Edited by: Laura Zucconi, University of Tuscia, Italy
Reviewed by: Lotta Purkamo, Geological Survey of Finland, Finland; Monica Tolotti, Fondazione Edmund Mach, Italy
This article was submitted to Extreme Microbiology, a section of the journal Frontiers in Microbiology
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.
Glacial meltwater drains into proglacial rivers where it interacts with the surrounding landscape, collecting microbial cells as it travels downstream. Characterizing the composition of the resulting microbial assemblages in transport can inform us about intra-annual changes in meltwater flowpaths beneath the glacier as well as hydrological connectivity with proglacial areas. Here, we investigated how the structure of suspended microbial assemblages evolves over the course of a melt season for three proglacial catchments of the Greenland Ice Sheet (GrIS), reasoning that differences in glacier size and the proportion of glacierized versus non-glacierized catchment areas will influence both the identity and relative abundance of microbial taxa in transport. Streamwater samples were taken at the same time each day over a period of 3 weeks (summer 2018) to identify temporal patterns in microbial assemblages for three outlet glaciers of the GrIS, which differed in glacier size (smallest to largest; Russell, Leverett, and Isunnguata Sermia [IS]) and their glacierized: proglacial catchment area ratio (Leverett, 76; Isunnguata Sermia, 25; Russell, 2). DNA was extracted from samples, and 16S rRNA gene amplicons sequenced to characterize the structure of assemblages. We found that microbial diversity was significantly greater in Isunnguata Sermia and Russell Glacier rivers compared to Leverett Glacier, the latter of which having the smallest relative proglacial catchment area. Furthermore, the microbial diversity of the former two catchments continued to increase over monitored period, presumably due to increasing hydrologic connectivity with proglacial habitats. Meanwhile, diversity decreased over the monitored period in Leverett, which may have resulted from the evolution of an efficient subglacial drainage system. Linear discriminant analysis further revealed that bacteria characteristic to soils were disproportionately represented in the Isunnguata Sermia river, while putative methylotrophs were disproportionately abundant in Russell Glacier. Meanwhile, taxa typical for glacierized habitats (i.e.,
Glacial meltwater is highly influential to downstream ecosystems by creating a unique habitat template (
The Greenland Ice Sheet (GrIS) represents one of the largest bodies of ice on Earth, and its meltwater bears considerable influence on downstream freshwater and marine ecosystems (
Within the glacial environment, microbial communities differ between distinct habitat types (
While glaciers are a significant source of suspended cells in the water column of proglacial streams, other sources within the hydrological catchment may gain influence downstream. For example, eroding stream banks, inlets of tributary streams, and thawing permafrost may also contribute to the transported “seed bank” (
Here, we investigated how meltwater microbial assemblages from three rivers draining glaciers of different sizes, and with different glacierized: proglacial catchment ratios, differ between each other and over the hydrological rising limb of a summer melt season. To do this, we collected daily water samples simultaneously from three proglacial streams draining outlet glaciers of the GrIS for approximately 3 weeks over the 2018 summer. We hypothesized that assemblage structure and diversity should differ between the three catchments due to their catchment characteristics, as well as with the expansion of the glacierized and proglacial hydrological catchment areas. Understanding how the composition of microbial assemblages differs between catchments and over the course of the melt season helps us to understand both the temporal shifts in catchment connectivity, as well as downstream ecosystem changes to anticipate in the future.
Fieldwork was conducted on three proglacial streams draining outlet glaciers of the GrIS, including Isunnguata Sermia (IS), Leverett Glacier (LG), and Russell Glacier (RG), over the 2018 summer (June–July). We report glacierized and proglacial catchment areas as calculated in arcGIS using the methodology outlined in
Ratios of proglacial catchment sizes (“Proglacial”) and estimated glacier catchment sizes (“Glacier”). In the case of LG’s glacier catchment size, a mean value (840 km2) was used.
