leaf margins (**Figures 2De,f**).

## Epidermal Ornamentation and Undulation Patterns

The ornamentation and undulation patterns of epidermal long cells of synfloresence bracts have been put into three categories viz., -undulated, η-undulated, and n-undulated ornamentations (Lu et al., 2009). We propose another undulated ornamentation which can be represented by the symbol '3' (Greek-Lamda) and further categorize it into three subtypes: 3- I, 3- II, and 3- III. The 3-type of undulations were classified on the width of the base and the length of the lateral extensions. If the width of the base and its length was nearly equal, it was put as 3-I type and if the length was three times the base of lateral extensions, it was recognized as 3-II type. Similarly, if the length of the lateral extension is more than thrice the width of the base of the extension, it was put as 3-III. The and η-undulated ornamentations are generally present on the lemmas and palea and have been further put into subtypes based on the degree of undulations as -I, -II & -III and η-I, η-II, η-III respectively (Lu et al., 2009). n-undulated ornamentations were reported on the margins of lemmas and paleas (Zhang et al., 2011).

S. pumila showed columellate extensions of epidermal cells (**Figures 3Aa–c**) whereas they were absent in the other two species. In addition to columellate extensions, S. pumila showed the presence of η-I (**Figures 3Ad–g**), granulate (**Figures 3Ah**), and n-I (**Figures 3Ai**) type of undulated ornamentations. These types of ornamentations have been reported in some other species of Setaria including S. italica, (Lu et al., 2009; Zhang et al., 2011). In our sample, S. verticillata showed the presence of η-I (**Figures 3Ba–c**) -I (**Figures 3Bd**), 3-I (**Figures 3Be,f**) 3 –II (**Figures 3Bg,h**), 3 –III (**Figure 3Bi**) and n-I (**Figure 3Bj**) and n-II (**Figure 3Bk**). S. viridis showed -I (**Figures 3C a,b,g**), – II (**Figures 3Cc–e**) and granulate (**Figures 3Cf,g**). The epidermal elements also showed the presence of papillae on the surface of lemmas. Kealhofer et al. (2015) also reported the similar ( –II) type of epidermal undulated ornamentations in S. viridis.

## Phytolith Morphotypes

In the present study, a cumulative total of 58 phytolith morphotypes were identified with an individual distribution of 38 in S. pumila, 39 in S. verticillata, and 41 in S. viridis. These morphotypes were grouped into nine broad categories namely, bulliform cells, epidermal elements, hairs, long cells, short cells, tabular types, globular types, blocky types, and tracheids (**Table 1** and **Figures 4–6**, **7A–C**). The first seven categories are known to have their origin in the epidermis, blocky types in the endodermis and the last one in the vascular tissue system (Twiss et al., 1969; Lu and Liu, 2003).

Except for the blocky and globular types, phytolith morphotypes have been well reported in family Poaceae (Twiss et al., 1969; Bonnett, 1972; Prychid et al., 2004). Both

FIGURE 4 | Phytolith morphotypes from various parts of *Setaria pumila* (Poir.) Roem. & Schult. (A) (Root): Bilobate class I (a); Bilobate class VI (b,c); Bilobate class V (d,e); Polylobate (f); Nodular bilobate (g,h); Globular *(Continued)*

FIGURE 4 | polyhedral (i,j), Blocky irregular (k,l); Oblong (m); Trapezoid (n,o); Rectangular (p,q); Cuneiform bulliform (r); Tabular irregular (s–v); Scutiform (w). (B) (Culm): Blocky polyhedral (a,b); Trapezoid (c); Globular polyhedral (d); Echinate elongate (e,f); Sinuate elongate with concave ends (g); Tabular irregular (h); Sinuate elongate (i–k); Smooth elongate (l,m); Elongate irregular (n,o). (C) (Leaf): Tabular simple (a); Blocky irregular (b); Rectangular (c); Globular granulate (d,e); Blocky polyhedral (f); Parrellepedal bulliform cells (g); Trapezoid (h–l); Globular polyhedral (m); Clavate (n,o); Cuboid (p,q); Scutiform (r,s); Ovate (t,u); Cylindrical (v,w); Smooth elongate (x). (D) (synflorescence): Macrohairs (a–e); Cuneiform bulliform (f); Globular polyhedral (g); Epidermal elements (h); Echinate elongate (i–k); Clavate (i); Trapezoid (m,n), Tracheid (o), Blocky polyhedral (p,q); Elongate irregular (r,s); Horned tower (t,u); Blocky irregular (v); Globular granulate (w); Smooth elongate (x,y); Facetate elongate (z); Sinuate elongate (aa); Columellate elongate (ab); Stomata (ac); Prickle hair (ad); Bilobate class I (ae); Plates (af,ag).

blocky and globular (spherical) morphotypes are considered to be characteristic of forest trees (Runge, 1999). Even within monocots, spinulose to tabular spheres are typically associated with the arborescent (palm) family, Arecaceae (Kealhofer and Piperno, 1998) wherein these types are produced in great abundance (Albert et al., 2007). While the blocky morphotype has been reported in some grasses (Wang and Lu, 1993; Carnelli et al., 2004), we have not come across any report of the globular type in the family. However, in view of the reports of the globular type from the commelinid families, Zingiberaceae, Marantaceae, and Strelitziaceae (Tomlinson, 1956, 1961; Kealhofer and Piperno, 1998; Brilhante de Albuquerque et al., 2013) and the non-commelinid family Orchidaceae (Sandoval-Zapotitla et al., 2010), the recovery of the globular morphotype in Poaceae during the present studies was not entirely unexpected.

The present study marks a significant addition to information on phytolith profiles particularly from underground (root) parts of three species of genus Setaria. Most of the previous studies in grasses have documented phytoliths from above ground parts, mainly the leaf (Tomlinson, 1969; Twiss et al., 1969; Bonnett, 1972; Krishnan et al., 2000; Lu and Liu, 2003; Prychid et al., 2004; Fahmy, 2008; Barboni and Bremond, 2009; Rudall et al., 2014; Shakoor et al., 2014; Jattisha and Sabu, 2015). Only a limited number of reports are available on phytolith analysis of roots of plant species (Ezell-Chandler et al., 2006; Das et al., 2014; Soukup et al., 2014; Shakoor et al., 2016).

A comparison among the three congeneric species of Setaria revealed that some of the phytolith morphotypes were shared by all the three species while some others were restricted to only one or two of the three species in the present study (**Table 1**). At one extreme were some morphotypes which had a ubiquity value of unity, i.e. they occur in at least one plant part in all the three species. For example, bilobate class V, blocky irregular and blocky polyhedral were present at least in one plant part in all the three species and hence carried a ubiquity value of unity (**Table 1**). Such morphotypes have the lowest diagnostic value. Similarly, phytolith morphotypes with a ubiquity value of 0.66 indicates their presence in two out of the three species. These types could be utilized for taxonomic diagnosis and demarcation of pairs of species in the present sample from the one lacking these morphotypes (**Table 1**). For example, bilobate class III,

FIGURE 5 | Phytolith morphotypes from various parts of *Setaria verticillata* (L.) P.Beauv. (A) (Root): Cuneiform bulliform (a,b); Tabular simple (c); Blocky irregular (d–f); Cuboid (g); Globular echinate (h); Smooth elongate (i); Bilobate class VII (j); Blocky polyhedral (k–m); Crescent moon (n,o); Parrellepedal bulliform cells (p,q); Rectangular (r,s); Globular polyhedral (t–v); Elongate with concave ends (w); Cylindrical (x); Triangular (y); Trapezoid (z). (B) (Culm): Sinuate elongate (a); Ovate (b,c); Blocky crenate (d,e); Globular psilate (f); *(Continued)*

FIGURE 5 | Trapezoid (g,h); Clavate (i); Scutiform (j,k); Blocky irregular (l–n); Blocky polyhedral (o); Cuboid (p); Smooth elongate (q); Half-moon (r). (C) (Leaf): Globular granulate (a,b); Globular polyhedral (c); Rectangular (d,e); Blocky polyhedral (f); Elongate irregular (g,h); Horned tower (i–k); Tabular irregular (l); Trapezoid (m–o); Scutiform (p); Bilobate class IV (q); Bilobate class VII (r); Nodular bilobate (s); Cuneiform bulliform (t–v); Blocky irregular (w,x). (D) (Synflorescence): Epidermal element with columellate extensions (a); Cuneiform bulliform (b,c); Blocky polyhedral (d) Smooth elongate (e,f); Rectangular (g); Cuboid (h); Clavate (i); Acicular (j,k); Polylobate irregular (l); Rondel (m–o); Cross (p); Globular polyhedral (q,r); Scutiform (s–u); Columellate elongate (v); Echinate elongate (w); Trapezoid (x–z).

columellate elongate, cross, horned tower, oblong and tabular simple demarcate S. pumila and S. verticillata from S. viridis in the present sample. Similar is the case with other morphotypes with ubiquity value of 0.66 between other pairs of species within the three congenerics (**Table 1**).

Phytolith morphotypes with ubiquity value of 0.33 indicates their presence in only one of the three studied species. Within the limited context of the present work, these phytoliths marked the individual species from the other two and helped in their taxonomic demarcation. For example, bilobate class I (**Figures 4Aa**,**Da,e**) from roots and synflorescences, polylobate (**Figure 4Af**) from roots, sinuate elongate with concave ends (**Figure 4Bg**) from culms, stomatas (**Figures 4Dac**) facetate elongate (**Figure 4Dz**), and tracheids (**Figure 4Do**) from synflorescences have ubiquity values of 0.33 and diagnose S. pumila from the other pair of species (**Table 1**). The "marker" phytolith morphotypes yielded by various parts of S. verticillata included blocky crenate (**Figures 5Bd,e**) from culms, crescent moon (**Figures 5An,o**) and elongate with concave ends (**Figure 5Aw**) from roots, half-moon (**Figure 5Br**), epidermal element with columellate extensions (**Figure 5Da**) and polylobate irregular (**Figure 5Dl**) from the synflorescences (**Table 1**). Similarly, the "marker" morphotypes from S. viridis included bilobate class II (**Figures 6D-h-j**), epidermal element with short silica cells and stomata, epidermal papillate, and prickly elongate (**Figures 7Cw,y,ad,ac**) from synflorescences, bilobate class VIII (**Figure 6Bv**) from culms, tabular polyhedral from the culms and leaves (**Figures 6Ce, 7Cl**) and carinate (**Figures 6Bt,u, 7Cx**) from the culms and the synflorescences (**Table 1**). What adds to the diagnostic significance of these morphotypes is that these were recovered from all the plant parts ranging from roots to synflorescences. Hence, the present study reiterates the necessity and significance of analysis of phytoliths from all the underground and aerial plant parts before utilizing them for taxonomic diagnosis as suggested in some earlier studies as well (Kealhofer et al., 2015; Shakoor et al., 2016). Here, it may be emphasized that these morphotypes "mark" the individual species only from the other two in the present study. An unqualified use of the term marker phytolith for the types recovered from species in the present sample would be an overstatement implying that these types diagnose these species individually from rest of the species of the foxtail grass genus Setaria. The full potential of phytolith types for interspecific diagnosis can only be realized after phytolith analysis of the entire

FIGURE 6 | Phytolith morphotypes from various parts of *Setaria viridis* (L.) P. Beauv. (A) (Root): Blocky polyhedral (a,b); Triangular (c,d); Rectangular (e–g); Blocky irregular (h); Trapezoid (i–k); cuboid (l,m); Globular psilate (n); Globular granulate (o–q); Scutiform (r,s); Parrellepedal bulliform cells (t–v); Oblong (w). (B) (Culm): Globular polyhedral (a–e); plates (f,g); Triangular (h); Cuneiform bulliform (i); Rondel (j); Smooth elongate (k); Rectangular (l–n); Blocky irregular (o,p); Blocky polyhedral (q,r); Globular echinate (s); Carinate (t,u); *(Continued)*

FIGURE 6 | Bilobate class VIII (v); Clavate (w,x); Trapezoid (y–z1); Cuboid (z2, z3); Elongate irregular (z4). (C) (Leaf): Blocky irregular (a–d); Tabular polyhedral (e); Clavate (f); Echinate elongate (g,h); Sinuate elongate (i); Smooth elongate (j); Pickle hair (k); Globular granulate (l–o); Scutiform (p–s); Cuboid (t–v); Trapezoid (w–y); Ovate (z,z1); Bilobates class VII (z2) Bilobates class VIII (z3); Plates (z4,z5). (D) (Synflorescence): Cuneiform bulliform (a–c); Blocky polyhedral (d–g); Bilobate class II (h–j); Blocky irregular (k,l); Parrellepedal bulliform cells (m–o); Globular polyhedral (p–r); Ovate (s,t); Smooth elongate (u,v); Acicular (w); Prickle hair (x); Globular psilate (y,z); Cylindrical (z1).

genus. Similarly, "marker" types for the genus and suprageneric levels can only be identified by profiling all the taxa included in these ranks.

SEM of phytoliths of the three congenerics of Setaria revealed subtle differences in topography of phytolith morphotypes which was not clear in light microscopy (**Figures 7A–C**). SEM has helped to distinguish and segregate particular phytolith morphotypes into sub-types. For example, the globular morphotype was further resolved into globular crenate (**Figure 7Cr**) globular granulate (**Figures 7Aa,d,Bi**), globular echinate (**Figures 7Al,Co,p**), globular polyhedral (**Figures 7Aq,r,Bh,Ca,h,z**), and globular psilate (**Figures 7Be,Ck**) morphotypes based on the type and degree of surface ornamentations. Similarly, the tabular morphotype was segregated into tabular polyhedral (**Figure 7Cl**), tabular irregular (**Figure 7Cq**) and tabular polyhedral (**Figure 7Ct**). Earlier studies grouped all broad and multisided structures into trapezoid category (Piperno, 1988, 2001; Pearsall, 2000). But recent studies have distinguished two more categories within the trapezoid morphotype viz., blocky polyhedral and blocky irregular morphotypes (Traoré et al., 2014). We have also recognized blocky irregular (**Figures 7Ab,o,Bj,r,Cb**) and blocky polyhedral (**Figures 7Af,k,Ba,b,Cc,d,n,ab**) morphotypes. Additionally, SEM has revealed the interlocking patterns between epidermal elements (**Figures 7Cv,w,y**). It has also revealed the presence of silica short cells embedded with the epidermal elements (**Figure 7Cw**).Thus, SEM has been employed as an effective tool in elucidation of ultrastructural features of phytolith morphotypes and their classification into subtypes that have further helped in demarcation of the grass species under reference.

The coefficient of association of phytolith morphotypes based on Pearson's association revealed highest association among overground parts (Supplementary Table 4). The strongest association was found among the leaf and synflorescence of S. pumila and S. viridis whereas S. verticillata showed significantly lower values of association (Supplementary Table 4). The highest values of coefficient of association between leaf and synflorescence could be attributed to the anatomical similarities of leaf and synflorescence bracts that produce phytoliths. Similarly, insignificant association between the underground and overground parts could be explained by the anatomical, histological and physiological differences among these plant parts and hence the phytolith morphotypes produced by them.

Clustering of species on the presence/ absence data of bilobate classes, using Jaccard's similarity index was carried out. S. pumila belongs to one clade of Setaria whereas the other two species belong to the other clade (Doust and Kellogg, 2002). A similar trend was observed in clusters from the totality of morphotypes (**Figure 8**). S. pumila stood apart from the other two species as it has 33% similarity of phytolith profile with S. verticillata and 28.57% with S. viridis. However, S. verticillata and S. viridis showed 42.85% similarity and were grouped together (**Figure 8**).

## Frequency Distributions and Morphometric Measurements

Several studies in the past have utilized data on morphotypes for taxonomic characterization and identification of plant species (Twiss et al., 1969; Lau et al., 1978; Hodson and Sangster, 1988; Ollendorf et al., 1988; Whang et al., 1998; Krishnan et al., 2000; Ponzi and Pizzolongo, 2003; Piperno, 2006). However, recent studies have enlarged the scope of phytolith research by including data on morphometric measurements and frequency distributions of phytolith morphotypes for taxonomic demarcation of species down the taxonomic hierarchy from family, genus, and species levels (Strömberg, 2009; Jattisha and Sabu, 2012; Tripathi et al., 2013; Szabo et al., 2014; Shakoor et al., 2016; Ball et al., 2017; Out and Madella, 2017).

Setaria spp. showed considerable differences in the frequency distribution of various phytolith morphotypes (**Figure 9**). The most frequent in all the three species were the trapezoids. However, they differ significantly within and between the species with 19.47% frequency in S. pumila, 14.38% in S. verticillata and 7.91% in S. viridis (p ≤ 0.05; χ 2 -test). Acicular morphotypes present in both S. verticillata and S. viridis differed many fold in terms of their percentage frequency with 15.17% in the former and 2.18% in the later species. Bilobate classes also differ significantly with respect to frequency distributions. For example, bilobate class III were present in the leaves of S. pumila and S. verticillata with highly variable percentage frequency values of 9.44% and 3.10% respectively (p ≤ 0.05; χ 2 -test). Similarly, bilobate class IV occurred in the leaves of S. verticillata and S. viridis with a percentage frequency of 8.10 and 4.39% respectively (p ≤ 0.05; χ 2 -test). Similarly, other phytolith morphotypes revealed significant differences in percentage frequency providing a definite clue that frequency of occurrence of phytolith morphotypes provides an additional evidence for taxonomic characterization apart from qualitative differences in phytolith types (**Figure 9**).

Apart from frequency distributions, morphometric data on size dimensions and shape descriptors of morphotypes also revealed significant differences between the species (Supplementary Tables 5A–C). In the present analysis, we included data on size parameters (length, width, area and perimeter) and one shape descriptor, the aspect ratio. Further, length and width of the shank of bilobate types have been employed as additional characteristics to classify the bilobates into various subtypes in order to further supplement taxonomic diagnosis of species (Supplementary Table 3). The use of multivariate statistical approaches like principal component

(a) Blocky irregular (b); Bilobate class V (c). Culm: Globular granulate (d) Cuneiform bulliform (e). Blocky polyhedral (f); Trapezoid (g); Leaf: Trapezoid (h,i); *(Continued)*

FIGURE 7 | Blocky irregular (j); Blocky polyhedral (k); Globular echinate (l); Elonagate irregular (m). Synflorescence: Prickle hair (n); Blocky irregular (o); Epidermal element with undulated ridges (p); Globular polyhedral (q,r); Trapezoid (s); Prickle hair (t). (B) *Setaria verticillata* (L.) P.Beauv. Root: Blocky polyhedral (a,b); Cuneiform bulliform (c); Trapezoid (d); Globular psilate (e). Culm: Scutiform (f); Elongate irregular (g). Leaf: Globular polyhedral (h); Globular granulate (i); Blocky irregular (j); Blocky polyhedral (k). Synflorescence: Echinate elongate (l); Crenate elongate (m); Columellate elongate (n); Blocky papillate (o); Trapezoid (p); Acicular (q); Blocky irregular (r); Rugose elongate (s). (C) *Setaria viridis* (L.) P. Beauv. Root: Globular polyhedral (a); Blocky irregular (b); Blocky polyhedral (c,d); Trapezoid (e,f). Culm: Trapezoid (g). Globular polyhedral (h); Epidermal element with undulated ridge (i); Blocky irregular (j); Globular psilate (k); Tabular polyhedral (l). Leaf: Trapezoid (m); Blocky polyhedral (n); Globular echinate (o,p); Tabular irregular (q); Globular crenate (r); Bilobate class V (s); Tabular polyhedral (t). Synflorescence: Trapezoid (u); Epidermal element (v); Epidermal element with silica short cells & stomata (w,y); Carinate (x); Globular polyhedral (z); Triangular (aa); Blocky polyhedral (ab); Prickly elongate (ac); Epidermal papillate (ad); Scutiform (ae).

analysis has been recommended and employed in earlier studies for taxonomic demarcation of species (Benvenuto et al., 2015; Pearsall, 2015; Ball et al., 2016).

Joint PCA analysis of morphometric parameters of phytoliths from different parts of the three species led to overcrowding of the data and did not help in diagnosis of species. However, PCA analysis of morphometric parameters of phytoliths from different parts individually proved useful in taxonomic demarcation of the species. PCA analysis of root phytoliths clearly separated the three species on the basis of surface areas of different morphotypes (Supplementary Figures 1a,b). S. pumila was demarcated on the basis of surface areas of blocky irregular and tabular irregular, S. verticillata by blocky polyhedral and S. viridis by area of trapezoid morphotypes as revealed by PCA results of component 1 and 2 (Supplementary Figure 1a). However, the PCA plot between component 1 and 3 revealed more clear demarcation than obtained from components 1 and 2 (Supplementary Figures 1b). PCA analysis of phytolith morphotypes of culm of the three species brought about the taxonomic demarcation of species on the basis of the area of smooth elongates for S. verticillata, and tabular irregular for S. pumila (Supplementary Figure 2). Similarly, PCA analysis of leaf and synflorescence phytolith morphotypes of the three species lent further support to taxonomic analysis of the three species of Setaria (Supplementary Figures 3, 4).

#### Transmission Electron Microscopy

TEM allows visualization and microstructural examination through a combination of high magnification and resolution. It helped to distinguish various physical states including amorphous from the crystalline and helped to study their atomic planes, (columns of atoms in crystals). TEM images of phytolith morphotypes from leaves and synflorescences of S. pumila and S. verticillata showed macroscopic clusters and agglomerates of silica that were not distinguished into particles at nanoscale regime (**Figures 10Aa–d,Ba–d,Ca–c,Da,b**). However, phytoliths from leaves and synflorescences of S. viridis revealed silica particles of spherical and cubic morphologies of nanoscale regime and were clustered (**Figures 10Ea,b,Fa,b**). The presence of spherical and cubic nanoparticle clusters in the latter species clearly demarcates it from the other two congenerics. Gonzalez-Espindola et al. (2014) reported clusters and agglomerates of phytoliths as well as spherical particles of nanoscale regime from the leaves of the grass species Stenotaphrum secundatum. Palanivelu et al. (2014) reported agglomerated particles of silica nanoparticles from rice hulls collected from different geographical locations.

High resolution transmission electron microscopy (HRTEM) revealed the presence of ordered interplanar atomic layers of Si–O, Si–O–Si bonds in all the species except in the leaf phytoliths of S. verticillata (**Figures 10Ae,Be,Dc,Ec,Fc**), which did not possess regular ordering of local clusters of Si–O and the silica bodies were completely amorphous (**Figures 10Bd,e**). HRTEM analysis of phytoliths from leaves of S. pumila and S. viridis revealed microcrystalline structures with an interplanar distance (d-spacing) of 0.1 nm which was indicative of the presence of tetragonal cristobalite polymorph of silica (**Figures 10Ae,Fc**). Similarly, silica from the synflorescences of all the three species revealed microcrystalline structures with a difference of interplanar distance which was 0.08 nm for S. pumila and S. viridis and 0.083 nm for S. verticillata. These distances correspond to tetragonal stishovite polymorphs (**Figures 10Be,Dc,Fc**) whose formation was favored by the presence of localized crystallization centers such as extraneous cations dispersed throughout

the siliceous phytoliths (Mann and Perry, 1986). The link got substantiated and explained by the presence of cations like Al2+, Ba2+, Fe2+,Ca2+,Cu2+, Mg2+, Na+, and K<sup>+</sup> as

revealed by SEM-EDX analysis of phytoliths (Supplementary Table 6).

Selected area electron diffraction (SAED) reveals the chemical composition of different mineral phases by their different patterns generated by the impact of X-rays and fast moving electrons. Phytoliths from the leaves of S. pumila and S. viridis revealed well defined single crystalline lattices that could be resolved to hexagonal and orthorhombic lattices of SiO2. (**Figures 10Af,Ed**) that were continuous and unbroken in the former but lacked grain boundaries in the latter (cf. Reid et al., 2011). The amorphous structure of phytoliths was revealed by an absence of SAED patterns (**Figure 10C**). Similarly, phytoliths from synflorescences of S. pumila revealed single crystal lattices corresponding to SiO<sup>2</sup> (with cubic, tetragonal and orthorhombic morphologies) and zeolites with a cubic lattice system (**Figure 10B**). The SAED pattern of synflorescences of S. verticillata and S. viridis showed well defined rings confirming their polycrystalline nature (**Figures 10Dd,Fd**). The SAED patterns of phytoliths from S. verticillata correspond to orthorhombic ferrierite and tridymite and anorthic SiO<sup>2</sup> polymorphs. Similarly, SAED patterns of silica from S. viridis correspond to orthorhombic ferrierite and tridymite (**Figure 10Fd**). Apart from taxonomic resolution, the formation of nanoscale silica particles during dry ashing of the plant material has applications in nanotechnology, particularly synthesis of metal silicates (Neethirajan et al., 2009; Qadri et al., 2015).

#### Biosilica Content

Grasses deposit silica in varied amounts in different plant parts ranging from 1 to 11% (Jones and Handreck, 1967). In the present study, the three species of Setaria revealed considerable differences in terms of ash and silica content in their parts (Supplementary Figure 5). The species showed higher values of ash and silica in their foliar parts with 21.06% ash and 11.62% silica in S. pumila, 19.87% ash and 9.23% silica in S. verticillata and 16.43% ash and 6.24% silica in S. viridis. The ash and silica content in other parts were in the order of, synflorescences>roots>culms. Higher amounts of silica in the leaves and synflorescences of grasses are well reported (Lanning and Eleuterius, 1981, 1987, 1989). The differential amounts of silica within and between different parts of the plant body have been correlated to the differences in the targeted cellular sites of silicification. For example, in roots endodermal cells have been proved to be the usual targets of silicification while in the aerial parts of the plant body different epidermal cells and associated structures as well as the cells of vascular bundles, and the spaces between the cortical cells are believed to be the targeted sites of silicification (Kumar et al., 2017; Kumar and Elbaum, 2018).

Our results indicated significantly higher silica content in the leaves of the presently studied Setaria species as compared to some other species of the genus. For example a much lower amount (6.06%) was reported in S. magna Griseb. (Hodson et al. (1982)) and other members of tribe Paniceae (1.04% for Panicum reptans L., 3.7% for Digitaria macroblephara (Hack.) Paoli) and related tribes (1.34% for Imperata cylindrical (L.) Raeusch. and

HRTEM (f) SAED patterns (Figures in parenthesis indicate hkl values and for description of alphabets refer Supplementary Table 8) and Synflorescence (B) (a–d) (Clusters and agglomerates of silica (e) HRTEM (f) SAED patterns (Figures in parenthesis indicate hkl values and for description of alphabets refer Supplementary Table 8). (C,D) *Setaria verticillata* (L.) P. Beauv. Leaf (C) (a–c) Clusters and agglomerates of silica (d–e) HRTEM (f) SAED patterns and Synflorescence (D) (a–b) Clusters and agglomerates of silica (c) HRTEM (d) SAED patterns (Figures in parenthesis indicate hkl values and and for description of alphabets refer Supplementary Table 8). (E,F) *Setaria viridis* (L.) P. Beauv. Leaf (E) (a,b) Spherical silica particles (c) HRTEM (d) SAED patterns (Figures in parenthesis indicate hkl values and for description of alphabets refer Supplementary Table 8) and Synflorescence (F) (a,b) Cubic and agglomerated silica (c) HRTEM (d) SAED patterns (Figures in parenthesis indicate hkl values and for description of alphabets refer Supplementary Table 8).

2.7% for Themeda triandra Forssk.) of the subfamily Panicoideae (Lanning and Eleuterius, 1989; Quigley and Anderson, 2014).

The variation in silicification rates in underground and aerial parts (particularly leaf and synflorescence bracts) are known to be controlled by a multitude of extrinsic (availability of silica and water in the soil) and intrinsic factors including the extent and nature of silicon transporters and channels, sink strength and the functional anatomy of various plant parts (Motomura et al., 2002; Ma and Yamaji, 2006; Honaine and Osterrieth, 2011). Besides these factors of control, higher levels of silicification in leaf laminae and the synflorescence bracts of aerial plant parts have been correlated with higher evapotranspiration rates in these parts. Once absorbed, silica is transported via xylem to various plant parts through the transpiration stream. As water evaporates during transpiration, silicic acid solutes are progressively concentrated resulting in super-saturated concentrations of Si(OH)<sup>4</sup> and deposition in tissues as amorphous silica in the form of phytoliths; the extent of supersaturation being controlled by the concentration of silicic acid in soil water (Jones and Handreck, 1965; Rosen and Weiner, 1994; Raven, 2003; Exley, 2015).

#### Elemental Composition

Though mainly siliceous in nature, phytoliths deposit many other elements in addition to silicon and oxygen in varying proportions during the course of their development (Shakoor et al., 2016). The elemental composition of phytolith morphotypes is reported to be controlled by species characteristics, geochemistry and prevailing environmental conditions (Bujan, 2013; Kamenik et al., 2013; Hodson, 2016). However, the elemental composition of phytoliths in association with their morphology has proved useful for taxonomic diagnosis of species. Elemental composition has been shown to be stable enough to serve as definitive evidence of palaeo-environments by providing clues to the type of the soil in which a given species grew (Kamenik et al., 2013; Hodson, 2016).

The presence of different elements in phytolith morphotypes of the present samples reflect the availability of elements in the soil (Supplementary Table 7). However, the present study revealed some species-specific elements as well. The elemental composition of rhizospheric soil samples from the three sampling sites (**Figure 1**) revealed a cumulative number of fourteen (14) elements (aluminum (Al), carbon (C), calcium (Ca), copper (Cu), iron (Fe), magnesium (Mg), sodium (Na), phosphorous (P), potassium (K), oxygen (O), silicon (Si), sulfur (S), titanium (Ti), and zinc (Zn). Species wise characterization of the soil revealed 11 elements (Al, Ca, C, Fe, Mg, O, K, Si, Na, Ti, and Zn excluding Cu, P, and S from the cumulative list) from sampling sites of S. pumila, 10 elements (excluding Cu, P, S, and Zn) from the soil supporting S. verticillata whereas the rhizospheric soil samples from the S. viridis sampling site revealed 12 elements (excluding Na and Zn from the cumulative list).

SEM-EDX analysis of the phytolith morphotypes from different parts of the three species revealed a cumulative total of 16 elements with 12 in S. pumila 14 in S. verticillata and 11 in S. viridis (**Figures 11A–C** and Supplementary Table 6). A comparison of elemental composition data of soil samples and phytolith morphotypes revealed that soil geochemistry controls the composition of phytoliths. However, some elements were present in phytolith samples in traces but were not detected in soil samples. For example chlorine (Cl) was detected in phytoliths from all parts of S. pumila and S. verticillata. Similarly barium (Ba), copper (Cu), and sulfur (S) were detected in the latter named species and rubidium (Rb) and sodium (Na) in S. viridis. This unexpected difference in elemental composition of soil samples and phytoliths could be attributed to some sort of "accumulation" of these elements in the living cells producing phytoliths. Most importantly, some elements were unique to one or the other species: barium (Ba), phosphorous (S), and sulfur (S) were detected in S. verticillata and rubidium (Rb) in S. viridis Principal Component Analysis (PCA) of elemental composition data from different parts of the three congeneric

FIGURE 12 | PC analysis of elemental composition of data of phytolith morphotypes of *Setaria* spp. SPR, Setaria pumila root; SPC, *Setaria pumila* culm; SPL, *Setaria pumila* leaf; SPS, *Setaria pumila* synflorescence; SVCR, *Setaria verticillata* root; SVCC, *Setaria verticillata* culm; SVCL, *Setaria verticillata* leaf; SVCS, *Setaria verticillata* synflorescence; SVR, *Setaria viridis* root; SVC, *Setaria viridis* culm; SVL, *Setaria viridis* leaf; SVS, *Setaria viridis* synflorescence.

species led to demarcation of S.'verticillata from the other two congenerics with the first two components explaining 97.12% (85.12% component 1 + 15% component 2) variation in the data set (**Figure 12**). The present study has revealed higher atomic and weight percentage values for carbon (C), oxygen (O), and silicon (Si) in phytoliths whereas other elements were present in considerably lesser amounts. The occlusion of carbon in phytoliths has been compared to its sequestration in cellulose and lignin (Parr and Sullivan, 2005). However, EDX analysis revealed the element form of carbon in phytoliths rather than the organic form.

Biomineralization of silica in plants is known to ameliorate metal (Al, Cu, Fe) and salinity stress (Okuda and Takahashi, 1962; Matoh et al., 1986; Cocker et al., 1998; Yeo et al., 1999). The deposition of metals like Al, Cu, Fe in phytoliths possibly alleviates the toxicity associated with these elements. Similarly, salinity stress seemed to be ameliorated by the bioaccumulation of silicophytoliths as revealed by K, Ca, and Mg in phytoliths (Anala and Nambisan, 2015). The segregation and compartmentalization of phytoliths embodying Si and other minerals has made isolation of these elements possible (Raven, 1983). Thus, deposition and immobilization of these toxic elements in the silicification process may be a strategy of plant species to get rid of these materials via their transport along the transpirational route and final occlusion in phytoliths.

#### X-Ray Diffraction Analysis

Silica exists in diverse polymorphs and sub-morphs; crystalline forms include alpha and beta-quartz, cristobalite, tridymite, coesite, keatite, and stishovite. Amorphous silica has the same composition as SiO<sup>2</sup> but has a random structure of the crystal lattice. The presence of both types in our specimens can be attributed to the transformation of amorphous silica into different crystalline polymorphs during dry ashing of the material (Holm et al., 1967).

Powder diffractograms of phytoliths isolated from underground and aerial parts of Setaria showed peaks characteristic of different crystalline polymorphic phases (**Figures 13A–C**). The most frequent phases were silicon dioxide (SiO2) from all the parts of the species (except the leaf of S. verticillata) and quartz (except in leaves and synflorescences of S. verticillata. The other phases present in all the three species (at least in one of the parts) included zeolites, tridymite, stishovite, ferririte, coesite and cristobalite (**Figures 13A–C** and Supplementary Table 8). However, stishovite was diagnostic of roots and leaves of S. pumila whereas ferririte was restricted only to the roots of S. viridis, suggesting a role in taxonomic diagnosis as already reported for some of the grass species (Gonzalez-Espindola et al., 2010, 2014; Shakoor et al., 2016).

The polymorphic phases have been known to have an identical chemical composition (SiO2) but different physical properties and lattice symmetries. They show distinct lattice systems ranging from anorthic (triclinic), through monoclinic, orthorhombic, hexagonal, cubic, and tetragonal. The present studies lend further credence to the existence of polymorphic silica in plants (Ollendorf et al., 1988; Piperno, 1988, 2006; Lu and Liu, 2003; Lu et al., 2009; Zhang et al., 2011; Szabo et al., 2015; Ge et al., 2016). The diffractogram of phytoliths of S. viridis (root) and S. pumila (root and culm) showed a unique peak corresponding to ferrierite and zeolite respectively. (**Figures 13A,C**). Ferrierite is a zeolite (aluminosilicate mineral) that binds a number of cations viz., Na+, K+, Ca2+, Mg2<sup>+</sup> etc. The presence of these phases can be explained by elemental composition data.

Further, the FTIR analysis revealed a peak of Aluminosilicate minerals in these species, thus supporting our XRD results (**Figures 14A,C**). Earlier, Kow et al. (2014) confirmed the shift from amorphous to crystalline phases of silica in cogon grass (Imperata cylindrica (L.) P. Beauv.) in the presence of potassium (K). Similarly, the presence of other minerals like, Na, Ca, Mg, K etc. in phytoliths from the different parts of these congeneric species could afford a possible explanation (acting as a controlling factor) for the presence of different crystalline polymorphic phases of silica. Such an association is further indicated by the presence of only amorphous silica in the phytoliths from the culms of S. verticillata that harbor a smaller number of elements (only 4 besides C & H) as compared to phytoliths from other parts of the plant body (**Figure 13B** and Supplementary Table 7).

#### FT-IR Spectroscopy

FTIR spectroscopy of silica from different parts of Setaria spp. revealed several peaks that could be assigned to different structural units of silica with varied vibrational degrees of freedom (**Figures 14A–C** and Supplementary Table 9). The peaks between 445.67–472.00 cm−<sup>1</sup> , 637.48–699.54 cm−<sup>1</sup> , 712.70– 801.08 cm−<sup>1</sup> , 1080.06–1094.44 cm−<sup>1</sup> , 1602.17-1616.24 cm−<sup>1</sup> ,

FIGURE 14 | FTIR spectra of phytoliths from different parts of *Setaria sps.* (A) *Setaria pumila* (B) *Setaria verticillata* (C) *Setaria viridis* (for description of peak points, refer to Supplementary Table 9).

1628.50–1641.66 cm−<sup>1</sup> , 2339.32–2366.49 cm−<sup>1</sup> , and 3346.27– 3597.36 cm−<sup>1</sup> present in all the three species (**Figures 14A–C** and Supplementary Table 9) have earlier been variously ascribed to deformation vibration of O–Si–O group (Bertoluzza et al., 1982), symmetrical vibration of Si–O–Si (Gopal et al., 2004), symmetric vibration of Si–O (Brinker et al., 1990), asymmetric vibration of Si–O–Si (Karunakaran et al., 2013; Mourhly et al., 2015), inplane stretching vibration of C-C (Ou and Seddon, 1997), deformation vibration of H–O–H (Socrates, 2001), inplane stretching vibration of Si–C (Socrates, 2001) and O– H/Si–OH bonds (Brinker et al., 1990) bonds. Peaks between 530.39–563.18, 1164.92, 1218.93, 1323.08–1332.72, 1743.21– 1933.14, 2825.52, and 3006.82–3271.05 cm−<sup>1</sup> characteristic of S. verticillata (L.) P. Beauv. (**Figure 14B** and Supplementary Table 9) could be ascribed to stretching vibration of O–Si (SiO<sup>2</sup> defect) (Brinker et al., 1990), asymmetric vibration of Si–O– Si (Duran et al., 1986), inplane stretching of free Si–O (Chmel et al., 2005), symmetric deformation vibration of Si–R (Socrates, 2001), deformation vibration of R (alkyl group), symmetric vibration of C–H (Gunzler and Gremlich, 2002), and stretching vibration of O–H bonds (Brinker et al., 1990). Similarly, peaks at 1463.02 and 1701.84 cm−<sup>1</sup> characteristic of S. viridis (**Figures 14C** and Table 10) could be ascribed to asymmetric and symmetric deformation vibrations of hydrocarbons (–CH3- CH2)– (Watling et al., 2011) and inplane stretching vibrations of

## CONCLUSIONS

Si–C bonds.

Within the context, scope and parameters of reference samples used in the present work, the three congenerics of Setaria revealed a degree of similarity in phytolith profiles but each was found to be well demarcated from the other in the group by "unique" morphotypes and their characteristic assemblages and structures. The bilobate morphotypes aptly illustrate phytolithassisted taxonomic demarcation of the three species. In the present study, eight variants of the bilobate morphotype were recognized on the basis of the length of the shanks (the interconnecting segment between the lobes) and the shape of the outer margin of the lobes. S. pumila showed three of the eight structural variants of the bilobate phytoliths (III, V, and VI) in the costal region on the adaxial surfaces. In the same location, S. viridis also showed three of these variants (II, IV, and V) whereas S. verticillata had only two of them (VII and IV). Thus bilobate classes were found to be highly conserved and useful for identification of grass species. Quadrihedral and hexahedral cystoliths (calcium oxalate crystals) on the adaxial epidermal surfaces of S. verticillata emerged as another diagnostic feature of the species (a first report for the foxtail grass genus Setaria)S. verticillata was also marked out by the presence of a new undulation type, (the 3-lambda with three variants viz. 3-I, 3-II, and 3-III) in the long epidermal cells.

Besides qualitative differences, the present samples of the three species also revealed interspecific variations in frequency distribution and morphometric measurements of various morphotypes. For example, the frequency of trapezoids was significantly different in these species: 19.47% in S. pumila, 14.38% in S. verticillata, and 7.91% in S. viridis (p ≤ 0.05; χ 2 test). Acicular morphotypes were present in both S. verticillata and S. viridis but differed many fold in their percentage frequency (15.17 and 2.18% respectively).

Principal Component Analysis of morphometric parameters of phytoliths from different parts of the plant body proved useful in taxonomic demarcation of the species. PCA of root phytoliths clearly separated the three species on the basis of the surface area of different morphotypes. S. pumila was demarcated on the basis of the surface area of blocky irregular and tabular irregular, S. verticillata by the surface area of blocky polyhedral and S. viridis by the area of trapezoid morphotype.

TEM revealed three valuable distinguishing parameters of phytoliths namely, micro-structural details, degree of amorpho-crystalline nature and inter-atomic planer distances in crystalline samples. Secondly, ultramicroscopy has proved useful in comparing and collating phytolith profiles from different parts of the plant body to develop phytolith signatures for each species. SAED patterns revealed by TEM showed the polycrystalline nature of silica in the synflorescences of S. verticillata and S. viridis whereas single crystal systems were reported in other parts of the three species. Thirdly, indexing of SAED patterns revealed silica polymorphism. The polymorphs of silica revealed by TEM were further confirmed by XRD patterns, particularly the ferrierite in S. viridis (root) and zeolite in S. pumila (root and culm).

The elemental composition of phytolith morphotypes from different parts of the present samples of the three species has revealed a cumulative total of 16 elements with 12 in S. pumila 14 in S. verticillata and 11 in S. viridis. A comparison of elemental composition of soil samples and phytolith morphotypes revealed that soil geochemistry controls the composition of phytoliths. Powder diffractograms of phytoliths revealed a number of polymorphic phases of silica. Stishovite was diagnostic of roots and leaves of S. pumila whereas ferririte was restricted only to the roots of S. viridis, thus strengthening a case for their role in taxonomic diagnosis as already reported for some other grass species.

FTIR analysis has revealed diversity of functional groups and their modes of vibrations with some groups being exclusively species specific. S. verticillata showed stretching vibration of O–Si (SiO<sup>2</sup> defect), asymmetric vibration of Si–O–Si, inplane stretching of free Si–O bond, symmetric deformation vibration of Si–R, deformation vibration of R (alkyl group), symmetric vibration of C–H and stretching vibration of O–H bonds. Similarly, groups characteristic of S. viridis include asymmetric and symmetric deformation vibrations of hydrocarbons (–CH3-CH2) – and inplane stretching vibrations of Si–C bonds.

The multiproxy approach employed in the present work has led to anatomical and physico-chemical characterization of the phytoliths produced by the present samples of three related species of the foxtail genus Setaria Phytolith analysis seems to confirm the comparatively isolated position of S. pumila in the present triumvirate of species. S. pumila was marked by two unique bilobate types compared to only one each in the other two species, the absence of polycrystalline silica in the synflorescences and the presence of the polymorphic silica as stishovite in the roots and the leaves. Clustering of species using Jaccard's similarity index for presence/absence data of the entire data set of phytolith morphotypes also revealed that S. pumila had a low similarity (33%) of phytolith profiles with S. verticillata and S. viridis (28.57%). However, S. verticillata and S. viridis showed much higher similarity (42.85%) and were grouped together (**Figures 8**). A plausible explanation may lie in the difference in the centers of origin of S. pumila (Africa) and the other two species (Asia).

Even though the full potential of phytoliths in understanding the taxonomy and phylogeny of the foxtail grass genus (Setaria) must come through future research involving an assessment of inter-population and intra-population variations and construction of representative master profiles for each species, the paper has made an initial contribution. We have made plant collections from single locations and homogenized the material part-wise but this limitation has been partly made good by following a multiproxy and multi-organ approach in carrying out the present work. In the larger context of plant systematics, concerted and coordinated efforts of a multidisciplinary nature are required to develop integrated and robust phytolith profiles of different groups of plants and their application in the characterization and diagnosis of plant taxa.

#### REFERENCES


## AUTHOR CONTRIBUTIONS

MB collected the material, conducted leaf epidermal studies, and wrote the initial draft of the manuscript. SS and PB carried out experimental work. AS designed the work, guided the conduct of experiments and checked the final manuscript.

#### ACKNOWLEDGMENTS

The authors are grateful to Professor Incharge, Emerging Life Sciences Laboratory, Guru Nanak Dev University, Amritsar, Punjab (India) and Director Indian Institute of Integrative Medicine, Jammu (Jammu and Kashmir) for SEM The help received from Mr. Musaib Ahmad Wani (Guru Ram Das School of Planning, Guru Nanak Dev University, Amritsar) in designing the map of the study area is gratefully acknowledged. The first author is thankful to the University Grants Commission, New Delhi for financial assistance under a major UGC project on phytolith studies on the grasses of the North West Himalayan region. The authors wish to thank the two reviewers whose comments and suggestions have vastly improved the quality of presentation and contributed to the logical coherence of the paper.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018. 00864/full#supplementary-material


a temperate forest ecosystem. Geochim. Cosmochim. Acta 72, 741–758. doi: 10.1016/j.gca.2007.11.010


Webster, R. D. (1987). The Australian Paniceae (Poaceae). Stuttgart: J. Cramer.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Bhat, Shakoor, Badgal and Soodan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Palm Phytoliths of Mid-Elevation Andean Forests

#### Seringe N. Huisman, M. F. Raczka and Crystal N. H. McMichael\*

Department of Ecosystem and Landscape Dynamics, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, Netherlands

Palms are one of the most common tropical plant groups. They are widespread across lowland tropical forests, but many are found in higher altitudes have more constrained environmental ranges. The limited range of these species makes them particularly useful in paleoecological and paleoclimate reconstructions. Palms produce phytoliths, or silica structures, which are found in their vegetative parts (e.g., wood, leaves, etc.). Recent research has shown that several palms in the lowland tropical forests produce phytoliths that are diagnostic to the sub-family or genus-level. Here we characterize Andean palm phytoliths, and determine whether many of these species can also be identified by their silica structures. All of our sampled Andean palm species produced phytoliths, and we were able to characterize several previously unclassified morphotypes. Some species contained unique phytoliths that did not occur in other species, particularly Ceroxylon alpinium, which is indicative of specific climatic conditions. The differences in the morphologies of the Andean species indicate that palm phytolith analysis is particularly useful in paleoecological reconstructions. Future phytolith analyses will allow researchers to track how these palm species with limited environmental ranges have migrated up and down the Andean slopes as a result of past climatic change. The phytolith analyses can track local-scale vegetation dynamics, whereas pollen, which is commonly used in paleoecological reconstructions, reflects regional-scale vegetation change.

#### Edited by:

Terry B. Ball, Brigham Young University, United States

#### Reviewed by:

Jose Iriarte, University of Exeter, United Kingdom Rosa Maria Albert, Catalan Institution for Research and Advanced Studies, Spain

#### \*Correspondence:

Crystal N. H. McMichael c.n.h.mcmichael@uva.nl

#### Specialty section:

This article was submitted to Paleoecology, a section of the journal Frontiers in Ecology and Evolution

Received: 04 August 2018 Accepted: 06 November 2018 Published: 27 November 2018

#### Citation:

Huisman SN, Raczka MF and McMichael CNH (2018) Palm Phytoliths of Mid-Elevation Andean Forests. Front. Ecol. Evol. 6:193. doi: 10.3389/fevo.2018.00193 Keywords: Andean ecosystems, Arecaceae, Ceroxylon, Dictyocaryum, paleoecology, palms, phytoliths

## INTRODUCTION

Arecaceae (palms) is a family of monocotyledonous plants that are important components of tropical ecosystems (Kahn and Mejia, 1990; Henderson et al., 1997; Phillips and Miller, 2002; Kahn and De Granville, 2012). Many palm species are widely distributed in the Neotropics, and belong to the most commonly found plants in Amazonian rainforests (Pitman et al., 2001; Vormisto et al., 2004; ter Steege et al., 2013). Many palms are also economically important for people, both in the modern era and in the past (Smith, 2014). Some of the earliest archaeological sites in the Andes and the Amazon contain evidence of a wide variety of palms consumed by people (Morcote-Rios and Bernal, 2001; Gnecco, 2003; Mora and Camargo, 2003). Some palm species, however, have constrained environmental tolerances or tend to be quite rare in the landscape, particularly in the mid-elevation regions (i.e., 1,000–3,000 m above sea level, hereafter masl) along the eastern Andean flank (Moraes et al., 1995; Henderson et al., 1997). Because the specific environmental optima vary between Andean palm species, palms are also important indicators of past climatic and ecosystem change in palaeoecological records (Bush et al., 2005, 2011; Schiferl et al., 2017); (Huisman et al., in revision).

Phytoliths are silica microfossils produced by many plant groups that can preserve in soils and lake sediments for millions of years (e.g., Strömberg, 2004; Piperno and Sues, 2005; Prasad et al., 2005; Piperno, 2006; Strömberg and McInerney, 2011). Phytoliths are commonly used in paleoecological and archaeological reconstructions, and provide evidence of localscale vegetation dynamics because they are typically deposited directly beneath the parent plant when it dies and decays (Piperno, 2006). Palms are known to be particularly prolific phytolith producers, and can often be identified to the subfamily or genus level (Piperno, 2006; Morcote-Ríos et al., 2016).

There have been recent advances in the ability to identify palm phytoliths in tropical settings, particularly in the Amazonian lowland forests, and in an archaeological context (Morcote-Ríos et al., 2016). The phytoliths of Andean palm species, however, remain relatively unstudied. Most palaeoecological reconstructions of climatic change in Andean systems use pollen, which can travel >10 km from the source plant. Thus, the ability to identify phytoliths of Andean palms with narrow environmental ranges would be highly advantageous in generating local-scale reconstructions of past climatic change in Andean systems. Here, we characterize the phytolith morphology of 12 species of Andean palms using herbarium specimens, to provide a foundation for future paleoecological and archaeological reconstructions in these highly diverse and vulnerable ecosystems.

#### METHODS

We collected herbarium species of Andean palm species from the Naturalis Herbarium in Leiden, The Netherlands. We obtained leaves, seeds, wood, and flowers (based on availability) of 12 palm species known to occur in mid-elevation Andean forests from 1,000 to 3,000 masl (Moraes et al., 1995; Henderson et al., 1997; **Table 1**). If the Andean species were unavailable, other available species within the same genus were collected (**Table 1**).

Prior to preparation, the dry plant material was ground and heated to 450◦C for 10 h. The samples were prepared by soaking in 33% hydrogen peroxide (H2O2), followed by 10% hydrochloric acid (HCl), and then potassium manganate (KMnO4) to break down the organic material. They were mounted in Naphrax and heated on a boiling plate to stabilize the material.

All samples were analyzed under a Leica Axiophot microscope with differential interference contrast (DIC) at 630x magnification using immersion oil. Categorization was based on Morcote-Ríos et al. (2016), but new (sub)categories were created for newly identified morphotypes. A total of 300 phytoliths was counted per sample to quantify the relative abundances of morphotypes. At least 30 phytoliths were measured per morphotype to obtain size ranges. Phytoliths were photographed using a Fujifilm X-E2 camera and Zeiss Universal microscope (DIC, Plan-Neofluar 63/1.4) and edited in Adobe Lightroom CC, Adobe Photoshop CC, CorelDraw and Helicon Focus.

TABLE 1 | Herbarium material sampled from palm species occurring on the eastern Andean flank between 1,000 and 3,000 m above sea level (masl).


The types of plant parts sampled are shown. An \*Indicates species that were used if Andean specimens were not available.

## RESULTS

All species and plant parts sampled produced abundant phytoliths, and contained one to four morphotypes (**Table 2**). We identified several new subtypes of globular echinate and conical morphotypes, which were previously characterized by other researchers (e.g., Piperno, 2006; Morcote-Ríos et al., 2016; **Figure 1**). The newly identified subtypes were defined by the following characteristics:



\*indicates species that were used if Andean specimens were not available. +indicates direct correspondence with the categories of Morcote-Ríos et al. (2016). GE, globular echinate; LGG, large globular granulate; GEE, globular echinate elongate; RE, reniform echinate; C, conical.

relatively few in number and of different sizes on the same phytolith, and irregularly placed on the surface, giving a starry rather than round overall appearance. Size: 4–15µm. This subtype was only found in Euterpe (**Table 2**). Photo: Euterpe precatoria, leaf.

4. Globular echinate variant 4: Short, bold projections (**Figure 1D**). This subtype appeared similar to the "globular echinate with dense short projections" described by Morcote-Ríos et al. (2016). It also appeared similar to the "large globular echinate" described by Dickau et al. (2013) and Watling et al. (2016), but smaller in size. This phytolith type contained many very small, bold projections. Due to the high number and stubby shape of the projections it did not have the typical thorny echinate outline. Size: 6–14µm. This subtype was found in Ceroxylon and Euterpe (**Table 2**). Photo: Ceroxylon alpinum, wood.


the projections were generally shorter than in Hyospathe and Euterpe. Photo: Euterpe precatoria, wood.