Site | Proglacial catchment | Glacierized catchment | Proglacial: Glacial | Glacial: Proglacial | Latitude | Longitude |
---|---|---|---|---|---|---|
IS | 560 | 14,000 | 0.04 | 25 | 67.166° N | 50.889° W |
LG | 11 | 840 | 0.01 | 76.36 | 67.066° N | 50.215° W |
RG | 40 | 81 | 0.49 | 2.03 | 67.104° N | 50.217° W |
Map of the three Greenland Ice Sheet (GrIS) catchments evaluated in the study. The northernmost green catchment is Isunnguata Sermia (IS), the middle purple catchment is Russell (RG), and the southernmost orange catchment is Leverett (LG). Solid colors indicate the glacierized catchment, while outlines indicate the proglacial catchment area. The left panel is a composite image of Greenland from Landsat (US Geological Survey, via Google), and the right panel was created with Sentinel 2 imagery (Level 2a, 16 August 2020, downloaded from
The intermediate-sized glacierized catchment drains LG and is the southernmost of the three sites. The LG glacierized catchment extends ~80 km into the GrIS interior (
The third proglacial stream drains RG and was sampled ~10 km from the glacier terminus (67.104° N, 50.217° W). This glacier is located between IS and LG, and extends ~10–15 km inland (
Physical and chemical analyses of streamwater were conducted as described in
Water samples were collected for approximately 3 weeks over the 2018 summer coinciding with the hydrological rising limb of the melt season (i.e., preceding peak melt), with one sample collected each day to measure temporal changes in the exported microbial assemblages (
The sampling was performed between 22 June and 14 July for IS, between 21 June and 13 July at LG, and between 21 June and 15 July at RG. At each stream, one sampling site was chosen to monitor and used every day to collect water samples at the same time. This was 11:00 for RG and LG, and at 15:00 for IS. The delayed timing for IS was due to the sampling site being located further downstream than in the other two, and in order to sample a comparable parcel of water on the stream’s daily hydrograph. Three 24-h sampling campaigns were also conducted, with one sample taken every 2 h over a full diel cycle to explore diurnal patterns in assemblage structure. The 24-h sampling campaigns took place on 24–25 June, 4–5 July, and 10–11 July 2018.
Water for microbiological characterization was filtered directly from the stream using a sterile syringe and was passed through a Sterivex filter (0.22 μm, Millipore, Billerica, MA, United States) until it was clogged (200–600 ml, with most <500 ml). This way, a similar amount of suspended sediment, the primary source of microbial cells, was collected on each filter. Filters were flushed of water, preserved with LifeGuard® preservation solution (MO BIO, Carlsbad, CA, United States), and frozen as soon as possible (within 2 h of collection at −20°C).
DNA isolation was performed directly on thawed Sterivex filters using the PowerWater® Sterivex™ DNA Isolation Kit (MO BIO) following the manufacturer’s instructions. DNA concentration was measured by using Invitrogen™ Qubit™ 4 Fluorometer with Qubit™ 1X dsDNA HS Assay Kit (Invitrogen, Carlsbad, CA, United States). The V4 region of 16s rRNA gene was PCR amplified using primers 515F (GTGYCAGCMGCCGCGGTAA) and 806R (GGACTACNVGGGTWTCTAAT) and sequenced at SEQme s.r.o. (Dobříš, Czechia) on the Illumina MiSeq platform using 2× 250 bp arrangement.
Sequence data were analyzed through the pipeline SEED v2.1.2 (
Approximately one third of samples had a number of reads << 7,853, and were predominantly from IS and LG during the early part of the melt season, which had high suspended sediment loads which may have interfered with extraction. It was therefore not possible to use all samples due to the low numbers of reads from this subset. The total number of samples used in this study was 112 (IS:
For statistical analysis, all samples available for each catchment (including the 24 h samples) were used together unless stated otherwise. All statistical analyses were performed using R (
Alpha diversity metrics (number of OTUs, Shannon Diversity) were calculated in the
To visualize differences in assemblage structure among the three catchments, principal coordinates analysis (PCoA) on Bray-Curtis distances were performed in the
To investigate relationships between discharge, EC, pH, TSS, and day of the year (plotted as decimal days, hereafter “DOY”) with individual OTUs, we identified differentially abundant taxa through a random forest analysis using the
Seasonal changes in our most abundant metabolism types as defined by the selected OTUs were also investigated. For this analysis, we used only samples collected each day at the same time (i.e., 24 h samples were omitted). However, if a sample from a particular timepoint was missing, we used a 24 h sample collected within 2 h from the regular sampling time if one was available. Specifically, this concerned two samples of IS from DOY 176 and 186. For the following analysis, we chose all OTUs which were more abundant than 0.99% in at least one sample, which altogether was 33 OTUs. To determine their potential metabolisms, these OTUs were thoroughly compared with the NCBI nt/nr database, while only BLAST hits consistent with the previous identification of the OTUs using the Silva database and, concurrently, belonging to well-characterized isolates, were considered. Most of the abundant OTUs were closely related to isolates characterized by cultivation or genome sequencing. Thus, the basic physiological traits were derived from these microorganisms. We assume that these traits are conserved sufficiently to be adopted for genotypes sharing ca 98–100% homology of their 16S rRNA genes (
In this study, we investigated the differences in suspended microbial assemblages between three proglacial streams during a three-week period of the 2018 melt season. These streams differ in the size of their glacierized catchment, but also in the proportion of their glacierized to proglacial catchment areas. Here, we hypothesize that the proportion of catchment type (glacial vs. non-glacial), as well as the stage of the melt season, are likely the major controls explaining differences in the structure of transported assemblages. This work is an important first step in understanding the temporally- and spatially-dynamic hydrological connectivity of proglacial streams to their catchments from glacier to estuary.