Huisman et al. Andean Phytoliths

overall shape. Its projections were few in number (3–6), and they were arranged around the top in a roughly circular arrangement. In most cases the projections were pronounced and easily discernible. Diameter/base length: 5–18µm. This subtype was found in Bactris, Dictyocaryum, Iriartea, and Socratea (**Table 2**). In Dictyocaryum and Socratea, sizes were fairly consistent, with ca. 10 and 15µm base length, respectively. Iriartea sizes were consistently 15–18µm, and Bactris presented highly variable sizes. Photo: Dictyocaryum fuscum, leaf.


6–15µm. This subtype was restricted to Geonoma. Photo: Geonoma paradoxa, leaf.

Most samples contained multiple phytolith subtypes, and there was variation of subtypes within and between species (**Table 2**). The Ceroxylon alpinum samples produced one or more of the globular echinate or reniform subtypes (**Table 2**). The seed samples, however, produced only globular echinate variant 2. The Cocoseae tribe samples also contained the globular echinate and reniform subtypes, but only in the Euterpeae and Geonomateae subtribes (**Table 2**). The leaves of Euterpe produced predominantly globular echinate variant 3, which was absent in Ceroxlyon alpinum. Euterpe seed and wood samples, and Hyospathe elegans leaf samples, had assemblages dominated by globular echinate variant 1 (**Table 2**). Reniform phytoliths occurred in low abundances in Hyospathe elegans and Geonoma samples (**Table 2**).

Conical phytoliths occurred primarily in the Iriarteeae and Chamaedoreeae tribes, and the Bactridinae and Geonomatae subtribes of the Cocoseae (**Table 2**). Leaf samples from Dictyocaryum, Iriartea, and Socratea contained 100% conical variant 1 phytoliths. Bactris simplicifrons also contained conical variant 1 phytoliths, but in lower abundances (**Table 2**). Chamaedorea pinnatifrons, Aiphanes aculeata, Bactris simplicifrons, and Wettinia hirsuta contained conical variant 2 phytoliths. Those same species, Geonoma undata, and Dictyocaryum fuscum (seeds only) contained conical variant 3 phytoliths (**Table 2**). Conical variant 4 phytoliths were produced primarily by Geonoma (**Table 2**).

## DISCUSSION

We defined several new subdivisions among the most recently described globular echinate and conical palm phytolith morphotypes (Morcote-Ríos et al., 2016; **Figure 1**). Our subdivided morphotype categories were based on DIC microscopy using 630x magnification, which was necessary to characterize differences within the broader globular echinate and conical categories, particularly when the phytoliths were ca. 10µm or smaller (**Figure 1**). Even with this magnification, it can still be difficult to distinguish surface ornamentation on some of these small phytoliths. We suggest that future identifications of palm phytoliths should be performed using a minimum magnification of 630x, but preferably a magnification of 1,000x. Scanning electron microscopy can also be particularly useful at identifying the nuances of phytolith morphotypes (Bowdery, 2015), though this approach is much more time consuming and may not be feasible when examining more than 250 phytoliths per sample and tens to hundreds of samples per sediment core.

The angle of view can also complicate identification of a single phytolith, especially in fossil samples. Silica-based microfossils are often mounted solidly and cannot be rotated once the mounting solution has dried, which fixes the angle of view. Our observations from the herbarium specimens of Geonoma indicated that the conical with basal projections type can resemble a globular echinate when viewed straight from the top, but it can also appear as an arboreal rugose sphere if positioned upside-down (**Figure S2**). The reference material from Euterpe precatoria and E. catinga also indicated that symmetry, a feature that was earlier used to distinguish echinate types (Morcote-Ríos et al., 2016), might be hard to determine when the phytoliths dry at different angles in the mounting solution (**Figure S3**). We therefore suggest to allow for rotation of the phytoliths by analyzing the slides before the mounting solution has fully dried, as we partly did in this study, or by using a nonsolidifying mounting medium such as liquid Entellan, glycerine or immersion oil (Cabanes and Shahack-Gross, 2015). However, the refractive indices of these media are similar to phytoliths and can make their overall features less distinguishable (Piperno, 2006). Because symmetry exists along a gradient, and identifying symmetry can also be troubled by damage or deterioration of the phytoliths in fossil samples, we did not include it as a diagnostic feature in our categorization.

Though we have classified new subtypes of phytoliths, the examination of the herbarium material suggested that phytolith morphologies occur as a gradient as opposed to clearly defined differences in types. For example, the newly characterized conical variants presented here are named "conical" following the previously described general category, which has also been called hat-shaped in some literature (Piperno, 2006; Tomlinson et al., 2011; Morcote-Ríos et al., 2016). Our examination of the conical phytoliths, however, revealed that the conical variants 2 and 3 do not always exhibit a true conical shape, and can also exhibit a range of shapes on their base (**Figures 1I–K**, **Figure S1**). These variations are mostly found in Chamaedorea, Aiphanes, and Wettinia (**Table 2**). The flowers of Wettinia hirsuta in particular exhibited a wide variation in shapes of the conical base and in the configuration of projections.

There seems to be a particularly strong gradient in the reniform echinate phytoliths, which ranged from small, thin and highly curved specimens (**Figure S4**) to large, thick and minimally curved ones (**Figure 1H**). Ceroxylon, Hyospathe, and Geonoma contained thick reniform shapes, which we identified as reniform echinate variant 1 (**Table 2**). The smaller, thinner reniform shapes described by Morcote-Ríos et al. (2016) were very rare in our samples, and we suggest those should be considered a different variant (i.e., reniform variant 2). We encountered a few thin reniform-like shapes in the leaves of Geonoma paradoxa; however, some of them appeared more randomly kinked than evenly curved and resembled caterpillar shapes (**Figure S4**). We are therefore uncertain if these "caterpillar" reniform phytoliths should be considered a separate variant (e.g., reniform variant 3), but we have also encountered them in fossil samples. Future studies focused on these reniform echinate phytoliths will be able to further determine subdivisions.

Geonoma was the only palm genus where conical and reniform echinate phytoliths co-occurred in the same sample. Geonoma was also the only genus where previous studies have reported the rare co-occurrence of conical phytolith with nonconical types (Tomlinson et al., 2011; Morcote-Ríos et al., 2016). In both of those cases, the co-occurrence happened with an echinate subtype that we did not encounter in our samples. We very rarely encountered echinate and conical types in the same phytolith sample apart from in Geonoma samples, and did not include these exceptions in **Table 2** because the echinates were so rare and we cannot rule out the possibility of contamination of the herbarium sheets that we sampled. Even though Geonoma produced an array of phytolith subtypes, it was the only genus to produce high amounts of conical variant 4 phytoliths (**Table 2**). Thus, in eastern Andean forests, high amounts of conical variant 4 phytoliths likely represent Geonoma, a genus of primarily understory palms whose Andean species grow larger than their lowland counterparts (up to 12 cm diameter and 13 m height) (Moraes et al., 1995). Andean Geonoma species are also economically important (e.g., Bernal et al., 2011), so the presence or absence of Geonoma phytoliths from archaeological settings may provide information on its past use and dispersal.

In many samples we encountered individual phytoliths that appeared deformed. This occurred primarily in seeds and woody parts, and much less frequently in leaf samples. Deformations ranged from completely randomly shaped bodies (e.g., wood/seeds of Ceroxylon alpinium, Euterpe precatoria, and Geonoma) (**Figure S5**) to still recognizable diagnostic types with smaller deformations, e.g., conical phytoliths that were extensively elongated or "tailed" at the base in Dictyocaryum seeds (**Figure S6**). We did not examine the ratios of the deformed phytoliths within samples, primarily because these ratios would not be applicable in a paleoecological or archaeological context where all phytoliths are intermixed.

Ceroxylon is one of the only palm genera that is more common at higher elevations (i.e., >2,000 masl) than at lower elevations (Moraes et al., 1995). Ceroxylon produced primarily globular echinate phytoliths and fewer reniform echinate phytoliths, as did some of the Euterpeae, which typically grow at lower elevations (Moraes et al., 1995; **Table 2**). Ceroxylon, however, was the only species to produce the globular echinate variant 2 phytoliths (**Table 2**, **Figure 1B**). Thus, the globular echinate variant 2 phytoliths in paleoecological contexts can be used to identify Ceroxylon, and indicate colder conditions than most other palm species prefer. Ceroxylon is also an economically important plant species (e.g., Bernal et al., 2011), and its phytoliths (or the absence of them) can now also be used in archaeological settings to reconstruct past palm management practices and the dispersal of economically important taxa. Ceroxylon also produced other types of unique silica structures that were not found in any other species that we sampled (**Figure S7**). These silica structures may come from the "waxy wood" that has commonly been described in Ceroxylon, which does not occur in most palm species (Moraes et al., 1995).

Size variation can be a key distinguishing feature in phytolith identification. In the case of Wettinia hirsuta, the variation of phytoliths between plant parts was mostly in terms of size. The seeds produced ca. 2–6µm larger phytoliths than the flower. Perhaps more importantly for paleoecological and archaeological reconstructions, we found that the size variation of phytolith subtypes can be used to distinguish species from each other. For example, Iriartea deltoidea and Dictyocaryum lamarckianum produce 100% the same subtype of conical variant 1 in their leaves. The phytoliths produced by Iriartea, however, are consistently and significantly larger than those produced by Dictyocaryum. Conical phytoliths from Iriartea typically range from 15 to 18µm, whereas those found in Dictyocaryum range from 5 to 10µm. Iriartea deltoidea is one of the most common trees in the Amazonian lowlands (Pitman et al., 2001; ter Steege et al., 2013), and rarely exceeds 1,300 masl (Henderson et al., 1997). In contrast, Dictyocaryum ranges primarily from 1,000 to 2,000 masl (Moraes et al., 1995). The ability to distinguish the conical phytoliths of these two species is therefore particularly advantageous in reconstructions of past climatic conditions.

Our newly characterized palm phytolith types (**Figure 1**, **Table 2**) provide a foundation to identify palm genera with narrow altitudinal ranges in mid-elevational Andean settings. Mid-elevation Andean ecosystems are some of the most diverse and threatened ecosystems on Earth (Olson et al., 2001; Olson and Dinerstein, 2002), and past climatic change has helped shape these systems into their current configuration (e.g., Flenley, 1979; Bush, 2002; Bush et al., 2007). Identification of an increased number of Andean palm phytolith subtypes, and the increased ability to associate them with a specific palm genus or set of genera, can provide more detailed reconstructions of past ecological dynamics in paleoecological and archaeobotanical contexts than was previously possible. Because phytoliths reflect local-scale vegetation dynamics (e.g., Piperno, 2006), ecological responses to past climatic changes or human activity can also be reconstructed at a higher spatial resolution than when using only pollen data. If paired, pollen and phytolith analyses could be used to determine regional vs. local vegetation change in Andean settings, as has been done in the Amazonian lowlands (e.g., Carson et al., 2014).

#### CONCLUSIONS

Mid-elevation Andean forests are some of Earth's most diverse and threatened ecosystems, yet their ecological history remains understudied. Phytoliths, which represent local-scale vegetation, are a valuable tool in reconstructing vegetation dynamics through time. Our study provides a more nuanced categorization of palm phytoliths than was previously available, and also demonstrates their potential to create more comprehensive paleoecological reconstructions of local-scale

#### REFERENCES


vegetation dynamics than previously possible. Phytoliths are becoming much more commonly used in paleoecological and archaeological reconstructions because of their potential in reconstructing local-scale vegetation patterns, and we stress the need to continue referencing regionally-specific palm (and also arboreal) phytoliths.

## DATA AVAILABILITY STATEMENT

The phytolith reference material generated for this study can be found in the phytolith reference collection at the University of Amsterdam.

## AUTHOR CONTRIBUTIONS

CM and SH designed the study. SH and MR carried out the phytolith analyses. CM and SH contributed equally to the writing of the manuscript.

## FUNDING

This research was part of the Masters Programme in the Institute for Biodiversity and Ecosystem Dynamics at the University of Amsterdam. The Netherlands Organisation for Scientific Research (NWO) award ALWOP.322 to CM also provided partial funding.

#### ACKNOWLEDGMENTS

We thank Naturalis Biodiversity Centre in Leiden, the Netherlands, for providing the herbarium material. We also thank Annemarie Philip for phytolith sample preparation, and Jan van Arkel and Britte Heijink for the phytolith photographs.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fevo. 2018.00193/full#supplementary-material


upper Amazonian terra firme forests. Ecology 82, 2101–2117. doi: 10.1890/ 0012-9658(2001)082[2101:DADOTS]2.0.CO;2


Smith, N. (2014). Palms and People in the Amazon. Heidelberg: Springer.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Huisman, Raczka and McMichael. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Bulliform Phytolith Size of Rice and Its Correlation With Hydrothermal Environment: A Preliminary Morphological Study on Species in Southern China

#### *Can Wang1\*, Houyuan Lu2,3,4\*, Jianping Zhang2,3, Limi Mao5 and Yong Ge6,7*

#### *Edited by:*

*Rivka Elbaum, The Hebrew University of Jerusalem, Israel*

#### *Reviewed by:*

*Wricha Tyagi, Central Agricultural University, India Dongying Gao, University of Georgia, United States Ofir Katz, Dead Sea and Arava Science Center, Israel*

#### *\*Correspondence:*

*Can Wang shandawangcan@163.com Houyuan Lu houyuanlu@mail.iggcas.ac.cn*

#### *Specialty section:*

*This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science*

*Received: 02 May 2019 Accepted: 24 July 2019 Published: 22 August 2019*

#### *Citation:*

*Wang C, Lu H, Zhang J, Mao L and Ge Y (2019) Bulliform Phytolith Size of Rice and Its Correlation With Hydrothermal Environment: A Preliminary Morphological Study on Species in Southern China. Front. Plant Sci. 10:1037. doi: 10.3389/fpls.2019.01037*

*1 Department of Archaeology, School of History and Culture, Shandong University, Jinan, China, 2 Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China, 3 Center for Excellence in Tibetan Plateau Earth Science, Chinese Academy of Sciences, Beijing, China, 4 College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing, China, 5 Nanjing Institute of Geology and Palaeontology, Chinese Academy of Sciences, Nanjing, China, 6 Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China, 7 Center for Excellence in Life and Paleoenvironment, Chinese Academy of Sciences, Beijing, China*

In the last decade, our understanding of rice domestication has improved by new archaeological findings using advanced analytical techniques such as morphological and morphometric analyses on rice grains, spikelet bases and phytoliths, and ancient DNA analysis on rice remains. Previous studies have considered the size of rice bulliform phytoliths as a proxy for tracking the domestication process. These phytoliths are often abundant and well preserved in sediments, and their shape is under the control of numerous genes, which may shift toward larger sizes by genetic mutation in domestication. Therefore, it has been assumed that the bulliforms of domesticated rice are usually larger than those of wild ones; however, morphometric data supporting this assumption are lacking in the literature, thereby requiring additional evidence to test its veracity. In this study, the vertical and horizonal lengths of bulliform phytoliths were measured in four rice species (domesticated *Oryza sativa* and wild *Oryza rufipogon*, *Oryza officinalis*, and *Oryza meyeriana*) from different regions of southern China. We found that the bulliform morphometric data of wild and domesticated rice overlapped and that there was no statistically significant difference between them. Therefore, bulliform size could not be used as a diagnostic indicator to distinguish domesticated rice from wild species and is a supporting rather than conclusive proxy for determining the domesticated status of rice in archaeological research. We further found that larger rice bulliform sizes likely occurred at the locations with higher temperature, precipitation, and water levels, indicating hydrothermal environment is an alternative factor influencing the size of rice bulliform phytoliths. For further archaeological use of an increasing size trend of bulliform phytoliths to reveal the process of rice domestication, we present some suggestions for controlling the influence of hydrothermal factors. Even so, the combination of bulliform phytolith size with other established criteria is strongly suggested to provide precise identification of wild and domesticated rice in future research.

Keywords: rice, bulliform phytolith, *Oryza sativa*, *Oryza rufipogon*, domestication, morphometric analysis

## INTRODUCTION

Asian rice (*Oryza sativa* L.) is one of the most important crops and forms a staple food for more than half of the global population (Nayar, 2014). Understanding its origins and domestication from wild rice (*Oryza rufipogon* Griff*.*) is thus an important aim for researchers. There are two major subspecies of domesticated *O. sativa*, *Oryza sativa indica,* which is thought to have originated in the Himalayan region, and *Oryza sativa japonica,* which is thought to have originated in China (Londo et al., 2006). *Oryza rufipogon* is generally recognized as the ancestor of *Oryza sativa* (Wei et al., 2012), with two distinct domestication events leading to the two subspecies of domesticated rice (Londo et al., 2006). However, this hypothesis is still being vigorously debated as other evidence supports a single origin of Asian rice (Huang et al., 2012). There are more than 20 wild rice species recognized in the genus *Oryza* (Nayar, 2014; Stein et al., 2018), which belongs to the family: Poaceae and tribe: Oryzeae. Three of these species are found in China, *Oryza rufipogon*, *Oryza officinalis*, and *Oryza meyeriana* (Fan et al., 2000).

In the last decade, scientific understanding of rice domestication has greatly improved by new archaeological and genetic evidence using advanced analytical techniques such as flotation combined with morphometric analysis on rice grains and spikelet bases (Liu et al., 2007; Fuller et al., 2009; Gross and Zhao, 2014; Zheng et al., 2016), phytolith analysis (Wu et al., 2014; Huan et al., 2015; Zuo et al., 2017), pan-genome analysis (Wang et al., 2018a), and genome-wide association studies (Huang et al., 2012; Civáň and Brown, 2017; Choi and Purugganan, 2018). According to recent findings, a general consensus among scholars has been reached that rice was first domesticated in the middle and lower Yangtze River regions of southern China (Molina et al., 2011; Gross and Zhao, 2014; Silva et al., 2015; Choi et al., 2017; Zuo et al., 2017), although some consider the Pearl River region to also be a part of the original area where rice domestication occurred (Huang et al., 2012; Wei et al., 2012).

The estimated time of origin of rice domestication is before 13,000 BP, based on molecular clock analysis (Molina et al., 2011; Choi et al., 2017; Choi and Purugganan, 2018), which is much older than the earliest archaeological date of domestication (<10,000 BP) (Fuller et al., 2014; Larson et al., 2014). This disagreement may result from genetic studies that mainly focused on identifying the origins of alleles associated with domestication (e.g., *sh4*, *rc*, *laba1*, *prog1*), which likely emerged in wild rice prior to domestication (Choi et al., 2017; Civáň and Brown, 2017). On the other hand, archaeological studies attempted to detect the first appearance of morphological traits associated with domestication in archaeobotanical remains (Fuller et al., 2009; Jones and Liu, 2009). The molecular and archaeological chronologies may date two main phases in the macroevolutionary process: the emergence of a trait and the success of that trait (the trait becomes quantitatively significant within a population) (Katz, 2019). However, the most notable chronological dispute over the rice domestication is between two archaeological opinions: one suggests that the process of rice domestication may have begun around 10,000–9,000 BP (Liu et al., 2007; Zheng et al., 2007; Wu et al., 2014; Zheng et al., 2016; Zuo et al., 2017), while the other suggests that domestication of rice did not occur until around 8,000–6,000 BP (Fuller et al., 2007; Fuller et al., 2008; Fuller et al., 2009; Fuller et al., 2010; Larson et al., 2014). This debate is largely attributable to the differences in the methods employed, and the criteria used by various authors to identify domestication in rice remain from early sites in the Yangtze River region such as Shangshan, Kuahuqiao, and Xiaohuangshan. Establishing accurate and feasible criteria for distinguishing between domesticated and wild rice is thus of prime importance.

Three important lines of archaeological evidence have frequently been used in China, including grain size and morphological characteristics, spikelet bases, and phytoliths (Liu and Chen, 2012; Fuller and Castillo, 2014; Fuller, 2018). Grain morphometrics are considered to be semi-domestication traits and may not be diagnostic indicators of early domestication (Fuller, 2007; Fuller et al., 2007), mostly due to the considerable variation and overlap in length between domesticated and wild populations which leads to some proportion of false assignments in ancient rice grains (Fuller, 2007). Many studies have focused on the form of spikelet base, which is thought to be the most diagnostic trait in rice remains in terms of identifying domestication status (Fuller, 2018). However, these are insufficient, because the distinctive characteristics of immaturity, shattering, and nonshattering states and/or wild, *japonica*, and *indica* rice based on spikelet bases are divergent among the criteria provided by different researchers (Zheng et al., 2007; Pan, 2008; Fuller et al., 2009; Fuller et al., 2010; Pan, 2011; Gross and Zhao, 2014; Zheng et al., 2016), and their diagnostic power for domestication has the potential to be more reliable. More importantly, these macrobotanical remains do not preserve well in early sediments with acid soil; therefore, very few have been recovered from sites dated earlier than 9,000 BP (Zhao, 2011; Qin, 2012; Zhao and Jiang, 2016).

Phytoliths have played an important role in the identification of rice remains recovered from early archaeological sites, due to their high resistance to decomposition (Piperno, 2006; Ball et al., 2016). Double-peaked phytoliths from husks and bulliform phytoliths from rice leaves are both certainly diagnostic indicators of *Oryza* and show variation within and between species. A number of identification criteria based on these phytoliths have been suggested and widely used in the last 20 years (Zhao et al., 1998; Lu et al., 2002; Gu et al., 2013; Wu et al., 2014; Huan et al., 2015; Hilbert et al., 2017; Zuo et al., 2017; Wang et al., 2018b). Although the utility of these methods for distinguishing domesticated from wild rice is still under debate (Fuller et al., 2010; Fuller, 2018), they are recognized as key alternative methodologies besides using morphological domestication data of rice macroremains. Three-dimensional measurements and discriminant function analysis of double-peaked phytoliths are useful in determining the wild/domesticated nature of rice remains; however, double-peaked phytoliths usually present their side and top view under the microscope, which does not meet the requirements of morphometric analysis (present in front view) (Zhao et al., 1998; Gu, 2009), making the work arduous in most cases.

Rice bulliform phytoliths are abundant in rice leaves and are often well preserved and represented in archaeological sediments (Fujiwara, 1993; Wang and Lu, 1993). Generally, bulliform phytoliths in *Oryza* have a distinctive fan shape with numerous scale-like decorations on the half round side (lateral side) (Lu et al., 2002; Wang and Lu, 2012; Gu et al., 2013). Morphological measurements and number of scale-like decorations along the scalloped edge have been employed to distinguish wild *Oryza* species from domesticated ones. Studies on modern rice plants and paddy surface soils have suggested that bulliform phytoliths with ≥9 scale-like decorations were likely domesticated, while those with <9 were generally wild (Lu et al., 2002; Huan et al., 2015). Whether this feature is a useful domestication indicator remains inconclusive and requires further validation; in addition, genetic explanatory mechanisms of bulliform scale-like decoration variation between species remain unclear and deserve further study.

Bulliform shape of rice appears to be under the control of 16 genes (QTLs) (Zheng et al., 2003a) and phytoliths may shift toward larger sizes as a result of genetic mutation during the domestication process (Zheng et al., 2003b; Piperno, 2006; Luo et al., 2016). Some researchers have, therefore, assumed that the bulliforms of domesticated rice are usually larger than wild ones and that the trend of increasing size in rice bulliform phytoliths could reflect domestication of rice (Zheng et al., 2003b, Zheng et al., 2004; Fuller et al., 2007). In recent years, vertical and horizontal lengths (i.e., sizes) of rice bulliforms have increasingly been used as a proxy for tracking the domestication process or determining the degree of domestication at different sites, such as the Tanghu (Zhang et al., 2012), Zhuzhai (Wang et al., 2018b), Shunshanji (Luo et al., 2016), Shangshan, Hehuanshan, and Huxi sites (Zuo et al., 2017; Qiu et al., 2019). However, to date, morphometric data from modern rice plants supporting this method and their assumptions are missing (Pearsall et al., 1995; Zhang and Wang, 1998; Ma and Fang, 2007; Gu et al., 2013), and thus additional evidence is required to test its veracity.

Moreover, some studies have argued that changes in phytolith size may not only be triggered by domestication but also influenced by environmental factors, such as CO2 concentrations (Ge et al., 2010), evapotranspiration rates (Issaharou-Matchi et al., 2016), and water levels in the growing habitat (Fuller, 2018). Fuller (2018) indicated that the 16 genes suggested by Zheng et al. (2003a) only explained between 37 and 54% of bulliform variation, suggesting that the environment or growing conditions also play an essential role. These authors further speculated that if shifting bulliform morphology was merely a phenotypic response to environmental conditions, it would be a less useful indicator of domestication. Thus, without data supporting the exclusion of environmental factors, bulliform phytolith measurements alone may not be an accurate identification tool for distinguishing between domesticated and wild *Oryza* species.

In the present study, we tested whether the size of bulliform phytoliths is an effective statistical indicator for distinguishing between wild rice and domesticated rice, based on the comparative analysis of morphometric data from different rice species. In addition, we attempted to examine how growing conditions, especially climate and water levels, influence bulliform size. Finally, we discussed how rice bulliform phytolith morphometry can be used in archaeological research.

## MATERIALS AND METHODS

In the present study, a total of 24 specimens of *Oryza* were sampled. The samples consisted of six specimens of the domestic *O. sativa*, and for the wild species, 16 specimens of *O. rufipogon*, 1 specimen of *O. officinalis*, and 1 specimen of *O. meyeriana* all collected in southern China (**Table 1**; **Figure 1**). All 6 specimens of *O. sativa* and 8 of the 16 specimens of *O. rufipogon* were sampled from the test paddy field belonging to Wuhan Botanical Garden, Chinese Academy of Sciences (CAS), at Huazhong Agricultural University, Wuhan, Hubei Province. Another eight specimens of *O. rufipogon* were sampled from Hainan, Yunnan, Hunan, and Jiangxi Provinces. The specimens of *O. officinalis* and *O. meyeriana* were sampled from Hainan Province.

Field collection of rice plants was assisted by a team of investigators from Wuhan Botanical Garden and Nanjing Institute of Geology and Palaeontology, CAS, with the permission of the owner or regulatory body for each location. The criteria for categorizing and naming the species collected were accepted from the classification scheme of the genus *Oryza* in the Flora of China (Liu and Phillips, 2006) (available at http://flora.huh.harvard.edu/china/PDF/PDF22/Oryza.pdf). The specimens of *O. sativa* and *O. rufipogon* in Wuhan were both mature when we collected them between September 21 and 23, 2011, but the ripening rate of *O. rufipogon* was very low due to low temperatures. In Hainan Province, the specimens of *O. rufipogon* were in anthesis and immature while the specimens of *O. officinalis* and *O. meyeriana* were mature and had begun shattering when we collected them between December 1 and 7, 2012. The specimens of *O. rufipogon* in Chaling, Hunan Province, were immature when we collected them between September 18 and 28, 2010. The specimens of *O. rufipogon* in Jiangxi and Yunnan Province were already mature when we collected them during October and November 2010, respectively. All plant samples were preserved at the Institute of Geology and Geophysics, CAS, Beijing.

There are different hydrological environments among our sampling sites. In Wuhan, specimens of *O. sativa* and *O. rufipogon* were simultaneously cultivated at paddy fields with shallow water (**Figures 2a**, **b**), which sometimes needs draining to achieve moderate draught during the pustulation and fruiting stage; *O. rufipogon* from Xishuangbanna Tropical Botanical Garden, CAS, Jinghong County, had a similar habitat. In Chaling, *O. rufipogon* grows in the Huli Marsh where there is perennial stagnant water (**Figure 2c**); *O. rufipogon* from Yuanjiang has a similar habitat. In Anjiashan and Shuitaoshuxia, Dongxiang County, *O. rufipogon* grows in a seasonal wetland with 0–100 cm water depth. In Hainan Province, specimens of *O. rufipogon* from Wenchang and Wanning Cities grow in permanent wetlands where there are ponds filled with deep water with 30–150-cm depth (**Figures 2d**–**g**, **j**); *O. officinalis* in Lingshui County grows in a ravine stream in the valley and prefers shady and wet habitats


TABLE 1 | Information on the rice plants studied and measured data of vertical length and horizontal length of bulliform phytoliths from the studied samples.

FIGURE 1 | Locations of sample collection sites. Test paddy field in Huazhong Agricultural University, Wuhan, Hubei Province (1); Shuitaoshuxia, Dongxiang, Jiangxi Province (2); Anjiashan, Dongxiang, Jiangxi Province (3); Huli Marsh, Chaling, Hunan Province (4); Yuanjiang protection area of wild rice, Yunnan Province (5); Xishuangbanna Tropical Botanical Garden, CAS, Yunnan Province (6); Hulu village, Wenchang, Hainan Province (7); Tanshen village, Wenchang, Hainan Province (8); Mingxing village, Wanning, Hainan Province (9); Zhangxian village, Lingshui, Hainan Province (10); Luhuitou Park, Sanya, Hainan Province (11).

(**Figures 2h**, **k**); *O. meyeriana* in Sanya City grows in an understory bush on a hill and prefers shady and dry habitats (**Figures 2i**, **l**).

For each rice specimen, we selected all leaf blades from the bottom to the top of a single plant, making sure to keep the leaf blade intact for phytolith extraction. This is because there is significant difference in bulliform phytolith size among different leaf blades of the same plant and different parts of the same leaf blade, with the smaller bulliform phytoliths from the lower leaves (Fuller and Qin, 2009). Bulliform size tends to decrease from leaf base to leaf apex of the same leaf blade (Wang et al., 1997). Therefore, variation in bulliform phytoliths from a few randomly selected rice leaves does not reflect the overall data, and only the selection of all intact leaves can guarantee the representativeness and reliability of the data.

Leaf blades were cleaned with distilled water in an ultrasonic cleaner, and then prepared for wet oxidation: 1) all samples were cut into 1–3-cm pieces and placed in 20 ml of 65% saturated nitric acid for over 12 h, then heated in a water bath for 20 min to oxidize organic materials completely. 2) The solutions were centrifuged at 3,000 rpm for 6 min, decanted and rinsed three times with distilled water, and then rinsed with 95% ethanol until the supernatants were clear. 3) The extracted phytoliths were mounted onto microscopic slides in neutral resin, and the residual samples were transferred to storage vials.

A Leica DM750 light microscope at 600× magnification was used for photomicrography and phytolith counting. One hundred or more bulliform phytoliths, including asymmetric types, were counted in each sample, except for the sample of *O. meyeriana* which produced only a few bulliform phytoliths. Two morphometric parameters, vertical and horizonal lengths (VL and HL), were measured to describe the size of bulliform phytoliths. The measurements were taken from images using the ImageJ software (version 1.48r.). Descriptive statistics of morphometric data were performed using Excel software, and the mean ± SD of each sample was plotted using the Grapher software to perform a comparative analysis. Discriminant function analysis in SPSS 24.0 software was then used to statistically determine the differences in bulliform sizes in different species.

In order to understand whether bulliform phytolith size correlated with environmental conditions and to what degree, a Pearson correlation analysis was performed. Eight environmental variables were chosen: altitude, mean annual precipitation (MAP), July precipitation (MP7), January precipitation (MP1), mean annual temperature (MAT), July temperature (MT7), January temperature (MT1), and relative humidity (HHH). Modern climatic data for the 11 sampling sites (**Figure 1**) were obtained from the nearest meteorological station to each site, since the spatial variation of climatic parameters exhibits a clear gradient across these locations. These data can be collected from the databases (1981–2010) of the National Meteorological Information Center, China (http://data.cma.cn/). Origin 8.5.1 software was used to conduct the correlation analysis of bulliform morphometrics and environmental variables, which were plotted into scatter plots. A linear regression was inserted into these scatter plots, and then the Pearson correlation coefficients (*r*) and significances (*P*) were taken to statistically evaluate the correlation.

#### RESULTS

#### Morphological Contrast of Bulliform Phytoliths in the Four *Oryza* Species

Overall, the bulliform phytoliths in *O. sativa* and *O. rufipogon* have similar shapes with an intact circular part of the fan, round arc of the scalloped edge, and ridge-like tubercle on the lateral side (**Figure 3**). Significant intraspecific morphological variation, however, was also found in both *O. sativa* (**Figures 3a**–**h**) and *O. rufipogon* (**Figures 3i**–**p**). Moreover, we noted that the bulliform phytoliths from the Xishuangbanna *O. rufipogon* had a very specific shape with a small circular part of the fan, angular scalloped edge, large deep decorations, and without the ridgelike tubercle on the lateral side (**Figures 3q**–**t**). This shape was not only different from that of *O. sativa* but also distinct among other *O. rufipogon* specimens.

Bulliform phytoliths of *O. officinalis* are mostly large and full in shape with a rounded arc of the scalloped edge, irregular large deep decorations, longer handles, and a shorter intact circular part of the fan but without the ridge-like tubercle on the lateral side (**Figure S1**). From an overall perspective, this shape is similar to that of *O. sativa* and *O. rufipogon*.

Bulliform phytoliths of *O. meyeriana* are generally long and thin and not full in shape like the other varieties. These phytoliths are very small and appear similar to a teardrop or nail, with an angular scalloped edge, small irregular, but deep decorations, a

FIGURE 2 | Photos of parts of sampling locations and rice plants. Domesticated rice paddy in Huazhong Agricultural University, Wuhan, Hubei Province (site 1 in Figure 1) (a); wild rice paddy in Huazhong Agricultural University, Wuhan, Hubei Province (site 1 in Figure 1) (b); site of *Oryza rufipogon* population in Huli Marsh, Chaling, Hunan Province (site 4 in Figure 1) (c); site of *O. rufipogon* population in Hulu village, Wenchang, Hainan Province (site 7 in Figure 1) (d); site and plants of *O. rufipogon* population in Mingxing village, Wanning, Hainan Province (site 9 in Figure 1) (e, f); site of *O. rufipogon* population in Tanshen village, Wenchang, Hainan Province (site 8 in Figure 1) (g); site of *O. officinalis* population in Zhangxian village, Lingshui, Hainan Province (site 10 in Figure 1) (h); site of *O. meyeriana* population in Luhuitou Park, Sanya, Hainan Province (site 11 in Figure 1) (i); plant of *O. rufipogon* in Hainan Province (j); plant of *O. officinalis* in Hainan Province (k); plant of *O. meyeriana* in Hainan Province (l); the pictures of rice plants were taken by Dr. Limi Mao. The individual in Figure 2c was Dr. Jianping Zhang who had approved the publication of this image.

longer handle, and shorter circular part of the fan not intact, without the ridge-like tubercle on the lateral side (**Figure S2**). This shape is different from that of the other three *Oryza* species.

#### Morphometric Analysis of Bulliform Phytoliths in the Four *Oryza* Species

Overall, our morphometric data demonstrated that the bulliforms of domesticated rice were not always larger than wild ones, and there was no significant difference in size. **Table 1** shows the mean values of vertical length (VL) and horizontal length (HL) of bulliform phytoliths from the studied samples. For original measured data, see the **Datasheet S1.**

In the six *O. sativa* specimens, the maximum mean of VL of bulliform phytoliths was 44.52 ± 6.22 μm (AA37), while the minimum was 37.77 ± 4.74 μm (AA38); the maximum mean of HL of bulliform phytoliths was 38.31 ± 6.26 μm (AA31), while the minimum was 32.78 ± 4.47 μm (AA38). In all 609 bulliform phytoliths from *O. sativa*, the maximum VL was 60.27 μm occurring in sample AA36, while the minimum was 25.25 μm occurring in sample AA38; the maximum HL was 61.96 μm occurring in sample AA31, while the minimum was 20.94 μm occurring in sample AA48.

In the 16 *O. rufipogon* specimens, the maximum mean of VL of bulliform phytoliths was 52.23 ± 6.84 μm (WN-7), while the minimum was 29.89 ± 4.01 μm (Z13); the maximum mean of HL of bulliform phytoliths was 43.05 ± 6.74 μm (WN-7), while the minimum was 24.19 ± 3.99 μm (Z13). In all 1,649

(XSBN) (q–t); scale bar = 30 μm. The red arrows point to the ridge-like tubercle.

bulliform phytoliths, the maximum VL was 70.11 μm occurring in sample WN-7, while the minimum was 21.30 μm occurring in sample Z13; the maximum HL was 59.81 μm occurring in sample WN-7, while the minimum was 15.54 μm occurring in sample Z13.

In the only specimen of *O. officinalis*, the mean values of VL and HL of bulliform phytoliths were 51.13 ± 8.77 and 40.43 ± 7.34 μm, respectively. In all 100 bulliform phytoliths, the maximum VL was 80.00 μm, while the minimum was 32.94 μm; the maximum HL was 62.76 μm, while the minimum was 25.17 μm.

In the only specimen of *O. meyeriana*, the mean values of VL and HL of bulliform phytoliths were 38.02 ± 5.35 and 29.37 ± 6.29 μm, respectively. In all 34 bulliform phytoliths, the maximum VL was 50.59 μm, while the minimum was 30.49 μm; the maximum HL was 44.26 μm, while the minimum was 18.18 μm.

**Figure 4** presents a comparison of the sizes of bulliform phytoliths from the studied rice species. The results revealed that the values of VL and HL from *O. rufipogon* were scattered and widely overlapped with the other three *Oryza* species. Although the bulliform sizes of *O. sativa* were larger

compared with *O. meyeriana*, they partly overlapped with *O. rufipogon* (mean VL: 37–45 μm; HL: 32–40 μm) and were significantly smaller than those of *O. officinalis*. The bulliform size of *O. meyeriana* was smaller compared with other *Oryza*

species and only larger than a few specimens of *O. rufipogon*. The bulliform size of the specimen of *O. officinalis* was larger, exceeding most of the studied samples.

Parameters VL and HL were important and used in the discriminant function analysis (**Table S1**). The following two canonical discriminate functions were used in the analysis: function 1 explained 89.5% of the variance, and function 2 explained 10.5% of the variance. Parameter VL had the largest absolute correlation with function 1, indicating that it contributed most to function 1; parameters HL had the largest absolute correlation with function 2, indicating that it contributed more to function 2 (**Table S1**). These two functions were used to plot the data (**Figure 5**). Four groups without distinct centroids were obtained; *O. sativa* had a clear intersection with *O. rufipogon*, whereas there was a slight distinction for *O. officinalis* and *O. meyeriana* with regard to the other species. The accuracy of the classification was ascertained by cross validating the results (**Table 2**). Only 36.8% of the original data and 36.7% of the cross-validated data were correctly classified, suggesting that the discriminant functions obtained using parameters VL and HL could not be successfully used to discriminate *O. sativa*, *O. rufipogon*, *O. meyeriana*, and *O. officinalis*. Thus, it supported the conclusion that there were no significant differences in bulliform sizes between wild and domesticated rice.

The 24 specimens of rice were then divided into two groups: mature and immature (**Figure 6**). As can be seen, the sizes of bulliform phytoliths from mature rice were scattered and partly overlapped with the immature rice. In contrast, bulliform phytoliths of immature rice were slightly larger than most mature species. There was no significant difference in bulliform size between mature and immature rice species.

We further compared the bulliform size of rice species in terms of their growing region (**Figure 7**). It was found that the bulliforms of *O. rufipogon* and *O. officinalis* growing in the tropical Hainan were the largest with mean vertical and horizontal lengths greater than 45 and 40 μm, respectively. The bulliforms of *O. rufipogon* growing in Chaling, Dongxiang, and Yuanjiang and *O. sativa* growing in Wuhan were similar in size, with mean VL and HL ranges of 40–45 and 33–40 μm, respectively. *O. rufipogon* growing in Wuhan and Xishuangbanna and *O. meyeriana* growing in Hainan, have the smallest bulliform phytoliths with mean VL and HL ranges of 33–39 and 29–34 μm, respectively. It is noted that *O. rufipogon* and *O. officinalis* in the Hainan population have the most favorable habitat in terms of water availability, with permanent deep water or perennial ravine streams; *O. rufipogon* in Chaling, Dongxiang, and Yuanjiang grow in marshes and seasonal wetlands where stagnant water converges at the roots of rice; *O. rufipogon* in Wuhan and Xishuangbanna grow in paddy fields with shallow water, which is occasionally drained to maintain relatively dry habitats; *O. meyeriana* in Hainan prefers a drier environment.

In addition, the effect of habitat wetness on bulliform phytolith size was investigated. We found that the bulliform phytolith size of specimens of *O. rufipogon* native to the warmer and wetter sites of Wenchang, Hainan (field no. Z23), and Dongxiang, Jiangxi (field no. Z1), and cultivated in the paddy fields in Wuhan, was


TABLE 2 | Classification results of the discriminant function analysis.

*Thirty-six point eight percent of original grouped cases correctly classified. Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the functions derived from all cases other than that case. Of cross-validated grouped cases, 36.7% are correctly classified.*

smaller than that of native species (**Figure 8**). Similarly, in the Wuhan paddy field, the species of *O. rufipogon* native to the warmer and wetter sites of Guangdong (field nos. Z13, Z15), Guangxi (field nos. Z12, Z14, Z27), and India (field no. Z41) also have smaller bulliforms (**Figure 8**). In the tropical Hainan, under parallel climate conditions, *O. officinalis* in the aquatic environment has a significantly larger bulliform size than those of *O. meyeriana* in a dry habitat (**Figure 8**). Therefore, the sizes of bulliform phytoliths from wild rice with preferable water habitats were mostly larger than those of wild rice under relatively dry conditions.

Finally, **Figure 9** shows the comparison of bulliform phytolith sizes between *O. sativa* and *O. rufipogon* growing in the adjacent test paddy field in Wuhan. For *O. sativa*, the ranges of mean VL and HL of bulliforms fell into 37–45 and 32–48 μm, respectively. For *O. rufipogon*, the mean VL and HL ranges were 29–41 and 24–34 μm, respectively. Thus, our data also indicated that the bulliform phytoliths of *O. rufipogon* may be generally smaller than those of *O. sativa* if they were artificially grown in the same environment.

#### Correlation Analysis of Bulliform Morphometrics and Environmental Variables

Summary statistics for the eight environmental variables of different sites are given in **Table 3**. The VL and HL values of bulliform phytoliths plotted against the different environmental variables and the results are shown in **Figure 10** and **Figure 11**. MAP, MAT, MT1, and HHH had a positive correlation (*r =*  0.438–0.610) with VL and HL, and the linear regression analysis revealed that these correlations were significant (*p* < 0.01 or 0.05) (**Figure 10**), demonstrating that the bulliform phytolith sizes are affected by changes in these climatic parameters. In contrast, other environmental characteristics, including MP7, MT7, MP1, and altitude, were not significantly correlated with changes in bulliform sizes (**Figure 11**).

## DISCUSSION AND CONCLUSIONS

Xishuangbanna; YNY-1, *O. rufipogon* from Yuanjiang.

#### Causes of Variations in Rice Bulliform Phytolith Morphometry

The results of the present study indicate that morphometric measurements of bulliform phytoliths from wild and domesticated rice widely overlap (**Figure 4**), exhibiting little diagnostic potential for taxonomic identification at the species level. These results thus support the conclusion that the morphometry of bulliform phytoliths is not as informative for distinguishing between domesticated rice and wild rice, as suggested by previous studies (e.g., Pearsall et al., 1995; Wang and Lu, 2012; Gu et al., 2013). However, the above morphometric data have often been overlooked, and an increasing number of studies have recently used bulliform phytolith size as a proxy to track the rice domestication process (Zhang et al., 2012; Luo et al., 2016; Zuo et al., 2017; Qiu et al., 2019). The bulliform phytoliths from domesticated rice really were larger than those from wild ones in the same test paddy field in Wuhan city (**Figure 9**), possibly indicating that domestication may result in an increase in bulliform phytolith size. The genetic and phylogenetic signal for this bulliform size variation has not been well revealed to date. Furthermore, this increase did not necessarily result from domestication and may be caused by other factors.

Previous studies have suggested that two factors, plant maturity and environmental conditions, may affect rice bulliform phytolith size. It is suggested that the bulliform phytoliths from mature rice leaves are usually larger than those from immature leaves (Zheng et al., 2003a; Qin et al., 2006; Fuller et al., 2007). This hypothesis may be real for the plants growing in the same location, but is not supported when comparing bulliform size of rice species from

different sites (**Figure 6**). Because environmental conditions are seemingly much more important, and the effect of degree of maturity could be ignored when multiple environmental factors are considered. According to the results of the present study (**Figures 10** and **11**), from a statistical point of view, we can conclude that the larger rice bulliform phytolith sizes, as defined by the higher VL and HL values, likely occurred at the locations with higher temperature and precipitation. Therefore, the increasing trend in rice bulliform phytolith size in some archaeological records (Zheng et al., 2003b; Zheng et al., 2004; Luo et al., 2016; Zuo et al., 2017; Qiu et al., 2019) may also be caused by climatic changes during the early and middle Holocene when temperature and precipitation were gradually rising.

The present study also revealed that the growing microenvironment, such as water environment, can also influence the size of rice bulliform phytoliths. Rice growing under wetter conditions usually produced larger bulliform phytoliths than those growing under drier conditions (**Figures 7** and **8**). Therefore, except climate regimes, changes in wet/dry habitat for rice should be considered for the use of bulliform size to track the process of rice domestication.

It should be pointed out that the present study just revealed the hydrothermal condition as one of the environmental factors influencing rice bulliform size. Some factors such as plant genotypes, soil fertility, light period length on photosynthesis, and other abiotic factors may also cause these variations, which were

FIGURE 9 | Contrast of bulliform phytolith sizes from domesticated and wild rice growing at the adjacent test paddy field in Wuhan. Demonstrating that the bulliform phytoliths of these domesticated rice specimens were generally larger compared with wild specimens (*Oryza rufipogon*) in the same environment. WH, Wuhan.

not controlled or excluded for this study. Conditional plantation experiment under controlling environments and genotypes in test paddy field is needed to further test if and how water levels and temperature can affect the bulliform phytolith size.

#### Implications for Archaeology of Rice Domestication

Changes in bulliform phytolith size of rice are regulated not only by domestication, which possibly represents genetic changes, but also by environmental factors. Given that environmental factors influence bulliform phytolith size of rice and that the role of genetic background has not yet been firmly established similar to established domestication traits such as non-shattering and increased seed size, bulliform measurement was considered as a semi-domestication trait (Fuller and Qin, 2009). Therefore, the use of rice bulliform phytolith size as an index for determining domesticated plants from their wild ancestors should be conditional. In other words, if the increasing size trend of bulliform phytoliths is used to reveal the process of rice domestication, the influence of hydrothermal conditions should be excluded first.

For further archaeological use of this index, we suggest that: 1) if the time series of rice bulliform phytolith size from a region is long, then the climatic changes (fluctuations in temperature and precipitation) through time should be considered, and the results from quantitative reconstructions of paleoclimate could be used as an independent variable to explain bulliform size variation; 2) the spatial scale of studied regions should be small and without a clear climatic gradient; 3) parallel comparison

TABLE 3 | Summary of the environmental variables in the 11 sampling sites used for correlation analysis.


*MAP, Mean annual precipitation; MP1, January precipitation; MP7, July precipitation; MAT, mean annual temperature; MT1, January temperature; MT7, July temperature; HHH, relative humidity. Data of MAP, MAT, and HHH are obtained from the dataset of annual surface observation values in individual years (1981–2010) in China (http://data.cma.cn/data/ cdcdetail/dataCode/A.0029.0005.html); data of MP1, MP7, MT1, and MT7 are obtained from the dataset of monthly surface observation values in individual years (1981–2010) in China (http://data.cma.cn/data/cdcdetail/dataCode/A.0029.0004.html). These datasets belong to the National Meteorological Information Center, China, and are available online.*

of rice domestication processes in different regions using bulliform size should consider climatic differences between the regions; and 4) the changes in rice arable systems (wet/dry growing conditions) in any studied archaeological sites should be first revealed using the promising sensitive/fixed phytolith morphotype model defined by Weisskopf et al. (2015).

Even though the influence of environmental factors has been controlled or excluded, and rice bulliform phytoliths shifting toward larger sizes are interpreted as reflecting the domestication process, it is still not possible to provide a determinate range of bulliform size for identifying domesticated rice, due to the wide overlap observed in the bulliform morphometric data between modern wild and domesticated rice (**Figure 4**). Thus, rice bulliform phytolith size is a supporting rather than conclusive proxy for determining the domesticated status of rice in archaeological research. Combination of bulliform phytolith size with other established criteria can provide precise identification of wild and domesticated rice.

Finally, notably, frequent gene exchange occurs between domesticated and wild rice, and there is a co-evolutionary relationship between them (Song et al., 2006; Zhao et al., 2010; Ge and Sang, 2011; Choi et al., 2017). Recent large-scale genomic analysis showed that *O. rufipogon* populations are widely affected by the gene flow of domesticated rice, and that the existing *O. rufipogon* are actually a hybrid swarm (Wang et al., 2017). This indicates that it is difficult to rule out the interference of the domesticated rice gene flow when using the existing *O. rufipogon* species as a reference for phytolith morphological analysis. Wild rice populations in the region with higher rice farming intensity are more affected by the introgression of the domesticated rice gene, and the genetic relationship with domesticated rice is closer (Song et al., 2003; Song et al., 2006), leading to the possibility of bias in the morphometric measurements of their bulliform phytoliths. Further research utilizing archaeological rice remains combined with ancient DNA analysis (e.g., Tanaka et al., 2010; Castillo et al., 2016), may reduce the interference of domesticated rice gene flow, and thus generate credible results to establish suitable criteria for distinguishing between domesticated rice and wild rice.

#### REFERENCES


## DATA AVAILABILITY

The datasets generated during the current study are available from the corresponding author on reasonable request.

## AUTHOR CONTRIBUTIONS

HL and CW designed research. JZ, CW, and LM collected samples. CW and JZ performed experiments. CW and YG analyzed data. CW and HL wrote the paper. All authors read and approved the final manuscript.

#### FUNDING

This work is supported by the National Natural Science Foundation of China (Grant Nos. 41701233, 41830322 and 41430103), the Strategic Priority Research Program of Chinese Academy of Sciences (Grant No. XDB26000000), and the Youth Top Talent Program of the Education Department of Hebei Province (Grant No. BJ2018118).

## ACKNOWLEDGMENTS

We are grateful to Dr. Wen Zhou and Yumei Li for their assistance in the field work on rice sampling; Dr. Yajie Dong and Deke Xu for kind help with data processing. We appreciate the constructive comments from the editor and three reviewers. We also would like to thank Editage (www.editage.cn) for English language editing.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01037/ full#supplementary-material

DATASHEET S1| Morphometric data of bulliform phytoliths from the 24 studied samples.


multilocus analysis of nucleotide variation of *O. sativa* and *O. rufipogon. Mol. Ecol.* 21 (20), 5073–5087. doi: 10.1111/j.1365-294X.2012.05748.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Wang, Lu, Zhang, Mao and Ge. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Silicon Supplementation of Rescuegrass Reduces Herbivory by a Grasshopper

*Showkat Hamid Mir1 , Irfan Rashid1 \*, Barkat Hussain2 , Zafar A. Reshi1 , Rezwana Assad1 and Irshad A. Sofi1*

*1 Department of Botany, University of Kashmir, Srinagar, India, 2 Division of Entomology, Sher-e-Kashmir University of Agricultural Sciences and Technology of Kashmir, Srinagar, India*

The theory of coevolution suggests that herbivores play an important role in the diversification and composition of plant communities. A prevalent idea holds that grasses and grazing animals participated in an evolutionary "arms race" as grassland ecosystems started spreading across the continents. In this race, besides other things, silicification in the form of phytoliths occurred in the grasses, and the graminivorous herbivores responded through specialized mandibles to feed on plants rich in phytoliths. It is important to understand whether these mandibles equip the herbivores in different environments or the grasses can augment their defense by channelizing their energy in high resource milieu. Here we used rescuegrass (*Bromus catharticus*; Family: Poaceae), an alien species of South America, to understand the mechanism of resistance offered by this species against a local insect herbivore (*Oxya grandis*; Family: Acrididae), graminivorous grasshopper, in different silicon-rich environments. We used different concentrations of silicon and observed the types of phytoliths formed after Si amendments and studied the effect of phytoliths on mandible wear of the grasshopper. Silicon concentrations increased ca. 12 fold in the highest supplementation treatments. The results reveal that higher foliar silica concentration in Si-rich plants did not result in changing the morphology of the phytoliths; still the leaf tissue consumption was lower in higher Si treatments, perhaps due to mandibular wear of the grasshoppers. The study opens a new dimension of investigating the role of Si amendments in reducing herbivory.

#### Keywords: herbivory, phytolith, grass, mandible wear, silicon

#### INTRODUCTION

The grasses (Poaceae) being the fifth most diverse family of angiosperms (800 genera and more than 11,000 species) have attracted the attention of paleoecologists, particularly in respect of their evolution and diversification (Strömberg, 2005, 2011; Bouchenak-Khelladi et al., 2010; Strömberg et al., 2013; Chen et al., 2015). One prevalent idea is that the grasses and their herbivores diversified by participating in an evolutionary "arms race" during the late Cretaceous – Cenozoic era (Stebbins, 1981). The theory of coevolution proposes that the adaptations between plant species and their herbivores are reciprocal, wherein the plant anti-herbivore traits play a major role in determining the host preference and community structure (Endara et al., 2017).

#### *Edited by:*

*Martin John Hodson, Oxford Brookes University, United Kingdom*

#### *Reviewed by:*

*Scott Nicholas Johnson, Western Sydney University, Australia Paul-andré Calatayud, Institut de recherche pour le développement (IRD), France Gerald Juma, University of Nairobi, Kenya*

> *\*Correspondence: Irfan Rashid ecoirfan@yahoo.co.in*

#### *Specialty section:*

*This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science*

*Received: 20 December 2018 Accepted: 03 May 2019 Published: 24 May 2019*

#### *Citation:*

*Mir SH, Rashid I, Hussain B, Reshi ZA, Assad R and Sofi IA (2019) Silicon Supplementation of Rescuegrass Reduces Herbivory by a Grasshopper. Front. Plant Sci. 10:671. doi: 10.3389/fpls.2019.00671*

According to this hypothesis, the open-habitat grasses significantly augmented silicon accumulation in the form of phytoliths over time, and to counter tooth wear from grass phytoliths, the mammalian herbivores evolved hypsodont teeth (Stebbins, 1981; McNaughton et al., 1985).

Phytoliths are microscopic amorphous silica bodies that occur as individual cell infillings to wholly silicified tissue sections, which toughen the plant tissues (averting food intake and digestion) and wear the herbivore mouthparts (affecting their normal life) (Strömberg et al., 2016). Although, the actual capacity of grass phytoliths to wear dental tissues and their linkage to hypsodonty evolution has limited experimental evidence (Damuth and Janis, 2011) and has generated much debate more recently (Sanson et al., 2007; Lucas et al., 2013, 2017; Rabenold and Pearson, 2014; Rabenold, 2017), the role of silica-laden abrasive grass diet in the development of mandibles has been suggested in several insect taxa (Chapman, 1964; Dravé and Lauge, 1978; Patterson, 1983, 1984). Elevated mandibular wear due to increased hardness of leaves has been found in various beetles (Raupp, 1985; Wallin, 1988; King et al., 1998), bees (Michener and Wille, 1961; Kokko et al., 1993; Schaber et al., 1993), caterpillars (Korth et al., 2006), a locust (Zouhourian-Saghiri et al., 1983), a weevil (Barnes and Giliomee, 1992), and a bug (Roitberg et al., 2005); and the wear in lepidopteran larvae fed on rice cultivars has been specifically ascribed to differences in silica contents (Djamin and Pathak, 1967; Dravé and Lauge, 1978; Ramachandran and Khan, 1991). However, these experiments used the model interactions in which the plants commonly coevolved with their insect herbivore. Here we tested the model system in which there was a lack of shared evolutionary history between plants and herbivores, creating a novel interaction, wherein the insects were less equipped to face the evolutionary arms race. We selected *Bromus catharticus* Vahl., an alien grass species, and a native herbivore *Oxya grandis* Willemse for the study.

*B. catharticus* is a densely tufted, robust annual or shortlived perennial, native to South America, recently reported as an alien introduction to the flora of Kashmir Himalaya, with the potential to spread along the length and breadth of this biodiversity hotspot (Muzafar et al., 2016). Its large openly branched seed-heads have a nodding appearance, and the glumes generally do not have any awns, while the florets usually have short awns, which make it distinct from other members of genus *Bromus*. On the other hand, *O. grandis* is considered to be a grass-feeding generalist herbivore, hopping in and around the rice fields and grasslands of the study area (Reshi, 2007). *O. grandis* is a large species (over 30 mm) with fully developed tegmina which are extended beyond apices of hind femora. The supra-anal plate is flat, with the apical part lobe extended posteriorly. We hypothesized that:


3. Increased exposure of insect herbivores to silica-rich plants will lead to increase in deleterious effects by affecting the insect mandible wear.

Hence, we performed the experiment at various Si concentrations, and expected silica-polymerized phytoliths within *B. catharticus* in high Si environments will affect the mandibles of *O. grandis*.

## METHODOLOGY

## Plant Growth Conditions

*B. catharticus* seeds obtained from Integrated Grass Fodder Research Institute (IGFRI) Srinagar, India, were grown in seed trays for 2 weeks containing inert growth media (perlite). Then, the seedlings were transplanted into earthen pots (12 cm diameter × 18 cm height) filled with peat. Peat with its low silicon content is preferred as growth substrate as it provides a better control for the treatments in such environments where additional silicon can be supplemented (Nanayakkara et al., 2008). Four seedlings were planted in a pot at an equal distance from the edge of the pot. The experiment was conducted in a completely randomized design (CRD) under greenhouse conditions (15–25°C, 16:8 light:dark) for a period of 7 months till harvest.

After every 4 days, the plants from the high (T4), moderate (T3), and low (T2) soil-silicon treatments received 50 ml of 2,000, 1,000, and 500 mg/L sodium silicate (Na2SiO3·9H2O) aqueous solution respectively. Plants from the control treatment received the same amount of tap water. After the third week, plants in all treatments were supplemented with 100 ml of half-strength Hoagland's nutrient solution, which was continuously added till the end of the fifth week. After the sixth week, 100 ml of full-strength Hoagland's solution was given to all the plants. Throughout the experiment, all plants received tap water as per requirement.

## Si Analysis

The optical emission spectroscopy of atoms excited by inductively coupled plasma (ICP-OES), which is currently one of the most efficient methods for the quantitative determination of elements in materials, was used to detect the Si concentration. The method is characterized by low detection limits and a high selectivity combined with good reproducibility and accuracy. In the present study, the ICP-OES data were recorded by a SPECTRO ARCOS EOP (Germany), spectrometer. Plants were harvested at maturity and the oldest leaves were used for Si estimation. Leaf samples were ground in an electric grinder, and put in a crucible within an incinerator at 800°C for ashing. The ash was then dissolved with aqua-regia and diluted to a known volume using distilled water. A known quantity of the solution was taken in a beaker, and HNO3 was added and heated. When it started boiling, perchloric acid (HClO4) was added to it dropwise and heated till all the organic matter was destroyed. The solution was then diluted to a known volume, using distilled water. This analytical solution was directly

injected into the hot argon ICP plasma (6,000–8,000 K). The spectral line at 251.611 nm which is characteristic for Si was used for the determination of the element concentration. A commercial standard solution of Si was used for calibration of the different concentrations of Si in order to generate a standard curve.

#### Phytolith Types

The harvested plants from different treatments were washed with distilled water and chopped into small pieces and then placed in labeled centrifuge tubes (50 ml). The tubes were rinsed with double distilled water before oven drying the material to constant weight. The weighed samples (2 g) were transferred to porcelain crucibles. The plant material was burned for 4–6 h in a muffle furnace at 550°C. The ensuing ash was mixed with 10 ml of hydrogen peroxide (30%) and kept at 80°C for 1 h in a water bath. The mixture was washed twice with double distilled water (DW). The pellet was treated with 10 ml of 10% hydrochloric acid (1 M) and incubated at 80°C for 1 h. The mixture was washed with DW and centrifuged at 3,500 rpm for 15 min. The supernatant was poured off and the pellet was rinsed with DW, till the pellet became clear. Small amount of the dried ash was mixed with 10 ml of Gentian Violet and a drop of this mixture was put on a glass slide which was subsequently covered by a cover slip. Extra stain was drained off with a filter paper and the slide was heated gently. Phytolith morphotypes were observed under a compound microscope (Leica DM300, Wetzlar GMBH) fitted with a digital camera (DFC 320), and photographed at a uniform magnification (40×). Classification of the morphotypes extracted through this dry ashing method was done as per ICPN 1.0 (Madella et al., 2005).

#### SEM Analyses of Grasshopper Mandibles

The laboratory colony of grasshoppers (*O. grandis*) was maintained on an artificial diet in the insect-rearing conditions (25–27°C, 14:10 light: dark, 50–60% RH) of the Division of Entomology at SKUAST, Kashmir. These laboratory-reared grasshoppers (third instar stage) were individually caged (*n* = 10) in 1-L sandwich boxes and starved for 24 h, much longer than the clearance time of grasshoppers, so that all food eaten before would have passed through their guts.

Pots from different Si treatments (50 replicates × 4 treatments) were enclosed in muslin cloth bags. Half of the pots (*N* = 100) were exposed to herbivory (infested) by two individuals per pot of third instar nymphs of *O. grandis* for 20 days. After 20 days of infestation under different Si regimes, the adult grasshoppers were collected and treated with 70% ethanol. The preserved mandibles were detached from the mouth part and were cleaned in an ultrasonic shaker. Different concentrations of ethanol, i.e., 80, 90, and 95% were used for dehydration of mandible specimens. In order to make the samples conductive, they were mounted on sample stubs, and then coated with gold for 5 min using a gold sputter coater. Following coating, the samples were rounded to the sample stub using graphite paint, and the specimens were observed under the SEM (S-3000H, Hitachi, Japan) at constant magnifications.

#### Assessment of the Leaf Damage

Leaf damage due to herbivory was measured in all treatments by calculating the leaf area of all the leaves on each plant in each treatment. The leaves were put on a Leaf Area Transparent Belt Conveyor (LI-3050C), and it was made sure that the knob was tightened in a way so that the belt moves freely through the scanner head, after the scanner was fixed in the conveyor belt. Leaves were placed on the supporting platform so that they pass through the scanner head and the reading was noted from the display panel. In this way, the leaves from different Si and herbivore treatments were measured.

#### Statistical Analysis

Results are reported as means ± SE unless otherwise stated. Data were analyzed using the Student's *t*-test (*p* ≤ 0.01 and 0.001) and comparison of individual treatment groups was done with one-way ANOVA, and the multiple comparisons where each experimental mean was compared with the control mean were analyzed by Tukey's *post hoc* test after normality test by the Shapiro-Wilks method. Data showing deviation from a normal distribution were arcsine root transformed before statistical analysis. All the statistical analyses were carried out with SPSS 20.

## RESULTS

#### Si Analysis

Silicon addition to peat increased leaf silicon concentrations significantly (*p* < 0.001). In control treatments, the silicon concentration was 0.1 ppm which increased ca. 12 fold in the highest silicon treatment (**Figure 1**).

#### Phytolith Morphotypes and Epidermal Patterns

In the current study, we identified a total of 21 phytolith morphotypes in the leaves of *B. catharticus* that were classified into 6 broad groups namely, short cross shaped, epidermal

elements, long hairs cells, blocky types, globular, and bulliform cells (**Figure 2**), that usually originate in the epidermis and endodermis (Twiss et al., 1969; Lu and Liu, 2003). It is pertinent to mention that Si addition did not result in changing the morphology of phytoliths although the frequency of the phytoliths was insignificantly changed. In *B. catharticus,* both the surfaces of the epidermis, i.e., adaxial and abaxial are divided into costal and intercostal zones which differ from each other in cell composition as well as silica deposition. In *B. catharticus,* on the abaxial side, the costal zone is composed of 1–3 layers of cells, and the intercostal zone consists of 4–8 layers of cells (**Figure 3**). A single layer of cross-shaped silica cells was present in the costal region on the adaxial surface. A smaller number of short cells were present in the intercostal region while the long cells were abundant in both the costal and intercostal regions. On the abaxial side, the costal zone contains 5–6 layers of cells and the intercostal zone consists of 14–18 layers of cells (**Figure 4**). In the intercostal region, a few cells were shorter than usual long cells. Silicified prickle hairs were present on the leaf border in large numbers along the margin of the leaf.

FIGURE 2 | Different types of phytoliths present in *Bromus catharticus*, (A) cross shaped, (B) orbuscular, (C) pyramidal, (D) microhair, (E) oblong, (F) trapezoid, (G) oblong elongated, (H) globular, (I) trapezoid, (J) elongated irregular, (K) scutiform, (L) trapeziform sinuate, (M) smooth elongated, (N) elongated irregular, (O) blocky irregular, (P) long hair shaped, (Q) orbuscular, (R) acicular, (S) horn like, (T) undulated, (U,V) rectangular, (W,X) pyramidal.

FIGURE 3 | *In situ* location of phytoliths in epidermis of adaxial surface (tc, trichome; sc, short cell phytoliths; st, stomata; sc, silica cells; lc, long cells; scp, short cell phytoliths; cs, cross-shaped phytoliths).

FIGURE 4 | Abaxial surface showing *in situ* silica particles (ph, prickle shaped; lc, long cell; scp, short cell phytoliths) lining the margins of epidermis.

values with two or three asterisks are significantly different as determined by the Student's *t*-test (*p* ≤ 0.01 and 0.001, respectively). Error bars represent SE.

#### Effect of Si Concentration and Grasshopper Herbivory on Rescuegrass Leaf Area

Soil silicon addition had a positive effect on the overall leaf area; however, the consumption of *B. catharticus* by *O. grandis* was reduced in high Si treatments. Higher Si concentrations caused an approximately two fold decrease in herbivory, and the leaf area consumption was similar in the infested and uninfested treatments at higher Si treatments (**Figure 5**). Hence *B. catharticus* seemed to deter herbivore feeding in high silicon diets by making leaves less palatable for the herbivore to digest.

#### Effect of Si Amendments on *O. grandis* Mandible Wear

The herbivore feeding on a high Si diet showed deformation of the incisors (which is otherwise the strongest part of the mandibles). The results are evident in SEM micrographs (**Figure 6**).

## DISCUSSION

#### Effect of Silicon Concentration and Grasshopper Herbivory on Leaf Area of Rescuegrass

Rescuegrass accumulated a significant amount of Si in Si-rich environments. Based on Si accumulation potential, plants have been categorized into Si accumulators, intermediate type, and excluder species (Jones and Handreck, 1967; Takahashi et al., 1990), and three modes of Si accumulation in plants (active, passive, and rejective) have been proposed for these corresponding types (Takahashi et al., 1990). The resultant Si deposition has been ascribed to high-level phylogenetic position (Hodson et al., 2005). However, the present study clearly demonstrates that Si accumulation also depends on Si availability; although, the source of Si acquired by the plants depends upon the type of minerals absorbed through various processes (Sposito, 2008; Hiradate, 2012). Therefore, the estimation of the Si-supplying power of soils is mainly determined by the available silicon present in the soil (Kyuma, 2004; Sauer et al., 2006; Matsumori and Gunjikake, 2013), and as such the actual potential of accumulation is seldom realized.

## Phytolith Types and Epidermal Pattern

The occurrence of specific phytolith-forming cells in leaves of *B. catharticus* indicates that they are specialized for the defense against insect herbivory. Although Si addition did not change phytolith morphology in the present study, the phytolith frequency (particularly short cell phytoliths) was slightly higher

in higher Si treatments. Silicon reduces insect herbivory as it increases epidermal hardness and abrasiveness of the leaf and that provides resistances to the plant and reduces digestibility for the herbivore. The hardness of the plant parts on which herbivore are fed reportedly caused mandibular wear in various beetles (Raupp, 1985; King et al., 1998), bees (Kokko et al., 1993; Schaber et al., 1993), caterpillars (Korth et al., 2006), a locust (Zouhourian-Saghiri et al., 1983), a weevil (Barnes and Giliomee, 1992), and even in the stylet of a true bug (Roitberg et al., 2005). The short cell phytoliths present in the leaf epidermis may discourage both large and small herbivores by making plant tissues less palatable and/or digestible (Hunt et al., 2008; Reynolds et al., 2009) and by wearing down insect mandibles and teeth in mammals (Massey and Hartley, 2009; Müller et al., 2014).

#### Herbivore Deterrence

The concentrations of Si in the leaves of *B. catharticus* affected the feeding potential of *O. grandis*. This increased resistance to herbivory has been ascribed to increased abrasiveness and hardness of plant tissues (especially epidermal) due to deposition of silica, mostly in the form of opaline phytoliths (Kaufman et al., 1985; Salim and Saxena, 1992; Ma et al., 2001; Massey et al., 2006; Massey and Hartley, 2009), which might affect the grasshopper directly or indirectly. Si-mediated herbivore resistance acts by hindering the establishment of the insect and defense against plant penetration, which reduces the palatability and feeding efficiency. However, the demonstration of Si-laden plants acting as a mechanical deterrence due to opaline phytoliths is difficult to achieve, and there is significant scope in this research area.

#### Mandible Wear

The results of the current study reveal that the long, chiseledged incisor cusps suffered microwear once the insect fed on plants grown under high silicon treatments. In agreement with previous studies (Sasamoto, 1958; Djamin and Pathak, 1967; Hanifa et al., 1974; Dravé and Lauge, 1978; Zouhourian-Saghiri et al., 1983; Ramachandran and Khan, 1991; Goussain et al., 2002), which reported that insect herbivores feeding on elevated silicon diets suffered greater mandibular wear, the present study showed some microwear in the incisors of the herbivores fed on very high Si treatments, although no mandible wear was witnessed at lower Si concentrations. Studies of dental wear require a minimum of several days of mastication to report measurable mandibular wear, though it mostly depends on the abrasiveness of the foods ingested (Teaford and Glander, 1991, 1996; Gügel et al., 2001). The difference in dental wear could be attributed to higher phytolith content in plants that received higher Si concentrations, as the major mechanical properties controlling abrasiveness of particles are hardness, particle size, and geometry (Williams, 2005). As dental wear markers are often the only proxy system bridging extant biomes and the fossil record, this opens a new research area in the phytolith studies.

#### AUTHOR CONTRIBUTIONS

IR and ZR designed the experiment. SM, IR, RA, and IS carried out the experimental work. BH helped in insect identification and mandible wear study. IR and SM wrote the

#### REFERENCES


initial draft of the experiment. SM, IR, BH, and ZR checked the final manuscript.

#### ACKNOWLEDGMENTS

The authors are grateful to the three reviewers and the handling editor for their valuable comments that helped us to improve the quality of the manuscript. The authors also want to thank the Head, Department of Botany, University of Kashmir, for his support in completing this work.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Mir, Rashid, Hussain, Reshi, Assad and Sofi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Translocation of Phytoliths Within Natural Soil Profiles in Northeast China

*Lidan Liu1, Dehui Li2,3,4\*, Dongmei Jie2,3,4\*, Hongyan Liu5, Guizai Gao2,3,4 and Nannan Li2,3,4*

*1 College of Resources and Environmental Science, Hunan Normal University, Changsha, China, 2 Institute for Peat and Mire Research, State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration, Northeast Normal University, Changchun, China, 3 Key Laboratory of Geographical Processes and Ecological Security in Changbai Mountains, Ministry of Education, Changchun, China, 4 Key Laboratory of Vegetation Ecology, Ministry of Education, Changchun, China, 5 Resources Environment & Tourism, Anyang Normal University, Anyang, China*

Phytoliths are a reliable paleovegetation proxy and have made an important contribution to paleoclimatic studies. However, little is known about the depositional processes affecting soil phytoliths, which limits their use for paleoclimate and paleovegetation reconstructions. Here, we present the results of a study of the vertical translocation characteristics of phytoliths in 40 natural soil profiles in Northeast China. The results show that phytolith concentration decreases within the humic horizon of the soil profiles and that ~22% of the phytoliths are translocated below the surface of the studied soils. In addition, we find that the translocation rate of phytoliths varies markedly with phytolith type and that phytolith size and aspect ratio also have a significant effect. Phytoliths with length >30 μm and with aspect ratio >2 and those with length <20 μm and aspect ratio <2 are preferentially translocated compared to those with length >25 μm and aspect ratio <2. Our results demonstrate that differential translocation of phytoliths within soil profiles should be considered when using soil phytoliths for paleoclimate and paleovegetation reconstruction.

#### *Edited by:*

*Martin John Hodson, Oxford Brookes University, United Kingdom*

#### *Reviewed by:*

*Subir Bera, University of Calcutta, India Heloisa Helena Gomes Coe, Rio de Janeiro State University, Brazil*

#### *\*Correspondence:*

*Dehui Li lidh357@nenu.edu.cn Dongmei Jie jiedongmei@nenu.edu.cn*

#### *Specialty section:*

*This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science*

*Received: 10 May 2019 Accepted: 09 September 2019 Published: 17 October 2019*

#### *Citation:*

*Liu L, Li D, Jie D, Liu H, Gao G and Li N (2019) Translocation of Phytoliths Within Natural Soil Profiles in Northeast China. Front. Plant Sci. 10:1254. doi: 10.3389/fpls.2019.01254*

Keywords: phytolith, transport characteristics, paleovegetation, climatic proxy, Northeast China

## INTRODUCTION

Phytoliths are microscopic silica bodies that precipitate in or among cells of living plant tissues. Owing to their abundance and environmental sensitivity, the use of phytoliths as an environmental indicator has received increasing attention. Specifically, phytolith analysis has been widely used in paleovegetation reconstructions, such as monitoring shifts in forest–grassland boundaries, vegetation succession, and changes in alpine timberlines (Barboni et al., 2007; Ákos, 2013; Coe et al., 2013; Dickau et al., 2013; Song et al., 2016; Li et al., 2017; Novello et al., 2017). However, it has been observed that soil phytoliths are subject to preservation bias, and they can be dissolved from archaeological and sedimentary records under alkaline conditions, or due to mechanical abrasion, and partially dissolved phytoliths will more easily break into fragments (Fraysse et al., 2009; Tsartsidou et al., 2009; Cabanes et al., 2011; Novello et al., 2012; Albert et al., 2015; Prentice and Webb, 2016). In addition, under the influences of wind, surface runoff, and human activity, soil phytoliths can be horizontally migrated (Wallis, 2001; Farmer et al., 2005; Esteban et al., 2017; Bremond et al., 2017), and phytoliths maybe also translocated beneath the soil surface due to various taphonomic events (Osterrieth et al., 2009; Golyeva and Svirida, 2017). Such dissolution and translocation effects can result in the misinterpretation of poorly preserved phytolith assemblages, which reduces their reliability for palaeovegetation and paleoclimatic reconstructions. Therefore, when using soil phytoliths for paleoclimate and paleovegetation reconstructions, the effects of these processes must be considered, and a first step is to improve our understanding of modern processes affecting phytoliths by conducting a study of their translocation within soil profiles.

To date, research on the vertical translocation of soil phytoliths has been conducted in several geographical regions (Borba-Roschel et al., 2006; Bradford et al., 2006; Cabanes et al., 2012; Li et al., 2012; Boixadera et al., 2016; Inoue et al., 2016). These studies have mainly focused on characterizing changes in phytolith assemblages with soil depth, and the results indicate that phytolith quantity decreases with depth, principally within the surface layer of the soil profile (Bradford et al., 2006; Li et al., 2012). However, although the phenomenon of the vertical translocation of phytoliths can be found in undisturbed soils (Bradford et al., 2006), there are only a few studies about their translocation rates in natural soils. Experiments have been conducted using, for example, irrigation (a fluorescent labeling technique), in which phytoliths are added to phytolith-free sandy sediment or other soil types (Fishkis et al., 2009; Fishkis et al., 2010a). Changes in phytolith concentration with depth are then measured to determine the translocation rates of specific phytolith types (Fishkis et al., 2009; Fishkis et al., 2010a; Fishkis et al., 2010b). However, this experimental approach may overlook the complexity of factors influencing the vertical translocation of soil phytoliths under natural conditions, which results in an incomplete understanding of the processes involved. In addition, soil phytoliths are contributed by a wide variety of plant species, and even by mixtures of herbaceous plants and trees, but phytolith morphotypes studied so far are insufficient to be fully representative of soil phytoliths in different environments (Piperno, 2006). Moreover, most previous studies have focused on changes in phytolith morphologies with depth and sampling interval and have rarely considered the influence of soil formation on phytolith translocation.

Here, we present the results of a study of soil phytolith assemblages in 40 natural soil profiles in Northeast China and analyze the results from the perspectives of soil formation and soil horizonation. Our main aim is to investigate the rate of phytolith translocation within natural soils and the degree to which phytolith translocation depends on phytolith morphology. Our results potentially provide a basic scientific reference for the preservation characteristics of soil phytoliths, and they may help improve the reliability of phytolith-based paleoclimate and paleovegetation reconstructions in the temperate zone.

## STUDY AREA

The study area in Jilin province of Northeast China is located at 39°40ʹN–53°30ʹN, 115°05ʹE–135°02ʹE (**Figure 1**) (Ma et al., 2007). The modern climate of the area is influenced by the East Asian monsoon, which has four distinct seasons. NE China also exhibits a large variety of soil types along with the vegetation changes, although they are all characterized by a high organic matter content. The vegetation zones within the study area exhibit a northeast–southwest (NE–SW) distribution,

Liu et al. Translocation of Phytoliths Within Natural Soil Profiles

reflecting the orientation of thrust faults. In the Daxing'anling Mountains region, in the western part of NE China, which belongs to the cold temperate zone, its regional average annual temperature is −2.8°C, and average annual precipitation is 746 mm; coniferous forest is widely distributed, with *Larix gmelinii*  as the dominant species; and Brown coniferous forest soils dominate in this zone. In the Changbai Mountains region, in the eastern part of NE China, which belongs to the temperate zone, its regional average annual temperature ranges from 2 to 6°C, and average annual precipitation ranges from 400 to 700 mm (Li et al., 2001); the natural vegetation is typically mixed coniferous-broadleaved forest, characterized by *Pinus koraiensis* and *Betula costata*; and dark brown soils and black soils mainly occur in this region. Songnen Plain, in the western part of NE China, belongs to the temperate zone, situated along the eastern margin of the temperate steppe in North China; its regional average annual temperature ranges from 3.5 to 5.0°C, and the average annual precipitation ranges from 360 to 480 mm (70% of the region's precipitation falls in summer) (Li et al., 2017); locally, the vegetation changes to forest grassland, alternatively called meadow grassland, dominated for example by *Leymus chinensis* and *Stipa baicalensis*, with occasional trees such as *Populus davidiana* and *Ulmus pumila*; Chernozems and Dark brown soils mainly occur in this region. In addition, there are also several intrazonal soil types (e.g., albic, meadow, and peaty soils), which occasionally occur locally (Guo et al., 2008; Zhao et al., 2011).

#### MATERIALS AND METHODS

#### Sample Collection

We collected samples from various soil types (dark brown soil, chernozem, chestnut soil, black soil, alluvial soil, and albic soil) and corresponding topsoil samples from 40 sampling sites in NE China (**Figure 1**). Forty soil profiles were collected based on soil horizonation, and as far as possible, we sampled all the diagnostic horizons within each profile. Forty topsoil samples were also collected from the uppermost 2–3 cm of surface soil, excluding the surface litter layer. Vegetation and soil profile information for these samples are listed in **Table 1**.

#### Phytolith Extraction Methods

Phytoliths were extracted from topsoil and soil profile samples using the wet ashing method (Li et al., 2017). The soil samples were air dried overnight at 80°C and then pulverized into a powder, and 5 g of sieved soil was weighed and added to a 50-ml centrifuge tube. To remove carbonates, 10% HCl was added, and the samples were stirred regularly until the reaction ceased. Distilled water was then added, and the mixture was centrifuged three times at 2,000 rpm for 20 min. To remove organic matter, concentrated HNO3 was added and the samples heated in a water bath at 90°C until the reaction subsided. Distilled water was then added, and the samples were centrifuged at 2,000 rpm for 20 min. Phytoliths were then extracted by floatation using a ZnBr2 solution with a specific gravity of 2.38, together with centrifugation; the supernatant was collected and washed with distilled water. Next, a known number of *Lycopodium* spores was added to another centrifuge tube and mixed with 10% HCl, which was then added to the abovementioned supernatant, and the mixture was centrifuged twice. Absolute ethanol was then added to the centrifuge tube, and the mixture was centrifuged at 2,000 rpm for 20 min. Finally, one to three drops of the suspension were placed on a glass microscope slide, which was heated over a spirit lamp until all the ethanol was evaporated. Canada balsam oil (one to two drops) was added and a cover slip placed on top. Observations and identification were performed with an Olympus microscope at a magnification of ×600. At least 300 phytolith grains were counted for each sample.

In addition, a phytolith concentration is the amount of phytoliths per gram of dry soil, and the formula of phytolith concentration:

$$\omega = \frac{n \times M}{N \times m}$$

In the formula, *n* represents the number of the phytolith in each slide, *N* represents the number of lycopodium spores in each slide, *M* represents the number of lycopodium spore in a slice of lycopodium spores, *m* represents the weight of each experimental samples (g), and finally calculate the phytolith concentrations *w* (103 particles/g).

The physical composition was tested using a laser diffraction particle size analyzer (Microtrac S3500, Montgomeryville, Pennsylvania, USA), which can measure particle sizes from 0.02 to 2,800.00 μm. More detailed information about the procedure for determining the physical composition is given in Ahmed et al. (2016) and Ordóñez et al. (2016).

#### RESULTS

#### Distribution of Phytoliths Within Natural Soil Profiles

The phytolith distributions in relation to the horizonation of the studied soils are illustrated in **Figure 2** and **Table 2**. In most of the soil profiles, the depth distribution of phytoliths exhibits a consistent pattern. Phytolith concentration decreases systematically with depth, from the humic horizon (Ahorizon) to the parent material (Chorizon). However, there are several exceptions: at some sites, for example, the phytolith distribution exhibits the opposite distribution, e.g., in the profiles from Shuangyang (3), Shuangyang (4), and Fusong (6). In general, however, the depth distribution of phytoliths in the profiles exhibits a similar pattern, with highest concentrations occurring in the humic horizon.

To confirm the phytolith content of different soil horizons derived from the aboveground vegetation, we studied the relationship between phytolith concentration and soil organic matter content and found that there was a closely linear relationship between phytolith concentration and soil organic matter content, which was also found in previous studies (Zhang et al., 2011). Thus, we used the linear regression equation [*Y* =



*A, humus horizon (A horizon); B, illuvial horizon (B horizon); C, parent material horizon (C horizon); E, eluvial horizon (E horizon); R, bed rock horizon (R horizon).*

0.5098 + 2.7971*x*, where *Y* = phytolith concentration and *x* = soil organic matter content (*R* = 0.673, *F* = 123.448, *p* = 0.000)] to estimate the phytolith concentrations of the soil profiles derived from the aboveground vegetation and compared the results with the original phytolith concentrations (**Figure 3**). Except for a few sites, the predicted phytolith concentrations of the illuvial (B) and eluvial (E) horizons of the soil profiles are all lower than the original values. On average, the original values of phytolith concentration of the B and E horizons of the soil profiles are six and four times greater than the predicted values, respectively.

Based on findings of soil phytolith preservation, we also examined differences in the content of poorly preserved (shortcell phytoliths and Tabular) and well-preserved phytoliths (lanceolate, elongate, blocky, and bulliform) (Albert et al., 2006; Cabanes et al., 2011) within different horizons of the soil profiles. In ~32% of the sample sites, the depth distribution of poorly preserved phytoliths exhibits a similar pattern, with the highest content in the lower layer (**Figure 4**). Combined with **Figure 5**, specifically, in the lower soil layers, soil pH is high, whereas the content of poorly preserved phytoliths is also high.

To further assess vertical translocation of phytoliths within natural soil profiles, the proportions of large and small phytoliths are calculated. The small phytoliths mainly include short-cell phytolits (e.g., saddle, rondel, bilobate, and trapeziform sinuate) (Gu et al., 2013), whereas large phytoliths contain lanceolate, elongate, tabular, blocky, and bulliform. Subsequently, a comparison of the proportions of large and

TABLE 2 | Phytolith concentration within the horizons of the soil profiles from different sampling sites in Northeast China.


small phytoliths with depth was made (**Figure 6**). In the present dataset, the size distribution of phytoliths with depth within the profiles is consistent with pollen. In ~29% of the sample sites, the content of small phytoliths increases with depth, whereas the content of large phytoliths decreases. Thus, we conclude that, in natural soil profiles, small phytoliths prefer to distribute in the lower layers.

## Vertical Translocation Rates of Phytoliths in Natural Soil Profiles

#### Total Vertical Translocation Rate of Phytoliths

In the studied soil profiles, the original phytolith concentration of the illuvial horizon (Bhorizon) and eluvial horizon (Ehorizon) is respectively six and four times greater than the values estimated by the linear regression model (see Section "Distribution of

Phytoliths Within Natural Soil Profiles"). Here, the predicted values of phytolith concentration of the Bhorizon and Ehorizon are regarded as the actual phytolith concentration derived from the aboveground vegetation. In addition, using the above rations (i.e., six and four), we recalculated the phytolith concentration of the Bhorizon and Ehorizon of the soil profiles caused by phytolith translocation. Finally, the recalculated phytolith concentration of the B, E, and C (or R) horizons may be the result of phytolith transport from the upper layers of the soil profile. To determine the phytolith translocation rate in natural soils, we defined various phytolith translocation indices, including the total translocation rate (*T*), translocation rate of the Chorizon (*CT*), and the relative translocation rate of the Chorizon (*CT*′). The formulae are listed below.

$$T = P / (\mathcal{S} + P)$$

$$CT = P\_1 / (\text{S} + P)$$

#### *CT*' = *P P*/ <sup>1</sup>

Here, *S* is the phytolith concentration of the humic horizon (Ahorizon) (103 particles/g); *P* is the total phytolith concentration of the soil profile (including the current phytolith concentration of the B, E, and C (or R) horizons), which is transported from the surface layers, but excluding the phytolith concentration of the humic horizon (103 particles/g); and *P*1 is the phytolith concentration of the Chorizon (103 particles/g).

*T* is the total translocation rate of phytoliths in the soil profile, and it reflects the intensity of vertical translocation of phytoliths. As *T* increases, there is an increase in phytolith translocation from

FIGURE 5 | Distribution of well- and poorly-preserved phytoliths within different soil horizons in the studied soil profiles in Northeast China. A, B, C, E, and R are soil horizons.

the surface humic horizon to the lower layers of the soil profile. The larger the *T* value, the weaker are the preservation of soil phytoliths. *T* < 18 indicates that the translocation rate of phytoliths is relatively low, and thus, the phytoliths are better preserved; by contrast, *T* > 30 indicates that phytoliths are poorly preserved in the soil, and 18 < *T* < 30 represents an intermediate translocation rate. *CT* is the phytolith transport rate to the Chorizon, i.e., the translocation distance of phytoliths within the soil profile is increased. *CT* < 4 indicates that the phytolith translocation rate of the Chorizon is relatively low and that the translocation distance of phytoliths within the soil profile is relatively low, whereas *CT* > 12 represents a greater translocation distance, and 4 < *CT* < 12 represents an intermediate translocation rate to the Chorizon. *CT*′ is the relative transport rate of phytoliths to the Chorizon.

There are substantial differences in *T* values among the various sampling sites in NE China (**Figure 7**). *T* ranges mainly from 0 to 40%, and the mean transport rate of phytoliths is 22%. In addition, *CT* is 10%. Specifically, the *T* values of total phytoliths for the chernozem and chestnut soils are lower (designated "low"), with values of 16 and 17%, respectively. *T* values of total phytoliths for dark brown soils and albic soils are greater (designated "intermediate"), with values of 22 and 28%, respectively. The *T* values of total phytoliths for black soils and alluvial soils are the largest (designated "high"), with values of 30 and 37%, respectively. Thus, the translocation rates of total phytoliths are lowest in chernozem and chestnut soils and highest in black soils and alluvial soils; whereas intermediate rates occurred in dark brown soil and albic soils. The distribution

characteristics of the phytolith translocation rates among the different layers in the soil profiles are illustrated in **Figure 8**. On average, ~28% of the total translocation rate of phytoliths occurs within the Chorizon of the soil profiles, i.e., ~72% of the total translocation rate of phytolith occurs within the upper horizons. These findings demonstrate that translocation of phytoliths occurs in natural soils in NE China, but the transport distance is minor, and only a relatively small number of phytoliths are transported to the Chorizon.

#### Vertical Translocation Rates of Different Phytolith Types in Natural Soil Profiles

The translocation rates of the main phytolith types among the different sampling sites in NE China are illustrated in **Figure 9**. Evidently, there are substantial differences in translocation rate among the different phytolith morphotypes. In general, the translocation rates of short-cell, lanceolate, and elongate phytoliths are greater (designated "intermediate"), with respective rates of 21%, 25%, and 27%. The translocation rates of tabular, blocky, and bulliform phytoliths are lower (designated "low"), with respective rates of 18%, 16%, and 17%. The *CT* values of the main phytolith morphotypes also vary. Specifically, the *CTs* of lanceolate and elongate phytoliths are the largest (designated "high"), with respective rates of 14% and 13%; the *CT* values of short-cell, tabular, and blocky phytoliths range mainly from 4% to 12% (designated "intermediate"); and the *CT* of bulliform phytoliths is low (designated "low"), only 3%. These findings indicate that phytolith morphotype significantly affects the translocation behavior, with small phytoliths being translocated preferentially.

In addition, we randomly measured the maximum length and width of short-cell, lanceolate, elongate, blocky, tabular, and bulliform phytoliths. For each phytolith type, 40 phytolith particles were measured. The maximum length and width of phytoliths were measured using the measuring tools provided by MOTIC software. The method used to determine the size parameters for different types of phytoliths is illustrated in Liu et al. (2016) and Gao et al. (2017). Scatter plots of the results are illustrated in **Figure 10**. For lanceolate and elongate phytoliths, their lengths are >30 μm, and their aspect ratios (namely, length/ width ratio) are >2.5. The average length of short-cell phytoliths is 14 μm, which is smaller than that of the other phytolith types, and their average aspect ratio is 1.86. By contrast, the lengths of tabular, blocky, and bulliform phytoliths are mainly >25 μm, and their aspect ratios are 1.56, 1.53, and 0.79, respectively. Therefore, the soil phytoliths can be grouped into three categories according to their lengths, widths, and aspect ratios (**Table 3**). Combined with the results shown in **Figure 10**, it is evident that phytolith size and aspect ratio significantly affect their translocation behavior. Phytoliths with an aspect ratio >2 (e.g., lanceolate and elongate phytoliths) are all preferentially translocated, at rates mainly >18%, and the translocation rates of phytoliths with length <20 μm (e.g., short-cell phytoliths) are also mainly >18%, indicating preferential translocation. Specifically, phytoliths with length >30 μm and aspect ratio >2 and those with length <20 μm and aspect ratio <2 are preferentially translocated compared to those with length >25 μm and aspect ratio <2. Thus, it can be concluded that phytolith size and aspect ratio have a significant effect on phytolith translocation and that these attributes should be considered in future research on phytolith translocation.

Northeast China.

TABLE 3 | Definition of size and aspect ratio in soil profiles in Northeast China.


## DISCUSSION

#### Phytolith Translocation Phenomenon in a Natural Soil Profile

Several researchers have suggested that phytolith translocation within soil profiles should be considered in paleoenvironmental reconstruction (Alexandre et al., 1999; Humphreys et al., 2003), whereas others have regarded phytoliths to be immobile (Rovner, 1983). Alexandre et al. (1999) reported phytolith translocation to a depth of 2.2 m in a ferrallitic soil, with a minor accumulation of phytoliths above an impermeable clay layer at the depth of 1.3–1.4 m. Humphreys et al. (2003) attributed the distribution of phytoliths in podzolic soils mainly to their translocation by percolating water; however, by contrast, Rovner (1983) concluded that phytolith mobility could be regarded as negligible for the purpose of paleoenvironmental reconstructions, due to their weight and large size. Piperno (2006) pointed out that the magnitude of translocation was probably minimal because phytoliths typically occurred only in the upper part of recent soils and their concentration usually decreased in the B horizon. However, in recent years, it has been found that due to various taphonomic events, soil phytoliths could be translocated from the soil surface, resulting in differences in phytolith content from the surface to the deeper horizons of soil profiles (Wallis, 2001; Farmer et al., 2005; He and Zhang, 2010; Golyeva and Svirida, 2017). Denis (2017) observed a relative increase in the concentration of phytoliths in the E horizon (at the depth of 25–30 cm) of an eluvial soil within a catenary sequence under field conditions, compared with the E and A′ horizons at the same depths, which confirms the occurrence of the downward translocation of phytoliths. The results of the present study confirm the occurrence of vertical translocation of phytoliths in natural soil profiles, which is consistent with the results of previous studies. In our study, the depth distribution of phytoliths in most profiles exhibits a similar pattern, with highest concentrations occurring in the humic horizon. However, there are several exceptions: at some sites, for example, the phytolith distribution exhibits the opposite distribution. This may be a result of the combination of soil type and the climatic conditions of NE China. Winter arrives early in NE China, and the interval of freezing is long, which greatly inhibits soil biological activity. As a result, organic matter produced within a growing season is not decomposed completely, which results in the accumulation of organic matter and the formation of a thick humic horizon. Consequently, the organic matter content of the gleyed horizon is increased. Several other studies have also reported that soil phytolith concentrations were closely related to soil organic matter content (Zhang et al., 2011). For Shuangyang (4) and Fusong (7), their soil types are albic soil. Studies have also reported that the gleyed horizon (E) of albic soils is always dominated by SiO2 particles and is firmer and contains a low porosity, which would be expected to result in only a limited movement of phytoliths to the next horizon (Yan et al., 2018). For Shuangyang (3), it belongs to black soils. The degree of humification of the illuvial horizon of this soil was higher than that in the other soil types; to a certain extent, organic matter can absorb and polymerize phytoliths, resulting in the humified layer having high phytolith content. Accordingly, the phytolith concentration of the gleyed horizon at these sites is increased.

In recent years, phytolith translocation studies based on experiments (e.g., irrigation, a fluorescent labeling technique) have further confirmed the occurrence of phytolith translocation in soils (Fishkis et al., 2009; Fishkis et al., 2010a; Fishkis et al., 2010b). However, current research on the vertical translocation of phytoliths in soils is based mainly on experiments, which often overlook the complexity of factors influencing the vertical translocation of soil phytoliths under natural conditions. Consequently, this prevents a full understanding of the postdepositional processes affecting phytoliths in soils. Thus, the vertical translocation of soil phytoliths in natural soil profile should be assessed, and phytolith translocation rates in a natural soil profile should be confirmed.

In the primary stage of soil formation, soil material mainly consists of lithophytes such as lichen and moss. Phytoliths are particles of hydrated silica (SiO2•*n*H2O) of phytogenic origin present in the tissues of many vascular plants or bryophytes, and they are typically deposited in plant cells or in the intercellular spaces of plants (Piperno, 2006). In addition, phytolith fragments have been observed in bryophytes, but morphogenetic phytoliths do not exist in bryophytes (Piperno, 2006). In the primary stage of soil formation, morphogenetic phytoliths do not exist in the soil; that is say, if phytoliths do occur, their morphology should be markedly different from those observed so far because the phytolith morphologies observed so far have a constant species source. Therefore, the C horizons of natural soil profiles developed on bedrock and consisting of weathering products may not contain phytoliths, and any phytoliths present are possibly derived from phytolith translocation from the upper soil layers.

*Soil organic matter content*. Soil organic matter is the carboncontaining component in soil and consists of residues of various plants and animals, soil microorganisms, and decomposed and synthesizes substances. When the parent plants die and decay, the phytoliths are preserved in soils and sediments on timescales of up to millions of years. Thus, soil phytolith concentration is intimately related to soil organic matter, and previous research has shown that soil phytoliths are closely related to soil organic matter (Zhang et al., 2011). Generally, the predicted phytolith concentrations of the illuvial (B) and eluvial (E) horizons of the soil profiles are all significantly lower than the original values. It is likely that, over time, soil phytoliths are dissolved, broken, and lost due to various taphonomic processes, and therefore, the phytoliths of the soil profiles derived from the aboveground vegetation predicted by the regression relationship are larger than the actual values. Thus, we suggest that, when a soil horizon develops within a profile, the phytolith concentration derived from the aboveground vegetation is substantially less than the current measured value in that horizon. Hence, for a given soil horizon, its excess phytoliths are possibly caused by the translocation of phytoliths from the upper layers, rather than supplied from the aboveground vegetation, although its source is not the only translocation process. This suggests that the translocation of phytoliths possibly occurs in natural soils.

*Soil pH*. Soil pH is an additional factor affecting phytolith preservation. Soil pH varies with soil type, depth, and horizonation. In the studied soil profiles, pH increases with depth (**Figure 4**). The pH values range mainly from 3 to 9, and only a few sites have pH values exceeding 9. Several studies have indicated that phytoliths are well preserved within the soil pH range of 3–9, whereas when soil pH exceeds 9, they are readily dissolved (Bremond et al., 2017). In addition, it has been found that, when soil pH exceeded 8, there was an increase in the number of phytoliths affected by dissolution (Fraysse et al., 2006; Karkanas, 2010). Thus, soil phytoliths are poorly preserved under alkaline pH conditions, and hence, the pH of the lower layers within a soil profile inhibits phytolith preservation or results in their complete dissolution. In our study, in the lower soil layers, soil pH is high, whereas the content of poorly preserved phytoliths is also high. This trend is consistent with the influence of soil pH on phytolith preservation. The preservation of soil phytoliths is likely influenced by numerous factors, and the mechanisms involved are poorly understood. In the present study, soil pH is likely a major factor affecting phytolith preservation (see also Li et al., 2005; Fraysse et al., 2006). Hence, we infer that the high content of poorly preserved phytoliths in the lower layers of some studied soil profiles is at least partly the result of phytolith translocation.

*Phytolith size*. It has been suggested that the downward movement of pollen in soils results from the downward percolation of surface water and that if the process occurs to a significant extent, pollen grains will be separated by size, with the concentration of small pollen grains increasing with depth (Walch et al., 1970). By analogy, if there is substantial phytolith translocation within a soil profile, the content of small phytoliths should also increase with depth. In the present dataset, the size distribution of phytoliths with depth within the profiles is consistent with this inference. The content of small phytoliths increases with depth, whereas the content of large phytoliths decreases. Thus, we conclude that, in natural soil profiles, phytoliths may be translocated from the upper to lower layers.

In conclusion, potential translocation exists in soil phytoliths, and the translocation bias of soil phytoliths is a concern for deeptime studies, as this would improve their accuracy with respect to phytolith assemblage reflecting original ecosystem types. However, more investigations are needed to further understand how soil phytoliths translocate to lower layers of soil profile before conducting a phytolith-based paleovegetation reconstruction.

#### Phytolith Translocation Rates in a Natural Soil Profile

An experimental study of the translocation rates of phytoliths in loamy and sandy soils confirmed this phenomenon (Fishkis et al., 2009, Fishkis et al., 2010a; Fishkis et al., 2010b). Fishkis et al. (2009) investigated phytolith translocation in sandy sediments under different rainfall conditions and found that, under highfrequency irrigation, 22% of the applied phytoliths were removed from the application layer. In addition, the results of the present study demonstrate that ~22% of the phytoliths were transported below the surface of natural soils in NE China. The phytolith translocation rates observed in our study are consistent with the results of other studies of experimental studies of phytolith translocation within soils.

We also observe differences in the vertical translocation rate of phytoliths among the studied soil types. The total translocation rates in chernozem and chestnut soils, and to a lesser extent in dark brown soil and albic soils, are all significantly smaller than in black soils and alluvial soils. In dark brown soils, they have a loose and porous structure, which is conducive to the vertical translocation of phytoliths, while the stronger earthworm activity in the humus horizon of dark brown soils may enhance the leaching of phytoliths. For albic soils, studies have shown that the clay particles in albic soils can be transported downwards with percolating water (Xiu et al., 2019). Notably, we found that the content of clay particles in the B layer and lower layers of albic soils was higher than in the upper layer (**Table 4**). Therefore, mechanical leaching occurs in these soils, namely the displacement of clay particles, which is consistent with the results of previous studies (Institute of Forestry Soil, Chinese Academy of Sciences, 1980). Moreover, soil clay particles can adsorb silicic acid; within a specific pH range, as the soil pH and the content of soil clay particles increase, the adsorption of both clay particles and silicic acid also increases (Zhang and Zhang, 1996). Therefore, with the mechanical leaching of clay particles in albic soils, the vertical translocation rate of phytoliths may also be relatively high. For alluvial soils, these soils are coarse textured, containing gravel particles with large interstices (**Table 4**). In addition, the sampling sites of alluvial soils are mainly located in the eastern mountainous and forested region of Northeast China, where erosion by rainfall and flowing water is strong. These conditions favor a high translocation rate of phytoliths. Conclusively, these findings all demonstrate that soil type is an important factor in determining phytolith translocation rates.

#### Effect of Phytolith Size on Its Translocation

Research on the vertical translocation of different phytolith types has been carried out in several geographical regions, and it has concentrated on an experimental approach (Fishkis et al., 2009; Fishkis et al., 2010a; Fishkis et al., 2010b). In these studies, plant phytoliths were added to sandy sediment (free of phytoliths) or other soil types, and changes in phytolith concentration with depth were observed to determine the translocation rates of different phytolith types. However, soil phytoliths are derived from a wide variety of plant species (including both herbaceous plants and trees), and therefore, the limited number of phytolith morphotypes used in experimental studies does not enable a comprehensive analysis of the factors affecting phytolith translocation. Our study of the translocation rates of different phytolith morphotypes in natural soils has revealed contrasts in translocation rates among different phytolith types, with smaller phytoliths being preferentially displaced. Previous studies have also shown that phytolith shape (such as length/width ratio) had a significant effect on translocation, with small phytoliths being most affected (Locke, 1986; Fishkis et al., 2010a). Our results are also consistent with those of column experiments on soil colloids, microorganism, and biochar in packed sand or soil (Weiss et al., 1995). Gannon et al. (1991) observed the more rapid translocation of small microorganisms in soil columns


*A, humus horizon (Ahorizon); B, illuvial horizon (Bhorizon); C, parent rock horizon (Chorizon); E, eluvial horizon (Ehorizon).*

compared with large microorganisms. In addition, Zhang et al. (2010) reported that coarse biochar was readily deposited during mechanical filtration, whereas fine biochar was preferentially displaced. Similarly, Sun et al. (2012) reported that, in the same soil, there was a stronger surface adsorption effect between large soil colloids (2,049.9 nm) and the rest of the soil, compared to that observed for small soil colloids (246.15 nm), and the effect of this phenomenon was to reduce the movement of large soil colloids. As in the case of soil colloids, microorganisms, and biochar, phytolith size has a significant effect on translocation, with phytoliths of smaller diameter being preferentially translocated. Overall, our results, together with those of previous studies, emphasize the need to consider the effects of differential phytolith translocation in studies which attempt to use soil phytoliths for paleoenvironmental reconstruction.

Our results also demonstrate that aspect ratio has a significant effect on phytolith translocation: phytoliths with length >30 μm, aspect ratio >2 and those with length <20 μm and aspect ratio <2 are preferentially translocated compared to those with length >25 μm and aspect ratio <2. These findings contrast with those of previous studies. For example, Weiss et al. (1995) reported a higher fraction of round bacteria in effluent passing through a packed sand column compared to the inflowing suspension. Similarly, Salerno et al. (2006) observed the more rapid translocation of rounded polystyrene latex particles compared to elongated particles in columns composed of glass beads. We suggest that the discrepancies between our results and these other studies reflect the different transport mechanisms of bacteria or soil colloids compared to phytoliths. Whereas the transport of bacteria or soil colloids is mainly controlled by diffusion and surface interactions, that of phytoliths latter is strongly affected by hydrodynamic shear and mechanical capture in small pores (Bradford and Bettaha, 2005, Bradford et al., 2006; Foppen and Schijven, 2006). Hence, the preferential transport of circular bacteria or colloids could be attributed to the smaller specific surface of rounded versus elongated particles, whereas the preferential transport of elongated phytoliths maybe due to the higher probability of detachment by water flowing through different pores. In addition, the results of correlation analysis of phytolith translocation rate and soil clay content indicate that the relationship between translocation rates of different phytolith types and soil clay content varies: translocation rates of phytoliths with length >30 μm and aspect ratio >2 and with length <20 μm and aspect ratio <2 are significantly positively correlated with soil clay content (*p* < 0.05). In contrast, the translocation rates of phytoliths with length >25 μm and aspect ratio <2 are negatively correlated with soil clay content (*p >*0.05) (**Table 5**). Our study provides direct evidence for a close relationship between phytolith translocation rate and soil clay content, which indicates the preferential adsorption of elongated phytoliths and small phytoliths by soil clay particles. In other words, when clay particles are translocated within the soil profile, the translocation rate of elongated phytoliths and small phytoliths is increased. The pronounced differences in total translocation rates among different phytolith types results in differences in the phytolith characteristics of soil horizons. Hence, differences in the percentages of different phytolith types with depth within a soil may reflect not only changes in vegetation but may also reflect the differential translocation of different morphotypes. This effect clearly needs to be considered in paleoenvironmental studies using phytoliths.

## CONCLUSION




*\*Correlation is significant at the 0.05 level (two-tailed).*

with length >25 μm and aspect ratio <2. These results indicate that phytolith size and aspect ratio should be considered in future studies of phytolith translocation.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the manuscript/supplementary files.

#### AUTHOR CONTRIBUTIONS

DJ designed the study. LL, HL, GG, DL, and NL organized field work. LL and HL carried out phytolith analysis. LL analyzed

#### REFERENCES


statistics and designed most figures, except one which was designed by NL. All authors helped to write and proofread the manuscript.

## FUNDING

This study was supported by the National Science Foundation of China (grants 41901097, 41971100, 41771214), the Natural Science Foundation of Hunan Province, China (grant, 2019JJ50371), Scientific Research Project of Education Department of Hunan Province, China (grant 18B029), and the Construct Program of the First-class Discipline (geographic science) in Hunan Province, China.

de Janeiro, Brazil): evidence from soil phytolith assemblages. *Quat. Int.* 287, 63–72. doi: 10.1016/j.quaint.2012.02.044


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Liu, Li, Jie, Liu, Gao and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Influence of Moisture and Temperature Regimes on the Phytolith Assemblage Composition of Mountain Ecosystems of the Mid Latitudes: A Case Study From the Altay Mountains

Marina Y. Solomonova<sup>1</sup> , Mikhail S. Blinnikov 2,3 \*, Marina M. Silantyeva<sup>1</sup> and Natalya Y. Speranskaja<sup>1</sup>

*<sup>1</sup> Faculty of Biology, Altay State University, Barnaul, Russia, <sup>2</sup> Department of Geography and Planning, St. Cloud State University, St. Cloud, MN, United States, <sup>3</sup> Archaeometry Center, Kazan Federal University, Kazan, Russia*

#### Edited by:

*Terry B. Ball, Brigham Young University, United States*

#### Reviewed by:

*Rand Evett, University of California, Berkeley, United States Dongmei Jie, Northeast Normal University, China*

> \*Correspondence: *Mikhail S. Blinnikov msblinnikov@stcloudstate.edu*

#### Specialty section:

*This article was submitted to Paleoecology, a section of the journal Frontiers in Ecology and Evolution*

Received: *06 November 2018* Accepted: *04 January 2019* Published: *29 January 2019*

#### Citation:

*Solomonova MY, Blinnikov MS, Silantyeva MM and Speranskaja NY (2019) Influence of Moisture and Temperature Regimes on the Phytolith Assemblage Composition of Mountain Ecosystems of the Mid Latitudes: A Case Study From the Altay Mountains. Front. Ecol. Evol. 7:2. doi: 10.3389/fevo.2019.00002* Background and Aims: Reconstruction of past ecosystems requires a robust understanding of modern deposition patterns and taphonomy for the proxies utilized. Recent advances in phytolith analysis have contributed to improved understanding of these processes, but many gaps remain. This study aims to test a few specific hypotheses that have been proposed by research outside the tropics in the Northern Hemisphere. Our study area focuses on the Northern Altay, a culturally important region, entirely within Russia, north of China, and Mongolia. We collected 60 phytolith assemblages from modern soils at 300 to 2,300 m a.s.l. elevations, sampled from 20 plots in triple replicates within 13 different plant communities. Detailed releves of these plant communities, including forests, meadows, steppe, and alpine tundra, were obtained during the summer of 2017. We used a locally derived scheme of V. P. Sedelnikov to assign studied communities to ecological categories based on moisture and temperature availability.

Methods: Standard oxidation and heavy liquid flotation methods of extraction were used. Morphotypes were counted under 400–1,000x magnification on an optical microscope. We used a two-tier approach to phytolith morphotypes classification: a detailed one with over 40 morphotypes included and a shorter one with only sums of selected morphotypes. The former approach can produce some interesting results, such as using various types of rondels (e.g., pyramidal vs. keeled) or large vs. small lanceolate (trichomes). Using sums may be more widely applicable, because the researchers can replicate these results better and less training is needed. However, there are fewer diagnostic options with the sums approach.

Key Results: Our results, using discriminant analysis, canonical correspondence analysis and other multivariate statistical methods, confirm earlier studies, both in the region and elsewhere that despite redundancy in phytolith distributions in soils, there are some selected morphotypes that can reliably distinguish communities at various positions along elevational, moisture, and temperature gradients. We developed a regionally diagnostic key that allows researchers to quickly identify various plant communities based on their phytolith assemblages in soils.

Conclusions: Seven of 13 regionally important communities at medium elevations in the Altay Mountains can be distinguished by using aggregated and more detailed phytolith morphotypes.

Keywords: elevational gradient, modern soils, phytoliths, plant communities, the Altay

## INTRODUCTION

Despite much recent progress in describing modern phytolith assemblages from temperate soils around the world (Blinnikov et al., 2013; McCune and Pellatt, 2013; Traoré et al., 2015; Gavrilov and Loyko, 2016; Lada, 2016; Feng et al., 2017; Gao et al., 2018), including the Russian Altay (Speranskaja et al., 2018), a number of issues persist that hamper phytolith use in the identification of past communities. First, many morphotypes are highly redundant and are found across many communities of various composition at similar concentrations (Blinnikov, 1994, 2005; Fredlund and Tieszen, 1994). Second, almost all studies done in temperate regions demonstrate much lower phytolith accumulation under forests as compared to grasslands (Beavers and Stephen, 1958; Verma and Rust, 1969; Volkova et al., 1995; Hyland et al., 2013) and this has to be accounted for in the interpretation of paleoassemblages. Third, outside of grasses, sedges, conifers, ferns, and sunflower families, few truly diagnostic forms exist that allow unequivocal identification of taxa in the temperate regions, although "trees and shrubs" as a rule can be detected (Blinnikov, 1994, 2005; Yost et al., 2013; McCune et al., 2015). Fourth, conifers produce phytoliths that may be easily confused with grasses (Klein and Geis, 1978; An, 2016) and this may lead to misinterpretation of some assemblages. Fifth, certain morphotypes demonstrate higher solubility in sediments and may be therefore underrepresented in the paleoassemblages (Cabanes et al., 2011).

Our earlier paper (Speranskaja et al., 2018) demonstrated utility of phytoliths in differentiating forests, meadows, and steppes in the Altaysky Kray region of Russia, in the lowlands. This study uses a new dataset collected from the neighboring and considerably more mountainous Republic of Gorny Altay, a culturally important region of the world (Reich et al., 2010). We use bi-level classification of phytoliths across the main elevational gradient (and associated temperature and moisture gradients) to complement the previous study from the foothills and the plains. The following research questions were investigated:


specific types of rondels or lanceolate forms to distinguish communities)?


## MATERIALS AND METHODS

#### Area of Study

Sixty samples of modern topsoil were collected from 20 different sites (three replicates per site, same vegetation) in the northern Republic of Altay, Russia in the summer of 2017 (**Figure 1**). Samples were collected from four mountain ranges comprising Northern Altay within the Republic of Gorny Altay in Russia: Anuy (highest point 1,815 m a.s.l.), Iolgo (highest point 2,618 m a.s.l.), Seminsky (highest point 2,507 a.s.l.), and Cherginsky (highest point 2,014 m a.s.l.) (**Table 1**).

The mean January temperature ranges from −28 to −16◦C and the mean July temperature ranges from +8 to +20◦C. The mean annual precipitation ranges from 500 mm at lower elevations to 800 mm at higher elevations, especially on the slopes with western exposure. The elevations range from 340 to 2,400 m a.s.l. Elevation and locational data were obtained in the field using a hand-held Garmin GPS unit to ∼5 m horizontal accuracy and corrected using topographic maps, as necessary. Temperature of the warmest month (July mean 1970–2017) and mean precipitation values (1970–2017) were estimated using a proprietary gridded GIS dataset of the Altay State University physical geography and GIS department derived from both published and unpublished sources, and estimates of local orographic variation. Additionally, we relied on a locally derived scheme of Sedelnikov (1988), who classified all habitats in our region into a few distinct classes of longterm temperature and moisture regimes ranging from warm to cool to cold and from wet to semi-wet to semi-dry to dry.

To assign site to a class, we performed vegetation community identification based on plant composition, visual appearance, and slope position. The advantage of using this classification is that it reflects a long-term climatic signal of each site (e.g., hydric vs. xeric).

Twenty geobotanical relevees and herbarium sampling of all graminoids and many forbs were obtained, one for each site. Plant cover was estimated to about 5% accuracy, and below that value only presence was noted. When plant identification in the field was not sufficient, additional identification was made using botanical keys at the herbarium of Altay State University. Thirteen different plant communities were sampled: pine, spruce, larch, birch-larch, and Siberian cedar pine forests; dry meadow and steppe meadows; true steppe and meadowsteppe; subalpine meadow, alpine meadow, alpine birch-heath, and alpine heath. The data on phytolith production in plants from the region are available from the research team, but were published elsewhere (Speranskaja et al., 2018).

At each site, called Bigplot in the Results section, all above ground plant diversity was described by percent cover in mid-July of 2017, the peak flowering season (**Table 1**). Aggregated random pinch samples of the upper 1 cm of topsoil, cleared of all litter, were collected within each site on three 10 × 10 m plots randomly chosen within the larger site. Approximately 10 g of dry soil was collected. Care was taken not to include any large plant pieces. Soil was dried and sieved in the lab at a coarse sieve to remove smaller fragments of plant matter before being subject to chemical treatment.

#### Lab Procedures

Lab processing followed procedure of Golyeva (2007) as modified in Speranskaja et al. (2018) ∼40 g of soil was boiled in 15% hydrochloric acid for 1 h to destroy carbonates and most organics. After that, the residue was cooled to 20◦C and sand was removed by rapid sieving through a 250 micron sieve and settling for 30 s to the 15 cm depth in a graduated cylinder. The residue below the 15 cm mark in the cylinder and on the sieve was mainly sand fraction and was discarded. The remaining suspension of clay and silt fractions was then subjected to a few cycles of gravity sedimentation and decantation to remove suspended clays near the top (after 3 h, repeated 3–7 times) and the pH was neutralized. Phytoliths were floated in a heavy liquid solution of CdI2 and KI at 2.3 g/cm<sup>3</sup> . The samples were mixed thoroughly with a glass rod and centrifuged for 10 min at a slow speed (∼1,000 rpm). The floated phytoliths were collected by a Pasteur pipette from the top 5 mm of the solution, transferred to clean test tubes, sunk by adding distilled water in proportion of 3:1, and dried. The phytolith-rich residue was stored dry in glass vials.

Phytolith abundance was estimated as percent of the original dry weight of the soil sample. Phytoliths were counted in immersion oil under an Olympus optical microscope (x400-x1000) to examine 3D shapes under

#### TABLE 1 | Plant communities analyzed in this study.


#### TABLE 1 | Continued


*Habitat ranking on temperature and moisture regimes from Sedelnikov (1988).*

rotation. Between 500 and 600 phytoliths were counted per sample. Phytolith morphotypes were documented by light microphotographs and permanent reference slides.

All identifiable phytoliths larger than 10µm were counted, not only short cells (rondels, bilobates, polylobates, and saddles), but also long cells and other grains of identifiable shape. We followed the classification system of Blinnikov (2005), originally modified from Mulholland (1989), and Fredlund and Tieszen (1994), in describing grass morphotypes; and Bozarth (1992) and Piperno (2006) in describing non-grass morphotypes (**Figure 2**). We also provide descriptions following the Glossary for the International

FIGURE 2 | Microphotographs of morphotypes. Microphotographs of all phytolith morphotypes used in this study. Scale bar = 50µm and applies to all photographs. A1—polylobate trapeziforms; B2—wavy plates; C3—saddles; D4—True bilobates (Panicoid); E5—cross (symmetrical quadrilobate); F6—trapeziform bilobate ("*Stipa*-type"); G7 conical of *Carex*, G8—conical with wavy bottom, H—rondels, including H9—low conical rondel, H10—tall conical rondel, H11—spherical bottom rondel, H12—elongated rondel; H13—saddle-top rondel; H14—low trapezoid (pyramidal) rondel; H15—single-keeled rondel; H16—tall trapezoid (pyramidal) rondel; H17—multiple keeled rondel; I—lanceolates (trichomes), including I18—with large base and short awn; I19—triangular, I20—with small base and long awn; *(Continued)*

Frontiers in Ecology and Evolution | www.frontiersin.org

FIGURE 2 | J21—bulliform cells; K22—globular irregular; L23 spherical; M—conifer, including M24—multiangled ribbed; M25—blocky with pores (cf. conifer); M26—club-shaped with protrusions (cf. *Pinus*); M27—conifer tracheids; N—long cells of grasses (possibly some belong to sedges), including N28—psilate symmetrical; N29—psilate asymetrical; N30—psilate ribbed; N31 psilate ribbed asymmetrical; N32—psilate wavy symmetical; N33—psilate wavy asymmetrical; N34—three-angled; N35—papillate; N36—slightly indented; N37—perforated; N38—strongly indented symmetrical; N39—strongly indented asymmetrical; N40—dendritic asymmetrical; N41—dendritic symmetrical; O42—cuticle casts (not used in analysis); O43—microhairs; P44—irregular dentate; Q45—jigsaw dicot epidermis; Q46—angled dicot epidermis; Q47—irregular plates; Q48—oval plates.

Code for Phytolith Nomenclature 1.0 (Madella et al., 2005) for each morphotype.

#### Statistical Analyses

Assemblages were assessed at two different levels: most detailed (55 morphotypes; **Figure 2**) and less detailed (22 morphotypes, including sums of rondels, long cells, and lanceolates; **Figure 3**). The former classification required more careful identification by the analyst under the microscope. The latter classification was easier for other analysts to replicate (for example, instead of 8 different rondel types only the sum of all rondels was used). Whenever possible, we followed ICPN 1.0 (Madella et al., 2005) in describing morphotypes. In some samples, a limited number of large fragments of silicified epidermis were encountered. We did not include them in the analysis.

Canonical correspondence analysis (CCA) of multivariate aggregated morphotype data (ter Braak, 1986) was carried out with the help of paleontological statistics software (PAST 3.20; Hammer et al., 2001). CCA is a direct gradient analysis method, with three parametric environmental variables (elevation, temperature of the warmest month, and mean annual precipitation) used in our study to simultaneously ordinate both morphotypes and samples in one hyperspace. CCA assumes unimodal distribution of species (morphotypes, in our study) along an environmental gradient, which is true for almost all morphotypes in our study (**Figure 3**). CCA advantage over Principal Components Analysis is that the former allows direct assessment of each environmental factor contribution to the morphotype and site distribution, and also detrends values along each axis, thus removing arching of resulting scores, a problem with PCA.

Cluster analysis, discriminant analysis, and mixed-effects model using percentages of phytoliths was performed using MINITAB version 18 (MINITAB version 18., 2018). C2 data analysis software (Juggins, 2003) was used to plot the pseudostratigraphic diagram of phytolith frequency data. Discriminant analysis (Manly, 2004, ch. 8) classifies samples into groups, when you have a sample with known groups. In our case, groups were communities, and we used 22 aggregated morphotypes as predictors. The squared distance (the Mahalanobis distance) of sample x to the center (mean) of group t for linear discriminant is given by the following general form:

$$\mathbf{d}\_{\mathbf{t}}^{2}\left(\mathbf{x}\right) = \left(\mathbf{x} - \mathbf{m}\_{\mathbf{t}}\right)\mathbf{S}\_{\mathbf{p}}^{-1}\left(\mathbf{x} - \mathbf{m}\_{\mathbf{t}}\right)$$

where x is a sample, m<sup>t</sup> is column vector of length p containing the means of the predictors calculated from the data in group t, and S<sup>p</sup> is pooled covariance matrix for linear discriminant analysis.

## RESULTS

## Common Morphotypes and Phytolith Abundance

We primarily relied on the 22 aggregated phytolith morphotypes in reporting the results (**Figure 3**). Some inferences are made from the more detailed set of 55 morphotypes later in the paper. Of the aggregated morphotypes, 10 were explicitly grass morphotypes, while the rest were conifer, sedge, fern, or dicot tree, and shrub morphotypes. Lanceolate forms (trichomes) and some long cells can be produced by both grasses and sedges. All phytolith counts are reported as percentages of the total. The estimates of phytolith abundance are shown on **Figures 3**, **4**. Percentage of phytoliths extracted relative to dry weight of the original samples were the highest for steppe meadow (mean = 9.0%, n = 3), 2nd highest for meadow steppe (6%), 3rd highest for alpine meadow (5.7%), and the lowest for spruce swamp (1.4%). Most communities had values between 2 and 5%. The difference in phytolith extract means between communities is significant using one-way ANOVA test (F = 4.84, p < 0.001).

The most common aggregated morphotypes were long cells (mean value = 27.3%), rondels (mean value = 19.2%), trapeziform polylobates (13.0%), all plates except wavy (11.7%), lanceolates (6.7%), wavy plates (4%), trapeziform bilobates (3.75%), globular blocky (2.6%), true "Panicoid" bilobates (1.5%), and conical of Carex (1%). All other morphotypes, including some taxonomically important, such as conifer tracheids, Panicoid crosses, or bulliform cells of grasses, had mean values <1%. As expected, non-grass morphotypes were always a small minority of the total assemblage, even in wet spruce forest with sedges and very few grasses, or in pine forest with a heavy presence of ferns.

## Are Phytoliths on Local Plots More Similar to Each Other, Than to Other Communities?

We ran a MANOVA test using Bigplot as the model factor and 22 morphotypes as responses. The results were highly significant with Wilks' lambda F = 3.033 (p < 0.001) and Lawley-Hotteling F = 4.883 (p < 0.001). Mean values for the three local plots within each community (20 "big plots") were more similar to each other than to all other means. Not all morphotypes were statistically significant contributors to this effect. Wavy plates, polylobate trapezoids, bilobate Panicoid, bilobate trapeziform, rondels, conical, lanceolate, long cells, plates, and globular morphotypes were highly significant at p < 0.001. Conifer phytoliths, spherical psilate, and three-sided forms were less significant at p < 0.05. Rarer forms were significant, probably

morphotype values are in % of the total sum of all counted phytoliths per sample (between 500 and 600 were counted). GSSC—phytoliths of grass silica short cells. LCSum is the sum of all long cells. TP, Ic, and Ix indices are based on phytolith data and are explained in text. Temp and hydro regime values are based on the scheme of Sedelnikov (1988) and are explained in text.

because they were only found in a few plots. For example, bulliform cells and saddles, both of which can be very important indicators of ecological conditions in other regions (Fredlund and Tieszen, 1994; Brémond et al., 2005b), were very scarce in our study.

## Discriminant Analysis

Discriminant analysis (DA) indicated that all 13 communities in our study could be distinguished based on their soil phytolith assemblages (**Table 2**). Of the 60 samples, 57 (or 95%) were correctly classified when using the aggregated classification with 22 morphotypes. Of the three misclassified samples, one was a dry meadow misclassified as larch forest, and two were larch forest samples classified as a dry meadow. These two communities have the shortest squared Mahalanobis distance of 9.516 in the set compared to the longest distance of 214.558 between pine forest and alpine meadow. Based on DA, the most similar assemblages are produced by: (a) the TABLE 2 | Results of discriminant analysis performed on aggregated phytolith morphotype percentages using squared Mahalanobis distance between groups (shortest distance in bold shows most similar assemblages).



group of high-elevation communities including alpine meadow, alpine heath, and subalpine birch; (b) Siberian cedar pine forest (typically found near the upper treeline) and subalpine meadow (intermediate elevations); and finally (c) true steppe and meadow steppe of low elevations. Spruce forest and pine forest appear individually as distinct assemblages with longer distances to other, because they have unique conifer and moss morphotypes.

The linear discriminant function (**Table 3**) suggested that the following morphotypes were important for distinguishing plant community types: three-sided forms—alpine meadows, conical—birch heaths and heaths, spherical psilate—heaths, TABLE 3 | Results of discriminant analysis showing linear discriminant function for groups (highest score in each community for each morphotype is shown in bold).



#### TABLE 3 | Continued


lanceolate sum of grasses and sedges—birch-larch forest, conifer tracheids—dry meadow (see Discussion for possible explanation), globular and polygonal ribbed—larch forest, indented irregular of dicots—meadow steppe, a few different forms including conifer with pores and club-shaped—pine forest, conical, and coniferous tracheids—spruce forest, saddles, and conical with wavy base—steppe meadow, and trapeziform and true bilobates, as well as crosses and sum of all plates—true steppe. Subalpine meadows and cedar pine forests did not have one predominately discriminant form. However, these two assemblages were similar to each other. It is important to note that the Siberian cedar pine community sampled near the treeline in our study produces few phytoliths, and its forest understory is frequently similar to that of the surrounding subalpine meadows.

#### What Factors Determine Phytolith Assemblage Composition on Plots? Results of Mixed-Effects Model

A mixed effects model with 60 plots as random and other factors as fixed (20 big plots, communities, temperature regime, hydro regime, and elevation) was run on the aggregate morphotype dataset. Testing the significance of contribution of each factor to the overall assemblage composition, elevation was a significant factor (p < 0.05) for seven morphotypes (wavy plate, Panicoid and trapeziform bilobates, rondel, conical, conical with wavy base, and long cell sum), while hydro regime was an important, but not statistically significant, component of the variance for the two kinds of bilobates (Panicoid and trapeziform), for long cells, plates, and lanceolate forms. Temperature was important (but not statistically significant) for the conical morphotype. All other phytoliths were not numerous enough to produce significant results with this method.

#### Canonical Correspondence Analysis

The main gradient in our study is the elevation, which is inversely correlated with temperature (higher is colder, ∼6 ◦C MAT decrease per 1,000 m) and mostly correlated with moisture (higher is wetter, ∼150 mm increase per 1,000 m), although the spruce forest sample is the wettest habitat in the middle of the gradient due to local soil conditions. The first axis (eigenvalue 0.0367) accounts for 73% of all variability in the data and represents elevational-temperature gradient (on the left are high elevation, cold communities, on the right are low elevation, warm communities). The second axis (eigenvalue 0.0153) accounts for most of the remaining variability in the data (29%) and is mostly related to the moisture signal. The axes were significantly related to the environmental data as tested with permutation technique at 999 permutations, pvalues were <0.001. The upper left corner of the CCA diagram (**Figure 5**) is occupied by high elevation alpine heath and meadow communities characterized by rondels, polylobates, three-sided, and spherical psilate morphotypes. The lower left is occupied by mid-elevation moist spruce forest characterized by polylobates, conical, coniferous tracheids, microhairs, and

plates. The upper right is occupied by meadow steppe and true steppe characterized by wavy plates, true and trapeziform bilobates, crosses, and saddles. These communities have the assemblages most dominated by grass morphotypes (>90%). The lower right are larch and birch-larch communities with some conifer phytolith presence.

It should be noted that birch-larch and larch forests look similar to some dry meadows, and rather different from steppe communities. These are characterized by the presence of conifer phytoliths, conical with wavy bases, lanceolate,

#### globular, polygonal ribbed, and indented irregular forms. Most of the phytolith morphotypes found in the soils under these communities are from non-grasses even though ∼80% of the total phytoliths in the soil assemblage are derived from grasses in other communities.

#### Cluster Analysis

Cluster analysis (**Figure 6**) suggests two major groups of assemblages: those from true steppe, meadow steppe, steppe meadow, alpine heath, alpine meadow, birch subalpine heath

and cedar pine forests near the treeline (right side of the dendrogram) and those from pine, spruce, larch, and birchlarch forests, subalpine meadow, and dry meadows at lower elevations (left side of the dendrogram). The first group lacks conifer phytoliths, has low proportion of lanceolate forms (<5%) and has high values for rondels and some other grass silica short cells (GSSC). The second group has some conifer phytoliths in most samples, higher values for lanceolate forms (>8%), as well as lower values for GSSC. Subclusters are well defined for pine forests, birch and birch-larch forests, meadows steppes, true steppes, and heaths. In contrast, dry meadows, spruce forests, subalpine meadows, and alpine meadows do not fit into tight

clusters, probably reflecting the divergent composition of these communities.

## Can We Use Phytolith Indices (T/P, Ic, Ix) to Distinguish Specific Communities?

We propose a new index Tree/Poaceae as a modified version of the well-known D/P◦ index (Brémond et al., 2005a). In our region there are few phytoliths produced by dicotyledonous trees originally used for D in D/P◦ index in Africa, but there are several produced by conifer trees. Our T/P index is the sum of all tree and shrub phytoliths divided by grass phytoliths from short cells:

between 30 and slightly over 100. The three wettest samples from spruce forest (4) have slightly higher values (45–50) than some of the samples in categories 2 and 3. We found that aggregating morphotype sums in different ways to define this index did not improve its performance. Thus, given the paucity of chloridoid and panicoid phytoliths, the xeric signal is harder to detect than the temperature signal in our study area.

### Can Rondels and Lanceolate Subtypes Be Used to Distinguish Specific Communities?

**Table 4** shows the results of the discriminant analysis performed using only nine different rondel morphotypes:


T/P ranges from 0.01 to 0.33 with the median value 0.078 and the mean value 0.091 (**Figure 7**). The value 0.08 distinguishes all forests at lower elevations from all subalpine and alpine communities. Cedar pine forests near the treeline cannot be distinguished from non-forested communities of the alpine and subalpine zone using this index. Likewise, dry meadows of lower elevations cannot be distinguished from the forests.

Ic (%) is the ratio of phytoliths mainly produced by Pooideae to the sum of phytoliths mainly produced by Pooideae, Chloridoideae, and Panicoideae and is typically used to detect a temperature signal, with high values corresponding to lower temperatures:

Pyramidal low rondel looks like a truncated pyramid, more or less square in top view, and with height < length in side view (**Figure 2**). Sometimes, the same morphotype is called short cell square trapeziform. It is moderately common in many communities and is most common in alpine meadow, alpine birch heath, and alpine heath (values between 6 and 9% of all phytoliths), which is confirmed by the high linear discriminant scores for this morphotype in these three communities and suggesting it as a characteristic morphotype of alpine habitats.

Pyramidal tall rondel has height > length in side view, and is a rare form, most common in alpine meadow (0.9%) and for which it has the highest discriminant value, although it is also found at even higher values in some steppe meadow samples.

```
Ic(%) =
                    (
                      ′Plates_wavy′ + ′Polylobate′ + ′Bilob_trapezif′ + ′RondelSUM′
                                                                                       )
(
  ′Plates_wavy′ + ′Polylobate′ + ′Saddles
                                          ′ + ′Bilob_Panic′ + ′Crosses′ + ′Bilobtrapezif′ + ′RondelSUM′
                                                                                                           )
                                                                                                            x100
```
In our study area there are few Panicoid and no Chloridoid grasses. Nevertheless, saddles, bilobates, and crosses (quadrilobates) are in fact produced by either Stipa or wild Panicum related species that are more common at lower sites with higher temperatures. The index can be used to distinguish true steppes, meadow steppes, and pine forests of lower elevations (Ic < 90%, warmer conditions) from high elevation communities (Ic > 96%, cold conditions; **Figure 8**). However, there is a large group of communities at intermediate elevations that include larch and birch-larch forests, that have Ic (%) values similar to the alpine and subalpine communities and making it of limited use to detect the intermediate temperature signal in our area.

Iph index was not very useful in our study, because there were few saddles in most samples (the index uses the ratio of saddles to bilobates to distinguish dry from wet conditions). A more useful index for our area to detect xeric signal is:

$$\text{Ix} = \frac{\text{saddle } + \text{ trapezoidal billodes} + \text{roundelSUM}}{\text{LCSUM} + \text{LancolelateSUM} + \text{Conica}} \times 100$$

This index is able to easily distinguish the two most xeric communities (group 1) in our study: true steppes and meadow steppes, with values exceeding 120 (**Figure 9**). However, its utility for separating communities at a higher level of moisture is low. Both mesoxeric (2) and xeromesic (3) communities have values

Conical low rondel looks like a truncated cone, more or less round in top view, and with height < length in side view (**Figure 2**). It is the most common rondel type, ranging from 5 to 17.5% in most communities. Its discriminant value is highest in meadow steppe and true steppe, but because it is so common, its overall diagnostic utility is low.

Keeled type 1 rondel has a keel, instead of flat bottom, and its distribution is similar to pyramidal low rondel because it is most common in alpine meadows and alpine birch heaths (about 2%) and has high discriminant values for those community types. However, it is also slightly common (1%) in larch and pine forests and in meadow steppe.

A ratio of pyramidal low and keeled type 1 rondel relative to keeled low rondel may help distinguish alpine communities (high values) from steppes (medium values) from forests (low values; **Figure 10**).

$$\text{RR} = \text{('Rondel\\_pyramidal\\_low}' + \text{'Rondel\\_keeld\\_type1')} / \text{'} \\ \text{Rondel\\_conical\\_low}'$$

Keeled type 2 rondel is most common in birch heath and true steppe (about 2% in each). Based on DA, its absence may be characteristic of alpine meadows and help distinguish that community from birch heath.

Conical high rondel (round in top view, height > length in side view) has a high occurrence (2–12%) and a high discriminant TABLE 4 | Discriminant analysis results for nine morphotypes of rondels.


*The highest values for each morphotype are highlighted.*

value in steppe meadows. These communities are more mesic than meadow steppes and are generally found at higher elevations (>1,200 m).

Oblong rondel (elongated in top view with usually flat or slightly keeled bottom) is the rarest rondel and is only found in true steppe and meadow steppe (0.4–0.5%) as also confirmed by DA scores.

Saddletop rondel has wavy, saddle-like top, but is trapeziform in side view. It is most common in alpine heath, birch heath and slightly less common in alpine meadows and cedar pine forests near the treeline (1.5%). The morphotype is also found in true steppe (1%), at much lower elevations. Its discriminant value is highest for the two types of heath.

Round bottom rondel appears round in top view, but are not trapeziform or conical in side view, rather, their bottom half is a hemisphere (**Figure 2**). It is common only in alpine heaths (2%), but its absence is particularly indicative of the true steppe.

We distinguished three kinds of lanceolate cells (trichomes) (**Figure 2**). They are common in both grasses and sedges. Large base lanceolate is most common in birch, birch-larch and pine forests, and dry meadows (3%), while steppe and alpine communities have <1% of this morphotype (**Figure 11A**). Lanceolate form with long awn is most common in pine forests (8%), but also in larch, birch-larch, and dry meadow communities (5%) (**Figure 11B**). Finally, triangular lanceolate form is most common in spruce forests (2–5%). The lanceolates as a group are good indicators of many different forests, but not meadows or steppe, except in the forest zone (dry meadows in our study).

We also distinguished 15 morphotypes of long cells (LC), which mostly come from grasses, but some are found in sedges and conifers (**Figure 2**). One LC with three ribs apparently comes from ferns. Due to the ubiquity of long cells and their higher degree of silicification with increased moisture, we did not expect their high utility in detecting specific communities.

The following observations can be made: long cells with smooth parallel walls (LC psilate) were most common in spruce forests (12%), and in cedar pine forests and nearby subalpine meadows (∼10%). They were also common in larch and birchlarch forests and alpine heaths (6%). Most other types were rare in all communities (<1%), for example, dendritic cells of Triticeae tribe were found only in alpine meadows and heaths, dry meadows, and true steppe.

Short plates with parallel walls were by far the most common in the spruce forest (25%) as well as polygonal ribbed phytoliths (2–4%). Some rare types in this community may be contributed by mosses.

#### Diagnostic Key for Plant Communities (All Percent Values Are From the Total of All Morphotypes)

Based on the results, we developed a simple diagnostic key to quickly identify each of the 13 communities in our study area.

	- a. Coniferous tracheids may help detect spruce forest
	- a. Conifer with pores, club shaped and lanceolate with long awn may be characteristic of this community
	- a. Indented irregular plates of dicots may help detect this community
	- a. Saddles and conical with wavy base may help detect this community

numbers).......................... **Alpine meadow, alpine heath, and alpine birch-heath**

	- a. Both of these communities may also be detected by presence of polygonal ribbed and globular forms
	- b. Lanceolate may help distinguish birch-larch forests

#### DISCUSSION

Our study contributes to the growing body of knowledge regarding the ability of phytoliths in subrecent assemblages in modern soils in temperate regions of the world to detect vegetation zones and specific plant communities (Kiseleva, 1982; Blinnikov, 1994, 2005; Fredlund and Tieszen, 1994; Volkova et al., 1995; Kerns, 2001; Blinnikov et al., 2013; McCune and Pellatt, 2013; Traoré et al., 2015; Gavrilov and Loyko, 2016; Lada, 2016; Feng et al., 2017; Gao et al., 2018). In the Republic of Gorny Altay of Russia this is the first attempt of its kind and complements our study of the lowlands in the Altaysky Kray (Speranskaja et al., 2018). Overall, MANOVA results from our study confirm that neighboring samples have more similar phytolith assemblages compared to those further apart, even from similar communities.

Our study also confirms the earlier findings that some morphotypes (e.g., rectangular plates) or their sums (long cells) are highly redundant across important environmental gradients, such as elevation, temperature or moisture (Blinnikov, 2005; Speranskaja et al., 2018); yet, combinations of phytoliths can detect relatively specific community types, such as true steppes (more grasses) vs. meadow steppes (more forbs), pine vs. spruce forests, subalpine meadows vs. alpine meadows, etc. The diagnostic key we developed for this study should be more widely tested across similar communities in Central Eurasia, including for example Altay extensions in Mongolia and Kazakhstan, the Sayan mountains of Russia and possibly even Central Asian mountains in Kazakhstan, Kyrgyzstan and Tajikistan.

Some of our findings pertaining to specific mountain communities corroborate earlier work in the Caucasus (Kiseleva, 1992; Blinnikov, 1994; Volkova et al., 1995), NE China (Traoré et al., 2015; Gao et al., 2018), the western European mountains (Carnelli et al., 2001; Delhon et al., 2003), and temperate North America (Kerns, 2001; Blinnikov, 2005; McCune et al., 2015). For example, the high incidence of rondels in grassland communities and high incidence of lanceolate forms in forests reported in many of these studies is also supported by our study. The Volkova et al. (1995) study of subalpine and alpine communities of Teberda Nature Reserve in the northwestern Caucasus

demonstrated distinct assemblages in eight communities, of which five are broadly analogous to this study: alpine heath, alpine meadow with forbs, alpine tussock grass community, subalpine meadow, and mid-elevation pine forest. Many species of plants in the Altay and the Caucasus are related vicariant species (Körner, 2003). Therefore, we expect to see some broad similarities in their phytolith production. In the Caucasus study, the community with the highest proportion of rondels ("hats" in Volkova et al., 1995) was the tussock grassland (57%) and it is the "true steppe" with tussock grasses in this study (25%). The alpine forb-rich meadow had high proportion of wavy forms in the Caucasus (mean value = 11%), similar to our findings (mean value = 12%). The alpine heath community had the highest proportion of conical phytoliths of Carex in the Caucasus study (6%); in our study, this community has more of this morphotype than any other community, varying from 2 to 5%. The pine forest assemblage from the Caucasus had 7% Pinus club-shaped phytoliths. In our study, pine forests have only 0.5%, but this is the highest percentage of the three communities in which this morphotype is found.

Traoré et al. (2015) found that in NE China over 50% of phytoliths in broadleaf forests may be of tree origin, and about 20% in pine forests at higher elevations. We did not observe values that high in any of our forests, only in single percentage points; the production of tree phytoliths in the humid subtropical trees is evidently much greater than in the cold continental species.

Gao et al. (2018) studied modern phytolith assemblages in soils at 108 sites from Changbai and Lesser Kingan Mountains and Songnen grassland in NE China. They listed 5 communities, including a mixed pine forest with Pinus koraiensis, but also with oaks; larch-birch forest with Larix olgensis and Betula platyfilla; broadleaf forest with Quercus and Juglans; low elevation steppe, and sparse mixed parkland. Of the five, pine and larch-birch forests in their study are very similar to ours, albeit with different species (but the same tree genera), and their parkland is similar to our open Siberian cedar pine forests and subalpine meadows mix. Similar to our study, the proportion of woody phytoliths in their assemblages rarely exceeds 10%, while grasses account for 80–90%. Their grassland assemblage is dominated by rondels and bilobates, as is the steppe in our study. Based on discriminant analysis, larch and broadleaf forest have distinct assemblages in their study (like larch forest in ours) due to presence of some tree phytoliths, but their pine forest assemblage can be confused with grassland or parkland because of low production of distinct morphotypes in Korean pine. Interestingly, this is similar to our cedar pine forests in this study, but is different for regular pine forests of Pinus sylvestris, which has more distinct phytolith assemblages, including pine club-shaped phytoliths (also see Kerns, 2001; McCune et al., 2015 for the description of similar phytoliths in Pinus ponderosa).

Delhon et al. (2003) looked at various Mediterranean communities in the lower Rhone valley in France, including a pine forest, a reed patch, a Pooid grassland, an oak forest and a riparian forest. Their results generally correspond to ours for pine forest and for grassland: pine phytoliths do occur in the forest, but not in grassland, and rondels may represent up to 50% of all phytoliths in the grassland.

The novel aspect of our work is the utilization of specific rondel and lanceolate types that can be specific to particular communities. While usually found in small quantities, some of these morphotypes proved very useful in detecting subtle community differences, such as alpine meadow vs. alpine heath or birch-larch vs. larch forest. Some physiognomically different

communities located at comparable elevations yield similar phytolith records in this study (e.g., dry meadows vs. pine forests and subalpine meadows vs. cedar pine forests). Two explanations come to mind: first, and most obvious, is that these communities have similar sets of phytolith producers, their main dominants do not produce phytoliths and are therefore "silent" or "quiet" taxa. For example, Scotch pine produces only very small amounts of its diagnostic club shaped form, making it a "quiet taxon" (Volkova et al., 1995; Delhon et al., 2003), but grasses of the pine forest have copious production and are the same species as those in the surrounding dry meadows. Cedar pine produces almost no phytoliths ("silent taxon"), but its understory grasses are very similar to those of the surrounding subalpine meadows. Therefore, grass signal masks pine presence and is almost the same in forest and nonforest in both cases as was also noted by Gao et al. (2018) for NE China forests. A second explanation is that a high level of spatial heterogeneity creates a temporally shifting mosaic commonly observed at treelines (Körner, 2003; Onipchenko, 2004). In this case, subrecent phytolith signal reflects prior inheritance and a mixture of both communities (for example, cedar pine vs. subalpine meadow in this study), demonstrating that there may be limits to the usefulness of phytolith analysis.

Our results confirm previous research that certain narrowlydefined morphotypes, such as various kinds of rondels (Kiseleva, 1992), and to a lesser extent lanceolate forms (Golyeva, 2007), can be used to distinguish plant communities. Rondel subtypes may be particularly useful in the temperate zone of the world dominated by the Pooideae subfamily of grasses and the consequent lack of Panicoid or Chloridoid forms. Low conical rondels found in steppes can be produced, for example, by Helictotrichon pubescens (Huds.) Pilg., Stipa capillata L., and Phleum phleoides L. In cedar pine forest the same morphotype is likely from Poa sibirica Roshev. In alpine lichen heath it may be from Festuca ovina. In contrast, low pyramidal rondel in the alpine and subalpine communities may be contributed by other Festuca species, and in lowland communities by Elymus-Leymus group of species (Speranskaya et al., 2018).

Another important morphotype that could prove useful is trapezoidal bilobate ("Stipa-type" of Mulholland, 1989; Fredlund and Tieszen, 1994). In North America it was originally defined from what is now considered a separate genus Achnatherum (Barkworth, 1981), but is ironically relatively rare in true Stipa in Eurasia, which produces more saddles or tall rondels (Speranskaya et al., 2018). However, in this study, many trapezoidal bilobates were found in communities dominated by Brachipodium pinnatum (L.) Beauv., an important dominant grass of pine forests of Eurasia. This is a novel result, because this species has not been evaluated for phytolith production. This morphotype is found in small quantities in many grasses with trapezoidal polylobate forms (e.g., Agrostis and Calamagrostis).

Lanceolate forms earlier described as "forest trichomes" (those with a large base and a short awn) or "meadow trichomes" (those with a small base and a long awn), have been used by Golyeva (2007) to distinguish certain communities in European Russia. We found that large base lanceolate forms are indeed more common in birch-larch, pine, and larch forests, but not in spruce or cedar pine forests (**Figure 11A**). They are also common in dry meadows, where they may be either inherited from a prior forest or produced locally by species similar to those growing in pine forests nearby (e.g., Calamagrostis langsdorffii (Link) Trin., C. arundinacea (L.) Roth, Brachypodium pinnatum (L.) Beauv.). Many of these forms likely come from upland sedges (e.g., Carex muricata L.), not from grasses. Long-awn form is also common in the same four communities, as can be seen on the boxplots. It is not common in any meadows, except dry meadow at lower elevations near pine or larch forest, where its presence can be again explained in the same fashion as their long-base cousins.

A third, triangular lanceolate type, was found primarily in spruce forest (which is actually a swamp in this study, not an upland spruce community), which were reported also in soils under spruce forests in western Siberia (Gavrilov and Loyko, 2016); it may be contributed by sedges found in this specific habitat. We found stronger linkage between hydro regime and lanceolate abundance than with specific community type in our study. Mesophytic communities have more large base, short awn ("forest type" of Golyeva) than small base, long awn type ("meadow type" of Golyeva). More studies of lanceolate production as it related to moisture regime is needed. In a recent study of common reed from China, Liu et al. (2016) found that lanceolate forms tend to be larger in plants growing in a higher evapotranspiration regime, an effect also earlier reported from West Africa (Brémond et al., 2005b). Attempts to distinguish taxa based on identification of many different types of long cells, except dendritic forms of Triticeae tribe, were not successful.

We found utility in three phytolith indices: T/P (this study, Delhon et al., 2003) to detect forests at values >0.08, Ic (Brémond et al., 2005a) to detect temperature signal (Ic>96% are cold alpine communities), and Ix (this study, to replace Iph, which is not applicable in our region) to detect aridity (values>120 indicate xeric communities). Given the coarse resolution of available climate data and the relatively short climatic gradient of our study, we cannot recommend relying on these for derivation of reliable climate transfer functions for the Altay, although this may be possible when research is performed along longer gradients and in different settings.

## CONCLUSIONS

Sixty total samples from 20 sites (triple replicates at each site) reliably differentiated 7 of 13 studied communities, including similar communities such as true steppe vs. meadow steppe or alpine meadow vs. alpine heath. Replicate samples from each site were more similar to each other than to other samples even from similar communities on different sites.

Our study failed to distinguish dry meadow from nearby pine and larch forest, two kinds of larch forest from each other, and cedar pine forest near the treeline from subalpine meadow, even when a very detailed classification scheme of rondels and lanceolate forms was followed. However, the phytolith approach was successful distinguishing other communities. Using only rondel, lanceolate and long cell sums is a more easily replicable, practical approach for most researchers and that can still detect fairly small changes in community composition. In our study area true bilobate and saddle forms were very rare but were occasionally found among the weedy species. Bilobate trapezoids could be contributed by Stipa, as well as by Brachypodium of pine forest (a novel finding). Lanceolate forms were contributed by both grasses and sedges. The relationship between their size and abundance with moisture regime should be subject of further study using morphometric approach.

## DATA AVAILABILITY STATEMENT

Datasets are available on request. The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

## AUTHOR CONTRIBUTIONS

All authors contributed equally to this research. MMS and NS designed the study and organized field work and supervised students. MYS carried out microscopy analysis. MB analyzed statistics and designed most figures, except 1 and 2 which were designed by MYS. All authors helped to write and proofread the manuscript.

#### ACKNOWLEDGMENTS

We acknowledge financial support from the Russian Foundation for Basic Research (RFBR) 17-04-00437 Influence of ecological

#### REFERENCES


and climatic factors on the composition of phytolith assemblages in modern soils under main types of plant communities of the Northern Altai to MMS and Saigo Fund and the Foundation of St. Cloud State University to MB. Climate data were provided by N. F. Kharlamova of the Physical Geography and GIS department of Altay State University. **Figure 1** was made by M. A. Borisenko of the same department.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Solomonova, Blinnikov, Silantyeva and Speranskaja. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Soil Warming Accelerates Biogeochemical Silica Cycling in a Temperate Forest

*Jonathan Gewirtzman1,2,3\*, Jianwu Tang1, Jerry M. Melillo1, William J. Werner1, Andrew C. Kurtz4, Robinson W. Fulweiler3,4 and Joanna C. Carey1,4,5*

*1 The Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA, United States, 2 Institute at Brown for Environment and Society, Brown University, Providence, RI, United States, 3 Department of Biology, Boston University, Boston, MA, United States, 4 Department of Earth and Environment, Boston University, Boston, MA, United States, 5 Division of Math and Science, Babson College, Babson Park, MA, United States*

Biological cycling of silica plays an important role in terrestrial primary production. Soil warming stemming from climate change can alter the cycling of elements, such as carbon and nitrogen, in forested ecosystems. However, the effects of soil warming on the biogeochemical cycle of silica in forested ecosystems remain unexplored. Here we examine long-term forest silica cycling under ambient and warmed conditions over a 15-year period of experimental soil warming at Harvard Forest (Petersham, MA). Specifically, we measured silica concentrations in organic and mineral soils, and in the foliage and litter of two dominant species (*Acer rubrum* and *Quercus rubra*), in a large (30 × 30 m) heated plot and an adjacent control plot (30 × 30 m). In 2016, we also examined effects of heating on dissolved silica in the soil solution, and conducted a litter decomposition experiment using four tree species *(Acer rubrum, Quercus rubra, Betula lenta, Tsuga canadensis)* to examine effects of warming on the release of biogenic silica (BSi) from plants to soils. We find that tree foliage maintained constant silica concentrations in the control and warmed plots, which, coupled with productivity enhancements under warming, led to an increase in total plant silica uptake. We also find that warming drove an acceleration in the release of silica from decaying litter in three of the four species we examined, and a substantial increase in the silica dissolved in soil solution. However, we observe no changes in soil BSi stocks with warming. Together, our data indicate that warming increases the magnitude of silica uptake by vegetation and accelerates the internal cycling of silica in in temperate forests, with possible, and yet unresolved, effects on the delivery of silica from terrestrial to marine systems.

Keywords: silica, climate change, soil, warming, phytoliths, plants, biogeochemistry

#### INTRODUCTION

Climate change is expected to cause pervasive alterations to ecosystem structures, functions, and processes in the coming decades (Grimm et al., 2013; Scheffers et al., 2016), resulting in complex and varied feedbacks to the climate system (Shaver et al., 2000; Field et al., 2007; IPCC, 2013). Climactic warming affects ecosystem processes, such as carbon storage in plants (Melillo et al., 1993; Chapin et al., 1995; Lin et al., 2010; Melillo et al., 2011) and carbon release from soils

#### *Edited by:*

*Eric Struyf, University of Antwerp, Belgium*

#### *Reviewed by:*

*Frederic Gerard, Montpellier SupAgro, France Ivika Ostonen, University of Tartu, Estonia*

*\*Correspondence: Jonathan Gewirtzman jgewirtz@bu.edu*

#### *Specialty section:*

*This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science*

*Received: 29 April 2019 Accepted: 09 August 2019 Published: 11 September 2019*

#### *Citation:*

*Gewirtzman J, Tang J, Melillo JM, Werner WJ, Kurtz AC, Fulweiler RW and Carey JC (2019) Soil Warming Accelerates Biogeochemical Silica Cycling in a Temperate Forest. Front. Plant Sci. 10:1097. doi: 10.3389/fpls.2019.01097*

**190**

(Peterjohn et al., 1994; Rustad et al., 2001; Melillo et al., 2002; Carey et al., 2016; Melillo et al., 2017). Much research has been conducted regarding the interaction between climate change and the biogeochemical cycling of elements, such as nitrogen and carbon (Rustad et al., 2001; Melillo et al., 2011; Bernal et al., 2012; Butler et al., 2012), and the plant stoichiometry of carbon, nitrogen, and phosphorus (An et al., 2005; Elser et al., 2010; Dijkstra et al., 2012; Sardans et al., 2012). However, less attention has been paid to the effects of soil warming on biogeochemical cycling of other globally important elements, such as silicon.

Silica [silicon dioxide (SiO2)] is the most abundant compound in the Earth's crust (Brown and Mussett, 1981) and soils (Epstein, 1994; Tréguer and De La Rocha, 2013). Silica and carbon are coupled in terrestrial and marine ecosystems through processes such as mineral silicate weathering, phytolith-occluded carbon storage in soils, and primary production by terrestrial and marine silica-accumulating organisms (Street‐Perrott and Barker, 2008; Carey and Fulweiler, 2012a; Song et al., 2012). Therefore, understanding the impacts of climactic warming and environmental change on silica cycling is important for modeling and predicting future global carbon cycling.

Over geological timescales, the weathering of silicate minerals consumes carbon dioxide (CO2), making the process a significant control on atmospheric CO2 and planetary climate (Urey, 1952; Berner et al., 1983; Berner, 1990; Street‐ Perrott and Barker, 2008). Mineral silicate weathering is driven by complex interactions between the climate and lithosphere (Bluth and Kump, 1994; Hilley and Porder, 2008), as well as the biosphere (Drever, 1994; Berner, 1997; Conley, 2002; Derry et al., 2005). The terrestrial biosphere also acts as a filter for weathering-derived silica before its eventual export to oceans (Struyf and Conley, 2012). Plants take up silica as dissolved silicic acid (DSi, H4SiO4), the dominant form of Si in soil solutions (Epstein, 1994). They convert DSi to biogenic silica (BSi, hydrated SiO2) whereupon it is deposited in siliceous structures, primarily phytoliths, in the plant biomass (Sangster and Parry, 1976; Canny, 1990; Fu et al., 2002; Trembath-Reichert et al., 2015). Silica concentrations in terrestrial vegetation vary widely, with some plants taking up silica in greater proportion than macronutrients like nitrogen or phosphorus (Epstein, 1994; Hodson et al., 2005). Carey and Fulweiler (2012a) estimate that active Si-accumulating plants are responsible for more than half of terrestrial net primary production (NPP), linking atmospheric CO2 and terrestrial silica cycling on biological timescales.

BSi accumulation in plants and soil has been shown to regulate the magnitude and phenology of forest silica cycling and watershed export in some systems (Meunier et al., 1999; Fulweiler and Nixon, 2005; Struyf et al., 2009a; Clymans et al., 2016). Plants return BSi to soil chiefly as fine litterfall (Bartoli, 1983; Struyf and Conley, 2012; Clymans et al., 2016), and phytoliths from this pool can accumulate throughout the topsoil (Struyf et al., 2009b; Cornelis et al., 2011). While many factors, such as pH and species differences in phytolith structure, can affect BSi dissolution rates in soil (Wilding and Drees, 1974; Fraysse et al., 2009), BSi is 7–20 times more soluble than mineral silicates in soils, often resulting in efficient recycling to DSi (Farmer et al., 2005; Fraysse et al., 2006; Fraysse et al., 2009; Struyf et al., 2009a; Cornelis et al., 2011). Thus, the soil BSi pool can be an important supply of DSi to soil solution and streams in diverse ecosystems, particularly high-Si accumulating systems, such as grasslands and deciduous forests, and highly weathered (Si-depleted) systems, such as tropical forests (Derry et al., 2005; Gérard et al., 2008; Struyf et al., 2009a; Lugolobi et al., 2010; Struyf and Conley, 2012). In a North American temperate deciduous forest watershed, Clymans et al. (2016) found that a minimum of 50% of annual soil DSi production derives from BSi dissolution, with 98% of that supply deriving from fresh leaf litter.

Terrestrial systems supply ~78% of annual silica inputs to oceans (Tréguer and De La Rocha, 2013), where silica is essential for a wide range of species including diatoms, which are responsible for approximately half of marine primary production (Tréguer et al., 1995; Rousseaux and Gregg, 2013). Because diatom productivity can be limited or co-limited by silica availability (Nelson and Treguer, 1992; Leynaert et al., 2001; Brzezinski et al., 2008), the magnitude of silica delivery from terrestrial to marine systems can impact marine and global NPP.

Anthropogenic perturbations, such as deforestation, urbanization, and agriculture, are known to alter terrestrial silica biogeochemistry (Conley et al., 2008; Laruelle et al., 2009; Vandevenne et al., 2012; Carey and Fulweiler, 2012b; Carey and Fulweiler, 2016; Unzué-Belmonte et al., 2017). However, the role of climate change on terrestrial silica biogeochemistry remains less well known. Recently, experimental CO2 and nitrogen enrichment was shown to increase forest silica uptake (Fulweiler et al., 2015), while experimental snowpack reduction and induced soil freezing was shown to impede plant silica uptake capacity (Maguire et al., 2017). Still, while there has been much research and discussion of the impact of temperature on longterm silica geochemistry (Brady, 1991; Velbel, 1993; Brady and Carroll, 1994; Turner et al., 2010), there has been no study to date addressing the impact of temperature on terrestrial silica biogeochemistry.

To explore the effects of soil warming on terrestrial silica cycling, we analyzed BSi in soils, foliage, litter, and soil solution samples taken over 15 years of a long-term soil warming experiment (Melillo et al., 2011). We also conducted a litter decomposition experiment to explore dynamics of plant silica release under ambient and warmed conditions. We hypothesized that soil warming would increase tree silica uptake as a result of increased productivity. We also hypothesized that soil warming would increase decay rates thereby accelerating the release of silica from decomposing litter. Finally, we hypothesized that these changes would increase forest silica recycling with minor, if any, net effect on soil BSi storage over the timespan of the study.

## MATERIALS AND METHODS

#### Site Description

The Barre Woods soil warming experiment is located in an evenaged, mixed deciduous stand at Harvard Forest in Petersham, MA (42° 28′ N, 72° 10′ W). Tree species composition is dominated by oak (*Quercus rubra and Quercus velutina*), red maple (*Acer rubrum*), and American ash (*Fraxinus americana*), comprising 42%, 29%, and 11% of basal area, respectively (Melillo et al., 2011; Melillo et al., 2017). The site was historically used for pastureland or low-intensity agriculture, and then came to be dominated by white pines in the first half of the 20th century. In 1938, a hurricane destroyed much of the stand, which was then left to regrow to its current state (Melillo et al., 2011). Soils at the study site are of the Canton series, with O horizon pH of 5.2 and subsurface mineral horizon pH of 5.5 (Melillo et al., 2011). Mean weekly air temperature at the site varies from a high of approximately 20°C in July to a low of approximately −6°C in January, and mean annual precipitation is approximately 1080 mm, distributed evenly throughout the year (Melillo et al., 2011).

## Soil Warming Experiment

Complete descriptions of the warming experiment methods have been previously published (Melillo et al., 2011) and are detailed in the Harvard Forest Data Archive (Melillo et al., 2017). Briefly, in 2001, heating cables were buried in a 30 × 30-m area at 10-cm depth with 20-cm spacing. An adjacent 900 m2 plot serves as the control area, separated from the heated plot by a 5-m buffer. Starting in 2003, the heating cables were cycled on and off to maintain soil temperatures in the heated plot elevated at 5°C above control plot soil temperatures. Soil warming continued for the duration of the measurements and sample collection in this study.

#### Soil and Vegetation Sampling

Soils were sampled to 10 cm in the control and heated plots, divided visually into organic and mineral horizons, sieved to 2 mm, and dried. We subsampled soils for silica analysis from a pre-treatment year (2002), and from three other years during the study (2005, 2010, and 2016). We analyzed three samples per layer in each plot for each of the 4 years (n = 48). We subsampled only from cores taken early in the growing season.

Foliage (green leaves) was sampled by shotgun during the summer between June and August. Four to five trees from each plot were sampled, preferentially selecting sunlit leaves, which were then bulked together, dried, homogenized, and milled. Each year's foliage sample, thus, represent a homogenized sample for a given species and plot. We measured triplicate subsamples of samples from 7 years between 2003 and 2016 (approximately every 2 years during the course of the study, n = 28). We report green leaf BSi values as the mean of those three subsamples.

Leaf litter was collected in baskets installed in each plot. Wire baskets dispersed throughout the plots were used from 2003 to 2006; thereafter, laundry baskets clustered in the center of the plots were used for litter collections. The litter was collected regularly from each basket during the fall, dried, and sorted by species. Fresh leaf litter samples were kept separate by collection basket (n = 3 per plot) for all years except 2008; in 2008, samples were bulked across baskets by species and treatment. For all years, the bulked litter samples were homogenized and milled. We analyzed subsamples from the same 7 years as the green foliage (n = 84). For all years, except 2008, we analyzed samples from each of three collection baskets for each species–treatment. For 2008, we analyzed three subsamples of the bulked sample from each species × treatment and calculated a single mean BSi value for that species × plot × year.

## Soil Solution and Stream Water Sampling

In 2016, we collected soil water samples using lysimeters previously installed in the plots. Six porous cup high-tension lysimeters were installed in each plot at a depth of 50 cm and evacuated. We sampled at approximately monthly intervals from May to December 2016. Lysimeters were evacuated to ~380 mm Hg the day before sampling. We retrieved as many samples as possible on each sampling occasion; however, we often recovered fewer than six samples per plot per sampling interval due to low soil water content. Soil water was filtered through a polypropylene syringe using a 0.45-μm nitrocellulose filter immediately upon retrieval. We retrieved and analyzed a total of 40 samples (21 from the heated plot and 19 from the control plot).

On each day that we sampled from lysimeters, we also sampled stream water from a nearby stream in the Prospect Hill tract of Harvard Forest, Bigelow Brook, at the Lower Pipe stream gauge. Three stream samples were collected at each of the seven sampling time points (n = 21), within an hour after lysimeter sample collection. To do this, 60 mL of water was drawn in a polypropylene syringe and filtered using 0.4-µm polycarbonate filters to isolate suspended silica. We measured DSi in the filtered water samples and suspended BSi captured on the filters to estimate total stream water silica. All lysimeter, stream water, and filter samples were kept refrigerated until analysis.

## Litter Decomposition Experiment

In 2016, we also conducted a litter decomposition experiment where we collected litter from four common species at the site: *Quercus rubra* (red oak)*, Acer rubrum* (red maple)*, Betula* spp. (mixed birch; mostly *Betula lenta*), and*, Tsuga canadensis*  (eastern hemlock). We chose these species to align with an earlier wood decomposition study in the same soil warming plots (Berbeco et al., 2012). We collected litter from the forest floor, outside of the study plots, in October 2015. We homogenized litter by species, rinsing with deionized water to remove soil, and placed them in a drying oven overnight at 60°C. We placed approximately 5 g of litter of single species into 20 × 20 cm bags made of 5-mm fiberglass mesh. The bags were closed on all sides with an impulse sealer and tagged with unique plastic identification tags.

We placed 21 bags of each of the four species in transects in each of the two plots (21 × 4 × 2 = 168 bags total) during May 2016. We placed litterbags between the Oi and Oe horizons, with at least 10-cm spacing between bags. We tied the bags to plastic stakes with nylon string for future retrieval. Subsamples of the four litter types were used and was kept in the laboratory for analysis of their initial chemical composition. We collected three litterbags from each species at each plot at approximately monthly intervals from May through December of 2016 (3 replicates × 4 species × 2 plots = 24 bags). Upon collection, we gently rinsed and dried bags at 60°C to constant mass. We then ground the contents of bags using a Wiley mill for subsequent silica analysis.

#### Chemical Analysis

We measured BSi concentrations in all of the aforementioned samples using a wet alkaline chemical extraction in a 1% Na2CO3 solution (DeMaster, 1981; Conley and Schelske, 2001). Duplicate subsamples were taken from each sample and weighed to approximately 30 mg. We digested samples in flat-bottomed polyethylene bottles in a shaking water bath at 85°C and 100 rpm. For leaf and litter samples, a single aliquot was taken from each digestion bottle after 4 h for analysis. No separation into mineral and amorphous/biogenic fractions was necessary given that all silica contained in those samples is by definition biogenic. For soils and stream water filters, the fraction of DSi released from BSi was determined by time-course extraction (aliquots taken at 3, 4, and 5 h), followed by a linear extrapolation to the intercept (DeMaster, 1981; Saccone et al., 2007). For all digested samples, aliquots taken were of 1 mL and were neutralized in 9 mL of 0.021 M HCl.

All extracted samples, as well as soil water and stream water, were analyzed for DSi using the molybdenum blue colorimetric method (Strickland and Parsons, 1968). Standards made of sodium hexafluorosilicate, as well as external standards, were used throughout the analysis to check accuracy. All errors between duplicate samples were less than 5%. Digestions were conducted at Brown University (Providence, RI), and colorimetric analyses were conducted at Boston University (Boston, MA) using a Seal AA3 flow injection autoanalyzer.

#### Statistical Analysis

We applied a pretreatment correction factor to our soil BSi data, following Melillo et al. (2011). The pretreatment correction factor scales the initial heated data to equal the control. Pretreatment samples were only available for soil, so only the soil data are presented with this correction.

All statistical analyses were conducted using R Version 3.4.4 (R Core Team, 2018). Soil data were analyzed using linear mixed-effects models in the "nlme" R package (Pinheiro et al., 2018), with year, layer, and treatment as fixed effects and subplot as a random effect. Green leaf BSi was analyzed using linear regression with single annual foliar concentrations as a function of sampling month, species, and treatment. Leaf litter BSi was analyzed using linear mixed-effect models with year, species, and treatment as fixed effects, and collection basket as random effect. Random effects were nested within litter basket to account for repeated measures and autocorrelation.

Using public litter mass data from the soil warming experiment (Melillo et al., 2017) and our measured BSi concentrations, we calculated annual litterfall masses and litterfall BSi fluxes. We fit the data to a linear mixed effects model, with litter mass per area (log-transformed) as product of species, treatment, year, and collection basket type, with specific basket location as a random effect. We excluded the data from 2015, when a major summertime hailstorm caused widespread defoliation prior to autumn leaf senescence and abscission.

For the decomposition experiment data, we calculated the percent of initial mass remaining for each litter bag as the quotient of final dry mass and initial dry mass. We then used a single exponential model to calculate a decay constant (Olson, 1963; Berbeco et al., 2012):

$$\mathbf{k} = [\ln(\mathbf{M}\_0 - \ln(\mathbf{M}\_t)) / \text{tr}]$$

where *k* is the decay constant, Mt is the percent of remaining biomass at time, *t*, and *M*0 is 100%. We also calculated the estimated time to decompose 95% of matter (Berbeco et al., 2012), using the equation:

$$\mathbf{t}\_{0.95} = -\ln(0.05)/\,\mathrm{k}$$

We analyzed the effects of species and treatment on mass loss, elemental composition, and elemental ratios using linear regression, with each response as a function of species, treatment, time, and interactions among these variables. Decay constants were regressed against species, treatment, and interactions. Data were rank-transformed prior to analysis to meet assumptions regarding homoscedasticity.

Stream and porewater dissolved silica were analyzed using a two-way ANOVA with sampling date and DSi pool as the main effects, followed by a *post hoc* Tukey's HSD test. Data were also rank-transformed prior to analysis.

We evaluated the normality of model residuals using visual inspection and Shapiro-Wilk normality tests. Significance for all statistical tests was judged using an alpha of 0.05. We report silica concentrations by percent dry weight as SiO2 (%BSi) unless otherwise specified, and all errors reported in text and figures are standard errors of the mean.

#### RESULTS

#### Soil BSi

Soil BSi concentration across all samples (**Figure 1**) averaged 0.95% ± 0.34%, and was higher in the organic layer (1.03% ± 0.04%) than in the mineral layer (0.87% ± 0.05%).

We applied a pretreatment correction factor to our soil BSi data, following Melillo et al. (2011), to account for the fact that soil BSi concentrations were higher in the heated plot compared to the control plot before treatment began (17% higher in organic soil and 21% higher in mineral soil; uncorrected data available in **Supplementary Tables 1** and **2**). The pretreatment correction

factor scales the initial heated data to equal the control, so that we can appropriately compare control and treated samples.

and three treatment years. Error bars represent standard errors of the mean.

Pretreatment-corrected soil BSi concentrations (**Table 1**) varied significantly between organic and mineral layers (p = 0.001), but did not vary between control and heated treatments (p = 0.485) or with warming duration (p = 0.623).

Mean bulk density at the Barre Woods site was previously reported to be 0.37 g cm−3 in the organic layer and 0.78 g cm−3 in the mineral layer, with mean organic layer depth of 1.4 cm (Melillo et al., 2011). Using these values and treatment mean BSi


*BSi, biogenic silica. BSi concentrations are reported as percent dry weight BSi (SiO2). Heated plot concentrations are pretreatment-corrected.*

concentrations, we calculated BSi storage in the top 10 cm of soil for the control and heated plots (**Table 2**).

#### Foliar and Litter BSi

Silica concentrations were significantly different in foliage vs. leaf litter (p < 0.001), with litter having consistently higher BSi concentrations than green leaves. Green leaf BSi concentrations varied significantly by species (p < 0.001; **Table 3**) and by sampling month (p < 0.001). However, green

TABLE 2 | Soil BSi stocks. Soil BSi stocks were calculated for the top 10 cm in each plot, and the data reported here are means across all samples analyzed from all years during experimental treatment. Heated plot values are pretreatment-corrected.


TABLE 3 | Foliar and litter BSi concentrations. The values reported are mean concentrations of BSi (percent dry weight) across all samples analyzed (all sampled during experimental treatment, from 7 years between 2003 and 2016; further detailed in methods above).


leaf BSi concentrations did not vary between years (p = 0.817) nor with warming treatment (p = 0.149).

Red maple foliar BSi was more than double that of red oak. Red maple foliage contained 1.29% ± 0.06% BSi (by dry mass) in the heated plot and 1.16% ± 0.05% in the control plot. Red oak BSi was 0.38 ± 0.01% in the heated plot and 0.37 ± 0.01% in the control plot. Foliar concentrations varied significantly by sampling month (p < 0.001) and were higher in years when sampling was conducted in the late growing season (August) compared with years where sampling was conducted earlier in the growing season (June or July; **Figure 2A**).

Similar to foliar BSi, litter BSi varied by species (p < 0.001) but did not vary by treatment (p = 0.588). Mean litter BSi concentrations were ~1.5× higher than green foliar concentrations across years, species, and treatments (**Figure 2B**). As in green foliage, substantially higher BSi concentrations were observed in red maple litter compared with red oak leaf litter.

Although leaf BSi concentrations did not vary between treatments, litterfall production was significantly elevated in the heated plot relative to the control (p < 0.001, **Figure 3A**. Across all years analyzed, mean litter mass for red maple and red oak was elevated 29% from 227 g m−2 in the control plot to 293 g m-2

in the heated plot. Due to the increase in litter production, leaf litter BSi mass per area was significantly higher in the heated plot compared to the control (p = 0.008; **Figure 3B**). Across all years for which we analyzed samples, mean litter BSi masses were 7.96 kg BSi ha−1 in control red maple and 8.91 kg BSi ha−1 in heated red maple. In red oak, mean litter BSi masses were 10.30 kg BSi ha -1 in the control and 14.03 kg BSi ha−1 in the heated plot.

#### Si Dynamics in Litter Decomposition

In each species and treatment, the litterbags decayed to roughly half of their initial masses over the 212 days for which they were allowed to decompose (**Figure 4**). Final litterbag masses were significantly affected by time (p < 0.001), species (p < 0.001) and treatment (p < 0.001), as well as interactions between treatment and time (p = 0.002) and between species and time (p < 0.001).

Across all samples, the modeled time to decay to 95% of original biomass (t95) varied between 1.8 and 5.5 years. Decay constants *k* (**Table 4**) varied according to species (p < 0.001) and treatment (p < 0.001). Heating increased mean decay rates

by 47% for red maple, 40% for red oak, and 37% for birch. In contrast, decay rates decreased by 12% in the heated treatment for hemlock; however, it should be noted that several hemlock litterbags were either lost or damaged during retrieval, so the number of replicates was reduced and error was highest among hemlock samples.

The percent of litter composed of BSi (**Figure 5**) varied with species (p < 0.001), but we did not observe a significant effect of time (p = 0.174) or heating (p = 0.211), indicating that silica losses tracked mass losses over time.

#### Soil Solution and Stream Water DSi

Mean soil solution DSi in the control plot and stream water DSi+BSi tracked closely with one another, whereas mean soil solution DSi in the heated plot was elevated above the other two pools (**Figure 6**). Across sampling dates, mean stream DSi+BSi concentrations were 181.53 ± 8.51 μM (mean DSi = 167.47 ± 7.64 μM; mean BSi = 14.06 ± 1.50 μM). Mean control plot soil solution DSi was 182.45 ± 7.18 μM, whereas mean

form y = 100e−kt are plotted for the control and heated litterbags for each of the four species, where *k* is the mean decay constant for each species × treatment, and *t* is the number of days deployed.

TABLE 4 | Decay constants and t95 values for litterbag mass. For each set of litterbags (species × treatment), mean decay constant (k), time in years to decompose to 5% of initial mass (t95), and number of litterbags successfully retrieved and analyzed (n). Errors reported are standard error of the mean.


heated plot soil solution DSi was 253.37 ± 15.57 μM, 39% greater relative to the control. There was a seasonal pattern to porewater and stream silica concentrations, with a peak in concentrations in August. The variation between sampling dates (p < 0.001) and silica pools (p < 0.001) were significant. The heated plot concentrations differed significantly from the control (p < 0.001) and the stream (p < 0.001), whereas the control plot soil solution was not significantly different than the stream water (p = 0.244).

## DISCUSSION

This is the first study, to our knowledge, to estimate the effect of soil warming on the biogeochemical cycling of silica in a temperate forest, highlighting the impacts of long-term experimental manipulation on forest silica dynamics. We find evidence supporting our three hypotheses: first, soil warming increased net tree silica uptake, due to elevated biomass production and relatively constant leaf tissue silica concentrations. This increase in plant silica uptake was balanced by the increased return of BSi to the soil through litterfall. Second, soil warming led to an acceleration of silica release from decomposing litter: warming increased litter decomposition rates, and soil solution DSi concentrations were elevated in the heated plots. Third, these changes had no net effect on soil BSi stocks over time, likely due to silica

inputs to the soil pool (i.e., litterfall and decay) balancing outputs (i.e., silica uptake by vegetation). Together, these data indicate faster internal silica cycling in temperate forests with warming. Below, we identify the likely mechanisms driving the observed changes.

#### Soil Warming Effects on Plant BSi Production and Return Through Litterfall

We hypothesized that we would see an increase in plant silica uptake due to increases in productivity, which has been observed for other elements, such as carbon and nitrogen with warming (Melillo et al., 2011; Butler et al., 2012), as well as for silica under free-air CO2 enrichment (Fulweiler et al., 2015). We also expected to see differences in leaf concentration between species and possibly changes in concentrations or canopy storage over time as a result of long-term soil warming.

In this study, we did find significant differences in leaf BSi concentrations between species, with red maple consistently exhibiting silica concentrations two- to three-fold higher than those of red oak under ambient and warmed conditions. Both maple and oak foliar BSi concentrations were within the range previously reported. Fulweiler et al. (2015) measured 1.06% ± 0.12% BSi in red maple, and Clymans et al. (2016) measured 1.24% ± 0.42% BSi in sugar maple. Geis (1973) measured 0.327% BSi in red oak leaves, and Hodson et al. (2005) estimated 0.37% ± 0.01% BSi by in red oak leaves in a meta-analysis. We also observed significantly higher silica concentrations in fresh leaf litter compared to green foliage, consistent with the previous observations that leaf litter is enriched in silica compared to green leaf tissues (Lovering, 1959; Geis, 1973).

We found no difference in green leaf tissue BSi concentrations or leaf litter tissue BSi concentrations between the control and heated plot. However, we measured a 29% increase in litter production with warming, resulting in greater total litter BSi production.

This increase in plant productivity was within the expected range. Melillo et al. (2011) reported a 45% increase in vegetative carbon storage over the first 7 years of this soil-warming experiment, as measured by radial growth, and attribute the productivity enhancement to warming-driven increases in available nitrogen. The discrepancy of 16% between increases in litter production versus radial growth could be explained

by many factors, such as mixing of litterfall between plots or differential impacts of warming on leaf and wood productivity.

The increase in litter mass was greater for red oak (31%) than that for red maple (19%). Year-to-year variation in the magnitude of litter productivity in the control and heated values closely tracked each other, indicating that local climate, rather than the warming treatment, drove inter-annual differences. Litterfall mass increased in both plots after 2006; however, this increase is likely a result of the aforementioned change in litterfall sampling procedures. Regardless, heated plot litter production was greater relative to the control in all years for red oak, and in all but 2 years for red maple.

We found a significant effect of sampling month (June through August) on green leaf silica concentrations, with the highest foliar silica concentrations observed in samples collected in August During each year of this study, foliar samples were collected during only 1 month of the growing season, making it difficult to distinguish between effects of phenology (early vs. late growing season) versus inter-annual variability or longterm changes on leaf silica concentrations. However, we find phenology to be the more likely explanation, as plants have been shown to continually accumulate silica in leaves throughout the growing season (Struyf et al., 2005; Carey and Fulweiler, 2013). Plant silica accumulation is driven by transpiration, and silica is immobile in plants after bio-silicification, leading to increased silica concentrations in older plant tissues (Ma and Yamaji, 2006). Moreover, we observed no trend across years in silica concentrations in fresh litter, which was collected at consistent times each year.

Because pre-treatment samples were not collected for green leaves or litter, we could not apply a correction factor to these data. However, given that we found no treatment effect on green leaf or litter BSi, we expect that pre-treatment correction would have been unlikely to affect our results.

#### Scaling to Canopy BSi Production

To obtain total canopy BSi production by red maple and red oak, we multiplied annual mean species concentrations in each plot by the mass of each species' litter in each plot. For years in which we did not measure foliar BSi concentrations, we used the mean concentration for the species × plot across all years analyzed. We used data only from years in which litter was collected in laundry baskets to eliminate discrepancy between methods. We estimate that 20.2-kg BSi ha−1 year−1 is fixed in the control plot canopy by red oak and red maple combined, and 27.2 kg BSi ha−1 year−1 fixed in the heated plot canopy by red oak and red maple combined. This constitutes an increase of 35% of canopy silica fixation in the heated plot relative to the control, and consequently, a 35% increase in the fine litterfall flux of BSi to the forest floor. Our data thus support our first hypothesis: due to an increase in tree productivity at constant foliar silica concentrations, soil warming increased tree silica uptake, and release through litterfall.

We took our measurements of red oak and red maple canopy BSi production and estimated concentrations for the remaining species using literature values (**Supplementary Table 3**) to estimate total canopy BSi production for each plot. The remaining species (making up a combined 29% of basal area) consisted of: *Acer pensylvanicum*, *Acer saccharum*, *Betula* sp., *Castanea dentata*, *Fagus grandifolia*, *Fraxinus americana*, *Populus grandidentata*, *Prunus serotina*, and *Quercus alba*. For *A. pennsylvanicum, C. dentata*, *F. grandifolia*, and *P. grandidentata*, species-specific concentrations were not available, so we used the mean of concentrations of the smallest available taxonomic classification containing each species. The published concentrations were for live biomass, so to obtain estimates for leaf litter concentrations, we applied an adjustment factor that equaled the mean of all leaf litter BSi concentrations measured in our study divided by the mean of all green foliage BSi (adjustment factor = 1.48). Given that we saw no effect of heating on red maple or red oak leaf BSi concentrations, we assumed the same to be true for the other species. We then multiplied our estimated litter BSi concentrations by mean litterfall mass per area for each species × plot, and took the sum of these values all species for each plot to obtain total litter BSi masses per plot. For all species combined, we estimate that soil warming increased canopy BSi production from 26.9-kg BSi ha−1 in the control plot to 30.4-kg BSi ha−1 in the heated plot (**Figure 7**). While we could not measure woody biomass silica concentrations for this study, we expect that the increase in woody biomass growth in the heated plot would likely further increase plant silica uptake and storage.

It should be noted that because the soil warming experiment consists of a single control plot and single heated plot, we cannot report error bounds for our estimates of plot-level phenomena. However, our estimates for canopy BSi production and return through litterfall are within the range of those previously reported for temperate forests: Cornelis et al. (2010a) report 4.5 to 90.3 kg BSi ha−1 in litterfall among a range of temperate forest types. Sommer et al. (2013) and Clymans et al. (2016) calculate overall uptake rates of 35 kg BSi ha−1 and 39 kg BSi ha−1, respectively, in similar deciduous forests.

It should be further noted that the experiment in our study involved soil warming only and should not be perceived as a simulation of whole-ecosystem climate change. In previous ecosystem-scale experiments, free-air CO2 enrichment (200 ppm) resulted in 20% increase in tree BSi uptake at the Duke Forest (Fulweiler et al., 2015), and soil freezing induced by snowpack manipulation resulted in a 28% decrease in sugar maple fine root Si uptake (Maguire et al., 2017). Interactions among these and other dimensions of global change may interact with soil warming in additive, synergistic, or antagonistic ways (Templer et al., 2017). Furthermore, given that the increase in silica uptake observed here depends on a nitrogen-mediated productivity enhancement, we note that soil warming could lead to contrasting or no effects of soil warming on silica cycling outside nitrogen-limited ecosystems. Understanding interactions between the biogeochemical cycles of multiple elements may be important for determining future alterations to terrestrial silica cycling.

#### Soil Warming Effects on Silica Release From Decaying Litter and Retention in Soil

In our decomposition experiment, we found an acceleration of litter decay rates with soil warming for red maple, red oak, and birch leaves. The acceleration of litter decomposition in the heated plot was of the same order of magnitude as the increase in litterfall inputs. For each of the species we studied, there was no effect of warming on litter silica concentration over the course of decomposition, indicating that loss of BSi from decomposing litter tracked total biomass loss during decay, and that silica losses from decomposing litter were likewise accelerated by warming. This confirmed our second hypothesis: the increase in litterfall BSi inputs, together with the increased rate of silica losses from decomposing litter, suggests an increase in magnitude and rate of silica release from decaying biomass.

Both the relative decay rates among species and magnitude of warming-driven decay acceleration were in line with previously published decomposition studies. For example, a previous decomposition study in the same experimental plots found that 5°C of soil warming resulted in a 50% increase in decay rate constant for small red maple debris, and a 32% increase for small red oak debris (Berbeco et al., 2012), quite similar to the increases we observed of 47% increase for red maple leaves and 40% for red oak leaves.

In some cases, leaves (Eleuterius and Lanning, 1987) and coarse woody debris (Clymans et al., 2016) have been shown to exhibit preferential silica retention over decomposition. We note here that our litterbag incubation times were relatively brief and probably represented primarily decomposition of relatively labile carbon fractions. A longer-term incubation (and over-winter measurements) would likely have revealed slower overall decay rates (Harmon et al., 2009), and possibly different patterns in carbon-silica coupling.

Despite the increase in litterfall BSi flux to soils, we found no change in soil BSi concentrations over time in the organic or mineral layers between the control and heated plot. This again highlights faster internal recycling of BSi with warming, but no changes in net silica retention. Similar to concentrations, we saw no change in the overall stocks of BSi in soils with warming. However, soil horizon depths and bulk density were not measured so were assumed to be consistent over time. Further, due to the large stocks of BSi in the soils, detecting a change in BSi concentrations may take longer than 15 years. Nonetheless, we observe no measurable effect of warming on soil BSi concentrations or stocks over the duration of this experiment, as predicted in our third hypothesis.

While soil BSi appeared unchanged, DSi was elevated in the soil solution heated plots. The DSi concentrations we observed were within the typical range for forest soils (100–600 µM; Cornelis et al., 2010b), and the increase in soil solution DSi concentrations we observed in the heated plot reveals a probable impact of faster BSi decomposition in heated plots and a proximate source for the additional silica taken up by plants. This also supports the lack of change in soil BSi with warming, as increased inputs from litterfall appear balanced in part by increased dissolution and movement of Si from the soil BSi pool to the soil solution DSi pool.

#### Connections to the Global Terrestrial Silica Cycle

In our study, we found that heating increased the fluxes of plant silica uptake and release through litterfall, with no net effect on soil silica stocks. There are two possible explanations for these trends. First, it is possible that enhanced plant silica uptake was balanced with enhanced BSi dissolution/release from decomposition, indicating accelerated internal Si recycling through the plant soil system, without changes to the weathering or leaching fluxes. Second, it is also possible that enhanced plant uptake exceeded the enhanced BSi dissolution rates, but increased mineral Si weathering inputs and/or reductions in leaching losses maintained the constant soil BSi stocks. In this case, an imbalance between changes to weathering and leaching fluxes could lead to enhancement or reduction of Si export to coastal receiving waters, with potential effects on marine primary productivity by diatoms and other silica-requiring species (Bristow et al., 2017).

The tight coupling between DSi observed in control plot soil solution and stream water in this study suggests to us that soil solution DSi is a dominant control on stream DSi in this system. Given that soil solution DSi was substantially elevated in the heated plot, we think it is likely that at least some portion of that additional DSi would be leached and delivered to streams. We also suspect that weathering inputs to soil DSi are increased by warming, as mineral silicate weathering typically increases with temperature as a result of reaction kinetics (Velbel, 1993; Brady and Carroll, 1994; White and Blum, 1995). Plants also influence mineral weathering through the physical and chemical alteration of soils (Lovering, 1959; Drever, 1994; Berner, 1997; Porder, 2019); thus, NPP enhancements may potentially lead to weathering increases in certain ecosystems (Kelly et al., 1998; Brault et al., 2017). In our study, this could lead to greater weathering inputs and leaching outputs of silica with warming, in addition to our observed enhancement of internal silica recycling.

Overall, our results indicate that soil warming can accelerate the biogeochemical cycling of silica in forests and increase the magnitude of the terrestrial silica pump (i.e., the uptake of DSi by land plants; Carey and Fulweiler, 2012a). The impacts of such changes on net vegetative silica storage and silica export from terrestrial to marine systems remain unresolved, but are likely important over the long term.

#### SUMMARY AND CONCLUSIONS

Our study indicates that the biogeochemical cycle of silica, like that of carbon, nitrogen, and other nutrients, can be altered by soil warming, and thus is likely to be affected by changes in global surface temperatures with climate change. We find that soil warming increases plant silica uptake as a result of increased overall productivity at constant tissue silica concentrations. We further find an increase in the magnitude and return of silica from plants to soil through litterfall and litter decay. We additionally find that soil BSi stocks remained constant over the 15-year duration of this study, indicating a balance of increased silica inputs and outputs from the soil BSi pool. Our results confirm that soil warming increases the extent of internal silica recycling within a temperate forest ecosystem, with potential implications for the global terrestrial silica pump, and land and ocean carbon cycling. These results underscore the need to further explore the interactions between geology and biology with climatic change to understand and predict future alterations to the global silica cycle.

#### DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

JG, JC, and JT conceived of the project. JG collected silica data with support from JC, JT, RF, and AK. JM conceived of the soil warming experiment and JM and WW provided samples, longterm data, and critical insight from the experiment. JG analyzed data and led the writing of the manuscript, and all authors contributed to drafts and the final paper.

#### FUNDING

This research was supported by the National Science Foundation (NSF PLR-1417763 to JT), the Geological Society of America (Stephen G. Pollock Undergraduate Research Grant to JG), the Institute at Brown for Environment and Society, and the Marine Biological Laboratory. Sample analysis and Fulweiler's involvement were supported by Boston University and a Bullard Fellowship from Harvard University. The soil warming experiment was supported by the National Science Foundation (DEB-0620443) and Department of Energy (DE-FC02- 06-ER641577 and DE-SC0005421).

#### ACKNOWLEDGMENTS

We thank Stephen Porder for his advice, Michael Bernard and Elizabeth de la Reguera for field assistance and Dave Murray, Ruby Ho, Rich McHorney, Alia Al-Haj, and Seth Berger for laboratory assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01097/ full#supplementary-material


Brown, G. C., and Mussett, A. E., (1981). *Inaccessible Earth*. London: Allen & Unwin.

Brzezinski, M. A., Dumousseaud, C., Krause, J. W., Measures, C. I., and Nelson, D. M. (2008). Iron and silicic acid concentrations together regulate Si uptake in the equatorial Pacific Ocean. *Limnol. Oceanogr.* 53, 875–889. doi: 10.4319/lo.2008.53.3.0875


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Gewirtzman, Tang, Melillo, Werner, Kurtz, Fulweiler and Carey. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The Role of Macrophytes in Biogenic Silica Storage in Ivory Coast Lagoons

#### Yefanlan Jose-Mathieu Koné1†, Bart Van de Vijver 2,3 and Jonas Schoelynck <sup>3</sup> \*

<sup>1</sup> Department of Environment, Centre de Recherches Océanologiques, Abidjan, Côte d'Ivoire, <sup>2</sup> Meise Botanic Garden, Meise, Belgium, <sup>3</sup> Ecosystem Management Research Group, Department of Biology, University of Antwerp, Antwerp, Belgium

Lagoons are shallow aquatic environments that typically show a broad variety in colonization by macrophytes. We present the biogenic silica (BSi) data obtained from 11 macrophyte species randomly collected in three small lagoons (Ono, Kodjoboue, and Hebe) of Ivory Coast during 12 consecutive months. BSi concentrations were different between species and between lagoons with average values ranging from 2 to 36 mg g−<sup>1</sup> . The highest values were found in Hebe and Kodjoboue lagoons due to the dominance of emergent plant species belonging to Poaceae and Cyperaceae families. However, because total plant coverage was low (5% of the lagoon surface), the total BSi stock in vegetation was low (0.2 and 6.1 t, respectively). Oppositely, lower BSi concentrations were found in plants from Ono lagoon, yet the abundance of macrophytes covered 66% of its surface area which resulted in a larger vegetation BSi stock (17.4 t). Dissolved silica in surface water varied seasonally between 1.7 and 10.8 mg L−<sup>1</sup> , and variation is assumed to be linked to diatom blooms rather than to macrophyte uptake. Sediment data showed that the three lagoons store a large quantity of BSi in their sediments with values ranging from 2 to 8 t BSi ha−<sup>1</sup> . Because of macrophyte influence in these lagoons, macrophyte phytoliths were expected to contribute significantly to this sediment BSi stock. However, microscopic analysis revealed that this stock is absolutely dominated by diatom frustules and sponge spicules rather than plant phytoliths. We conclude that macrophytes in these lagoons contribute only marginally to BSi storage in sediments but that fragile phytogenic silica structures may affect local silica cycling.

Keywords: lagoons, phytogenic silica, sediment, diatom frustules, phytoliths, tropical wetland

## INTRODUCTION

Lagoons are transitional water systems between land and ocean, characterized by a shallow depth (<5 m) and occasional water exchanges with the adjacent ocean (Boynton et al., 1996). They are subjected worldwide to increased nutrient inputs due to anthropogenic activities such as land use change, effluent disposal and aquaculture (Caumette et al., 1996). The resulting eutrophication is characterized by enhanced phytoplankton blooms and increased macrophyte biomass (Sidinei et al., 2001). Macrophytes provide food and shelter for aquatic organisms and regulate nutrient and element cycling dynamics within these systems (Rejmankova, 2011). The distribution of different aquatic macrophyte species and growth forms in lagoons is related in various ways to the physical characteristics of the system (depth, bottom slope, type of sediments, wind regime. . . ), to riverine inputs of nutrients, and to the frequency of salt ocean water intrusions (Kjerfve and Magil, 1989; Aloo et al., 2013).

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Joanna Carey, Marine Biological Laboratory (MBL), United States Daniel Puppe, Leibniz Center for Agricultural Landscape Research (ZALF), Germany

\*Correspondence:

Jonas Schoelynck jonas.schoelynck@uantwerpen.be

†This manuscript is dedicated to a young and promising scientist who passed away too early

#### Specialty section:

This article was submitted to Geochemistry, a section of the journal Frontiers in Earth Science

Received: 18 December 2018 Accepted: 04 September 2019 Published: 26 September 2019

#### Citation:

Koné YJ-M, Van de Vijver B and Schoelynck J (2019) The Role of Macrophytes in Biogenic Silica Storage in Ivory Coast Lagoons. Front. Earth Sci. 7:248. doi: 10.3389/feart.2019.00248

**205**

Aquatic macrophytes are also known to store significant amounts of silica (SiO2), thus potentially affecting silica storage, fluxes and turnover rates in freshwater aquatic ecosystems (Schoelynck and Struyf, 2016). Silica in plants is present in an amorphous and hydrated form (SiO2·nH2O), usually referred to as biogenic silica (BSi) and mostly deposited as phytoliths (siliceous plant bodies) or associated with cell wall molecules (Broadley et al., 2012). Studies have demonstrated the significant roles of plant silicification in general (i.e., not only aquatic plants, but also terrestrial vegetation) on the silica cycle. It is well-established that 55–113 Tmol Si yr−<sup>1</sup> is fixed globally in the terrestrial biosphere (Avg. 84 ± 29 Tmol Si yr−<sup>1</sup> ; Carey and Fulweiler, 2012). This annual ecosystem biogeochemical uptake exceeds the annual export from continents to the ocean through physicochemical weathering of silicates by two orders of magnitude (Conley, 2002), and thus exerts a strong control on global land-ocean interactions in the silica cycle. Specifically for freshwater aquatic ecosystems, studies have pointed to a significant effect of phytoliths on the silica cycle, and strong potential storage of phytolith-silica in soils and sediments (Cary et al., 2005; Struyf et al., 2007). The carbon occluded in the phytoliths of wetland plant species potentially even forms an important sink to consider in the carbon cycle (Li et al., 2013). Exemplary to this is the strong (permanent) silica sink in the Okavango Delta sediments (Botswana) that can be associated with a dominance of silica-rich tropical giant-grasses such as Cyperus papyrus L. (Struyf et al., 2015; Mosimane et al., 2017; Schoelynck et al., 2017). BSi accumulation in macrophytes, and thus potentially in sediments, varies strongly between species. Aquatic vegetation shows apparent plasticity regarding silica uptake, adaptive to silica availability, water and wind dynamics, light interception, herbivory and nutrient stress (Schoelynck and Struyf, 2016). High silica uptake results in high BSi concentrations in plant litter (Struyf et al., 2005; Carey and Fulweiler, 2014), which can impact on aquatic decomposition processes (Schaller and Struyf, 2013) and thus delays in and/or reductions of the transfer of dissolved silica (DSi) from land to ocean. Besides macrophytes, diatoms (Bacillariophyta, unicellular algae), testate amoebae (polyphyletic assemblage of protozoa with a shell) and sponges (Porifera) also contribute significantly to the BSi storage in wetland ecosystems [see Puppe et al. (2015) for a classification of BSi pools in terrestrial ecosystems]. All diatom species and a large number of the testate amoebae synthesize amorphous siliceous shells (Aoki et al., 2007; Smol and Stoermer, 2010) or have an endoskeleton stiffened by (sponge) spicules (Maldonado et al., 2010).

It has been shown that wetlands, as is the case for many other nutrients, play an important role in the biogeochemical silica cycle (Struyf and Conley, 2009). In temperate and subarctic climates, both riparian (e.g., Struyf et al., 2009, 2010) and tidal wetlands (e.g., Struyf et al., 2006; Müller et al., 2013) play an essential role in Si cycling, but tropical wetlands have not yet received similar attention. Tropical rivers deliver about 70–80% of the global DSi load into the ocean (Beusen et al., 2009), implying it is crucial to assess environmental factors that can influence its transport. Studies have indicated the potential influence of Si uptake by giant grasses and sedges (e.g., Ding et al., 2008; Cardinal et al., 2010) on Si-isotope biogeochemistry in tropical rivers, but did not consider the importance of the wetlands in large scale Si balances.

In this study, we hypothesized that macrophyte phytoliths play an important role in silica storage in three tropical lagoons of Ivory Coast. The observed plant species are typical for eutrophic waters of numerous tropical rivers, streams, reservoirs, and natural lakes. After vegetation senescence, plant BSi (phytoliths) may contribute to the BSi stock in lagoon sediments. We quantified this contribution by counting the relative abundance of plant phytoliths, diatom frustules and sponge spicules using a light microscope. After dissolution, DSi may again be taken up by vegetation or be transported downstream. Therefore, we made conservative estimates of the storage of BSi in the vegetation and sediments and compare it between the different lagoons.

## MATERIALS AND METHODS

### Description of the Study Area

Kodjoboue, Ono and Hebe are three small lagoons located in the South-East of Ivory Coast with a joined surface area of about 1,150 ha (**Figure 1**). These lagoons are mainly fed by ground water seepage and flow into the Comoé River before reaching the Atlantic Ocean through different small channels. The catchment area of Ono lagoon has been dominated by pineapple cultivation since 1960 while the watersheds of Hebe and Kodjoboue lagoons are influenced by palm and coconut crops. Today, the three lagoons behave as freshwater systems due to the absence of seawater input since the opening of the Vridi channel in the near of Abidjan in 1950, that discharges most of the flow. The lagoons became gradually dominated by freshwater macrophytes following the reduced salt water inputs. The development of plants is particularly high in Ono lagoon where it starts hampering water transport and fishery activities. Between 1986 and 1989, only Salvinia molesta D.S.Mitch. and Eichhornia crassipes (Mart.) Solms were reported occasionally along the shores of Ono lagoon (Amon et al., 1991; Guiral and N'da, 1991). Today, Ono lagoon is 66% covered by different macrophyte species while the two other lagoons remain more or less in their original state with (emergent) plants limited to the littoral zones. The climate of the area is sub-equatorial with bimodal character. Two rainy seasons, a long rainy season (April–July) and a short rainy season (October to November), are intercepted by two dry periods: a long dry season (December– March) and a short dry season (August–September). The average annual precipitation is 1,704 mm (referred to the years 1970– 2014); average annual temperature is 26.3◦C with March as the hottest month (27.8◦C) and August as the coolest month (24.3◦C) of the year. Humidity is high with average values of 85%.

#### Sampling and Analytical Techniques

The following physicochemical parameters were measured in situ with a probe (Ap-5000, AQUAREAD Limited, Broadstairs, UK): depth (m), pH, temperature (◦C), and electric conductivity (µS cm−<sup>1</sup> ). Subsurface waters (depth ∼30 cm) were sampled with a 1.7 L Niskin bottle. Water samples for DSi analyses were filtered through Sartorius cellulose acetate filters and refiltered

through 0.2µm pore size polysulfone filters. DSi concentrations (mg L−<sup>1</sup> ) were determined with a spectrophotometer (DR 6000) according to standard techniques (Grasshoff et al., 1983). DSi samples were collected only in Ono and Kodjoboue lagoons during eight sampling campaigns at three different stations in each lagoon (**Figure 1**). For Ono lagoon, station 1 was close to dense Echinochloa pyramidalis (Lam.) Hitchc. and Chase vegetation, station 2 was close to Hydrilla verticillata (L.f.) Royle vegetation, and station 3 was devoid of macrophytes and located in the main channel connecting Ono lagoon to the Comoé River. In Kodjoboue lagoon, station 1 was located in the center of the lagoon, station 2 was further to the west and close to the main channel, and station 3 was to the east. All three stations were devoid of macrophytes because plants are limited to the littoral zones.

In parallel, sediment samples were randomly collected at three occasions (April, July, and September) in Ono and Kodjoboue lagoons (same locations as for water samples) and one time (April) in Hebe lagoon (sampling station devoid of macrophytes because plants are limited to the littoral zones). Sediment samples were taken in the organic rich top layer of the sediment using a sediment corer of 3.3 cm diameter and 15 cm length. Sediment cores were subsampled at three different depths (0–5, 5–10, and 10–15 cm). Sediment samples were dried at 75◦C for 5 days and manually homogenized afterwards. Sediment subsamples of 30 mg were mixed with 25 mL of NaOH solution (0.5 M) and incubated in a water bath maintained at 85◦C for 5 h. Subsamples taken at 3, 4, and 5 h were filtered through 0.45µm nitrocellulose Chromafil syringe filters (A-45/25) and analyzed for silica using the spectrophotometric molybdate—blue method (Grasshoff et al., 1983) on a colorimetric segmented flow analyser (SAN++, Skalar, Breda, The Netherlands). Sediment BSi concentration is derived from the intercept of the linear regression between sampling time and measured concentrations, as adapted from DeMaster (1981).

Macrophytes were collected randomly on a monthly base from October 2015 to September 2016 in the three lagoons (12 sampling campaigns). Free floating macrophytes were collected with a net by sweeping a demarcated quadrant (surface area: 0.315 m<sup>2</sup> ). The other plants were cut manually from a fixed area of 1 m<sup>2</sup> . After collection, macrophytes were thoroughly rinsed to remove sediments, algae and macroinvertebrates, sorted per species and total fresh mass was weighted. A subsample of ±100 g fresh mass of each plant species was oven dried at 75◦C for 5 days and weighed again. This conversion factor between fresh and dry mass of the subsample enabled to calculate total macrophyte dry mass per unit surface area. The dried samples were then ground to 300µm. BSi was extracted from 25–30 mg of dried plant material by incubation in a 0.5 M NaOH solution at 80◦C for 5 h (DeMaster, 1981), and analyzed for silica using the spectrophotometric molybdate—blue method (Grasshoff et al., 1983) on a colorimetric segmented flow analyser (SAN++, Skalar, Breda, The Netherlands).

## Diatom, Sponge Spicule, and Phytolith Fixation Protocol

The presence/absence of BSi particles (primarily diatom frustules, sponge spicules and phytoliths) was determined in 12 sediment samples and 6 (non-rinsed) dominant plant samples (see **Table 1** for species list) across all three lagoons (0–15 cm depth). Samples were prepared for light microscopy following the method described in Van der Werf (1955). Small parts of the sample were cleaned by adding 37% H2O<sup>2</sup> and heating to 80◦C for about 1 h. The reaction was completed by addition of saturated KMnO4. Following digestion and centrifugation (three times 10 min at 3,700 rpm), cleaned material was diluted with distilled water to avoid excessive concentrations of BSi particles on the object slides. Cleaned material was mounted in Naphrax. The object slides were analyzed using an Olympus BX53 microscope, equipped with Differential Interference Contrast (Nomarski) optics and the Olympus UC30 Imaging System. The variation in BSi particles in both sediment and plant samples was visualized and BSi particles were quantified (up to 100 particles) in sediment samples on random transects at 400x magnification and the relative abundance of each group (diatoms, sponge spicules, and phytoliths) was determined. Diatom species were identified up to genus level using Lange-Bertalot et al. (2017) and phytolith morphotypes were described according to Madella et al. (2005).

## Stock Calculations

To calculate the BSi stocks in macrophyte biomass, we selected the dominant macrophyte species from each lagoon (see **Table 1** for species list) and multiplied the average calculated dry biomass of each plant species (g m−<sup>2</sup> ) with their respective average BSi concentration (mg g−<sup>1</sup> ). These values were converted to t of BSi by multiplying with the respective vegetation coverage to obtain the total stock in vegetation in each lagoon. For Ono lagoon, the area covered by macrophytes was obtained by a satellite image of 2016 (Landsat 8-OLI/TIRS) using ArcGIS. For the two other lagoons, we did not get the images but field observations (Dr. Egnankou from the SOS Forêts, pers. obs.) show that the area covered by macrophytes is small and represents only about 1 and 5% for Hebe and Kodjoboue lagoon, respectively. To calculate the BSi stocks in the sediment, BSi concentrations in sediments were converted to t of BSi by multiplying the average concentrations (mg g−<sup>1</sup> ) by the density of the sediment (g m−<sup>3</sup> ), a depth of 10 cm (ca. rooting depth of macrophytes) and by the total surface area of the respective lagoon.

### Statistical Analyses

By means of an Anova One Way test, average values of macrophyte BSi were statistically compared between species within the same lagoon, and average values of water DSi were statistically compared between months within the same lagoon [using Prism 5.00 (GraphPad)]. P-values are not explicitly mentioned hereafter but "significant(ly)" refers to p < 0.05 and "not significant(ly)" refers to p ≥ 0.05. Using linear regression, we did not find relationships between plant BSi, sediment BSi and water DSi, hence these results were excluded from the text.

TABLE 2 | Physicochemical characteristics (average ± SD) of the three lagoons.


TABLE 1 | Characteristics and occurrence of macrophyte species collected in the three lagoons (×, presence of the species; #, species used for phytolith visualization; \*, species' values used in BSi stock calculation).


## RESULTS

## Water Quality Parameters and DSi Concentrations

In general, pH was low with average values below 7 (**Table 2**). Conductivity was also low with average values ranging from 13.5 to 36.7 µS cm−<sup>1</sup> showing the highest values in Hebe lagoon and the lowest in Kodjoboue lagoon. Both parameters are characteristic for freshwater systems and demonstrate the present-day absence of marine influence. Temperature varied slightly between the three lagoons with average values ranging from 27.5 to 29.9◦C. DSi showed a strong temporal variability with a similar pattern in Ono and Kodjoboue lagoons (**Figure 2**). The highest values were observed between March and April and the lowest in September. When comparing the two lagoons, the highest DSi values were generally observed in Ono lagoon. Overall, DSi concentrations ranged from 1.7 to 10.8 mg L−<sup>1</sup> .

## Biogenic Silica Accumulation in Macrophytes

Eight plant species were found in Ono lagoon, seven in Kodjoboue lagoon and only three in Hebe lagoon (**Table 1**). The most dominant species were Echinocloa pyramidalis followed by Hydrilla verticillata and Eichhornia crassipes for Ono lagoon, E. pyramidalis, Panicum parvifolium, and Acroceras zizanoides for Kodjoboue lagoon, and Salvinia molesta, E. pyramidalis, and Rhynchospora corymbosa for Hebe lagoon. All other species were less abundant.

BSi concentrations ranged from 0.8 to 54.9 mg g−<sup>1</sup> with the lowest concentration found in E. crassipes in Ono lagoon and the highest found in S. molesta in Kodjoboue lagoon (**Figure 3**). Generally, free floating species had low BSi concentrations and much lower than submerged and emergent species. Among the emergent species, Brachiaria villosa and A. zizanioides showed the highest values while the lowest were found in Jussiaea repens. BSi concentrations of B. villosa were significantly higher than those of the other plants collected in Ono lagoon. In contrast, the BSi concentrations of A. zizanioides in Kodjoboue lagoon did not differ significantly than those of S. molesta (a freefloating species), P. parvifolium and B. villosa. For the single submerged macrophyte (H. verticillata) found in Ono lagoon, the concentrations were relatively high (5.1–19.8 mg g−<sup>1</sup> ) and similar to those found in emergent species. In Hebe lagoon, BSi concentrations found in R. corymbosa were significantly higher than those obtained in S. molesta and E. pyramidalis.

Ono lagoon stored less BSi per m² macrophyte biomass than the two other lagoons. Average values for the three lagoons were 0.054, 0.102, and 0.288 t BSi ha−<sup>1</sup> for Ono, Hebe and Kodjoboue, respectively. By extrapolating these values to the total surface covered by macrophytes in each lagoon, the BSi stored in macrophyte biomass was 0.2, 6.1, and 17.4 t for Hebe, Kodjoboue and Ono, respectively.

#### Biogenic Silica Accumulation in Sediment

The concentrations of BSi generally decreased with depth (**Table 3**): the highest values were found near the subsurface (0– 5 cm) while the lowest occurred at the bottom of the sediment core. The lowest values were observed in July in Ono lagoon. Overall, average BSi concentrations in sediments ranged from 49.6 to 89.4 mg g−<sup>1</sup> . The BSi stock in the sediments over 10 cm depth was higher in Kodjoboue lagoon (8.2 t BSi ha−<sup>1</sup> ) in comparison to those observed in Hebe (4.2 t BSi ha−<sup>1</sup> ) and Ono (5.5 t BSi ha−<sup>1</sup> ). Extrapolating these values to the total surface of each lagoon, the BSi stock in the sediments was 1,025, 2,651, and 3,477 t for Hebe, Ono and Kodjoboue, respectively.

## Characterization of the BSi Particles in Sediment and Plant Samples

All sediment samples were dominated by diatom frustules, mainly belonging to the genera Aulacoseira (>90%), Pinnularia (2–3%), Eunotia (2–3%), and Diadesmis (<1%). In the samples from Kodjoboue lagoon, several sponge spicules were observed (**Figure 4a**, arrow). Only megascleres were found. Gemmoscleres and microscleres were never observed. In Ono and Hebe lagoons, the amount of spicules was extremely low and spicules were



usually not included in the 100 particle-counts. Phytoliths were only very rarely observed (**Figure 4b**). In the samples from Ono lagoon, no phytoliths were observed, even when scanning an entire object slide. In Hebe lagoon and Kodjoboue lagoon, occasionally, only fragments of phytoliths were recorded after scanning entire object slides but never during counting. The observed phytolith fragments were not identifiable. Intact shells of testate amoebae were not found, but very rarely idiosomes (siliceous platelets synthesized by testate amoebae for shell construction) were observed (**Figures 4c,d**).

In plant samples, phytoliths were observed in E. pyramidalis (**Figures 4e–h**) and P. parvifolium (**Figures 4i–k**) samples whereas in H. verticillata and S. molesta samples, no phytoliths were found. Most phytoliths were short bilobate or polylobate cells (**Figures 4e–g,j**). Occasionally, short cell cross (**Figure 4i**), globular and long echinate phytoliths (**Figures 4h,k**) were observed.

#### DISCUSSION

#### The Role of Macrophytes and Diatoms in Lagoon BSi Storage

BSi accumulation in vegetation varied per lagoon and was dependent on plant species dominance [see Prychid et al. (2003) and Hodson et al. (2005) for BSi contents in plants related to their phylogenetic position]. The BSi stock in macrophytes ranged from 0.054 to 0.288 t BSi ha−<sup>1</sup> , which is situated between values reported for submerged macrophytes in the Okavango Delta channels (0.008 t BSi ha−<sup>1</sup> ; Schoelynck et al., 2017) and values reported for emergent wetland vegetation in the Okavango Delta (ranging from 0.010 to 1.600 t BSi ha−<sup>1</sup> ; Struyf et al., 2015). The highest BSi concentrations were found in macrophytes collected from Kodjoboue lagoon and were related to the dominance of emergent species (A. zizanioides, B. villosa, R. corymbosa, E. pyramidalis, P. lanceolatus, and P. parvifolium) belonging to the Poaceae and Cyperaceae families. Species of these two families are often characterized by high BSi contents (Ma and Takahashi, 2002) and were also identified as a macrophyte growth form

FIGURE 4 | Silica particles in the sediment and plant samples. Sediment samples originate from Kodjoboue lagoon, and are exemplary to all examined samples from the three lagoons. (a) Large number of diatom valves, mainly belonging to the genus Aulacoseira and one sponge spicule (megasclere, see arrow). (b) Shows a few broken diatom valves and one phytolith (see arrow). Plant samples originate from dominant plant species from across all three lagoons. (c,d) Show testate amoeba idiosomes. (e–h) Show different phytoliths found in Echinochloa pyramidalis samples: short cell bilobate (e,f), cylindrical polylobate (g), and globular echinate (h) phytoliths were observed. (i–k) Show different phytoliths found in a Panicum parviflorum sample: short cell cross (i), short cell bilobate (j), and long cell echinate (k) phytoliths were observed. Scale bar represents 20µm for panels (a,b) and 10µm for panels (c–k).

with high BSi content in Schoelynck and Struyf (2016). The generally lower BSi concentrations found in Ono lagoon were counterbalanced by the high macrophyte coverage giving this lagoon the highest BSi stock in vegetation of all three lagoons: 17.4 t. We did not find a relationship between plant BSi and sediment BSi. Other studies (e.g. in the Okavango Delta, Struyf et al., 2015) identified an impact of vegetation on phytolith input to the soils and thus on BSi content, but results are not always consistent. In saltmarshes, de Bakker et al. (1999) found no relation between plant BSi content and sediment BSi content. Carey and Fulweiler (2014) concluded that silica uptake in saltmarsh plants is not always directly linked to sediment BSi and sediment DSi, since active uptake mechanisms stimulated by multiple abiotic factors (e.g., hydrological stress) can strongly impact plant silica uptake.

Light microscopy of the lagoon sediments revealed that the majority of BSi particles in the sediment of all lagoons comprise almost entirely of diatom frustules, occasionally sponge spicules and almost never phytolith particles. Tychoplanktonic diatoms such as Aulacoseira species form long filaments of large, heavily silicified, cylindrical cells, and are known to bloom mainly in spring and autumn (Siver and Kling, 1997). The light microscopy investigated samples in the current study were taken in November 2017 and April 2018, most likely in the middle of large Aulacoseira blooms. During this sampling period, stimulated by the nutrient rich character of these waters, the shallow depths and the abundant sunlight, Aulacoseira blooms may have produced large amounts of frustules (Conley et al., 1993). Additionally, non-rinsed plant samples were also dominated by large amounts of epiphytic diatoms, mainly belonging to the genus Eunotia, typically found in acidic environments such as these lagoons (Van de Vijver, pers. obs.). It is thus clear that diatoms could have depleted the plantavailable DSi in the water rapidly, before the plants could take up the necessary silica to produce phytoliths. This can be observed in **Figure 2** where the amount of DSi strongly decreased between April and May suggesting either a dilution effect in the rainy season (resulting in lower concentrations; Meunier et al., 2011), or an increased uptake by organisms compared to other months (Rabosky and Sorhannus, 2009).

The fact that we do find BSi in plants but rarely phytoliths, may be explained by different deposition mechanisms. Rapid deposition of silica results in lumen and cell wall phytoliths, whereas slow deposition results in silica deposition in intercellular spaces or in an extracellular (cuticular) layer (reviewed by Prychid et al., 2003; Hodson, 2019). These nonhomogeneous masses can be i.a. a silica ring formed around the periphery of the cell, deposited on a dispersed organic matrix, or laid down as deposits within the cell wall or between the cellulose wall and the plasma membrane or in cortical intercellular spaces within the cell (Prychid et al., 2003). In general, these fragile phytogenic silica structures as well as very small phytoliths (<5µm) are difficult to trace in sediments, but they potentially play an important role in silica cycling (Meunier et al., 2017; Puppe et al., 2017). Additionally, given the very large number of epiphytic diatoms on non-rinsed plant samples, it cannot be excluded that a fraction of plant-silica in our results is actually diatom-silica, even though plants were washed thoroughly before analysis, according to protocol. Furthermore, it could be hypothesized that plant organic matter may float away through the outlets of the lagoons, or accumulate on the shores, hence not contributing to the sediment silica storage (cf. Struyf et al., 2015). Alternatively, Puppe et al. (2017) suggest the formation of a layer of coarse organic matter on top of sediments from which phytoliths cannot be released easily and hence are missing in underneath sediment layers. We have no arguments supporting either of these hypotheses, none of our sediment cores contained such a distinct organic top layer and samples were visually homogeneous organic rich (fine material) over the entire depth of the core (15 cm).

#### Sediments as a BSi Sink

Overall, the BSi stocks in sediments of the three lagoons ranged from 4.2 × 10<sup>3</sup> kg ha−<sup>1</sup> to 8.2 × 10<sup>3</sup> kg ha−<sup>1</sup> and were much higher than those found in the vegetation, suggesting that lagoon sediments are important sites for long-term BSi storage. Similar studies on African wetland BSi accumulation are scarce. The best studied ecosystem in this regard is the Okavango Delta (Botswana) where similar values were found: between 3.8 × 10<sup>3</sup> kg ha−<sup>1</sup> and 36 × 10<sup>3</sup> kg ha−<sup>1</sup> in the riparian wetlands [depending on flooding frequency and in top 5 cm of sediment; Struyf et al. (2015)] and between 0.2 × 10<sup>3</sup> kg ha−<sup>1</sup> and 1.5 × 10<sup>3</sup> kg ha−<sup>1</sup> in the main channels [depending on vegetation cover and in top 10 cm of sediment; Schoelynck et al. (2017)]. Since phytoplankton production in the Delta is limited by low ambient nutrient concentrations, Okavango sediment BSi particles mainly originate from dead terrestrial and aquatic plant fragments (Struyf et al., 2015). In case of the analyzed groundwater-fed Ivory Coast lagoons, most of the BSi is produced within the lagoons and not brought in by any river. Overland runoff and topsoil erosion as in the Nyong basin in Cameroon (Cary et al., 2005) can potentially introduce allochthonous BSi particles, especially by transformation of an area from pristine to agricultural land use (cf. Smis et al., 2011). However, we have no indication at the moment that this is a significant process in the analyzed lagoons and no phytoliths with confirmed terrestrial origin were found in corresponding sediment samples. The investigated lagoons sediment BSi pool is thus dominated by diatom frustules, occasional sponge spicules and phytolith fragments, and an unquantified amount of fragile phytogenic silica structures and small phytoliths (<5µm). Entire shells of testate amoebae were not observed probably because the organic cement, which glues the idiosomes (i.e., the building blocks of the shell) together, was destroyed during the H2O<sup>2</sup> sample preparation step. But also single idiosomes were very rare in our samples, although other studies showed that they can be quite abundant in sediment samples in general (cf. Douglas and Smol, 1987; Cary et al., 2005).

## Implications for the Ecosystem Silica Cycling

The in situ produced BSi is deposited locally due to limited exchange with the adjacent coastal ocean (Boynton et al., 1996). This accumulation strongly influences BSi cycling and storage in the sediments. In general, plant and phytoplankton BSi is returned in detritus to soil and sediments, where decomposition processes and pedogenic transformations can result in either stored BSi or re-dissolved DSi (Frings et al., 2014). As the analyzed lagoons have an acidic pH (<6.4), the preservation of diatom frustules and phytoliths can be assumed to be quite good (Flower and Ryves, 2009) making silica only very slowly available again for uptake. The average DSi concentration for the Comoé river draining these lagoons is with 2.86 mg L−<sup>1</sup> rather low (Koné et al., 2008). Invertebrate fauna might be important contributors to remineralisation processes in general as was reported for other nutrients in Taabo Lake, which is similar to the studied lagoons (Kouamé et al., 2011). Invertebrate fauna, especially burrowers, are known to amplify BSi cycling in shallow environments like coastal lagoons and estuaries (Viaroli et al., 2013). The presumed fragile phytogenic silica in the plants (potentially explaining the relatively high plant BSi concentrations) may be very important in such conditions as they may be able to dissolve more easily than homogeneous masses (i.e., phytoliths and diatom frustules) and can hence contribute to local silica cycling. Plants may then be an important source of silica availability in the lagoons, although this warrants further investigation.

Generally, BSi stored in wetland sediments is several orders of magnitude more soluble than mineral silicates (Farmer et al., 2005) and its temporal or permanent storage vs. recycling or downstream transport exhibits an important control on silica export toward the ocean (e.g., Derry et al., 2005; Fulweiler and Nixon, 2005; Sommer et al., 2006), which is crucial for the silica cycling in these (sub) tropical regions (Beusen et al., 2009). Decreasing availability of silica compared to other nutrients can negate any competitive advantage the diatoms have and can lead to nuisance and toxic blooms of green and blue green algae in coastal ecosystems (Cloern, 2001). Diatoms need an optimal nutrient ratio of C:Si:N:P = 106:15:16:1, and diatom growth will cease when DSi supplies are depleted, allowing other phytoplankton classes to proliferate using any excess of N and P (Anderson and Burkholder, 2002). Because diatoms are primary producers at the base of the food-web, any change in that production might have knock-on effects to the entire ecosystem. Phytoplankton assemblages at Ivory Coast coastal environment show that, although diatoms are the most species-rich, cyanobacteria are the most abundant (up to 90% of the phytoplankton biomass), and silica was identified as predominant abiotic factor controlling phytoplankton dynamics (Osemwegie et al., 2016). Expected human population rise in the greater Abidjan area however, will probably put more pressure on local water bodies and nutrient levels may increase to levels which could escalate to dramatic levels of eutrophication [projections made for the nearby (and larger) Ebrié Lagoon which receives water from the Comoé River; (Scheeren et al., 2004)].

#### CONCLUSION

It can be concluded that relatively recent colonization of the lagoons by aquatic vegetation had no major effect yet on the lagoons' sediment BSi storage that is still dominated by phytoplankton, entirely comparable to any other open water system lacking plants. Perhaps it is just a matter of time before plants start having an impact here. Values reach up to 8.2 t BSi ha−<sup>1</sup> , which is similar to other tropical wetlands and may exhibit an important control on fluxes toward the

#### REFERENCES


tropical coastal zone. Relatively high plant BSi concentrations are likely explained by non-homogeneous fragile phytogenic silica structures that may be more able to dissolve and contribute to local silica cycling, rather than the more homogeneous diatom frustules. Using other techniques are advised measuring or visualizing these fragile phytogenic silica structures to prove this hypothesis.

#### AUTHOR CONTRIBUTIONS

YK and JS conceived the program research. YK collected the samples. BV analyzed the BSi particles in sediment and plant samples. YK, JS, and BV wrote the paper. JS and BV read and approved the final manuscript and we are pretty confident that YK would have agreed with our corrections after revision.

#### ACKNOWLEDGMENTS

The authors thank the Director of the Centre de Recherches Océanologiques d'Abidjan (CRO), B. Diara, and E. K. Sessou from the University of Félix Houphouët Boigny for assisting with the field work, Dr. Christine Cocquyt (Meise Botanic Garden) for her help in identifying silica particles in the sediment samples, and Dr. Eric Struyf (UAntwerpen) for proofreading the manuscript. JS is a postdoctoral fellow of FWO (project no. 12H8616N).


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Koné, Van de Vijver and Schoelynck. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Belowground Phytolith-Occluded Carbon of Monopodial Bamboo in China: An Overlooked Carbon Stock

Chen Chen1,2† , Zhangting Huang1,2,3† , Peikun Jiang1,2,3 \*, Junhui Chen1,2 \* and Jiasen Wu1,2

<sup>1</sup> State Key Laboratory of Subtropical Silviculture, Zhejiang A&F University, Lin'an, China, <sup>2</sup> School of Environmental and Resource Sciences, Zhejiang A&F University, Lin'an, China, <sup>3</sup> Zhejiang Provincial Collaborative Innovation Center for Bamboo Resources and High-efficiency Utilization, Lin'an, China

#### Edited by:

Zhaoliang Song, Tianjin University, China

#### Reviewed by:

Lukas Van Zwieten, New South Wales Department of Primary Industries, Australia Yong Ge, Institute of Geology and Geophysics (CAS), China Xing Sun, Chuzhou University, China

#### \*Correspondence:

Peikun Jiang jiangpeikun@zafu.edu.cn Junhui Chen junhui@zafu.edu.cn †These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science

Received: 20 July 2018 Accepted: 18 October 2018 Published: 06 November 2018

#### Citation:

Chen C, Huang Z, Jiang P, Chen J and Wu J (2018) Belowground Phytolith-Occluded Carbon of Monopodial Bamboo in China: An Overlooked Carbon Stock. Front. Plant Sci. 9:1615. doi: 10.3389/fpls.2018.01615 Phytolith-occluded carbon (PhytOC), a highly stable carbon (C) fraction resistant to decomposition, plays an important role in long-term global C sequestration. Previous studies have demonstrated that bamboo plants contribute greatly to PhytOC sink in forests based on their aboveground biomass. However, little is known about the contribution of belowground parts of bamboo to the PhytOC stock. Here, we reported the phytolith and PhytOC accumulation in belowground trunk and rhizome of eight monopodial bamboo species that widely distributed across China. The results showed that the belowground parts made up an average of 39.41% of the total plant biomass of the eight bamboo species. There were significant (p < 0.05) variations in the phytolith and PhytOC concentrations in the belowground trunk and rhizome between the bamboo species. The mean concentrations of PhytOC in dry biomass ranged from 0.34 to 0.83 g kg−<sup>1</sup> in the belowground rhizome and from 0.10 to 0.94 g kg−<sup>1</sup> in the belowground trunk across the eight bamboo species, respectively. The mean PhytOC stocks in belowground biomass ranged from 2.57 to 23.71 kg ha−<sup>1</sup> , occupying an average of 23.36% of the total plant PhytOC stocks. This implies that 1.01 × 10<sup>5</sup> t PhytOC was overlooked based on the distribution of monopodial bamboos across China. Therefore, our results suggest that the belowground biomass of bamboo represents an important PhytOC stock, and should be taken into account in future studies in order to better quantifying PhytOC sequestration capacity.

Keywords: PhytOC, aboveground biomass, belowground trunk and rhizome, carbon sequestration, phytolith

## INTRODUCTION

Increased greenhouse gas (GHG) emissions have been widely accepted as the main cause of climate change, which threatens the sustainability of terrestrial ecosystem (Kosten et al., 2010; IPCC, 2014). Among the GHGs, the CO<sup>2</sup> emission rate had increased to 3.11 × 10<sup>11</sup> Mg per year by 2010 at the global scale (DOE, 2008). Methods that can reduce the speed of rapidly rising CO<sup>2</sup> concentrations are urgently needed to contribute to climate change mitigation. Terrestrial biogeochemical carbon (C) sequestration is one of the most promising approaches for long-term atmospheric CO<sup>2</sup> sequestration (IPCC, 2014).

Occlusion of C within phytoliths (PhytOC) as an effective mechanism of biotic C sequestration has received much attention in recent years (Parr and Sullivan, 2005; Song et al., 2012a,b, 2016;

Yang et al., 2018). Phytolith, also referred to as plant opal, is an amorphous silica that formed in living plants (Wang and Lü, 1993; Parr and Sullivan, 2005). During the formation of phytolith, some organic C can be occluded in plant tissues. Previous studies demonstrated that PhytOC is highly stable and could be preserved in the soil for several 1000s of years after plant decomposition (Wilding et al., 1967; Parr and Sullivan, 2005; Santos et al., 2010). For example, Parr and Sullivan (2005) found that PhytOC could contribute up to 82% of the total soil C pool after 2000 years decomposition in Numundo oil palm (Elaeis guineensis) plantations. It is also suggested that PhytOC makes up between 15 and 37% of the estimated global accumulation rate (24 kg C ha−<sup>1</sup> yr−<sup>1</sup> ) of stable soil C, demonstrating the significant potential of PhytOC in the long-term terrestrial C sequestration (Parr and Sullivan, 2005; Song et al., 2012a).

The PhytOC concentrations in different plants vary greatly due to their differences in the capacity for phytolith accumulation (Parr et al., 2010; Song et al., 2012b, 2017; Yang et al., 2015; Xiang et al., 2016). Bamboo, a typical phytolith-accumulator (Parr et al., 2010), has been shown to have a greater production of PhytOC in comparison with other plants such as sugarcane (Parr et al., 2009), rice (Li et al., 2013), and millet (Zuo and Lü, 2011). Being predominantly distributed in the tropical and subtropical regions, bamboo has a global area of 2.2 × 10<sup>7</sup> ha by 2010 and is increasing at a rate of 3% annually (Cao et al., 2011; Zhou et al., 2011). It has been estimated that the present annual PhytOC sink in China's forests is 1.7 ± 0.4 Tg CO<sup>2</sup> yr−<sup>1</sup> , 30% of which is contributed by bamboo because the production flux of PhytOC through tree leaf litter for bamboo is 3–80 times higher than that of other forest types (Song et al., 2013).

The potential of PhytOC sequestration in bamboo species also varies depending on the rhizomatous forms (Li et al., 2014a; Xiang et al., 2016). Monopodial scattering bamboo (typically Moso bamboo and Lei bamboo) forests accounted for 77.71% of the total area of bamboo forests in China and were estimated to contribute 75% of the total PhytOC sequestration in Chinese bamboo (Li et al., 2014b). Yang et al. (2015) further demonstrated that the PhytOC production flux contributed by aboveground biomass (including branches and culms) was 1.18 to 1.78 times compared with those estimated by leaf samples for eight monopodial bamboo species due to their larger biomass. Although existing research suggest the significant role of global PhytOC sequestration through bamboo plants, their estimates were only based on the aboveground biomass and the contribution of belowground parts was never determined. Bamboo plants usually have vigorous rhizomes with high biomass. For example, the belowground biomass of Moso bamboo could account for more than one third of total stand biomass (Wang et al., 2013). Given the large phytolith accumulation in bamboo branches and culm in our previous studies (Huang et al., 2014; Yang et al., 2015), we infer that the phytolith could also be accumulated in the bamboo rhizomes and a large amount of PhytOC sequestered in the belowground biomass may have been overlooked in previous studies, leading to a severe underestimation of the PhytOC stock in bamboo forests. The purposes of this study are (1) to examine and compare the concentration of phytolith and PhytOC in belowground trunk and rhizome and (2) to estimate the PhytOC stocks in belowground biomass of bamboo species that widely distributed across China. We hypothesize that the bamboo species differ in PhytOC concentrations in their belowground trunks and rhizomes, and that the belowground parts make a significant contribution to the total PhytOC sequestration of bamboo plants.

## MATERIALS AND METHODS

## Experimental Site and Sampling

We selected eight monopodial bamboo species that account for more than 85% of the total area of monopodial bamboo forests in Zhejiang and Anhui Provinces, China. The eight bamboo species are Phyllostachys heterocycla (Carr.) Mitford 'Pubescens' (PHMP), Phyllostachys praecox C. D. Chu 'Prevernalis' (PPP), Phyllostachys prominens W. Y. Xiong (PP), Pseudosasa amabilis (McClure) Keng f (PAMK), Phyllostachys glauca McClure (PGM), Pleioblastus amarus (Keng) Keng f (PAKK), Phyllostachys heteroclada Oliver (PHO), and Bambusa piscatorum McClure (BPM). Detailed sampling site information is given in **Table 1**. For each species, four plots with an area size of 20 m × 20 m were established in the bamboo forest. The plots in each forest had similar site conditions including elevation, soil type, slope gradient and aspect. The average diameter at breast height (DBH) and stem density were determined. In each plot, one individual bamboo plant having an DBH similar to the mean values was selected and used to determine the biomass of organs including leaves, branches, culms and belowground trunk. The silicon content, phytolith content, C content of phytolith, and PhytOC content per dry biomass were also determined. The rhizome for each species was collected from four subplots of 1 m × 1 m randomly established in each plot. All leaves, branches, and culms of each sample plant, and the belowground trunk and rhizome were weighed separately.

#### Sample Measurements

Each sample was mixed, rinsed with ultrapure water and ultrasonic cleaning to clear all clays contaminated on the bamboo roots. The plant samples were oven-dried at 70◦C for 48 h to a constant mass and then ground to pass through a 0.25-mm sieve for chemical analysis. Phytoliths in samples were extracted using a microwave digestion method (Parr et al., 2001). The phytolith extracts were transferred into pre-weighed centrifugal tubes, dried at 65◦C for 48 h in an oven, and then weighed. A K2Cr2O<sup>7</sup> solution (0.8 M) was used to detect whether the organic matter surrounding the phytolith had been completely removed before the determination of PhytOC (Parr et al., 2010). The PhytOC was determined according to the PhytOC alkali spectrophotometry method (Yang et al., 2014). The accuracy and repeatability of this analytical method was well verified against the results obtained with acid dissolution-Elementar

TABLE 1 | Site information of sampling plots of eight monopodial bamboo species studied.


<sup>a</sup>DBH, diameter at breast height. PHMP, Phyllostachys heterocycla (Carr.) Mitford 'Pubescens'; PPP, Phyllostachys praecox C. D. Chu 'Prevernalis'; PP, Phyllostachys prominens W. Y. Xiong; PAMK, Pseudosasa amabilis (McClure) Keng f; PGM, Phyllostachys glauca McClure; PAKK, Pleioblastus amarus (Keng) Keng f; PHO, Phyllostachys heteroclada Oliver; BPM, Bambusa piscatorum McClure.

Vario MAX CN method (Germany) (Yang et al., 2014). In this method, a 0.01 g phytolith sample was placed into a 10 mL centrifuge tube, 0.5 mL 10 M NaOH added, and incubated for 12 h at 25◦C to dissolve the phytoliths. The extract was further treated with 1.0 mL of 0.8 M K2Cr2O<sup>7</sup> solution followed by addition of 4.6 mL of concentrated H2SO<sup>4</sup> to oxidize the released organic C. The obtained solutions were placed in a water bath at 98◦C for 1 h, and the concentration of PhytOC in the solutions was determined colorimetrically at 590 nm on a Hitachi 150-20 spectrophotometer (Hitachi, Ltd., Tokyo, Japan).

#### Data Calculation and Statistical Analysis

C concentration in phytolith, PhytOC concentration in dry biomass and PhytOC stock were calculated using the following formulas:

C concentration in phytolith (g kg−<sup>1</sup> ) = C content in

phytolith (g)/phytolith weight(kg) (1)

PhytOC concentration (g kg−<sup>1</sup> ) = C content in

phytolith (g)/ dry biomass(kg) (2)

PhytOC stock (kg ha−<sup>1</sup> ) = 6 [PhytOC

$$\text{concentration} \,(\text{g}\,\text{kg}^{-1}) \times \,\text{biomass} \,(\text{kg}\,\text{ha}^{-1}) \times \,10^{-3}\text{J} \,\quad \text{(3)}$$

MS Excel 2010 and SPSS 18 software were used to carry out data processing and statistical analysis. One-way ANOVA followed by LSD test (p < 0.05) were used to examine the difference in phytolith and PhytOC contents among different plant species.

#### RESULTS

#### Belowground Biomass of Eight Monopodial Bamboo Species

The total aboveground biomass (including leaves, branches, and culm) ranged from 20.82 to 48.68 t ha−<sup>1</sup> per dry weight across the eight species, with the highest in PAKK [Pleioblastus amarus (Keng) Keng f] and lowest in PP (Phyllostachys prominens W. Y. Xiong) (**Table 2**). The biomass of the rhizome was much smaller than that of the aboveground across the eight species with the exception that the biomass of rhizome of PP was almost three times higher than that in the aboveground. The biomass of the belowground trunk ranged from 1.19 to 7.18 t ha−<sup>1</sup> across the eight species. The proportion of belowground biomass to total biomass varied from 18.64% for PHO (Phyllostachys heteroclada Oliver) to 74.91% for PP, with a mean of 39.41%.

## Phytolith and PhytOC Concentrations of Bamboo in Belowground Biomass

There was a significant (p < 0.05) variation in the concentrations of Si, phytolith, C concentration in phytolith, and PhytOC in belowground biomass among the eight bamboo species (**Table 3**). The concentration of Si and phytolith in the rhizome ranged from 8.43 g kg−<sup>1</sup> for PHMP [Phyllostachys heterocycla (Carr.) Mitford'Pubescens'] to 21.05 g kg−<sup>1</sup> for PGM (Phyllostachys glauca McClure), and from 11.20 g kg−<sup>1</sup> for PAKK to 35.44 g kg−<sup>1</sup> for PGM, respectively. The C concentration in phytolith in the rhizome ranged from 11.02 g kg−<sup>1</sup> for BPM (Bambusa piscatorum McClure) to 80.42 g kg−<sup>1</sup> for PHMP. There were no significant differences in the C concentration in phytolith in the rhizome among the other seven bamboo species except PHMP.

Frontiers in Plant Science | www.frontiersin.org


<sup>a</sup>Data were cited from Huang (2014) and Yang (2016). The aboveground biomass includes leaves, branches and culm. PHMP, Phyllostachys heterocycla (Carr.) Mitford 'Pubescens'; PPP, Phyllostachys praecox C. D. Chu 'Prevernalis'; PP, Phyllostachys prominens W. Y. Xiong; PAMK, Pseudosasa amabilis (McClure) Keng f; PGM, Phyllostachys glauca McClure; PAKK, Pleioblastus amarus (Keng) Keng f; PHO, Phyllostachys heteroclada Oliver; BPM, Bambusa piscatorum McClure.

The concentration of Si and phytolith in the belowground trunk ranged from 2.30 g kg−<sup>1</sup> for PPP (Phyllostachys praecox C. D. Chu'Prevernalis') to 14.07 g kg−<sup>1</sup> for PHO, and from 5.88 g kg−<sup>1</sup> for BPM to 14.95 g kg−<sup>1</sup> for PHO, respectively. The C concentration in phytolith in the rhizome were generally higher than those in the belowground trunk, and both of them varied greatly among the bamboo species. The C concentration in phytolith was highest in PPP (179.99 g kg−<sup>1</sup> ), and lowest in BPM (23.44 g kg−<sup>1</sup> ). The concentration of PhytOC in dry biomass was also significantly higher in PPP (0.94 g kg−<sup>1</sup> ) than the other bamboo species, followed by PHO (0.61 g kg−<sup>1</sup> ) and was lowest in PP (0.10 g kg−<sup>1</sup> ).

## Estimation of PhytOC Stock of Bamboo in Belowground Biomass

The PhytOC stocks in the rhizome and belowground trunk varied among the bamboo species with the range of 2.30–23.58 and 0.13–3.73 kg ha−<sup>1</sup> , respectively (**Table 4**). The PhytOC stocks in the rhizome of PP were almost 10 times higher than those in the rhizome of PHO and BPM. The PhytOC stocks in the rhizome were much higher than those in the belowground trunk, accounting for more than 80% of the total belowground biomass across the eight species with the exception of PHO (69.48%). The PhytOC stocks in the aboveground and belowground biomass ranged from 13.00 to 90.36 kg ha−<sup>1</sup> and from 2.57 to

TABLE 3 | The concentrations of Si and phytolith, C concentration in phytolith and PhytOC/dry biomass in the belowground trunk and rhizome of eight monopodial bamboo species.


Values are means ± standard error of four replicates. Means followed by different letters within a column are significantly different at the p < 0.05 level. PHMP, Phyllostachys heterocycla (Carr.) Mitford 'Pubescens'; PPP, Phyllostachys praecox C. D. Chu 'Prevernalis'; PP, Phyllostachys prominens W. Y. Xiong; PAMK, Pseudosasa amabilis (McClure) Keng f; PGM, Phyllostachys glauca McClure; PAKK, Pleioblastus amarus (Keng) Keng f; PHO, Phyllostachys heteroclada Oliver; BPM, Bambusa piscatorum McClure.



#### DISCUSSION

The present study showed that the phytolith and PhytOC concentrations in belowground trunk and rhizome differed greatly between bamboo species. Our results were partly similar to the findings by Yang et al. (2015), who found that the phytolith and PhytOC concentrations vary across leaf, branch and culm, and also between bamboo species. However, in comparison with the range of phytolith and PhytOC concentrations in aboveground biomass found by Yang et al. (2015), the quantity of PhytOC in the rhizome and belowground trunk in our study was much smaller. Our study suggested that the PhytOC production capacities of different bamboo species and different organs of the same species vary substantially, which may be ascribed to differences in both physiological properties and the environments. Several studies suggested that the variation of PhytOC concentration in plant depends on the contents of phytolith and C concentration in phytolith, both of which are related to the plant absorption capacities of Si (Ding et al., 2008; Parr et al., 2010). It is well-known that although Si can be taken up by plant roots in the form of Si(OH)<sup>4</sup> (Gong et al., 2004; Ranganathan et al., 2006), the ability of transpiration for Si varies in bamboos of different species and within different organs (Leng et al., 2009; Li et al., 2014a; Yang et al., 2015). In addition, phylogenetic type, climate, soil, and the efficiency of C encapsulation by the silica are also important factors influencing the absorption and transpiration of Si (Song et al., 2013; Li et al., 2014a,b; Zhang et al., 2017). Among these factors, soil conditions such as water and pH could not only influence the accumulation of soil phytoliths by affecting the stability of soil phytoliths, but also influence plant Si uptake from soil solution by affecting the bioavailability of Si in soils (Parr and Sullivan, 2005; Li et al., 2014c; Yang et al., 2018). For example, plants in soils with low pH and high organic matter are reported to take up and accumulate more Si, and consequently higher PhytOC accumulation (Song et al., 2012b). Liu et al. (2017) found that the contents of Si and PhytOC in the Moso bamboo leaves differed between soils derived from different parent rocks. Similarly, Li et al. (2014c) observed that variation of bioavailable Si of soils developed on different parent rocks could lead to the differences in Si absorption from soil solution and phytolith accumulation in bamboo leaves. Nevertheless, we acknowledge that it is a limitation of our study that the soil properties were not examined, and the mechanisms of Si absorption and phytolith accumulation in belowground trunk and rhizome of bamboo deserves more studies.

fpls-09-01615 November 5, 2018 Time: 7:47 # 5

In previous studies, the potential of phytolith C biosequestration has been largely assessed based on the above- rather than belowground biomass across both agriculture, grassland and forestry ecosystems (Parr et al., 2010; Song et al., 2012b; Li et al., 2013; Ru et al., 2018). One of the important reasons is that the belowground biomass including shoot stumps and roots is usually smaller than the aboveground especially for some Si-accumulator plants, such as sugarcane, rice, and wheat. Another key reason is that some researchers believed that only the aboveground biomass, such as leaves and sheath, can accumulate phytoliths and have high PhytOC concentration (Parr and Sullivan, 2005; Parr et al., 2009; Song et al., 2012b), leading to the potential of belowground C sequestration by phytoliths being overlooked. Our study showed that the belowground trunk and rhizome of bamboo accounted for an average of 39.41% of the total plant biomass, while the belowground material of PP contributed 74.91% of the total weight. These results are consistent with the findings of Wang et al. (2013). The PhytOC stock in the belowground biomass of eight monopodial bamboo species ranged from 2.57 to 23.71 kg ha−<sup>1</sup> , which was comparable to those in the aboveground of grassland (1.64 to 10.36 kg ha−<sup>1</sup> ) (Song et al., 2012a), wetland (0.82 to 21 kg ha−<sup>1</sup> ) (Li et al., 2013) and wheat (1.64 to 10.36 kg ha−<sup>1</sup> ) (Parr and Sullivan, 2011). The PhytOC stock in the belowground biomass of PPP (Phyllostachys praecox C. D. Chu 'Prevernalis') was even larger than that in its aboveground biomass. These observations suggested that though the PhytOC concentrations in belowground biomass were relatively smaller than those in the leaves or branches of bamboo or in other plants, the large total belowground biomass per hectare of the monopodial bamboo could contribute greatly to the PhytOC stock in belowground. In contrast, the PhytOC stock in the belowground biomass of bamboo species was much lower than that in the aboveground of sugarcane (32.73 to 98.18 kg ha−<sup>1</sup> ) (Parr et al., 2009), which could be explained by the higher phytolith accumulation ability and greater biomass of aboveground per unit area of sugarcane compared to monopodial bamboo (Tu, 2011). Qi et al. (2017) observed that the PhytOC stock in belowground biomass was about 40 times of that in aboveground biomass in a typical steppe grassland due to the greater belowground PhytOC content and net primary productivity. In agreement, our study for the first time showed that the PhytOC stock in belowground biomass makes up an average of 23.36% of the total PhytOC stocks of plant biomass among the eight bamboo species, though the percentage value was much smaller than that reported by Qi et al. (2017). Taking the mean value (14.60 kg·ha−<sup>1</sup> ) of PhytOC stock in belowground biomass across the eight species and China's current

#### REFERENCES


monopodial bamboo area of 5.85 × 10<sup>6</sup> ha, we estimated that the belowground PhytOC stock of monopodial bamboo is 1.01 × 10<sup>5</sup> t, and approximately 3.69 × 10<sup>5</sup> t CO<sup>2</sup> would be sequestered in belowground phytoliths of Chinese monopodial bamboo forests. According to Huang et al. (2014) and Yang et al. (2015) who estimated that the total aboveground PhytOC stock of the eight species was 4.27 × 10<sup>5</sup> t, this study further showed that the total belowground PhytOC stock of the eight species was 9.14 × 10<sup>4</sup> t, accounting for 21.38% of the whole plant PhytOC stock in China. Our study provides an important finding that the belowground biomass of bamboo is a large PhytOC stock that should be taken into account when estimating the potential of PhytOC sequestration of the whole bamboo biomass accurately in future studies. Therefore, the findings here supported our hypothesis that the bamboo species differ greatly in PhytOC concentrations in their belowground biomass and between species, and that the belowground parts make a significant contribution to the total PhytOC sequestration of bamboo plants.

## CONCLUSION

Our study reveals that the PhytOC concentration in the belowground trunk and rhizome varied among the studied bamboo species. The PhytOC stock in belowground biomass makes up an average of 23.36% of the total PhytOC stocks of plant biomass among the eight bamboo species. Based on our results, approximately 3.69 × 10<sup>5</sup> t CO<sup>2</sup> would be sequestered in belowground phytoliths of Chinese monopodial bamboo forests, suggesting that the belowground biomass of bamboo represent a great PhytOC stock, and should not be overlooked in future studies in order to better quantify the PhytOC sequestration capacity.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

## FUNDING

This work was financially supported by the National Natural Science Foundation of China (Grant No. 31600494) and the Science and Technology Projects of Zhejiang Province, China (Grant No. 2016C33028).



carbon sequestration. Glob. Chang. Biol. 19, 2907–2915. doi: 10.1111/gcb. 12275


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Chen, Huang, Jiang, Chen and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Phytolith-Occluded Carbon Storages in Forest Litter Layers in Southern China: Implications for Evaluation of Long-Term Forest Carbon Budget

Xiaodong Zhang<sup>1</sup> , Zhaoliang Song<sup>1</sup> \*, Qian Hao<sup>1</sup> , Yidong Wang<sup>2</sup> , Fan Ding<sup>3</sup> and Alin Song<sup>4</sup>

1 Institute of Surface-Earth System Science, Tianjin University, Tianjin, China, <sup>2</sup> Tianjin Key Laboratory of Water Resources and Environment, Tianjin Normal University, Tianjin, China, <sup>3</sup> College of Land and Environment, Shenyang Agricultural University, Shenyang, China, <sup>4</sup> Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing, China

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Peikun Jiang, Zhejiang A&F University, China Paul Eric Reyerson, University of Wisconsin–La Crosse, United States Jonas Schoelynck, University of Antwerp, Belgium

\*Correspondence:

Zhaoliang Song songzhaoliang78@163.com

#### Specialty section:

This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science

Received: 17 December 2018 Accepted: 18 April 2019 Published: 03 May 2019

#### Citation:

Zhang X, Song Z, Hao Q, Wang Y, Ding F and Song A (2019) Phytolith-Occluded Carbon Storages in Forest Litter Layers in Southern China: Implications for Evaluation of Long-Term Forest Carbon Budget. Front. Plant Sci. 10:581. doi: 10.3389/fpls.2019.00581 Phytolith-occluded carbon (PhytOC) can be preserved in soils or sediments for thousands of years and might be a promising potential mechanism for long-term terrestrial carbon (C) sequestration. As the principal pathway for the return of organic matters to soils, the forest litter layers make a considerable contribution to terrestrial C sequestration. Although previous studies have estimated the phytolith production fluxes in the above-ground vegetations of various terrestrial ecosystems, the storages of phytoliths and PhytOC in litter layers have not been thoroughly investigated, especially in forest ecosystems. Using analytical data of silica, phytoliths, return fluxes and storages of forest litter, this study estimated the phytolith and PhytOC storages in litter layers in different forest types in southern China. The results indicated that the total phytolith storage in forest litter layers in southern China was 24.34 ± 8.72 Tg. Among the different forest types, the phytolith storage in bamboo forest litter layers (15.40 ± 3.40 Tg) was much higher than that in other forests. At the same time, the total PhytOC storage reached up to 2.68 ± 0.96 Tg CO<sup>2</sup> in forest litter layers in southern China, of which approximately 60% was contributed by bamboo forest litter layers. Based on the current litter turnover time of different forest types in southern China, a total of 1.01 ± 0.32 Tg of PhytOC per year would be released into soil profiles as a stable C pool during litter decomposition, which would make an important contribution to the global terrestrial long-term biogeochemical C sink. Therefore, the important role of PhytOC storage in forest litter layers should be taken into account in evaluating long-term forest C budgets.

Keywords: forest litter layer, phytolith, phytolith-occluded carbon, carbon sequestration, forest carbon budget

## INTRODUCTION

Global warming, as one of the major challenges facing human survival and development, is mainly caused by the rapid increases in greenhouse gas (e.g., CO2) concentrations in the atmosphere (IPCC, 2013; Fang et al., 2018). Terrestrial biogeochemical carbon (C) sequestration counteracts about 30% of the total anthropogenic CO<sup>2</sup> emissions to the atmosphere, and thus plays a

**223**

crucial role in mitigating long-term climate warming (Law and Harmon, 2011). Currently, one of the most promising mechanisms of terrestrial biogeochemical C sequestration is C occlusion within phytoliths (phytolith-occluded carbon, PhytOC), which has drawn the attention of many researchers (Parr and Sullivan, 2005; Zuo and Lü, 2011; Song et al., 2012b; Li et al., 2013).

Phytoliths, also called plant stones or plant opal, are the silicified features of plants and mainly take the shapes of plant cell walls, cell lumens and the intercellular spaces of the cortex (Piperno, 1988; Parr and Sullivan, 2011). Generally, silicon (Si) in the soil solution is taken up by plant roots in the form of Si(OH)<sup>4</sup> or Si(OH)3O−, then transported with the transpiration stream and finally deposited as phytoliths or nanostructures of silica bodies (Ma, 2003; Neumann, 2003). Compared with nanostructures of silica bodies, the size of phytoliths mainly ranges from 5 to 250 µm and phytolith morphotypes generally vary with plant species, which makes them more stable due to their microscale internal cavities (Piperno, 1988; Strömberg, 2004; Lu et al., 2007; Song et al., 2016). Phytolith contents depend not only on plant phylogeny (Hodson et al., 2005), but also on the type of plant tissues and the soil Si availability (Van Soest, 2006; Henriet et al., 2008; Li et al., 2013; Yang X.M. et al., 2015). As phytoliths consist mainly (66–91%) of silica (SiO2) and show a positive correlation with Si content in plant (Li B.L. et al., 2014), the phytolith content can be estimated directly or indirectly from plant Si content (Hodson et al., 2008; Parr et al., 2010; Song et al., 2012a, 2013; Anala and Nambisan, 2015).

During the formation of phytoliths, between 0.2 and 5.8% of organic C can be occluded within the phytoliths (Bartoli and Wilding, 1980; Parr et al., 2010; Santos et al., 2010; Li et al., 2013). Relative to other organic C fractions, PhytOC is stable and can persist in the soils at a millennial scale due to the strong resistance of phytoliths to decomposition (Wilding, 1967; Parr and Sullivan, 2005; Zuo et al., 2014). For example, Zhang et al. (2017) have estimated that soil phytolith turnover time in the subtropical and tropical areas ranged from 433 to 1018 years. Previous studies indicated that the PhytOC accumulation rate in tropics and subtropics was 7.2– 8.8 kg ha−<sup>1</sup> yr−<sup>1</sup> , which contributed to nearly 37% of the global mean long-term soil organic carbon accumulation rate (Parr and Sullivan, 2005). In addition, the average turnover time of soil phytoliths ranged from 200 years to longer than 1000 years for most terrestrial ecosystems (Borrelli et al., 2010; Parr et al., 2010; White et al., 2012). Although a fraction of phytoliths may be dissolved, many studies have demonstrated that most phytoliths are stable and could be conserved for hundreds of years. Thus, the potential of phytoliths for the long-term terrestrial biogeochemical sequestration of atmospheric CO<sup>2</sup> is quite considerable (Parr et al., 2010; Song et al., 2012b).

In global terrestrial ecosystems, approximately 50–90% of the total annual C flux occurs between forests and the atmosphere (Bonan, 2008; Beer et al., 2010), indicating a significant contribution of forests to the terrestrial biogeochemical C cycle (Fang et al., 2002). In China, the area of forest is approximately 2.08 × 10<sup>8</sup> ha according to the Eighth National Inventory of Forest Resources (Shen et al., 2017; Fang et al., 2018). In terms of geographical distribution, more than 35% of the forest resources in China are distributed in southern region. Previous studies have focused mainly on the production fluxes of phytoliths or PhytOC in above-ground vegetation of various forest types. For example, Song et al. (2013) indicated that the phytolith C sequestration in the above-ground vegetation in China's forest was about 1.7 Tg CO<sup>2</sup> yr−<sup>1</sup> , approximately 30% of which was attributed to bamboo due to its high PhytOC production. Li B.L. et al. (2014) indicated that the phytolith C sequestration by bamboo in China was equivalent to 0.29 Tg CO<sup>2</sup> yr−<sup>1</sup> , approximately 75, 3, and 22% of which was contributed by scattered, mixed and clustered bamboo communities, respectively. However, the contributions of phytoliths and PhytOC in forest litter layers as C storages have not been studied in depth.

As a principal pathway for the return of organic matter to soils, litter layers represent significant C stocks and have a distinct influence on the C dynamics in forest ecosystems (Harmon et al., 1986). Therefore, estimating the phytolith and PhytOC storages in forest litter layers at a regional scale is very essential and would play a significant role in predicting the future evolution of forest C storages under different climatic conditions. In the forests of southern China, previous studies have investigated litter and its C storages in the forest litter layer (Cornwell et al., 2010; Shen et al., 2017; Zhu et al., 2017). However, the extent of the PhytOC storages in forest litter layers and its distribution among different forest types remain unknown, although a few local-scale studies have estimated PhytOC storage in the litter layers (He et al., 2016; Xiang et al., 2016; Ying et al., 2016). In this study, we used data on silica, phytoliths, and forest litter return flux and storage in forests of southern China to estimate the phytolith and PhytOC storages in litter layers in different forest types in southern China, aiming to provide a reference for a future re-evaluation of forest C budgets.

## MATERIALS AND METHODS

### General Characteristics of the Forest Types in Southern China

In southern China, forests are categorized into six types, according to the principles of Chinese vegetation regionalization (Fang et al., 2002). They are subtropical and tropical coniferous forest (STC), subtropical coniferous and broadleaf mixed forest (SCB), subtropical evergreen and deciduous broadleaf forest (SEDB); subtropical evergreen broadleaf forest (SEB), subtropical and tropical bamboo forest (STB), and tropical monsoon forest (TM) (**Table 1**). Across the six types of forest, the mean annual temperature (MAT) varies from 2 to 25◦C, and the mean annual precipitation (MAP) ranges from 500 to 2000 mm. The

TABLE 1 | Properties of dominant forest types in southern China.


STC, subtropical and tropical coniferous forest; SCB, subtropical coniferous and broad-leaf mixed forest; SEDB, subtropical evergreen and deciduous broad-leaf forest; SEB, subtropical evergreen broad-leaf forest; STB, subtropical and tropical bamboo forest; TM, tropical monsoon forest.

main species composition of each forest type in this study are shown in **Table 1**.

#### combined data for SiO<sup>2</sup> content in the two pools from previous studies (Yang J. et al., 2015; He et al., 2016; Xiang et al., 2016; Ying et al., 2016).

### Phytolith Content–Silica Content Transfer Function and Phytolith Content Estimation

The data for SiO<sup>2</sup> content of mature leaves across different tree species in southern China were collected from published monographs (Hou, 1982; Chen et al., 1997; Song et al., 2013) and theses (He et al., 2016; Ying et al., 2016). To calculate SiO<sup>2</sup> content in forest litter, we constructed a transfer function (Eq. 1) between SiO<sup>2</sup> content in mature leaves and in forest litter by a regression analysis method (**Figure 1**), based on the

SiO<sup>2</sup> content in forest litter (wt. %) = 1.211 × SiO<sup>2</sup> content in mature leaves (wt. %)

$$(\mathbb{R}^2 = 0.8979, p \prec 0.01) \tag{1}$$

As phytoliths consist mainly of SiO<sup>2</sup> and the phytolith content in litter leaf is generally significantly positively correlated with SiO<sup>2</sup> content of the leaf litter (Parr and Sullivan, 2005; Song et al., 2012a, 2013), phytolith content of different forest litter layers can be estimated based on the phytolith content-SiO<sup>2</sup> content transfer function of the samples published in the paper by Song et al. (2013), as follows:

Phytolith content (wt. %) = 0.953 × silica content (wt. %)

$$(\mathbb{R}^2 \,:\, 0.96, p \,\, <\, 0.01) \tag{2}$$

#### Estimation of Phytolith and Return Fluxes and PhytOC Storages in Litter Layers From Different Forest Types

When plants or plant parts die and decay, phytoliths formed in plant tissues can return to the forest floor, along with litter, maintaining their morphological integrity and their chemical characteristics (Strömberg, 2004; McInerney et al., 2011). Therefore, phytolith return flux of litter layers in different forest types can be estimated based on the data of phytolith content and the return flux of forest litter:

phytolith return flux = litter return flux × phytolith content (3)

where phytolith return flux is the weight of phytoliths returned to the floor in a given forest type per area per year (kg ha−<sup>1</sup> yr−<sup>1</sup> ), phytolith content is the content of phytolith in the unit mass of

forest litter (wt. %), and litter return flux is the net return flux of forest litter in per area per year (kg ha−<sup>1</sup> yr−<sup>1</sup> ).

Phytolith-occluded carbon content is the organic C content occluded within phytoliths. When the organic materials wrapped on the surface of phytoliths are completely removed and the phytoliths remain intact, the values of PhytOC content are precise. Previous studies indicated that PhytOC content ranges from less than 0.1% to up to 10%, but mainly from 0.2 to 5.8% (Jones and Handreck, 1965; Parr and Sullivan, 2005; Santos et al., 2012; Song et al., 2016). Therefore, in this study, we used a median PhytOC concentration in phytoliths (3%) to estimate the PhytOC storages in different forest litter layers. The PhytOC storages in different forest litter layers were calculated based on the values for litter storage per unit area, phytolith content, PhytOC content and forest area as follows:

PhytOC storage = litter storage per area × phytolith content × PhytOC content × forest area × [44/12] (4)

where PhytOC storage is the total PhytOC amount in each forest litter layer (Tg CO2), litter storage per area is the storage of litter in per area of different forest floors (t ha−<sup>1</sup> ), phytolith content is the content of phytolith in different forest litter layers and can be estimated by Eq. 2 (wt. %), and forest area is the area of each forest type in southern China (10<sup>6</sup> ha). The equation is multiplied by [44/12] to transform the data from 'Tg C' to 'Tg CO2. '

#### RESULTS

#### Phytolith Concentration in Different Forest Litter Layers

There was a distinct difference in litter SiO<sup>2</sup> concentration among various plant species (**Table 2**). Phytolith concentration in forest litters ranged from 0.64 to 203.37 g kg−<sup>1</sup> across all the different plant species. At the same time, the phytolith concentration of forest litter layers also varied greatly among the different forest types (2.45–148.54 g kg−<sup>1</sup> ). Generally, the phytolith concentration of forest litter layers in STB was 148.54 ± 32.77 g kg−<sup>1</sup> , which was the highest among all forest type. In SEDB, SEB, and TM, the phytolith concentration of forest litter layers was 24.54 ± 16.34 g kg−<sup>1</sup> , 17.18 ± 9.49 g kg−<sup>1</sup> and 20.14 ± 14.56 g kg−<sup>1</sup> , respectively, which were moderate values among the various forest types. The lowest phytolith concentration of forest litter layers were found in STC (2.45 ± 1.21 g kg−<sup>1</sup> ) and SCB (4.29 ± 2.30 g kg−<sup>1</sup> ).

#### The Phytolith Return Fluxes and PhytOC Storages of Forest Litter Layers in Southern China

The phytolith return fluxes through forest litter were significantly different among the different types of forest (**Table 3**). Generally, the mean ± SD phytolith return flux for the six types of forest in southern China was 168.73 ± 58.67 kg ha−<sup>1</sup> yr−<sup>1</sup> . Phytolith return flux was the highest in STB (484.25 ± 106.93 kg ha−<sup>1</sup> TABLE 2 | SiO<sup>2</sup> and phytolith contents in litter of dominant tree species from six forest types in southern China.


#### TABLE 2 | Continued

fpls-10-00581 May 2, 2019 Time: 17:45 # 5


SiO<sup>2</sup> contents in forest litter layer are estimated from silica contents of mature leaves using forest silica content transfer function between silica content in litter layer and mature leaves of Eq. 1. At the same time, phytolith contents in forest litter layer are estimated base on forest phytolith content-silica content transfer function of Eq. 2 from Song et al. (2013).

yr−<sup>1</sup> ), moderate in SEDB (154.35 ± 102.94 kg ha−<sup>1</sup> yr−<sup>1</sup> ), SEB (145.34 ± 80.28 kg ha−<sup>1</sup> yr−<sup>1</sup> ) and TM (183.31 ± 132.86 kg ha−<sup>1</sup> yr−<sup>1</sup> ), and lowest in STC (9.94 ± 4.97 kg ha−<sup>1</sup> yr−<sup>1</sup> ) and SCB (35.17 ± 18.81 kg ha−<sup>1</sup> yr−<sup>1</sup> ) (**Table 3**). Furthermore, the total PhytOC storage in forest litter layer in southern China was 2.68 ± 0.96 Tg CO2. Similarly, PhytOC storage was the highest in STBF (1.69 ± 0.37 Tg CO2), which contributed to more than 60% of the total PhytOC storage in forests in southern China, followed by SEDB (0.38 ± 0.26 Tg CO2) and SEB (0.49 ± 0.27 Tg

CO2), with the lowest being STC (0.07 ± 0.03 Tg CO2), SCB (0.03 ± 0.01 Tg CO2), and TM (0.02 ± 0.01 Tg CO2) (**Figure 2**).

#### DISCUSSION

#### Impact of Different Factors on Phytolith Content in Forest Litter Layers

Previous studies have demonstrated that phytolith contents ranged from less than 0.5% in most dicotyledons to more than 15% in some Gramineae, such as bamboo (Epstein, 1994; Hodson et al., 2005; Seyfferth et al., 2013). When plants or plant parts die and decay, phytoliths present in terrestrial plants can be returned to the forest floors in the litters. In this study, the phytolith concentration in forest litter layers varied significantly among different forest types (2.45–148.54 g kg−<sup>1</sup> ), due mainly to differences in phytolith return flux, litter decomposition rate and phytolith stability (Parr et al., 2010; Song et al., 2012b, 2016).

Phytolith return fluxes showed significant differences among different forest types (p < 0.05), which was due to different litter return fluxes or to different phytolith contents in the litter (Song et al., 2013; Yang X.M. et al., 2015; **Table 3**). For example, phytolith return flux in the bamboo-dominated STB (484.25 ± 106.93 kg ha−<sup>1</sup> yr−<sup>1</sup> ) was significantly higher than


†The data of litter return fluxes and storages in different forest types are from Chen et al. (1997); Peng and Liu (2002), Guan et al. (2004); Lu et al. (2012), Guo et al. (2015); Jia et al. (2016), Ying et al. (2016), and Liu et al. (2017, 2018).

that in other forest types (p < 0.05). Song et al. (2013) estimated the phytolith contents in China's forests using the biogenic silica content-phytolith content transfer function and the results indicated that phytolith concentration in different forests in China ranged from 0.5 to 124.5 g kg−<sup>1</sup> with the average phytolith concentration in STB (105.2 g kg−<sup>1</sup> ) being between four and fifty times higher than that in other forest types (Song et al., 2013). This reflects the high Si content of bamboos. A recent study also indicated that the plant species composition of each forest significantly influenced the production and accumulation of phytoliths (Yang et al., 2018). In this study, the plant species compositions of the six types of forest were distinctly different (**Table 1**), which would be a major cause of their differences in phytolith return flux among different forest litter layers. From our investigation, there were distinct differences between the litter return fluxes in various forest types in southern China (**Table 3**), a variable which also plays a significant role in phytolith return flux (He et al., 2016; Ying et al., 2016).

Phytoliths accumulated in forest litter layers could be released into soil profiles by litter decomposition, which has an impact on the phytolith content of the forest litter layer. Therefore, litter decomposition rate is another factor influencing phytolith content in forest litter layers. The drivers of litter decomposition rate are multiple, including the effects of environment, composition of the decomposer communities, and the substrate characteristics of the forest litter (Cornelissen, 1996; Aerts, 1997; Cornwell et al., 2010). Previous studies have shown that the decomposition rate of forest litter can vary under different temperature and moisture conditions, as a result of changes in decomposer community composition and biological activities (Waltman and Ciolkosz, 1995; Liski et al., 2010). In this study, climatic conditions in the different forest types show distinct differences (**Table 1**), probably contributing to different decomposition rates. Furthermore, the main plant species composition of different forest types under various climate conditions in southern China show fundamental differences (**Tables 1**, **2**), which result in different plant species traits. Previous studies indicated that plant species traits were thought to be a major factor that determined the litter decomposition rates (Cornwell et al., 2010; Li Z.M. et al., 2014; Lu et al., 2017). For example, the nutrient chemistry, stoichiometry and physical features of the forest litter had marked effects on the activity and abundance of microbial decomposers (Melillo et al., 1982). At the same time, differences in the plant species composition of different forest types in southern China, with their associated differences in phytolith content (e.g., the phytolith-rich bamboos dominating the STB) could affect the phytolith content in different forest litter layers (Song et al., 2013).

Phytolith stability is another factor influencing phytolith content in the litter layers from different forests. Previous studies had demonstrated that the phytolith geochemical stability is mainly controlled by phytolith properties and climatic and edaphic conditions (e.g., pH, temperature, moisture, etc.) (Iler, 1979; Bartoli and Wilding, 1980; Li Z.M. et al., 2014). For example, Bartoli (1985) demonstrated that phytoliths from beech leaves had a lower degree of crystallization and a lower Al content than those from pine needles, with beech having a much higher equilibrium concentration of silicic acid (300 µmol Si L−<sup>1</sup> ) compared to pine (100 µmol Si L−<sup>1</sup> ). Furthermore, previous studies indicated that phytolith dissolution rate may increase with soil pH (Fraysse et al., 2006, 2009). Blecker et al. (2006) estimated the turnover times of soil phytoliths in the Central Great Plains across the bioclimosequence and found a distinct correlation of faster turnover with MAP increasing. In addition, Song et al. (2017) estimated the phytolith stability factors in different forest ecosystems based on the phytolith turnover time, and the results showed that phytolith stability factors in tropical forest, temperature forest and boreal forest ranged from 0.6 to 0.9 (Song et al., 2017). In this study, marked differences in plant species compositions and climatic edaphic conditions among the various forest types contributed to the differences in phytolith stability in the various forest litter layers. Although the phytolith contents in litter layers of different forest types are affected by many factors, such as microbial activity, temperature, moisture and phytolith stability so on, phytolith contents in different plant species play the most important role in controlling phytolith contents in litter layers of different forest types in this study.

## PhytOC Storages in Forest Litter Layers in Southern China

In this study, there were distinct differences in the litter storages among different forest types due to the differences in stand composition, MAT, MAP, and altitude of each forest type (**Table 3**). Based on the phytolith contents and forest litter storages in different forest ecosystems, the storages of phytoliths in the various forest types in southern China were calculated and the results showed that the total phytolith storages in forest litter layers in southern China was 24.34 ± 8.71 Tg. Among the different forest types, the phytolith storage of litter layers in bamboo forest (15.40 ± 3.40 Tg) was much higher than that in other forests. Assuming the median concentration of 3% C occluded during the formation of the phytoliths, the total PhytOC storage could reach up to 2.68 ± 0.96 Tg CO<sup>2</sup> in forest litter layers in southern China, approximately 60% of which was contributed by bamboo forest litter layer but which occupied only 9.5% of the forest area in this region (**Figures 2**, **3**). Global bamboo forest areas in 1990s and now are 1.75 × 10<sup>7</sup> ha and 2.2 × 10<sup>7</sup> ha, respectively, and mainly distributed in the subtropical and tropical regions (Liang, 1990; Zhou et al., 2011; Li B.L. et al., 2014). Based on the current PhytOC content in the litter layer of the bamboo-dominated STB and the global bamboo distribution area in 1990s and now, we calculated that the PhytOC storages of litter layers in the global bamboo ecosystem in 1990s and now were 4.12 ± 0.91 Tg CO<sup>2</sup> and 5.18 ± 1.14 Tg CO2, respectively. It is noted that although the forest area in some countries has obviously decreased, the distribution area of bamboo forest in the world has increased at a rate of 3% annually over the last decade and will continue to increase until 2050 due to bamboo afforestation in forest-priority land (Song et al., 2013; Shen et al., 2017). Previous studies have indicated that the global bamboo distribution area will increase from 25 × 10<sup>6</sup> ha to 100 × 10<sup>6</sup> ha by 2050, by which point it will occupy approximately 3% of the world's forest, as a result of bamboo afforestation/reforestation

in the subtropical and tropical regions of the world (Zhou et al., 2011; Song et al., 2013). Therefore, based on the rate of increase (3%) of bamboo forest area in the world, the potential size of the global phytolith carbon sink in the litter layer of the bamboo ecosystems would reach up to 13.33 ± 2.94 Tg CO<sup>2</sup> by 2050 (**Figure 4**), indicating that bamboo forest will play an increasingly important role in regulating atmospheric CO<sup>2</sup> sequestration in the form of the bamboo forest phytolith C sink. Although the end result (total PhytOC storage in forest litter layers in southern China) is not fixed and has some uncertainties which are mainly caused by land use changes, litter storages and PhytOC contents in various forest litter, our preliminary estimation is reasonable. In this study, we carefully calculated the minimum, maximum and mean values of PhytOC storages of forest litter layers in different forest types. The results will provide a baseline for evaluating forest carbon budget in the future.

## Implications for Evaluation of Forest Carbon Budget

Phytolith-occluded carbon, as one of the long-term global biogeochemical C sink mechanisms, has attracted the attention of many researchers. Although some phytolith particles (<2 µm) can be quickly dissolved due to their high surface area, approximately 80% of phytoliths released by litter decomposition can be preserved in soils or sediments for 400–3000 years due to their relatively intact surfaces (Parr and Sullivan, 2005; Song et al., 2016; Yang et al., 2018). For example, Zhang et al. (2017) have estimated that soil phytolith turnover time in subtropical and tropical areas ranged from 433 to 1018 years. In the present study, all of the forest types were natural forests with minimal human interference. Assuming that the litter storages in the various forest types have achieved a dynamic balance, the litter turnover time can be estimated from the litter storage size and litter return flux. Results showed that the turnover time in STB (4.42 ± 0.74 years) was the longest, with moderate rates in STC (2.15 ± 0.43 years), SCB (1.42 ± 0.26 years), SEDB (1.81 ± 0.15 years), SEB (1.42 ± 0.43 years), and the shortest rates in TM (0.90 ± 0.01 years), findings which were consistent with the results of Zhang and Wang (2015). Based on the litter turnover time and the size of the PhytOC storage in various forest litter layers, we estimate that a total of 1.01 ± 0.32 Tg CO2, in the form of long-term stable organic C components, are released into soil profiles in per year by litter decomposition in southern China. The size of the PhytOC storages in STC, SCB, SEDB, SEB, STB, and TM were estimated to be 0.03 ± 0.02, 0.02 ± 0.01, 0.21 ± 0.01, 0.34 ± 0.19, 0.38 ± 0.08, and 0.02 ± 0.01 Tg CO2, respectively. Furthermore, several very large national ecological restoration projects (e.g., Natural Forest Protection Program, the Desertification Combating Program around Beijing and Tianjin, the Sloping Land Conversion Program, etc.) have been implemented to slow climate change and to protect the global environment (Fang et al., 2018), and results show that the national forest litter stock has continuously increased at a steady rate over the past 20 years, mainly due to the expansion of the forest area (Zhu et al., 2017). This implies that increasing amounts of PhytOC will be stored in forest litter layers over the coming decades, while a small proportion of the phytoliths will be dissolved during the decomposition process of forest litter. Therefore, our findings highlight that the PhytOC storage in forest litter layers should be taken into account in the future in any evaluation of the forest C budget, which will play an increasingly important role in the global long-term biogeochemical C sink at a centennial scale.

#### CONCLUSION

In this study we mainly estimated the sizes of PhytOC storages in the litter layers of different forest types in southern China. The results showed that the PhytOC storage was the highest in STB (1.69 ± 0.37 Tg CO2), followed by SEDB (0.38 ± 0.26 Tg CO2)

and SEB (0.49 ± 0.27 Tg CO2), with the smallest storages being in STC (0.07 ± 0.03 Tg CO2), SCB (0.03 ± 0.01 Tg CO2) and TM (0.02 ± 0.01 Tg CO2). The total PhytOC storage in forest litter layers in southern China was estimated to be 2.68 ± 0.96 Tg CO2, approximately 60% of which was contributed by bamboo forest. In addition, the total amount of PhytOC, as a long-term stable organic C component, released into soil profiles per year by litter decomposition in southern China was estimated to be 1.01 ± 0.32 Tg CO2. Based on the current PhytOC content in bamboo litter layers, the potential of the phytolith carbon sink in the world's bamboo ecosystem litter layers could reach up to 13.33 ± 2.94 Tg CO<sup>2</sup> in 2050 by practices such as bamboo afforestation/reforestation in the subtropical and tropical regions of the world. Thus, the importance of the PhytOC storage in litter layers of the terrestrial forest ecosystems should be taken into account when evaluating forest C budgets since it plays

#### REFERENCES


a significant role in long-term C sequestration operating at centennial-millennial scales.

#### AUTHOR CONTRIBUTIONS

XZ and ZS analyzed the data. QH, YW, FD, and AS contributed to revise it for publication.

## FUNDING

We acknowledge the support from the National Natural Science Foundation of China (41571130042 and 41522207) and the State's Key Project of Research and Development Plan of China (2016YFA0601002 and 2017YFC0212703).




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor is currently organizing a Research Topic with one of the authors ZS, and confirms the absence of any other collaboration.

Copyright © 2019 Zhang, Song, Hao, Wang, Ding and Song. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Relative Importance of Cell Wall and Lumen Phytoliths in Carbon Sequestration in Soil: A Hypothesis

#### Martin J. Hodson\*

Department of Biological and Medical Sciences, Faculty of Health and Life Sciences, Oxford Brookes University, Oxford, United Kingdom

There has been much interest in the possibility that phytoliths might sequester substantial amounts of carbon and might continue to do so in soils and sediments after the death of the plant. This may contribute to mitigating climate change. However, this idea is controversial and it is unclear how much carbon is sequestered in phytoliths. High values would suggest that sequestration on a global scale could be significant, but low values would indicate insignificant sequestration. Different methodologies in preparing phytoliths give different carbon concentrations. Little interest has been shown in determining which types of phytoliths are most important for carbon sequestration. There are two main types of phytolith in plants, the cell wall types which are formed on a carbohydrate matrix, and the cell lumen types which are not. A literature survey of transmission and scanning electron microscopy studies to determine which phytoliths are cell wall phytoliths was carried out. Cell wall silicification was common in most plant organs and throughout the plant kingdom. Macrohairs, prickle hairs, and the wall protrusion of papillae are certainly cell wall types. The primary cell walls of many epidermal cells types are often silicified. Cell wall phytoliths have considerably higher carbon concentrations than lumen types. An attempt is made to model mixtures of cell wall and lumen phytoliths, containing different carbon concentrations. Literature data on carbon and nitrogen concentrations in phytoliths was used to produce C/N ratios. These showed that cell wall phytoliths had higher C/N ratios than lumen phytoliths, and that over-extraction of phytolith mixtures removes carbon preferentially from the cell wall types and leads to low C/N ratios. The dissolution of phytoliths in soils and sediments is considered, and it is unknown whether cell wall or lumen phytoliths break down faster. However, it is clear from the literature that cell wall phytoliths persist in soils and sediments for hundreds or thousands of years. The paper is brought to a climax with two hypotheses, one to explain what happens to carbon in phytoliths as they undergo preparatory procedures in the laboratory, and the other looking at dissolution and breakdown in the soil.

Keywords: phytolith, silica, carbon sequestration, cell wall, soil

#### Edited by:

Jon Telling, Newcastle University, United Kingdom

#### Reviewed by:

Ziming Yang, Oakland University, United States Zimin LI, Catholic University of Louvain, Belgium

> \*Correspondence: Martin J. Hodson mjhodson@brookes.ac.uk

#### Specialty section:

This article was submitted to Geochemistry, a section of the journal Frontiers in Earth Science

Received: 17 December 2018 Accepted: 14 June 2019 Published: 02 July 2019

#### Citation:

Hodson MJ (2019) The Relative Importance of Cell Wall and Lumen Phytoliths in Carbon Sequestration in Soil: A Hypothesis. Front. Earth Sci. 7:167. doi: 10.3389/feart.2019.00167

#### INTRODUCTION

feart-07-00167 June 29, 2019 Time: 17:6 # 2

The sequestration of carbon in soils has now become a topic of global significance. It is recognized that soils store very considerable amounts of carbon. If we could find ways of increasing that storage it might go some way toward stabilizing atmospheric carbon dioxide concentrations and thereby help in the fight against climate change. Powlson et al. (2011) pointed out that carbon sequestration in soil suffered from a number of constraints. Firstly, the quantity of carbon stored is finite. Secondly, the process is reversible. Finally soil organic carbon may be increased, but there may be changes in the fluxes of nitrous oxide and methane, important greenhouse gasses.

Parr and Sullivan (2005) first suggested the possibility that phytoliths could play a major part in carbon sequestration in soils. Their proposition was that so-called phytolith occluded carbon (PhytOC) might be locked up in phytoliths for centuries or longer, and not be returned to the atmosphere as quickly as other components of the soil organic matter. So the idea is that carbon sequestered as PhytOC would be less labile, and the reversible nature of sequestration mentioned by Powlson et al. (2011) would be reduced. In their abstract Parr and Sullivan wrote, "Estimated PhytOC accumulation rates were between 15 and 37% of the estimated global mean long-term (i.e., on a millennial scale) soil carbon accumulation rate of 2.4 g C m−<sup>2</sup> year−<sup>1</sup> indicating that the accumulation of PhytOC within soil is an important process in the terrestrial sequestration of carbon." If true, this would be a highly significant finding that could have very major implications for our understanding of the global carbon cycle and for methodologies to reduce global warming. Parr and Sullivan also suggested that it might be possible to select plant species that were particularly high in PhytOC to increase carbon sequestration. Their work stimulated the interest of researchers around the world, and there are now many publications on this topic. However, work on PhytOC has not been without controversy. As, we shall see below this has focused on methodology, with different methods of preparing phytoliths for analysis giving different values for PhytOC. Essentially, if a technique gives a high value for PhytOC then when the value is entered into the equations for estimating global carbon sequestration it will suggest that phytoliths are very important in this process. Conversely, if PhytOC values measured are low then the calculated sequestration at a global scale will be low. This has led to a vigorous debate: what is the "real" value of PhytOC? A related, and even more disputed, area of phytolith research at the moment is the whole topic of "old carbon" from the soil being taken up by plants, and deposited in phytoliths, causing problems in carbon dating. I have covered this area in two recent reviews (Hodson, 2016, 2018) and do not intend to look at it again here.

Phytoliths are morphologically diverse (Madella et al., 2005; Piperno, 2006), but it is becoming increasingly evident that they are also chemically diverse (Hodson, 2016). Kumar et al. (2017b) reviewed the various locations where phytoliths were found in grasses. It appears that silica deposition occurs in all tissues, including the roots, stems, leaves, inflorescence and seed (caryopsis), but that it is concentrated in certain organs and tissues. In the roots, silica is deposited in the endodermis, in the stems, leaves and inflorescence bracts the main deposition sites are in the epidermis, and small amounts are deposited in the seed in brush hairs and other locations. Less work has been done on species other than grasses and cereals, but in general the epidermis in leaves is the major location for most silica deposition (Piperno, 2006). There are three main types of silica deposition in plants: that where silica is deposited onto a carbohydrate matrix such as the cell wall; that where silica deposition lacks an obvious matrix onto which it is deposited, mainly in the cell lumen; and in intercellular spaces (Hodson, 2016). It does not appear that deposition in intercellular spaces is important in the production of recognizable phytoliths that survive once the plant dies, and so the two main types we need to consider are those in the cell walls and the cell lumen. I have previously assessed the evidence that the cell wall and lumen phytoliths have very different chemistries (Hodson, 2016, 2018). Here, we will concentrate on carbon within phytoliths. It would be expected that higher carbon concentrations will be found in cell wall types that are deposited on a carbohydrate matrix, and the evidence available suggests that this is the case.

There has been very little consideration of which phytolith types are the most important for PhytOC and carbon sequestration. This is, perhaps, surprising given the interest in this topic. The aim of this paper will be to bring together the available literature and to assess the relative importance of cell wall and lumen phytoliths in carbon sequestration. I will then develop a hypothesis concerning what happens to phytoliths as they are prepared for analysis and when they enter the soil environment.

#### A BRIEF HISTORY OF CARBON SEQUESTRATION IN PHYTOLITHS

Percentage carbon was first measured in phytoliths by Jones and Beavers (1963) who found that those isolated from a Cisne silt loam contained 0.86% carbon. They were also the first to suggest that carbon was occluded within phytoliths where it is protected from oxidation. For many years after that, papers emerged with varying estimates of the concentration of carbon in phytoliths. **Table 1** gives a selection of these publications arranged in date order. It was widely recognized that different methods of preparing phytoliths will give different results, but there was little controversy over this. Usually researchers were using the same method to investigate carbon concentration in a number of species or different organs from the same plant, and they were not comparing their results with other publications that used different methods. Often measuring percentage carbon was incidental to the main focus of the investigation with workers being more interested in carbon dating or carbon isotopes. It was only after Parr and Sullivan (2005) suggested that PhytOC might be important in helping to combat climate change that the controversy really began. It now very much mattered what concentration of carbon was to be found in phytoliths.

As can be seen in **Table 1**, in all of the early publications the preparatory techniques used by those wishing to measure

TABLE 1 | Selected examples of %C measurements in phytoliths.


PhytOC involved either wet ashing or dry ashing, and those are still the preferred methods for many workers. Essentially, wet ashing involves digestion of the plant material in strong acids and/or treatment with strong oxidizing agents such as hydrogen peroxide. In dry ashing, plant material is heated in a muffle furnace to a suitable temperature (often around 450– 500◦C), that will burn off the surrounding organic matter without damaging the phytoliths. The third method, microwave digestion, was first introduced by Parr et al. (2001). Parr and Sullivan (2014) compared wet ashing and microwave digestion, preferring the latter, as it kept more organic matter within certain classes of phytoliths (see below). A number of wet and dry ashing techniques were investigated by Corbineau et al. (2013), and all had some advantages and disadvantages. It would be fair to say that there is no consensus among scientists over which is the best method, and this has contributed to our problems in determining the "correct" value for PhytOC.

Parr, Sullivan and their co-workers set a firm basis for work on carbon sequestration in phytoliths. Importantly, they were able to show considerable differences between PhytOC in phytoliths from different sugarcane cultivars (Parr et al., 2009), bamboo species (Parr et al., 2010), wheat cultivars (Parr and Sullivan, 2011) and rice cultivars (Li et al., 2013b). Likewise, Zuo and Lü (2011) showed variation in PhytOC in different millet species. More recently, Sun et al. (2017) carried out an extensive survey of carbon sequestration in 51 rice cultivars, finding that there were significant differences between the amounts sequestered by different cultivars. All this work opens up the possibilities of planting certain species or cultivars which will increase carbon sequestration, and of breeding for this trait.

The work of Parr and Sullivan has been followed up and extended in China, mainly by Zhaoliang Song and his group. They have been particularly concerned to measure the potential for carbon sequestration in different environments in China: grasslands (Song et al., 2012a); wetlands (Li et al., 2013a); forests (Song et al., 2013); bamboo forests (Huang et al., 2014); and croplands (Song et al., 2014). In addition the group produced a number of reviews where they considered carbon sequestration at a global scale (e.g., Song et al., 2012b).

It seemed that the idea that phytoliths could sequester substantial amounts of carbon, and thereby help in combatting global warming, was becoming well established, but then Alexandre, Santos and their co-workers produced a series of papers reporting much lower values for PhytOC in their analyses (Santos et al., 2010; Alexandre et al., 2015, 2016; Reyerson et al., 2016). Like most of these papers Reyerson et al. (2016) concentrated on the "old carbon" hypothesis, but they did have a short paragraph looking at carbon sequestration in phytoliths. There they took the maximum PhytOC value that they found in their work (0.3%), and a phytolith stability factor of 10% (Alexandre et al., 2011), and calculated global annual carbon sequestration at 4.1 × 10<sup>4</sup> tC year−<sup>1</sup> . This is around 100 times lower than the 3.7 × 10<sup>6</sup> tC year−<sup>1</sup> suggested by Song et al. (2014). If Reyerson et al. (2016) are correct then the amount of carbon sequestered in phytoliths would be insignificant on a global scale.

The controversy came to a peak with the publication of two papers in Earth-Science Reviews in 2016 and 2017. Firstly, Song et al. (2016) reviewed the topic from their viewpoint, and then Santos and Alexandre (2017) responded with an almost point by point rebuttal. Song et al. then wrote a reply to Santos and Alexandre (2017), but this was soon "temporarily withdrawn" by Earth-Science Reviews in early 2017, and that remains the case (in June 2019). Clearly there are serious problems here, and it is a great pity that very good scientists have ended up in such a heated debate. I will not take sides here, but try to reconcile the conflicting opinions, and to introduce some new thinking which might help sort out a rather unfortunate situation. Where is the main point of contention? Song et al. (2016) and Santos and Alexandre (2017) disagree on a number of topics, and some of these will be touched on later, but the main one is undoubtedly the true concentration of carbon in phytoliths. Song et al. (2016) routinely use a figure of 3% for PhytOC, and Santos and Alexandre (2017) think this is too high and that the figure should

be 0.1–0.5%. The technique of extraction used by Song et al. (2016) is the microwave digestion technique developed by Parr et al. (2001), and Santos and Alexandre worry that this may leave organic contaminants on the surface of extracted phytoliths. The methods of preference for Alexandre, Santos and their coworkers are described in Corbineau et al. (2013), and involve dry ashing and acid digestion or alternatively acid digestion and alkali immersion. They then strongly advise checking the samples with scanning electron microscopy (SEM) and x-ray microanalysis for particles that have high C/Si ratios, and discarding those that have, thus eliminating contamination. Song et al. (2016) consider that the low PhytOC values preferred by Santos and Alexandre (2017) are caused by oxidation and over-extraction.

We appear to have reached an impasse with highly respected researchers taking up very entrenched positions. In the past this whole argument would not have happened, but it is now very important that we can determine PhytOC accurately if we are to assess its importance in carbon sequestration. The problem is that it is very difficult to obtain totally clean phytolith preparations without extracting some of the carbon from inside the structures. This may be even more difficult for cell wall phytoliths (see below). So I am uncertain that we will ever be able to give an exact figure for PhytOC, except in very rare circumstances. It is probably safer to give a range of potential concentrations for PhytOC, and to calculate carbon sequestration using a number of values. We should also remember that all the values in **Table 1** are estimates of the amount of carbon that is sequestered fairly tightly within the structure of the phytolith. How do they relate to the situation in the soil? Moreover, all of the data in **Table 1** are for mixtures of cell wall and lumen phytoliths. As, we shall see below the presence of two phytolith types with very different chemistries complicates matters even more.

## LUMEN AND CELL WALL PHYTOLITHS

Over many years of working on phytoliths I have talked with numerous scientists, seen many conference presentations, reviewed many papers, and read a lot more. I have noticed that some scientists concentrate on cell wall phytoliths and others on lumen phytoliths, and that this at least partly depends on the discipline of the scientist. Chemists and plant scientists have mostly worked on cell wall phytoliths and have a greater interest in what happens in the cell wall. It is possible that the chemists (e.g., Currie and Perry, 2007; Exley, 2015) prefer working on cell wall phytoliths because they have a matrix for deposition which makes the chemistry more interesting. The plant scientists (e.g., Coskun et al., 2018) see many important processes happening in the cell wall including transport, detoxification of metals and defense against pathogens. On the other hand, archeologists, palaeoecologists, and biogeochemists have sometimes seemed to downplay the significance of cell wall types, probably because of their perceived low stability in soils and sediments. For example, Song et al. (2017) stated that, "...C (carbon) from cell wall phytoliths may be quite labile and easily lost at an annual-decadal scale compared to C trapped in lumen phytoliths, which are likely to be much more stable at a centennial-millennial scale..." The otherwise excellent review of phytoliths in palaeoecology by Strömberg et al. (2018) paid almost no attention to cell wall phytoliths, even when considering the factors likely to increase the dissolution of phytoliths in soils and sediments.

If we are to understand this topic it is important that we have a clear idea of which phytoliths are cell wall types and which are lumen types. This can seem a simple question to answer, but experience suggests that it is not that easy, particularly just using light microscopy. Madella et al. (2005) gave us a nomenclature to classify phytoliths according to their morphology, but there was no mention of their chemistry as this was not the focus of the paper. In **Table 2**, I present some selected studies using transmission electron microscopy (TEM), where it is easy to determine which phytoliths are cell wall types and which are from the lumen. In TEM silica appears as electron opaque deposits, and the presence of silicon can be confirmed by x-ray microanalysis. In the context of this paper we are most interested in cell wall phytoliths, and that will bias the selection of publications in this section. The additional bias is that most of this work has been on grasses and cereals. The other methodology that is useful in some circumstances uses SEM in tandem with x-ray microanalysis (**Table 3**). Using SEM it is not always easy to distinguish between cell wall and lumen phytoliths and I have excluded any observations that I felt were uncertain from **Table 3**.

To facilitate further discussion it is helpful at this point to be able to visualize the different types of phytolith and how they develop. **Figure 1** showing five potential pathways of phytolith development. In **Figure 1A** only the primary cell wall is silicified, whilst in **Figure 1B** secondary cell walls develop to almost fill the lumen and silica is then deposited on them. In the third type (**Figure 1C**), silica is deposited in the space between the primary cell wall and the protoplast, and eventually this fills the lumen. In **Figure 1D** the protoplast breaks down and silica is subsequently deposited within the lumen, entrapping some organelles and membranes. Finally, in **Figure 1E** silica is first deposited in part of the primary cell wall, and this later grows into the lumen. These five types are not exclusive, and other types are possible. For example, in some cases both the primary (**Figure 1A**) and secondary (**Figure 1B**) cell walls are silicified. It is also possible for silicification to begin in the primary wall as in **Figure 1E**, and to continue into the space between the wall and the protoplast as in **Figure 1C**.

Almost all of the work on silica deposition in roots has concerned grasses and cereals. There seem to be two main types of deposition. In the roots of Phalaris canariensis (Hodson, 1986) and wheat (Hodson and Sangster, 1989) the cell walls of the endodermis become silicified (as in **Figure 1A**). The sorghum root has been the most studied system, and here deposition begins in the inner tangential wall of the endodermis and the deposit then grows into the space between the wall and the protoplast (Sangster and Parry, 1976). So part of the deposit is on a carbohydrate matrix and part is not (similar to **Figure 1E**). In general there have been few reports of silica deposition in tissues of the root other than the endodermis.

TABLE 2 | Some publications that have used transmission electron microscopy to investigate phytoliths.


TABLE 3 | Some publications that have used scanning electron microscopy and x-ray microanalysis to investigate phytoliths.


It seems that almost all of the silica deposition in grass culms (stems) is in the outer tangential wall of the epidermis (Gartner and Paris-Pireyre, 1984; Hodson, 1986; Hodson and Sangster, 1990). **Figure 2** shows a light micrograph of epidermal silicification from the wheat culm in a dry ashed preparation. The cells form a complete sheet or silica skeleton, and the long and short cells all have thin silicified walls similar to the situation in **Figure 1A**.

Silicification of grass and cereal leaves is quite varied, with both lumen deposition in silica cells and elsewhere, and silica deposition in the cell walls. There has been much attention given to the silica cells (e.g., Kaufman et al., 1985; Hodson and Sangster, 1990; Laue et al., 2007). Kumar et al. (2017a) investigated the development of silica cells in sorghum leaves, and found that the deposits developed in the apoplastic space between the cell wall and the protoplast (see **Figure 1C**).

The inflorescence bracts of grasses and cereals have received some consideration. In the 1980s three groups all worked on the macrohairs from the lemma of P. canariensis, and Hodson et al. (1984) looked at the development of the highly thickened and silicified hairs. By maturity the hairs had only a very small lumen, and the whole wall was silicified. The long cells of the outer epidermis also considerably thickened during development, and silica was then deposited in the thickened cell wall (see **Figure 1B**). So in this case, what appeared to be a lumen phytolith was actually a cell wall phytolith. In the Phalaris lemma almost all phytoliths isolated from the organ were cell wall phytoliths. It is important to note that using light microscopy and SEM on this system did not indicate that these long cells were cell wall phytoliths, and that the silica was deposited on a carbohydrate matrix (Sangster et al., 1983). It is not always easy to be sure whether a phytolith has such a matrix. Long cells are very important repositories for silica in the epidermis. How many other apparently lumen types are in reality cell wall phytoliths? The Phalaris glumes, the next layer of bracts outside the lemma, are a completely different structure, with silica cells, and several different types of wall phytolith (Hodson et al., 1985).

There has been less work on silica deposition in the grass caryopsis (seed), and where present the amount is low (Hodson and Parry, 1982; Parry et al., 1984). I have included a few examples of work on silica deposition in non-grass species in **Tables 2**, **3**: nettle, Equisetum, Cannabis, white spruce, and bracken. Recently, phytolith production in the bryophytes has also been investigated (Thummel et al., 2018), and most silica deposition seems to be in the cell walls of these plants. It seems that there is little epidermal lumen deposition outside the grasses.

This survey has, of necessity, been brief and I have left out many papers and just selected a few examples. We can conclude that the following phytoliths are definitely cell wall types: macrohairs, prickle hairs, papillae (wall). The cell walls of many epidermal cell types are often silicified, and not only in the grasses. It seems that there are some organs where cell wall silicification is the only type (e.g., grass roots and culms). Cell lumen deposition, particularly in the epidermis, is apparently more common in grasses and cereals than in the rest of the plant kingdom. Whilst it is not possible from this survey to quantify the relative importance of cell wall and lumen phytoliths it is clear that the former make up a very significant proportion of the total.

#### MODELING CARBON IN PHYTOLITHS

So far, we have seen that there is carbon in phytoliths, and that the amounts reported vary depending both on the sample analyzed and on the technique used to process the phytoliths. We have tried to determine precisely which phytoliths develop as silica is laid down onto a carbohydrate cell wall, and which are not. In this section we will use data that is available in the literature in an attempt to partition PhytOC between cell wall and lumen phytoliths. We will begin with cell wall phytoliths.

Perry et al. (1987) found that the macrohairs from the lemma of the grass, P. canariensis, consisted of 40% silica, 55% carbohydrate, and less than 5% protein. This is the only analysis of native cell wall phytoliths that I am aware of. It is an unusual situation where there are considerable amounts of large silicified hairs that are easy to harvest, and by maturity they consist only of cell wall phytoliths (Hodson et al., 1984). For our purposes, we need to convert the percentages for carbohydrate and protein to percentage carbon. So for 55% carbohydrate we multiply by 12/30 to obtain a value of 22% carbon. Let us then assume that the whole of the remaining 5% organic material is protein. Most

FIGURE 2 | Epidermal silicification from the wheat culm. A light micrograph of epidermal silicification from the wheat (Triticum aestivum L. cv. Brock) culm in a dry ashed preparation (Buchanan and Hodson, unpublished).

proteins consist of about 53% carbon. So the proteins in the Phalaris macrohair account for about 2.65% carbon. The overall PhytOC in these hairs is therefore 24.65%. This value may or may not be typical for cell wall phytoliths, but we would expect that these phytoliths would have significantly higher carbon than lumen phytoliths. Thus, we have PhytOC for native cell wall phytoliths before any treatment to remove carbon (e.g., wet or dry ashing or microwave digestion) or degradation in the soil. This is uncommon as all other literature analyses are for phytoliths that have been treated in some way to remove external carbon. It is also important to note that all of the values quoted in **Table 1** above represent those obtained from mixtures of cell wall and lumen phytoliths, and are bound to be lower than those for pure cell wall preparations.

We do not have similar data for native lumen phytoliths, but the percentage of carbon will undoubtedly be much lower than in cell wall phytoliths. The silica cell phytoliths in sorghum developed in the space between the protoplast and the cell wall (Kumar et al., 2017a). In cases like this (**Figure 1C**), we would expect that phytoliths would not only be low in carbohydrates, but also largely devoid of membranes, DNA and other organic compounds and have very low percentage carbon. Alexandre et al. (2015) found carbon and nitrogen spread evenly across short cell phytoliths from wheat, suggesting that there were no membrane remains, but the possibility that amino acids were present was raised (see below).

Therefore we potentially have a situation where there are two distinct classes of phytoliths that are very different in their carbon concentrations. What other evidence is there for this idea? Jones and Beavers (1963) separated phytoliths on the basis of their specific gravity, and found that those with specify gravity less than 2.10 had a carbon content of 1.6%, well above the overall sample (0.86%). In another approach, Yin et al. (2014) heated rice straw phytoliths and found that there were two pools of carbon within them. They attributed the carbon released at lower temperatures

to that in the cell wall phytoliths, and that at higher temperatures was suggested to come from the lumen phytoliths. Yin et al. (2014) estimated the ratio of cell wall to lumen carbon as 12 or 13 to 1. These two different approaches both confirm that there are two types of phytolith with different carbon concentrations.

Parr and Sullivan (2014) produced the single paper that comes closest to the overall hypothesis that I am setting out here. They compared two methods of preparing phytoliths from sugarcane and sorghum, microwave digestion, and a rapid digestion using H2SO4/H2O2. The microwave digestion method was less damaging for the phytoliths and retained much more carbon. Parr and Sullivan suggested that there were two main types of phytolith, cavate, and solid. Cavate phytoliths were essentially the cell wall phytoliths of the type often seen in the epidermal long cells where the thin walls form a hollow structure. These would be similar to the situation depicted in **Figures 1A**, **2**. Solid phytoliths were silica cells and other lumen types (**Figures 1C,D**). The amount of carbon found using microwave digestion was considered by Parr and Sullivan to give an accurate total value for carbon in their preparations, what they termed PhytOCTot (**Table 4**). They thought that the rapid digestion procedure removed all of the carbon from the cavate (cell wall) phytoliths, but left it in the solid (lumen) types. This was termed matrix carbon, and hence PhytOCMat. It is then a simple matter to deduct these matrix values from the total to give the cavate (cell wall) percentages (PhytOCCav). Neither value comes close to the 24.65% calculated for the Phalaris macrohair (above). This could either suggest that percentage carbon in cell walls varies considerably depending on the source, or that even the microwave digestion technique employed by Parr and Sullivan is over-extracting some carbon.

As an aside, I am not that keen on "cavate" and "solid" as descriptive terms, as many of the cell wall types mentioned in **Tables 2**, **3** above are solid. But as we have seen, even my preferred terminology of "lumen" and "cell wall" has some problems. It may be that we will need to classify phytoliths according to whether or not they are formed on a carbohydrate matrix.

At this point, I would like to introduce three concepts, PhytOCmax, PhytOCmin, and PhytOCprep. These are in some ways related to the terms suggested by Parr and Sullivan (2014). I hope that they will prove helpful in throwing some light on the problems we have encountered with interpreting carbon sequestration in phytoliths. Firstly, PhytOCmax is the maximum amount of carbon occluded within a phytolith as it drops from a plant into the soil. This will be the amount in native phytoliths before they begin to degrade in the soil or before any attempt at preparation in a laboratory. In practice this is usually very difficult to determine. When phytoliths drop into the soil they are generally surrounded by non-silicified organic material. The

TABLE 4 | Partitioning of carbon in sugarcane and sorghum samples (Parr and Sullivan, 2014).


aim of the preparatory techniques (dry ashing, wet ashing, and microwave digestion) is to remove all of the extraneous organic material without touching that which is bound within the phytolith structure. But it is only in very rare cases such as the Phalaris macrohairs described above (Perry et al., 1987) that we can be sure that we have accurately determined PhytOCmax. So PhytOCmax is an important, but largely theoretical, concept. Secondly, PhytOCmin is the amount of carbon remaining in a phytolith after all of the easily available carbon has been removed. Of course, this value is very likely going to differ for different types of phytoliths. We might expect that most of the carbon in cell wall phytoliths will be easier to remove, and maybe that in lumen phytoliths will be less labile. In the soil it may take a long time to reach PhytOCmin (see below for a discussion), but laboratory preparatory techniques may well approach this value very rapidly. A key question is how many of the measurements in **Table 1** represent PhytOCmin and how many are closer to PhytOCmax? This leads us on to the final concept, PhytOCprep. This is the amount of carbon left within a phytolith after it has been subjected to a suitable preparatory technique in the laboratory (the equivalent of PhytOCTot in Parr and Sullivan's terminology when using microwave digestion). Of course, PhytOCprep must lie between PhytOCmax and PhytOCmin, but exactly where is difficult to be certain, and will depend on the technique used. An important question is whether the PhytOCprep value of 3% used by Song et al. (2016) is close to PhytOCmax? If that is the case then is the PhytOCprep value of 0.1–0.5% used by Santos and Alexandre (2017) close to PhytOCmin? So are both Song et al. (2016) and Santos and Alexandre (2017) "correct," but the PhytOC values they give just represent what is present at different times in phytolith degradation and dissolution? We will have more to say on this point below.

Next we need to investigate mixtures of different types of phytoliths. As, we saw above there are some cases where we are fairly sure that nearly all of the phytoliths in an organ are cell wall types (e.g., the wheat root, and the Phalaris lemma), but in many cases, particularly in the grasses, there will be a mixture of both cell wall and lumen types. The evidence presented here and in my previous publications (Hodson, 2016, 2018) strongly suggests that the two types have very different chemistries, and that cell wall types have much higher carbon concentrations. So, when we have two phytolith types with different PhytOC, how do we calculate the overall PhytOC for the material? To do this, we need to have some estimates of PhytOC for both lumen and cell wall types, and some idea of the relative amounts of the two types of phytolith in a particular sample. Once, we have those estimates we can use the following equation:

$$\mathbf{(a \times y/100) + (b \times z/100) = Total Percentage Property} \quad \text{(1)}$$

Where:


We can then investigate a variety of potential scenarios:


**Figure 3** shows the effects of varying the ratio of cell wall to lumen phytoliths, moving from a situation where none of the phytoliths are cell wall types to where there are 100% in a sample. As would be expected, for all three scenarios, a higher percentage of cell wall types leads to a higher total PhytOC. In general Scenario 1 gives higher total PhytOC values under almost all conditions. It is interesting to calculate the percentage of cell wall phytoliths that would be required to reach the 3% total PhytOC figure that is given by Song et al. (2016) and that is used in many of the other papers from their group. For Scenario 1 only 11% of cell wall types would be needed, for Scenario 2 (sugarcane) the figure is 29%, and for Scenario 3 (sorghum) it is 88%. Clearly there is very big variation in these figures, but at least in some scenarios a relatively small percentage of cell wall phytoliths would be needed to bring us close to the 3% figure that Song et al. (2016) preferred.

As far as I am aware nobody has attempted to quantify the relative volumes of cell wall and lumen phytoliths in organs like the grass leaf, and this is an important topic for future research.

FIGURE 3 | The effects of mixing different amounts of cell wall and lumen phytoliths on total PhytOC. Potential scenarios: (1) PhytOC for the cell wall phytoliths is 24.65% and PhytOC for lumen phytoliths is 0.33%. (2) PhytOC for the cell wall phytoliths is 10.12% and PhytOC for lumen phytoliths is 0.15%. (3) PhytOC for the cell wall phytoliths is 3.37% and PhytOC for lumen phytoliths is 0.51%.

Another key subject arising from this work concerns differences in carbon allocation between species and cultivars. As we saw above, Parr, Sullivan and their team found major differences between PhytOC in phytoliths from different bamboo species, and sugarcane, wheat, and rice cultivars. It has been suggested that it might be possible to breed plants for high PhytOC. But at the cellular level what are we breeding for? Is it simply a change in the ratio of cell wall to lumen phytoliths? Or is it more complex than that?

### SOME THOUGHTS ON CARBON AND NITROGEN IN PHYTOLITHS

There have been few measurements of nitrogen in phytoliths to date. The presence of nitrogen would indicate that proteins, amino acids, and possibly nucleic acids had been incorporated into the phytoliths. Most proteins contain about 53% carbon and about 16.3% nitrogen so their C/N ratio will be about 3.25. Values higher than that would suggest that carbohydrates and/or lipids were a significant part of the carbon present in the phytoliths. **Table 5** shows the data that I have been able to locate concerning nitrogen concentrations in phytoliths.

Jones and Beavers (1963) were the first to measure nitrogen in phytoliths at 0.01%, which would give a C/N ratio of 86. The accuracy of this C/N ratio is probably somewhat questionable given the very low nitrogen concentration, and the age of the work. I was able to calculate the percentage nitrogen in the Phalaris macrohairs studied by Perry et al. (1987). They estimated that percentage protein in the hairs was less than 5% and so the maximum percentage nitrogen will be 0.82% at PhytOCmax in native hairs before treatment. Using the previously calculated value for percentage carbon of 24.65% we can determine that the C/N ratio of Phalaris macrohairs is a minimum of 30. Hodson et al. (2008) measured nitrogen in phytoliths extracted from various wheat organs using dry ashing followed by boiling in hydrogen peroxide. They found low, but detectable, amounts of 0.01–0.06% nitrogen. The calculated C/N ratios varied from 7 to 43, depending on the organ, with the bulk sample containing all organs giving a value of 41. Fragmented glycoproteins were found in wheat leaf phytoliths by Elbaum et al. (2009), confirming the presence of nitrogenous compounds, although they did not quantify the amounts. The leaf short cell phytoliths of Triticum durum were analyzed using nanoSIMS by Alexandre et al. (2015). They were not able to quantify carbon and nitrogen concentrations, but their C/N ratio was 3.7. In the following year, Alexandre et al. (2016) wet ashed the leaves of Festuca arundinacea and found that the phytoliths had a C/N ratio of 5.1.

How do we interpret the above data? Firstly, it is striking that the values for C/N ratio in **Table 5** fall into two groups with the rachis of wheat and the analyses conducted by Alexandre et al. (2015, 2016) giving markedly lower values than the rest. The value for native Phalaris macrohairs calculated from Perry et al. (1987) gives an approximate baseline for cell wall phytoliths with a minimum C/N ratio of 30. It seems that the C/N signature for cell wall phytoliths dominates the wheat samples analyzed by Hodson et al. (2008) even after they have undergone dry


TABLE 5 | Phytolith carbon and nitrogen analyses.

feart-07-00167 June 29, 2019 Time: 17:6 # 9

ashing and boiling in hydrogen peroxide. Alexandre et al. (2015) were quite correct to point out that their nanoSIMS analysis for leaf short cells strongly suggests the presence of amino acids with a C/N ratio of 3.7 in these lumen phytoliths. This forms a baseline for lumen phytoliths. When Alexandre et al. (2016) wet ashed F. arundinacea leaves, the phytoliths within them had a slightly higher C/N ratio of 5.1, again suggesting dominance of amino acids and proteins. It seems very likely that the extraction procedure used by Alexandre et al. (2016) was stronger than that used by Hodson et al. (2008), and that they removed most of the carbon from within the cell wall phytoliths, leaving that in the lumen phytoliths largely intact. This thinking is along similar lines to that of Parr and Sullivan (2014), where cell wall (cavate) phytoliths were considered to be more susceptible to extraction than lumen (solid) phytoliths. The wheat rachis sample had very low carbon and nitrogen in its phytoliths, and I suspect that this relatively lightly silicified organ was also over-extracted. More work is needed employing C/N ratios for phytolith analyses to confirm these ideas, but this ratio certainly seems to have potential for assessing the relative contributions of carbohydrates and amino acids within a processed sample.

#### THE LOSS OF CARBON FROM PHYTOLITHS IN THE SOIL AND SEDIMENTS

The evidence I have presented so far in this paper very strongly suggests that lumen phytoliths generally have low PhytOC. Even Parr and Sullivan (2014), the originators of the carbon sequestration in phytoliths idea, are proposing values as low as 0.15–0.51% (**Table 4**). As, we saw above, Reyerson et al. (2016) calculated global carbon sequestration using a PhytOC of 0.3%, assuming that this applied to all phytoliths, and concluded that sequestration would be insignificant. So if lumen phytoliths are not that important for carbon sequestration the whole hypothesis hangs on the cell wall phytoliths. However, the general assumption is that cell wall phytoliths are less likely to remain in soil as they are more easily broken down (Song et al., 2017). But is this really the case?

Strömberg et al. (2018) have produced a very detailed assessment of what happens to phytoliths when they enter the soil, and we will not go back over all of this material, but mostly concentrate on any differences between cell wall and lumen phytoliths. It is clear that, in many soil environments, a considerable amount of siliceous plant material, including phytoliths, breaks down fairly quickly. Indeed, phytoliths are often an important source of dissolved Si in soils as they are much more soluble than quartz, aluminosilicates, and other soil minerals. Working on a temperate coniferous forest, Gérard et al. (2008) showed that 60% of the biogeochemical cycle was controlled by biological processes, namely Si uptake by plants and dissolution of phytoliths. There is a large labile pool of phytogenic silica in soils (Strömberg et al., 2018), with values for this pool ranging from 69% in short grass prairie to 92% in tropical forest.

Puppe et al. (2017) conducted a detailed analysis of the contribution of biogenic silica to the soil soluble silicon pool at Chicken Creek in Brandenburg, Germany. They considered diatoms and sponge spicules in addition to phytoliths, but it was the latter that were by far the most important in contributing to soluble silicon concentrations in the soil. However, they discovered that small, delicate, phytolith fragments which were not usually quantified using standard extraction processes made up 84% of the phytogenic material and those larger than 5 µm represented only 16%. The authors stressed the importance of this large pool of small delicate material in contributing to soluble silicon in the soil. The micrographs of the fragile phytoliths they showed (their Figure 7) were not that dissimilar to my **Figure 2** with thin cell wall silicification. Clearly these structures would be highly susceptible to dissolution. Presumably the larger phytoliths that represented 16% of the phytogenic material would remain in the soil for much longer periods.

At the global scale, the phytolith stability factor was one of the disagreements between Song et al. (2016) and Santos and Alexandre (2017). The former suggested a stability factor of 0.8 to 1.0 as phytoliths in most systems are stable for 500 to 3000 years. However, Santos and Alexandre (2017) suggested a stability factor of 20%, which combined with their much lower value of PhytOC (0.3%), led them to suggest that carbon sequestration in phytoliths on a global scale was insignificant. As we saw above, it

seems very likely that the value of 0.3% PhytOC used by Santos and Alexandre is an underestimate due to over-extraction, and that particularly applies to cell wall phytoliths. It is difficult to speculate on the influence of cell wall phytoliths on the stability factor as we lack even basic data.

Next we will investigate the dissolution of lumen and cell wall phytoliths. I recently reviewed this topic (Hodson, 2018), and a number of factors seem to be important. Of the soil chemical factors, high pH was the most significant, causing increased phytolith dissolution. It is possible that aluminum in phytoliths may decrease their dissolution, but their carbon content hardly seems to have been considered. Cabanes and Shahack-Gross (2015) carried out the most detailed work so far on this topic but, with the exception of the double peaked glume phytoliths from rice husks, most of their work concentrated on lumen types. The key factor in increasing phytolith solubility was geometric surface to bulk ratio. There was no indication that cell wall phytoliths were either more or less soluble.

We should now consider what is known about cell walls that have undergone silicification. The small number of measurements so far available for carbon in cell wall phytoliths shows considerable variability (see above: Perry et al., 1987; Parr and Sullivan, 2014). If the percentage carbon is high then percentage silicon must be low and vice versa. We would not necessarily expect all cells walls to be silicified to the same extent, but this will mean that they will vary in their chemical properties, and potentially in how susceptible they are to breakdown processes in the soil. I previously discussed evidence that suggests that after the organic matter is removed from cell wall phytoliths the remaining silica has a porous structure (Hodson, 2016). Since that publication, Sola-Rabada et al. (2018) have published the first estimate of the size of the pores that I am aware of. In phytoliths isolated from Equisetum myriochaetum using wet ashing the silica had a surface area of ∼400 m<sup>2</sup> · g −1 and a pore size of ∼5 nm. Presumably, in the native state these pores will have been filled by carbohydrates and other organic compounds. Almost certainly pore size will vary, and we might expect more lightly silicified material to have larger pores. But we should remember that these cell wall phytoliths are only porous after most of the organic matter has been removed with drastic treatment. Will cell wall phytoliths necessarily be more susceptible to breakdown in the soil than lumen types just because they have higher organic matter within them? Does being encrusted by silica slow down the microbial degradation of organic matter in phytoliths? Conversely, does being so intimately associated with organic matter impede the dissolution of silica from phytoliths? We do not know the answers to these questions yet.

There is not very much known about how silica and organic matter are associated in the soil. However, the work of Watteau and Villemin (2001) on the breakdown of leaves and roots soils of a beech forest is important in this respect. Using TEM and electron energy loss spectroscopy (EELS) they found silica was deposited in beech leaves in the walls of the epidermal and parenchymatous cells, in the middle lamellae, against the walls in the cells, or in cell intersections. The authors stressed the close relationships between biogenic silica and cellulose, hemicellulose, and pectic substances in these samples. Deposition was in similar locations in the cell walls of beech roots, but also in the root cortical cells closely associated with polyphenolic substances. In the soil the leaf and root tissues were broken down primarily by fungi, but bacteria were also present. The fungi attacked the carbohydrates in the cell walls, leaving the silica largely intact, particularly that in the cell intersections. More recently, Turpault et al. (2018) also investigated silicon cycling in beech forests. Much of the silicon was associated with cell walls in the beech tissues. Turpault et al found that fine beech roots were particularly important in cycling as they had a high Si content and were rapidly broken down and recycled. Very little Si was lost from the system through deposition in perennial tissues or leaching from the soil, and it was an almost closed system. It is clear from both Watteau and Villemin (2001) and Turpault et al. (2018) that cell wall Si deposition is the most important in beech, and in the soils beneath the forests. The papers also give us some insights into breakdown of cell wall phytoliths in soils and how rapid this can be. The beech cell walls investigated appear to be fairly thin and relatively lightly silicified (similar to **Figure 1A**), and it would be unwise to extrapolate from this situation to others where heavier silicification has occurred.

As I was writing this paper, the Intergovernmental Panel on Climate Change (IPCC) brought out their 2018 report on the feasibility of keeping the global temperature rise under 1.5◦C above the pre-industrial temperature. The report has a section (4.3.7.3) which considers increasing carbon sequestration in soils as one of the means of tackling the problem (de Coninck et al., 2018). At a local scale the benefits of increasing carbon sequestration in soils are clear, but there is much uncertainty about how much carbon can be sequestered at a global scale and what the costs might be. The section does consider work on the use of biochar to increase sequestration but, rather like work on phytoliths, there is considerable debate about its potential. Not surprisingly, the idea that phytoliths might be involved in carbon sequestration in soils has not yet impinged on the IPCC. We will need much more work and much greater certainty before that might happen.

Before we leave this topic, I have one more question to raise. How long do we need to sequester carbon? There seems to be a general assumption in the phytolith literature that we need to sequester carbon for hundreds or thousands of years, and that sequestration for shorter periods is not worthwhile. Parr and Sullivan (2005) found phytoliths from 8710 BP at Byron Bay in Australia still contained PhytOC, and so sequestration is possible for very long periods of time. However, I would argue that the problems that we are having with climate change are so severe that we need to maximize short term sequestration, and that even locking away carbon in phytoliths for 50 or 100 years might make a valuable contribution. In that time we might hope that the world will make the switch to renewable technologies, and that we might have developed other methods for sequestering carbon. The IPCC have made it very clear how urgent the problem of climate change is, and the short time scales involved to reduce what could be very serious impacts. We need to keep this in mind as we investigate carbon sequestration in phytoliths.

## DO CELL WALL PHYTOLITHS REMAIN IN ARCHEOLOGICAL AND PALAEOECOLOGICAL SAMPLES?

In the previous section, we investigated what is known about the chemistry of phytolith breakdown and dissolution in the soil and sediments. It is highly unclear whether cell wall phytoliths are degraded faster than lumen phytoliths as there is little data available. Since this is the case, we will now turn to the archeological and palaeoecological literature to investigate whether cell wall phytoliths persist in soils and sediments.

As we have seen earlier it is not always easy to determine whether a phytolith has a carbohydrate matrix, so we will confine this survey mostly to macrohairs, prickle hairs and papillae (**Table 6**). The double-peaked rice husk phytoliths are also cell wall phytoliths occurring on the outer surface of the rice husk, and they are heavily silicified (Park et al., 2003). An additional category we will add are multi-celled phytoliths, also known as silica skeletons (Rosen and Weiner, 1994). These are groups of phytoliths frequently, but not only (see **Figure 2**), originating from the husks of cereals. They often contain papillae within their structures, and will inevitably enclose cell walls between the different cells. Included within the silica skeletons are the cut phytoliths which appear in archeological contexts, and are diagnostic for cutting and threshing activities (Cummings, 2007).

**Table 6** represents the results of a partial survey of the literature, and there are many other papers that could have been cited. However, it is clear that cell wall phytoliths can be found in samples that are hundreds or thousands of years old. In two cases (Prasad et al., 2005; Wu et al., 2018) they were found associated with dinosaur remains from the Cretaceous. Here, we would expect that the phytoliths discovered will be fossilized and have lost their original organic matter, but it does indicate that they persisted long enough to be preserved in this way. Cell wall types have been found in many different contexts and environments, from extremely arid to temperate, and in many different countries. It would not be wise to attempt any sort of quantification, particularly as we are uncertain how many long cell phytoliths have a carbohydrate matrix (see **Figure 1B** and discussion above). I have included both archeological and palaeoecological examples in **Table 6**. It could be argued that archeological contexts do not always replicate conditions from the natural environment. However, it is now recognized that agriculture is having a major impact on the global silicon cycle (Struyf et al., 2010), and so, to some extent, the work on past agricultural activity is an analog for what is happening today. Moreover, much of the work on increasing carbon sequestration in phytoliths in the future concerns agricultural crops (Parr and Sullivan, 2011).

In conclusion, the small survey shown in **Table 6** has strongly indicated that cell wall phytoliths can persist in soils and sediments for considerable periods of time. It is conceivable that these cell wall phytoliths may have lost much of their organic matter over time, but that their basic structure remains intact. It is, however, more likely that they still contain substantial amounts of carbon hidden deep within the phytolith structure. So we are faced with the possibility that some carbon may be sequestered in cell wall phytoliths for hundreds or thousands of years. But as we argued above, the more important issue is how much carbon is sequestered for short periods of time, maybe 100 years.

## A HYPOTHESIS

Having gathered together data and observations from many different sources I am now able to put forward a hypothesis which attempts to explain the overall picture. First, let us reflect on what happens when phytoliths are prepared for analysis by wet ashing, dry ashing or microwave digestion:


Now let us consider the situation in the soil:


From the work of Watteau and Villemin (2001) and Turpault et al. (2018) it is evident that cell wall phytoliths are the most important in the soils of beech forests. I am not aware of similar work for the soils of coniferous forests. However, my

TABLE 6 | Cell wall phytoliths in archeological and palaeoecological samples.


previous work on conifer needles has strongly indicated that cell wall deposition is important in this group. For example, Hodson and Sangster (1998) found that silica deposition was almost entirely confined to the hypodermal and endodermal walls of white spruce needles. Presumably this would be reflected in the phytoliths to be found in the soils of conifer forests. In grasslands it is probable that lumen phytoliths (particularly short cells) will dominate in most phytolith assemblages isolated from soils. In these cases we may need to balance a very large number of lumen phytoliths that contain small amounts of carbon against a smaller number of cell wall phytoliths that contain much more carbon. It may be that a modified form of Equation 1 could be used in these circumstances to determine the overall PhytOC percentage. Very recently, Zhang et al. (2019) showed the importance of bamboo litter layers in carbon sequestration, and demonstrated the considerable potential that exists to increase carbon sequestration in the future. But what types of phytoliths might we expect to dominate the litter layers? Firstly, it is clear that bamboo leaves contain much higher silicon concentrations than the other organs (Collin et al., 2012), and these will undoubtedly be the major contributor to phytoliths in the litter. Lux et al. (2003) investigated silicification of bamboo (Phyllostachys heterocycla Mitf.) leaves and found the highest Si concentrations were in the epidermal cell walls and short cells. This suggests that bamboo litter should contain both cell wall and lumen phytoliths. Which will be dominant in the litter is uncertain. We have seen that bamboo species differ in the amount of PhytOC within their phytoliths (Parr et al., 2010), and we might also expect that the litter will vary in the relative amounts of cell wall and lumen phytoliths, depending on the species involved.

I am not clear whether lumen phytoliths or cell wall phytoliths will be the more significant in sequestering carbon at a global scale. At the moment we have not got enough data even to make an informed guess on the relative importance of the two phytolith types in carbon sequestration at this scale.

## PRIORITIES FOR FUTURE RESEARCH

A number of important research topics have arisen from the present study:



#### MY CONCLUSION AND PERSPECTIVE

I said at the beginning of this paper that I would not take sides in what has become an acrimonious debate over carbon sequestration in phytoliths. However, having carried out a detailed analysis and weighed up all the evidence I conclude that the hypothesis that carbon sequestration in phytoliths is important on a global scale is probably correct, or at least cannot yet be discarded. I think it is likely that all workers in this area (and I include myself) have over-extracted phytoliths, and that we have not given an accurate representation of PhytOCmax. It is probable that Alexandre, Santos and their co-workers have over-extracted to the point where their preparations are approaching PhytOCmin. I am very clear that we have all not taken enough account of heterogeneity in phytolith chemistry. It seems very likely that cell wall phytoliths are important in carbon sequestration, and it may even prove to be the case that they are more significant than the lumen types. In this paper I have considered lumen and cell wall phytoliths, but it is quite possible that this is an oversimplification and that there are more than two types or some gradation between the two (e.g., the situation in **Figure 1E**). There is no doubt that what I have presented here makes the whole topic of carbon sequestration in phytoliths even more complex than it was, but if we are to move this field forward then these complexities need to be accounted for.

I have worked on phytoliths for nearly 40 years. Much of my work has been what some people call "blue skies" research. That is it had no obvious immediate practical application. So, I have been quite surprised that some of my publications on microanalysis and phytolith development from the 1980s now have a new relevance in 2019 when we consider carbon sequestration and PhytOC. I suspect that quite a few of the authors cited in this paper will be equally surprised. I do worry that financial pressures mean that we are losing the possibility to research topics just because they are interesting.

Many scientists from very diverse disciplines have contributed to the picture I have painted in this paper. However, if we look specifically at the question of carbon sequestration in phytoliths a few people stand out. Foremost among these must be Jeffrey Parr and Leigh Sullivan who first had the idea that phytoliths might sequester substantial amounts of carbon. All of the data was already there for everyone to see, but they had the idea, and the sudden spark of brilliance that really created a whole new field of phytolith research. They then carried out a considerable amount of work to test their hypothesis, and particularly to look at variation in carbon sequestration in phytoliths from related species and cultivars. If Parr and Sullivan were the originators of the idea, then Zhaoliang Song and his team in China were those who tested it out in a whole string of investigations. We should also be grateful to Ann Alexandre, Guaciara Santos and their co-workers who "shook the tree" and made us all wonder if sequestration of carbon in phytoliths was an important phenomenon. I disagree with their overall conclusion on the importance of carbon sequestration in phytoliths, but they are very good scientists and have done some excellent work in this area. It was through their work that I hit on the concept of PhytOCmin, which I have described above. They also provided some useful data on nitrogen in phytoliths which was crucial in my thinking about C/N ratios. Finally, I must mention Carole Perry, whose work on the chemistry of phytoliths has been seminal. I used some of her early research to develop the idea of PhytOCmax, and in many ways her analysis of the Phalaris macrohair (Perry et al., 1987) was the key to unlocking this puzzle. I am sure that Carole would never have guessed back in the 1980s that her work would be used in this way. For me this is a fascinating story that has developed over more than 30 years. It is notable that much of the research I have based my ideas on was originally "blue skies," but now it makes a significant contribution to a very important topic.

As I write in 2019, the evidence for the effects of humaninduced climate change is all too obvious from around the world. The IPCC report (2018) that I mentioned above laid out what we need to do to avoid a very perilous future. For the last 15 years I have spent much time speaking and writing about climate change for general, non-scientific audiences (Hodson and Hodson, 2011, 2013, 2015). Every new talk I prepare or article I write about

climate change makes me aware of just how serious and urgent this issue is. Now I am approaching my 40th anniversary of working on phytoliths, and we can see that they might have a potential role in carbon sequestration in soils. I am not convinced that phytoliths will be a "silver bullet" for climate change, but the work described above suggests that they may have a role to play. We now really need a concerted and determined effort from the whole phytolith community to test out some of the ideas laid out above. There are key topics for scientists from many different disciplines to work on, from those investigating phytolith chemistry and formation at a molecular level right up to those studying biogeochemical cycles. It is extremely important that we maintain very good communications between all these scientists, and not end up in disciplinary boxes. There is a lot to recommend the phytolith superdiscipline idea of Katz (2018), where boundaries between disciplines are dissolved.

There has been considerable tension within the phytolith community over carbon sequestration in the last few years, and academic disagreements have turned to friction and friction to heat. I sincerely hope that all of the scientists working on carbon sequestration in phytoliths will one day be reconciled (and reconciliation is even more needed where dating of phytoliths is concerned). This issue is too important for personal rivalries to get in the way. I would appeal for all involved to work together toward a common goal. That goal is working out how important PhytOC is, and if it is important then finding ways of using

#### REFERENCES


the knowledge gained as quickly as possible. Put aside previous arguments and get on with the job. I will gladly work with anyone who wants advice or help, and I will not be upset if some of the thinking above is incorrect. I have put forward a hypothesis which seems to explain the available data, but it is a hypothesis and it needs testing. If, in 10 years' time, someone writes, "Hodson got it totally wrong, but he gave me some ideas, and now we have it right," then I will be very happy.

#### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

#### ACKNOWLEDGMENTS

I would like to dedicate this paper to the memory of my friend and colleague, Allan Sangster who died on September 6, 2018, aged 91. I published more papers with Allan than with anyone else, and it is fair to say that without him the present paper would never have happened. I am particularly grateful to my wife, Margot, for allowing me to bounce ideas off her for this paper. Even more grateful as sometimes our conversations were in the early hours of the morning! With all my love Martin.


glycoproteins but no DNA. Quat. Int. 193, 11–19. doi: 10.1016/j.quaint.2007. 07.006


for biogeochemical carbon sequestration. Earth Sci. Rev. 115, 319–331. doi: 10.1016/j.earscirev.2012.09.006


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Hodson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

feart-07-00167 June 29, 2019 Time: 17:6 # 16

# Phytolith Radiocarbon Dating: A Review of Previous Studies in China and the Current State of the Debate

*Xinxin Zuo1,2\* and Houyuan Lu3,4,5*

*1 State Key Laboratory for Subtropical Mountain Ecology of the Ministry of Science and Technology and Fujian Province, Fujian Normal University, Fuzhou, China, 2 School of Geographical Sciences, Fujian Normal University, Fuzhou, China, 3 Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China, 4 Center for Excellence in Tibetan Plateau Earth Science, Chinese Academy of Sciences, Beijing, China, 5 College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing, China*

Phytolith radiocarbon dating can be traced back to the 1960s. However, its reliability has recently been called into question. Piperno summarized recent dating evidence, but most phytolith dating results from China were not included in the review because they are written in Chinese. Herein, we summarize and evaluate previous phytolith dating results from China. We also review recent debates on the nature and origin of phytolithoccluded carbon (abbreviated as PhytOC), as well as the older age of phytoliths retrieved from modern plants. We conclude that although PhytOC includes a small amount of old carbon absorbed from the soil, this carbon fraction has not always biased phytolith ages, indicating that in certain situations, phytoliths can be tried as an alternative dating tool in archaeological and paleoecological research when other datable materials are not available.

#### *Edited by:*

*Martin John Hodson, Oxford Brookes University, United Kingdom*

#### *Reviewed by:*

*Dolores R. Piperno, Smithsonian Institution, United States Rand Evett, University of California, Berkeley, United States*

#### *\*Correspondence:*

*Xinxin Zuo zuoxinxin@fjnu.edu.cn; zuoxinxin@live.cn*

#### *Specialty section:*

*This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science*

*Received: 15 July 2019 Accepted: 18 September 2019 Published: 16 October 2019*

#### *Citation:*

*Zuo X and Lu H (2019) Phytolith Radiocarbon Dating: A Review of Previous Studies in China and the Current State of the Debate. Front. Plant Sci. 10:1302. doi: 10.3389/fpls.2019.01302*

Keywords: older carbon, PhytOC, radiocarbon dating, phytolith age, phytolith

## INTRODUCTION

Phytoliths are noncrystalline SiO2 ·nH2O that are deposited within the cells and cell walls in different parts of plants (Piperno, 2006). The morphology of a phytolith often resembles the shape of the cell in which it is formed and can be used in plant taxonomy. Phytoliths occlude a small amount of carbon during their deposition [phytolith-occluded carbon (PhytOC)] (Smith and Anderson, 2001; Parr and Sullivan, 2005). When a plant dies and decays, phytoliths and their occluded carbon can persist in the soil for a long time owing to the high resistance of phytoliths against decomposition. Phytolith analysis has been applied to environmental, anthropological, and geological research. Radiocarbon dating of phytoliths is a long-established technique that can be traced back to the 1960s (Wilding et al., 1967; Kelly et al., 1991; Piperno and Becker, 1996; Piperno and Jones, 2003). During the past decades, several researchers have attempted to date phytoliths, and some of them have achieved reasonable results. However, some of them have failed, because they found that phytolith carbon comes from multiple sources (either photosynthetic or soil carbon) (Reyerson et al., 2016). Moreover, the carbon in phytoliths that is taken up from the soil is variable and generally unknowable, which limits phytoliths carbon as a reliable dating material (Alexandre et al., 2015; Alexandre et al., 2016; Santos et al., 2018). Consequently, along with organic matter in pottery, phytoliths are considered as problematic samples for radiocarbon dating (Taylor and Bar-Yosef, 2014).

Recent debates in phytolith carbon dating research include the following topics: Is phytolith dating reliable? Is all phytolith carbon encapsulated *via* photosynthesis from atmospheric CO2 during plant growth, or is some absorbed from soil, which might distort phytolith dating? These questions are relatively new and were widely discussed recently (Hodson, 2018). Researchers have so far failed to reach an agreement on the reliability of phytolith carbon dating, largely because the scientific study of the nature, content, and status of PhytOC is still in its infancy.

In a review article, Piperno (2016a) summarized and evaluated almost all previous phytolith dating results of studies from different regions of the world. However, the results of several phytolith dating studies from China were not included, possibly because they are written in Chinese. Herein, we briefly review the history of phytolith carbon dating research. We then introduce and summarize the history of phytolith carbon dating research in China. Finally, we will discuss the main focus of current debate and the issues associated with phytolith carbon dating.

#### A BRIEF HISTORY OF PHYTOLITH CARBON DATING RESEARCH

Jones and Beavers (1964) were the earliest researchers to discover that phytoliths can occlude carbon during their formation in plants (Wilding et al., 1967). The earliest attempt to date carbon in phytoliths was published in 1967 by Wilding (1967), who extracted approximately 75 g of phytoliths from 45 kg of a prairie surface soil horizon, isolated the occluded carbon, and obtained a date of 13,300±450 a BP. Since Wilding's pioneering research on phytolith carbon dating, three stages of phytolith carbon dating research can been identified according to the total annual citations of Wilding' s 1967 article (**Figure 1**).

First is the early research period, from around 1970 to 1990. As shown in **Figure 1**, although Wilding's phytolith dating results received some attention sporadically, only a few studies used phytolith dating to construct chronological sequences, mainly because of the time-consuming phytolith extraction process and the large sample size required for conventional radiocarbon dating.

Second is the revived period of research, from 1990 to 2010. The development of accelerator mass spectrometry (AMS) technology has enabled the measurement of very small samples containing trace amounts of carbon. Utilizing this technique, a much smaller amount of phytoliths would yield sufficient carbon for dating, greatly reducing the amount of phytolith extraction required. Mulholland and Prior (1993) summarized the process of AMS-based radiocarbon dating of phytoliths by presenting details of extracting and dating phytoliths. The initial application of phytolith carbon dating during this period was performed by Kelly et al. (1991). They applied phytolith carbon dating into three soil profiles from the northern Great Plains. The results showed that there may be some serious problems with dating phytoliths because two of the three soils they examined showed the phytoliths were younger at deeper horizons in the soil profile, contrary to expectations. Piperno and Stothert (2003) used phytolith carbon to date *Cucurbita* domestication through phytolith carbon-14 study during the early Holocene in Southwest Ecuador (Piperno and Becker, 1996; Piperno and Jones, 2003).

Third is the period of controversy in phytolith carbon dating research after 2010. Recent studies on phytolith dating of modern plants have argued that old carbon absorbed by plants from soils

distorts the accuracy of phytolith carbon dating, with modern plants producing phytoliths radiocarbon dates up to several thousand years (Santos et al., 2010; Santos et al., 2012; Yin et al., 2014; Reyerson et al., 2016). Because the age of the phytoliths is overestimated compared with that of other dating materials, phytolith carbon is considered problematic for dating by several researchers (Taylor and Bar-Yosef, 2014; Santos et al., 2018). Other researchers argue that some reasonable phytolith dates have been measured from both modern plants and paleo-soils (Sullivan and Parr, 2013; Piperno, 2016a; Asscher et al., 2017; Zuo et al., 2017). Meanwhile, the soil phytolith ages extracted from different cultural layers of several archaeological sites have shown good consistency with their paired dating samples collected from the same depth (Asscher et al., 2017; Zuo et al., 2017) (**Table 1**).

## PHYTOLITH CARBON DATING RESEARCH IN CHINA

Phytolith research began in the late 1980s in China (Wang and Lu, 1989; Lu and Wang, 1990; Wang and Lu, 1992), which is over 150 years after the first report of phytoliths in living plants by Struve in 1835. The first report of phytolith carbon dating in a Chinese journal was published by Wang and Lu (1997), two pioneer phytolith researchers, in 1997 (Wang, 1998). They introduced the idea of radiocarbon dating of PhytOC to China, as summarized in the review of Mulholland and Prior (1993). Wang aimed to determine the chemical composition of phytoliths extracted from 16 species using an electron microprobe. Although the method used could not accurately measure the chemical composition of phytoliths, Wang was the first scholar in China who realized the importance of chemical aspects of phytoliths. However, both Wang and Lu did not actually date phytoliths.

It was only after 2010 that PhytOC and phytolith radiocarbon dating were studied again in China. To test the importance of carbon sequestration in phytoliths (Parr and Sullivan, 2005; Parr et al., 2009; Parr et al., 2010), we used the wet oxidation method to extract phytoliths from eight species of millet and showed a significant variation in PhytOC in different millet species (Zuo and Lü, 2011). Song et al. (2014; Song et al., 2017a) evaluated PhytOC and estimated the PhytOC accumulative rate in different ecosystems in China and even at the global scale. Zuo et al. (2014) then focused on soil phytoliths in the Chinese Loess Plateau, developing a wet oxidation method, modified from previous phytolith extraction processes, which can extract pure phytoliths from the soil (Piperno, 2006; Carter, 2009; Santos et al., 2010).



Zuo and Lu Phytolith Radiocarbon Dating—A Review

In 2013, Wu, an expert in archeometry from Peking University, cooperated with us in phytolith carbon dating by providing secure cultural layers rich in phytoliths. We then used the modified wet oxidation method to extract phytoliths, and the recovered phytoliths were sent to the Peking University Radiocarbon Laboratory for radiocarbon measurement. Wu also sent her students to our laboratory to learn how to extract pure phytoliths from soil. One of them, Jin, extracted phytoliths from the early cultural layers of Tianluoshan site. The results showed that the phytolith date (4,550±35 a BP) was marginally older than their paired seeds age (4,400±40 a BP). They speculated that the organic material with carboxyl groups that were not completely removed during the extraction processes might cause phytolith dates older than its paired seed date (Jin et al., 2014). Another student, Yan (2013), further compared different dating substances, such as charcoal, phytoliths, fatty acids, and total organic carbon, collected from the same depth of storage pits in Cishan site and paleo rice fields in Shanlonggang site. Among the five paired dating samples, two phytolith dates overlapped with their paired charcoal ages within ±2σ uncertainty; one was almost 5,000 years older than its paired charcoal age, and the remaining two were approximately 100 years older than the charcoal ages (**Table 2**). She concluded that the phytolith age is usually older than the charcoal age, while the fatty acid age was closer to the charcoal age, as it is relatively stable among all the dating substances (Yan, 2013).



Furthermore, Yin, an expert in quaternary geochronology from the Institute of Geology, Chinese Earthquake Administration, joined us in phytolith carbon dating. He and his colleague developed a new AMS graphite target preparation line in their 14C laboratory. They dated phytoliths extracted from paleo-loess with an OSL date of 71 ka. The results showed that the phytolith date (42,380±180 a BP) was close to the background date of the graphite system (42,750±190 a BP), suggesting that not only was soil PhytOC not contaminated by exogenous organic materials, but also very limited modern carbon was introduced during phytolith extraction, AMS graphite sample preparation, and radiocarbon measurement (Yang, 2013). They then combusted phytoliths extracted from modern rice and millet at different temperatures and the results showed that phytoliths combusted at lower temperatures (≤900°C) yielded more reasonable ages than at higher temperatures (≥1,100°C) (Yin et al., 2014). Given older phytolith ages at higher combustion temperatures, they speculated that there are probably two fractions of organic carbon in phytoliths, namely, labile and recalcitrant carbon.

As mentioned above, several Chinese research groups have shown great interest in phytolith carbon dating; however, only a few have provided images of phytoliths extracted from the soil to validate the efficiency of their extraction methods in completely eliminating all exogenous organic materials and other minerals. In this regard, we used our modified oxidation method to extract phytoliths from the cultural layers of several archaeological sites in China. Before sending the phytolith samples to Beta Analytic for radiocarbon measurement, we used scanning electron microscopy, energy-dispersive X-ray spectroscopy, and X-ray refraction to check the purity of phytoliths. The preliminary results showed that most of the phytolith ages were generally consistent with that of other dating materials collected from the same depth as phytolith samples, except for one outlier (Zuo et al., 2016). We attributed the inconsistency to the postdepositional processes of soil phytoliths. This suggests that each step of phytolith dating, including sampling, extracting, and measurement, should be carefully carried out to ensure that phytolith carbon dating is based on a secure archaeological context (without postdepositional processes) and appropriate chemical preparation (without exogenous organic materials). Our results showed that, for these sites, phytolith ages were consistent with those of other dated materials at the same level or context, suggesting that phytolith radiocarbon dating can be reliable and accurate at some sites (Zuo et al., 2017).

The reliability of phytolith dating will be discussed with respect to the following three aspects: 1) Is old carbon from the soil occluded into the phytoliths? 2) If so, how much will the old carbon skew the phytolith age determination? 3) Do the different methods (both for phytolith extraction and radiocarbon measurement) affect the phytolith dating results?

#### THE NATURE AND SOURCE OF PHYTOC: OLDER CARBON OR PHOTOSYNTHETIC CARBON?

Although there has been considerable discussion, researchers began to pay attention to the nature of PhytOC in the early stage of phytolith carbon dating research in the 1960s. Infrared spectral data of phytoliths suggested that PhytOC is composed of a variety of cell-derived substances, such as humic acid, amino acids, and amines (Wilding et al., 1967). The significantly depleted δ13C in phytoliths relative to that in the host plant tissue indicated that PhytOC might include lipids and lignin, which might have a depleted carbon isotope (Kelly et al., 1991). Smith and Anderson (2001) also found lipids in phytoliths, but no lignin. Masion et al. (2017) detected several carbohydrate components in phytoliths, such as sugars, adenosine triphosphate, and sodium pyrogluconate, using a new technique of dynamic nuclear polarization nuclear magnetic resonance. Raman spectrum analysis of single dumbbell phytoliths from sorghum also revealed that phytoliths contain carbohydrates, lipids, and other organic substances (Gallagher et al., 2015). Although there are differences in the understanding of the nature of PhytOC, previous studies assumed organic matter from plant tissue is the only source of PhytOC.

Santos, an expert in isotopic analysis, was the first to question the reliability of phytolith carbon dating. Initially, Santos et al. (2010) performed radiocarbon AMS measurement of carbon occluded in phytoliths from living plants and unexpectedly obtained dates that were several thousand years old. They suggested that there are some possible sources of carbon contamination, which needed further investigation (Santos et al., 2010). In 2012, they further suggested that soil-derived carbon (older carbon) absorbed by plant roots is a possible reason for the old phytolith ages obtained for living plants (Santos et al., 2012), although they lacked direct evidence showing that phytoliths can occlude older carbon from the soil. If older carbon is occluded in phytoliths, not only is the use of phytolith carbon for dating called into question, but it also reduces the importance of PhytOC in global carbon sequestration (Santos and Alexandre, 2017), and phytolith carbon sequestration might not be as significant as that reported by Song et al. (2016). While the contribution of old soil carbon to PhytOC was debated by several researchers interested in PhytOC (Piperno, 2016b; Santos et al., 2016; Santos and Alexandre, 2017; Song et al., 2017b; Zuo et al., 2017; Santos et al., 2018), Santos and her group were seeking direct evidence of soil-derived C in phytoliths. Using the comparative isotopic analysis of PhytOC, host tissues, atmospheric CO2, and soil organic matter, they found that PhytOC is partially obtained from soil carbon (Reyerson et al., 2016).

It is now clear that small amounts of soil carbon are occluded in phytoliths, as well as in plant tissues, as some hydroponic experiments have indicated that plants can absorb a small amount of sucrose or glucose from the source medium (Wu et al., 2015; Zhang and He, 2015; Chen et al., 2016). Because it is not possible to estimate the percentage of PhytOC that is of soil origin and the age of the soil carbon occluded by phytoliths is unknown, Santos et al. (2018) suggest that radiocarbon dating of phytoliths is highly problematic and not trustworthy.

## CONTRIBUTION OF OLDER SOIL CARBON TO PHYTOLITH AGES

With further understanding of the nature of phytoliths and PhytOC, we now realize that although most of the PhytOC is

from atmospheric CO2 fixed by photosynthesis, a small amount of carbon is not photosynthetic, likely derived from soil organic carbon. Because plants absorb old soil carbon through the roots, this carbon should be homogenously distributed in different tissues (Gallagher et al., 2015), and the roots, stems, leaves, and other parts will contain old carbon from the soil. If the phytolith ages are skewed by older carbon from the soil, one would expect the same effect when dating plant tissue, but this is clearly not the case, because plant debris is one of the best dating materials in sediment. Santos et al. (2018) noted that compared to PhytOC 14C results, plant-C 14C results were not biased by old soil carbon, suggesting the asymmetric 14C effects of soil carbon contribution to plant debris and PhytOC. They speculated that there must be some unknown processes that allow most of the soil carbon absorbed by the roots to accumulate in phytoliths (Alexandre et al., 2016; Santos et al., 2018). However, due to the limited knowledge about the relocation of soil carbon in plants, further studies are needed to investigate whether asymmetric relocation of soil carbon exists in plants. However, Hodson (2012; 2015) has stated that no mechanism can explain why soil carbon preferentially accumulates in phytoliths, while large amounts of photosynthetic carbon in plant tissues are excluded during the deposition of silica.

It is unreasonable to attribute all questionable phytolithdating results to distortion by soil carbon. Other possible factors influencing the process of sampling, phytolith extraction, and radiocarbon measurement cannot be ignored when evaluating phytolith ages. Studies on contamination effects on 14C dating showed that the introduction of 1% dead carbon can only result in an increase in the age by approximately 80 years (Taylor and Bar-Yosef, 2014). With the isotopic-labeled analysis of the silicon-rich hydroponic solution of grass, it was revealed that soil-derived carbon in phytoliths might constitute 0.15% of the PhytOC (Alexandre et al., 2016). Even though the actual percentage is likely considerably higher in natural soil conditions, such a small amount of older carbon will not yield phytolith ages thousands of years older than expected if assuming a 1.5% soil carbon contribution to PhytOC (10 times higher than under hydroponic conditions).

## INFLUENCE OF DIFFERENT EXTRACTION AND RADIOCARBON MEASUREMENT METHODS ON PHYTOLITH CARBON DATING

The wet oxidation method is the main phytolith extraction method in phytolith carbon dating research, and the difference among different extraction methods is mainly in the oxidation stage before heavy-liquid flotation of phytoliths. One method uses H2SO4+H2O2, known as rapid oxidation or over oxidation, and the other uses HNO3 or HNO3+KClO3. Researchers who used the latter method suggested that the oxidation process should remove as much exogenous organic matter as possible; however, rapid oxidation is so harsh that it not only can remove the exogenous organic matter, but also might change the nature of phytoliths, thus skewing the phytoliths ages (Sullivan and Parr, 2013; Song et al., 2016; Zuo et al., 2016). Whether the rapid oxidation method will change the nature of PhytOC remains unclear, but the PhytOC content will decrease significantly after rapid oxidation (**Table 3**) (Santos et al., 2010; Zuo and Lü, 2011; Santos et al., 2012; Corbineau et al., 2013), indicating that the carbon occluded in cavities of phytoliths is likely to be removed and that the integrity of PhytOC is destroyed (Sullivan and Parr, 2013; Parr and Sullivan, 2014).

The overoxidation method is so strong that it might cause phytolith ages older than the expected ages because of changes in the nature and structure of PhytOC, while the underoxidation method and incomplete removal of organic material could cause older phytolith ages (Zuo et al., 2018). We compared the influence of two different phytolith extraction methods on radiocarbon dating of phytoliths. The results showed that phytolith ages acquired using the conventional extraction method that does not exclude all exogenous organic materials were substantially older than those obtained using improved extraction methods.

Nondestructive phytolith extraction methods to extract phytoliths without using a strong acid not only can yield pure phytoliths, but also can maintain the integrity of PhytOC. Asscher et al. (2017) only used HCl (1N) to exclude calcium carbonate in the phytolith extraction process. Before acid treatment, they used a heavy liquid (2.4 and 1.6 g/ml) to remove quartz, calcite, and carbonized organic matter. There was no heating in any step of the phytolith extraction process. The results showed that several phytolith ages were consistent with the age of carbonized seeds within the ±1σ correction interval at the same level; the others have slightly older ages (Asscher et al., 2017). These phytolith dates were challenged by Santos et al. because they were obtained from phytoliths whose purity was not assessed by scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX) (Santos et al., 2018). The non-heating method used by Asscher et al. (2017) is not likely to produce a pure phytolith extract and thus the remaining exogenous organic matter might cause phytolith ages older than the paired seeds ages. This method was also used to extract phytoliths in the analysis



of DNA in phytoliths, in order to avoid the influence of strong acids and high temperatures on DNA information that might be preserved in phytoliths (Elbaum et al., 2009).

## POSSIBLE REASONS FOR OLDER PHYTOLITH AGES IN SOIL PROFILES AND LIVING PLANTS

After carefully reviewing previous phytolith dating reports and other studies where older carbon may have biased PhytOC dates (Kelly et al., 1991; Piperno and Becker, 1996; Kelly et al., 1998; Krull et al., 2003; Piperno and Jones, 2003; Piperno and Stothert, 2003; Mcmichael et al., 2012; Sullivan and Parr, 2013; Madella et al., 2014), we speculate that older phytolith ages in soil profiles could be explained by the following two aspects. First, if the extracted phytoliths are pure after checking with SEM and EDX, then the postdepositional processes of phytoliths should be considered. Second, if the phytolith dating results are older than expected (even ten thousand years older), one should repeat the experiment and revaluate if the protocols used could exclude all carbonate and other minerals from the samples. Incompletely excluding sources of dead carbon can lead to phytolith ages hundreds to thousands of years older than expected. The introduction of 5% old carbon would make the dating sample (true age is 10,000 a BP) approximately 400 years older, and adding 50% very old carbon would make the age only about 5,000 years older. Thus, if no more than 1% PhytOC is taken up from the soil, one would not expect phytolith dates to differ from expected ages by thousands of years.

The unexpectedly older ages dated for extracted phytoliths from modern plants (Reyerson et al., 2016) may be caused by phytolith extraction procedures such as overaggressive digestion protocols that alter the structure, nature, and yield of PhytOC (Sullivan and Parr, 2013; Parr and Sullivan, 2014). Recently, some detection techniques, such as Raman spectroscopy and nanoscale secondary ion mass spectrometry, have been used to determine the location, distribution, and chemical structure of PhytOC (Alexandre et al., 2015; Gallagher et al., 2015), showing a continuous but nonhomogeneous distribution. The amount and nature of PhytOC might vary considerably depending on phytolith morphology and different allocations within phytoliths. Using harsh protocols to extract phytoliths from modern plants (Santos et al., 2010; Santos et al., 2012; Corbineau et al., 2013; Yin et al., 2014; Reyerson et al., 2016) might damage the integrity of PhytOC. Moreover, the carbon in cavate or surface of phytoliths might be consumed by the harsh digestion protocols, which would lead to underestimation of the total amount of PhytOC (Parr and Sullivan, 2014). The consumed carbon might be isotopically rich in 14C; however, the residual carbon fraction might be highly depleted in 14C (**Figure 2**). Given that the lipids within phytoliths are depleted in 13C (Smith and Anderson, 2001), this is also probably true for 14C (Hodson, 2016).

## RECONCILING OR REBUTTING?

As mentioned above, it is difficult for researchers to reconcile on the reliability of phytolith carbon dating. The focus of discussion is mainly on the nature of PhytOC, the actual contribution of soil carbon to phytolith dates, and influences of different extraction and measurement processes on phytolith dates. Santos et al. based their older carbon theory on the following four aspects: 1) the age of phytoliths from modern living plants are decades to thousands of years older than their sampling time (Santos et al., 2010; Santos et al., 2012); 2) over 200 comparative isotopic measurements of PhytOC and isotopic-labeled experiment provide evidence of soil carbon in PhytOC (Reyerson et al., 2016); 3) although soil carbon can be absorbed by the roots, it does not skew plant-C 14C results, but only the PhytOC 14C results; and 4) no matter how soil phytolith dates match their expected ages, they are all questionable due to the variability of soil carbon contribution to PhytOC (Santos et al., 2018).

As discussed in the beginning of the review, not all phytolith dating results are older than the expected results. Several phytoliths extracted from modern plants, dated by Piperno (2016a) and Sullivan and Parr (2013), have either returned postbomb 14C ages or are very close to the modern dates. Most of the older modern phytoliths were dated by Santos et al. (2010). Phytolith dates from modern plants processed with the harsh techniques (Santos et al., 2012; Reyerson et al., 2016) are often considerably older than on plants processed with less harsh methods (Piperno, 2016a; Asscher et al., 2017; Zuo et al., 2017). Moreover, the harsh techniques typically leave much less carbon for dating than less harsh methods, partially due to leakage and the dual source of carbon—one labile and the other resistant (Hodson, 2019). Although researchers have stated that they have carefully dated PhytOC, Santos et al. (2012) might have only dated the carbon in lumen phytoliths, while Piperno et al. (2016a) might have dated not only the carbon in lumen phytoliths but also a part of carbon in cell wall phytoliths. A high amount of carbon processed by less harsh methods might preserve the integrity of PhytOC, but a less amount of carbon processed by harsher methods should not be preferred for dating.

Another key point that must be considered is that phytoliths differ in several aspects from other datable materials such as charcoal and seeds. Dating phytoliths and charcoal from the same stratigraphic/sedimentary level does not mean that they should have exactly the same dates, since phytolith age is the average age of all phytoliths in that level, whereas macro-plant/ charcoal dates from a single sample represent a single moment in time. It is unreasonable to expect that a piece of charcoal or seed deposited at a single moment can completely fall within the age of a collection of phytoliths (Piperno, 2016a). A difference of hundreds of years between the dating results of soil phytoliths and other datable materials when sampling a thick soil layer of 5 to 10 cm is generally acceptable and reasonable (Zuo et al., 2016). Considering the depositional processes of phytoliths in soil, PhytOC should not be used for answering high-resolution chronological questions. However, it can be tried as an alternative dating method when other datable materials are absent.

#### CONCLUSIONS AND REMARKS

As an unconventional 14C dating material, phytoliths have been widely used during the past half century. Radiocarbon dating PhytOC has played an important role in constructing the chronological sequence of some key scientific issues, such as when pumpkin and rice domestication began (Piperno and Stothert, 2003; Zuo et al., 2017), but at the same time, the technique has also been criticized (Santos et al., 2016; Santos and Alexandre, 2017; Santos et al., 2018). A review of the phytolith dating literature revealed that not all phytolith dating results are inconsistent with expected ages. The poor results cannot be entirely attributed to the influence of older carbon absorbed from the soil, because most of the PhytOC is obtained from the atmospheric CO2 synthesized by photosynthesis. Phytolith ages thousands of years older than expected are probably due to impure phytolith extracts not completely cleaned of extraneous carbon rather than phytolith occluded carbon obtained from the soil.

Compared with other conventional dating materials, research on the mechanisms, methods, and results of phytolith dating is limited. There are considerable empirical data showing that at many sites, PhytOC dating provides reasonable dates. However, concerns about extract purity, as well as the variable nature of the PhytOC carbon pool, suggest that the reliability of phytolith dates is questionable in many cases. Whether different phytolith extraction methods will inevitably lead to differences in the dating results remains an open question. Whether the difference in the PhytOC content obtained using the rapid oxidation method and the conventional oxidation methods is due to PhytOC being destroyed or the organic matter in plants being incompletely removed is important for evaluating the phytolith dating

#### REFERENCES


results and key to reconcile the conflicting opinions. Phytolith researchers working with PhytOC urgently need to agree on a standardized extraction procedure that produces a phytolith extract verified by SEM and EDX to be free of extraneous carbon while using the least harsh chemicals possible. We expect that more data on phytolith dating in other regions and laboratories will be published in the future and will further clarify issues relating to 14C dating and will allow the expansion of the application of phytolith dating to the construction of chronological sequences.

#### AUTHOR CONTRIBUTIONS

XZ designed and wrote the manuscript. HL contributed to discussion and approved the final manuscript.

#### FUNDING

This work is jointly supported by the National Natural Science Foundation of China (41771241 and 41830322) and the Innovation Research Team Fund of Fujian Normal University (IRTL1705).

#### ACKNOWLEDGMENTS

We thank Prof. Dolores R. Piperno for her encouragement and guidance during the study of phytolith carbon dating; two reviewers for their valuable comments on the manuscript; Prof. Martin Hodson for his invitation to join the topic of *Frontiers in Phytolith Research*. We thank the Editage (www.editage.cn) for the English improvement.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Zuo and Lu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Plant Silicon and Phytolith Research and the Earth-Life Superdiscipline

Ofir Katz\*

Dead Sea and Arava Science Center, Tamar Regional Council, Israel

Keywords: plants, silicon, phytolith, earth system, superdiscipline

### DISCIPLINARY ORIGINS OF PLANT SILICON AND PHYTOLITH RESEARCH

Plant silicon and phytolith research stands as a good example for how a single phenomenon or theme is studied by scholars from multiple disciplines, and for how knowledge flows among disciplines. At its very core and origins, plant silicon and phytolith research lies in traditional botany, since it studies the occurrence and role of silicon and phytoliths within plants and among plant groups (e.g., Hodson et al., 2005; Katz, 2014, 2015; Strömberg et al., 2016) and can be potentially used to improve taxonomy and systematics by providing more characters to be included in analyses (e.g., Prychid et al., 2004; Katz, 2014, 2018a). Nevertheless, plant physiologists study the mechanisms of silicon uptake, transport and accumulation within plants (e.g., Peleg et al., 2010; Mitani-Ueno et al., 2014; Ma and Yamaji, 2015; Kumar et al., 2017), chemists study the mechanisms of its deposition (e.g., Currie and Perry, 2007; Patwardhan et al., 2012), ecophysiologists identify silicon's and phytoliths' functions within plant tissues (e.g., Fauteux et al., 2005; Liang et al., 2007; Epstein, 2009; Guntzer et al., 2012; Cooke and Leishman, 2016; Coskun et al., 2016) and ecologists study how silicon and phytoliths interact with herbivores (e.g., Massey and Hartley, 2006; Katz et al., 2014; Hartley, 2015; Frew et al., 2016) and shape plant communities (e.g., Jacobs et al., 2013; Schoelynck et al., 2014; Cooke et al., 2016), ecosystems (e.g., Cooke and Leishman, 2011; Cooke et al., 2016; Schoelynck and Struyf, 2016), and even biomes and the entire ecosphere (e.g., Carey and Fulweiler, 2012, 2016; Song et al., 2012, 2017; Katz, 2018b).

Within Earth sciences, biogeochemists study the physics and chemistry of plant silicon and phytoliths, including their dissolution (e.g., Fraysse et al., 2009; Cabanes and Shahack-Gross, 2015) and chemical and isotopic composition (e.g., Hodson et al., 2008; Kamenik et al., 2013; Alexandre et al., 2015). Others study the silicon cycle (e.g., Alexandre et al., 2011; Carey and Fulweiler, 2012, 2016; Song et al., 2012, 2017) and its connections with other biogeochemical cycles (e.g., Street-Perrott and Barker, 2008; Carey and Fulweiler, 2012, 2016; Song et al., 2012, 2017; Alexandre et al., 2015; Cornelis and Delvaux, 2016). Plant silicon and phytoliths are also often used in geoarchaeology to infer past human life (e.g., Tsartsidou et al., 2008; Lancelotti et al., 2014; Hart, 2016), as well as in paleontology to reconstruct ancient vegetation and ecosystems (e.g., Strömberg et al., 2007; Albert and Bamford, 2012) and to trace the evolution of plants and animals (e.g., Prasad et al., 2011; Strömberg, 2011; Katz, 2015; Strömberg et al., 2016).

Thus, plant silicon and phytolith research demonstrates the integration of knowledge from both Earth and life sciences. The plant silicon and phytolith research community studies the effects of plant silicon uptake on other organisms, ecosystems and biogeochemical cycles in tandem with the effects of other organisms, ecosystems and biogeochemical cycles on plant silicon uptake. Likewise, members of the community use geoarchaeological and palaeontological methods to understand the evolution and history of plants and animals, while using knowledge of the evolution of plants and animals to understand changes in the geosphere and Earth themselves. While all these themes are intimately connected, share many theoretical and methodological aspects, and constitute a single

#### Edited by:

Martin John Hodson, Oxford Brookes University, United Kingdom

#### Reviewed by:

Santosh Kumar, Weizmann Institute of Science, Israel Sheikh Abdul Shakoor, Guru Nanak Dev University, India Jean-Thomas Cornelis, Gembloux Agro-Bio Tech-University of Liège, Belgium

> \*Correspondence: Ofir Katz katz.phyt@gmail.com

#### Specialty section:

This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science

Received: 11 June 2018 Accepted: 15 August 2018 Published: 05 September 2018

#### Citation:

Katz O (2018) Plant Silicon and Phytolith Research and the Earth-Life Superdiscipline. Front. Plant Sci. 9:1281. doi: 10.3389/fpls.2018.01281

**259**

research topic, only rarely do we see a researcher or a research group that covers a considerable portion of this wide range. One possible reason for this is that the question one asks, the methods one employs to answer them, and the interpretations of these results are strongly influenced by one's parent discipline. Many of us study plant silicon and phytoliths as part or in addition to other themes within our parent disciplines, thus hindering the formation of a common meeting ground or language for plant silicon and phytolith researchers from various parent disciplines. The compilation of the International Code for Phytolith Nomenclature (Madella et al., 2005) is an advancement toward solving part of this problem, albeit somewhat limited to more technical rather than theoretical issues.

By remaining bound to parent disciplines, we sentence our field to remain adjacent to the mainstream (rather than within it) and led by parent disciplines and their agendas. Instead, we should form a greater integrated framework that links our parent disciplines, extends their scopes, increases dialogue among them, and achieves high-order knowledge transfers among them (**Figure 1**). This is possible now more than ever. Since Earth and life sciences are merging, our field that sits between them can gain a rightful place at the center stage of a new emerging superdiscipline.

## SILICON AND PHYTOLITH RESEARCH WITHIN THE INTERDISCIPLINARY EARTH-LIFE SCIENCES MERGER

As science progresses, so do models of knowledge transfer (Krishnan, 2009). The simplest model is cross-disciplinary knowledge transfer (**Figure 1A**), in which knowledge from one discipline is borrowed by researchers from another discipline, without true collaboration or dialogue among disciplines. This model is very uncommon because of its inherent shortcoming: that people outside a discipline use knowledge although they have lesser understanding of its underlying assumptions and theories or of specific methodologies (Keene, 1983; Krishnan, 2009). The use of this model nowadays is limited strictly to methodological technicalities.

A second, common model is of multidisciplinary collaboration (**Figure 1B**), in which one discipline initiates a research programme, on which research teams from other disciplines work independently. The initiating discipline is responsible for synthesis and gains most of the knowledge, while the other disciplines gain less knowledge (often methodological knowledge only). Examples for multidisciplinary collaborations in plant silicon and phytolith research include the use of phytolith analysis to increase botanical knowledge in archaeology (e.g., Albert et al., 1999; Tsartsidou et al., 2008; Lancelotti et al., 2014; Hart, 2016) and palaeoecology (e.g., Albert and Bamford, 2012). Others revealed parts of evolutionary history through new insights into plant physiology and ecology (e.g., Strömberg et al., 2007; Prasad et al., 2011; Strömberg, 2011; Katz, 2015, 2018a). Vice versa, some scholars use plant silicon and phytoliths to identify possible external evolutionary stimuli that may provide insight into the function of plant silicon

FIGURE 1 | Four models for transfer and sharing among disciplines (Klink et al., 2002; Krishnan, 2009). (A) Cross-disciplinary knowledge transfer: Scholars from one discipline (yellow) use knowledge or methods from another discipline asymmetrically and unidirectionally. (B) Multidisciplinary collaboration: One discipline (yellow) initiates a research programme, on which other disciplines work independently. Synthesis is carried out almost solely by the initiating discipline, and although knowledge transfer is not unidirectional, it is asymmetrical. (C) Interdisciplinary framework: Several disciplines share a theoretical framework. All disciplines contribute knowledge to the shared framework and take part in synthesis. Knowledge flows symmetrically, but through a mediating intersection. (D) A superdiscipline: Disciplines are rearranged by relaxing boundaries among them and thus looking at the union rather than at the intersection. Each discipline bears an equal weight and knowledge flows in all directions (ideally) free of constrains.

and phytoliths (e.g., Katz, 2014, 2015; Strömberg et al., 2016). Finally, phytolith chemistry contributes to our understanding of silicon dissolution in soil and transport in ecosystems (Fraysse et al., 2009; Alexandre et al., 2015; Cornelis and Delvaux, 2016).

A third, more complex model is the interdisciplinary framework (**Figure 1C**), in which researchers from various disciplines contribute and gain relatively equally, but all knowledge transfer is carried out through a shared theoretical framework. Two interdisciplinary frameworks that are of special interest for plant silicon and phytolith researchers are Earth System Science (ESS) and plant functional diversity. ESS is an interdisciplinary framework that attempts "to obtain a scientific understanding of the entire Earth System on a global scale by describing how its component parts and their interactions have evolved, how they function, and how they may be expected to continue to evolve on all time scales" (Earth System Science Committee, 1986) by applying methods and concepts from systems and complexity theories. ESS is therefore a merger of Earth and life sciences that uses systems and complexity theories as the common ground. Both paleontology and ecosystem ecology can be seen as subdivision of ESS, the former focusing on evidence for the evolution of the entire Earth System and the latter focusing on the direct interactions of Earth and life components within ecosystems, hence relying on emergence theory and ecosystem theory (respectively) as subsets of systems and complexity theories. Some studies of the silicon cycle and its interactions with the carbon cycle have quite explicitly used systems and complexity theories (Alexandre et al., 2011; Carey and Fulweiler, 2012, 2016; Cornelis and Delvaux, 2016), and thus represent an integration of plant silicon and phytolith research within the ESS framework.

Plant functional traits are quantitative traits whose values are affected by environmental variables and affect plant, community and ecosystem properties and functioning (Garnier et al., 2016). When discussing ecosystem functions like elemental cycling, plant functional diversity is an interdisciplinary framework that connects Earth and life sciences, with plant functional traits as the common ground that mediates the effects of Earth components on plants and the effects of plants on the ecosystem, again greatly relying on systems theory. Therefore, ESS studies and models can improve if they take into account plant functional traits and types (Beerling, 2007; Van Bodegom et al., 2012; Wullschleger et al., 2014). Although often ignored by mainstream plant functional diversity literature, plant silicon and phytolith contents are gaining increasing recognition as a plant functional trait, and are now known to be involved in plant responses to their environment and plant effects on the environment (Cooke and Leishman, 2011; Carey and Fulweiler, 2012; Song et al., 2012, 2017; Katz, 2014, 2015, 2018b; Schoelynck et al., 2014; Cooke et al., 2016; Schoelynck and Struyf, 2016). Therefore, plant silicon and phytolith research is a part of the interdisciplinary ESS framework.

#### AN EARTH-LIFE SUPERDISCIPLINE–A PROMISING FUTURE

These three aforementioned models, and especially multidisciplinary collaborations and interdisciplinary

frameworks, have served scientists very well, including in merging knowledge from Earth and life sciences and in plant silicon and phytolith research. However, they are not without shortcomings, including the asymmetry of knowledge transfer, the adherence to certain framing theories, and the limited integration that stems from maintaining boundaries among disciplines.

These shortcomings are overcome in the most advanced model of knowledge transfer and sharing among disciplines, the superdiscipline (**Figure 1D**), in which boundaries among disciplines are relaxed and knowledge flows freely within the greater domain of the superdiscipline, unbounded to any discipline or framing theory. Although relaxing disciplinary boundaries without the mediation of framing theories is difficult, it is very promising when attempting to answer big, complex, discipline-transgressing and irreducible questions (Krishnan, 2009). The seeds for a merged Earth-life superdiscipline have been sown many years ago. Ecosystem ecology, ESS and plant functional diversity represent great advancements in this direction, yet as interdisciplinary frameworks they are bound to the intersections of the parent disciplines and to the framing of systems and complexity theories. The road to a true Earth-life superdiscipline lies, at least in part, in removing these boundaries, as Beerling (2007) has nicely demonstrated in his book The Emerald Planet, which introduces a synthesis of plant physiology, paleontology and atmospheric sciences.

Somewhat ironically, the fact that plant silicon and phytolith research is adjacent to the mainstream means that it is less constrained than existing interdisciplinary frameworks, and therefore freer to achieve superdiscilinarity and have a leading role in the formation of an Earth-life superdiscipline. A key reason why plant silicon and phytoliths research can take a leading role in forming the new Earth-life superdiscipline is that this phenomenon inherently and intimately links Earth and life. Silicon is the second most abundant element in the Earth's crust, whose uptake by plants affects biotic (Massey and Hartley, 2006; Epstein, 2009; Cooke and Leishman, 2011, 2016; Strömberg, 2011; Guntzer et al., 2012; Schoelynck et al., 2014; Hartley, 2015; Schoelynck and Struyf, 2016; Frew et al., 2016) and abiotic (Street-Perrott and Barker, 2008; Alexandre et al., 2011; Carey and Fulweiler, 2012, 2016; Song et al., 2012, 2017) processes at multiple scales. Understanding some of these processes requires and benefits from understanding the variation of plant silicon uptake and accumulation across taxa (Hodson et al., 2005; Katz, 2014, 2015; Strömberg et al., 2016), habitats, ecosystems and biomes (Carey and Fulweiler, 2012; Katz et al., 2013, 2014; Schoelynck et al., 2014; Song et al., 2017) and geologic time (Prasad et al., 2011; Strömberg, 2011; Katz, 2015; Strömberg et al., 2016). The references cited in this paragraph alone (and throughout this manuscript) demonstrate that many of us already carry out studies that cross and relax disciplinary boundaries, either in a single study or in a person's or group's combined research portfolio. It seems that this attribute of our field puts it in a better and more developed and advanced position to intimately merge Earth and life sciences, possibly even compared to some fields of research that lay deeper within the mainstream and that are more intensively studied, such as photosynthesis and the carbon cycle.

Hence, embedding superdisciplinary thinking in plant silicon and phytolith research can not only advance our field, but increase its impact in the merger of Earth and life sciences into a single superdiscipline. Working toward this goal is a true new frontier for plant silicon and phytolith research, for Earth-life sciences and for science in general.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.

#### FUNDING

The writing of this manuscript was made possible thanks to the Israel Ministry of Science and Technology's support to the Dead Sea and Arava Science Center.

is reversed by silicon-based plant defences. J. Appl. Ecol. 54, 1310–1319. doi: 10.1111/1365-2664.12822


for biogeochemical carbon sequestration. Earth Sci. Rev. 115, 319–331. doi: 10.1016/j.earscirev.2012.09.006


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Katz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.