The three proglacial streams differed in terms of their monitored physical and chemical characteristics (
Hydrochemical measurements, including:
The size of the glaciated catchment had an influence on the hydrology of these three streams, with the greatest glacierized catchment sizes having the greatest discharge (i.e., IS > LG > RG), and the decrease in TSS and EC over the monitored period indicating solute dilution with increasing discharge as the melt season progressed. The monitored period did not comprise a period of pronounced “outburst” events described in other studies from other seasons investigating these sites (in particular LG; e.g.,
In terms of alpha diversity, both the number of OTUs and Shannon diversity were higher in IS (Richness: mean = 849.3, range 604–1,035; Shannon: mean = 4.31, range 3.98–4.66) and RG (Richness: mean = 821.6, range 689–973; Shannon: mean = 4.57, range 4.13–4.9) when compared to LG (Richness: mean = 536.8, range 428–663; Shannon: mean = 3.45, range 3.21–3.69,
Observed number of OTUs (left panel) and Shannon diversity (right panel) for the three proglacial streams. In each panel, Isunnguata Sermia (IS) is to the left in green, Leverett (LG) in the middle in orange, and Russell (RG) to the right in purple.
On the other hand, RG had several slow moving “pond” areas upstream of the sampling site, which likely had a strong influence on microbial assemblages (
However, it was not only the alpha diversity that differed between the three catchments, but also the assemblage composition. When Bray-Curtis distances were compared in a PCoA, the first two axes explained 80.5% of variability (the first axis explained 74.7%, and the second axis explained 5.8%). IS and LG clustered closely together along axis 1, while RG clustered away from the former two (
Principal coordinates analysis showing the difference between three GrIS sites on Bray-Curtis distances. Isunnguata Sermia (IS) in green, Leverett (LG) in orange, and Russell (RG) in purple.
In total, 8,172 OTUs were identified in the full rarefied dataset. The five most abundant OTUs across the whole dataset (from highest to lowest) were
RG on the other hand was disproportionately represented by methylotrophs such as
Microbial assemblages changed over the monitored period as the hydrologically active glacial catchment size expanded and as the hydrological connection with the proglacial environment likely developed. Interestingly, microbial richness (i.e., the number of OTUs) in IS and RG increased over the sampling period, while LG richness decreased (
Observed number of OTUs plotted against Julian day (2018) samples were taken from the three GrIS proglacial streams, with Isunnguata Sermia (IS) to the left, Leverett (LG) in the middle, and Russell (RG) to the right.
The declining richness for LG over the summer is more difficult to explain. We hypothesize the reason for this is likely related to changes taking place within the glacierized catchment (i.e., rather than the non-glaciated catchment) given the limited interaction of this river with the proglacial landscape above the sampling site. Specifically, given the well-known evolution of an inefficient distributed hydrological system to an efficient subglacial drainage system at LG (
Environmental correlations for individual OTUs were substantially different between the three catchments (
LG had the most significant associations with individual OTUs and hydrochemical variables, possibly due to streamwater originating disproportionately from one source (
That said, many of the OTUs themselves seemed influenced by environmental variables. These included methylotrophs such as
Lastly, no detectable changes were observed in richness nor community composition over diel timescales (i.e., the “daily” samples). While this could of course be due to our relatively low number of samples/replicates, as well as the early period of the melt season from which the samples were taken (i.e., when diurnal peaks in discharge are less pronounced than later in the season), we tentatively hypothesize that the seasonal variability associated with the melt season is much more important than changes that occur over daily timescales.
In order to further investigate how the putative functionality of microbial assemblages changed over the melt season, we selected the main metabolism types of the most abundant OTUs (reaching at least 0.99% abundance in at least 1 sample) in our dataset for further analysis (
Most of the relative abundance of all sites were captured within “aerobes and facultative anaerobes” and “heterotrophs/facultative autotrophs.” Heterotrophs and facultative autotrophs exceeded the number of “autotrophs” at all sites. For RG and IS, heterotrophs and facultative autotrophs decreased in relative abundance over the course of the study (
Relative abundance of OTUs for selected metabolism types plotted against Julian day (2018). Please note the different y-axis scales.
At the same time, “autotroph” relative abundances increased in RG and IS (both
Methylotrophs (as a subgroup of heterotrophs) constituted between 30 and 40% of the relative abundances calculated within the top OTUs at RG, though only about 5–10% for IS and LG. This comparatively high abundance of methylotrophs at RG further suggests that special conditions occur here which are suitable for this particular metabolism type (
Collectively, the lack of temporal trends in metabolism types at LG potentially indicates both a lack of connectivity with the surrounding proglacial catchment area, but also a relatively consistent range of habitats being drained by this river over the course of the study period. Overall, the limited effect of seasonality among the different metabolism types observed from these catchments might be due to the origin of the microbes collected from the water column. Since microbes are primarily flushed from upstream environments, temporal patterns in this study are more reflective of seasonal shifts taking place in these upstream habitats than within the microbial communities growing directly at the sampling site. In addition, the seasonal changes may reflect evolution of the drainage topology and advancing sediment erosion rather than temporal changes of microbial communities. With this in mind, greater metabolic shifts might be observed if sampling was conducted over a wider temporal interval. For example, microbes with much different metabolisms might make up a larger proportion of the assemblage if the sampling were to capture the onset of glacier melt, which would coincide with the drainage of potentially much different habitat types.
We investigated how catchment-level differences influence the structure of suspended microbial assemblages recovered from three rivers draining the GrIS through monitoring. Our results suggest that both glacial and proglacial hydrological processes are likely influential to some degree for the resulting assemblage structure. Specifically, the ratio of glacierized to proglacial catchment coverage and the hydrological connectivity with different habitat types seems to play a bigger role in assemblage structure than the absolute catchment size or discharge. Importantly, the interaction of stream water with corresponding non-glaciated proglacial catchments likely evolve over the course of the summer, leading to substantial changes to assemblage structure over the melt season. Although we argue that our interpretations for the observed patterns make ecological sense, it will be necessary to validate these interpretations with new field efforts to definitively isolate mechanisms. Specifically, further efforts to characterize and quantify microbial contributions from the proglacial watershed is much needed, as are quantitative metabolic profiling of microbial communities derived from these and other cold freshwater environments. Nonetheless, these results are important for hydrologists and biogeochemists using microbial cells as indicators for hydrological flow paths, which are likely to change in these sensitive landscapes with climate warming.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at:
MSt, TK, JH, and JW conceived of the project. TK, PV, MB, GL-G, AH, JH, AK, SH, KC, and MSt collected the samples. KV, TK, PV, LF, and GL-G performed the labwork. KV, LF, MSc, and TK performed the analyses. KV, TK, and MSt wrote the paper along with significant input and editing from all coauthors. All authors contributed to the article and approved the submitted version.
This research was supported by Czech Science Foundation (18-12630S) and the Czech Ministry of Education, Youth, and Sport (ERC CZ LL2004) grants to MS. Greenland fieldwork was additionally supported by a Leverhulme Trust Research Grant (RPG-2016-439) to JLW. TJK was supported by the Charles University project PRIMUS/22/SCI/001 and by the Charles University Research Centre program no. 204069. ADH was supported by Monica Cole and Gino Watkins Memorial grants from the Royal Geographical Society and the Scott Polar Institute, respectively.
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.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
We thank the Kangerlussuaq International Science Station and Air Greenland for support with field logistics. We would also like to acknowledge two reviewers whose comments greatly improved the manuscript.
The Supplementary material for this article can be found online at: