In total we used 42 retained winter wheat (T. aestivum, cultivar JB Asano, comprising 21 husk and 21 corresponding straw samples) and corresponding 21 soil samples for the current study. Winter wheat samples were taken in 2018 from an ongoing long-term field experiment in NE Germany (Puppe et al., 2021). Plant samples were washed to remove adhering soil minerals, oven-dried at 45°C for 48 h, and subsequently separated into straw, grain, and husk. The different plant materials of winter wheat were separately homogenized using a knife mill (Grindomix GM 200, Retsch) in two steps: i) for 1 min at 4,000 rpm and ii) for 3 min at 10,000 rpm. As grain samples showed only very low total Si contents (mean 0.2–0.4 mg g^-1), we used solely straw (mean total Si 6.3–10.7 mg g^-1) and husk (mean total Si 9.5–23.2 mg g^-1) samples for our method comparison (total Si concentrations obtained from HF digestion, see Puppe et al., 2021). Soil samples were also taken in 2018 at the identical plots of plant sampling (see Puppe et al., 2021 for details) and were air dried and sieved (2 mm) previous to the extraction of plant available Si.

HF digestion was performed in the course of a previous study and the corresponding results can be found in Puppe et al. (2021). Plant sample aliquots of approximately 100 mg were digested under pressure in PFA (Perfluoroalkoxy alkane) digestion vessels using a mixture of 4 mL distilled water, 5 mL nitric acid (65%), and 1 mL HF acid (40%) at 190°C in a microwave digestion system (Mars 6, CEM). A second digestion step was used to neutralize HF acid with 10 mL of a 4% boric acid solution at 150°C. We used the results of HF digestion as a reference (= total Si in a plant sample) for the evaluation of the efficacy of the two alkaline (i.e., Na₂CO₃ and Tiron) Si extraction methods tested in our study.

For Na₂CO₃ extraction 30 mg of plant material were mixed with 30 mL of Na₂CO₃ (1%) and heated at 85°C for 5 h (DeMaster, 1981). Samples were gently shaken by hand directly before heating and subsequently every hour while heating, i.e., after 1, 2, 3, and 4 h in the water bath. Finally, the extracts were filtrated at a pore-size of 0.2 µm (Whatman NL 16).

The Tiron extraction followed the method developed by Biermans and Baert (1977) and modified by Kodama and Ross (1991). It has been used to quantify amorphous biogenic and pedogenic Si (Kendrick & Graham, 2004), although a partial dissolution of primary minerals is well known (Kodama & Ross, 1991; Sauer et al., 2006). The extraction solution was produced by dilution of 31.42 g Tiron with 800 mL of distilled water, followed by addition of 100 mL sodium carbonate solution (5.3 g Na₂CO₃ plus 100 mL distilled water) under constant stirring. The final pH of 10.5 was reached by adding small volumes of a 4 M NaOH solution. For the extraction, 30 mg of plant samples were weighed into 50 mL centrifuge tubes and a 30 mL aliquot of the Tiron solution was added. The tubes were then heated at 85°C in a water bath for 1 h. The samples were gently shaken by hand directly before heating and after 30 min in the heated water bath. Finally, the extracted solutions were centrifuged at 4,000 rpm for 30 min and filtrated (0.45 μm polyamide membrane filters, Whatman NL 17).

Plant available Si in soil samples (upper 25 cm, Ap horizon) was extracted in the course of the study of Puppe et al. (2021) following the procedure described by Haysom and Chapman (1975); de Lima Rodrigues et al. (2003). Two-gram samples of soil were placed in 50 mL plastic centrifuge tubes, mixed with 20 mL of a 0.01 M calcium chloride (CaCl₂) solution, and agitated continuously on a swivel roller mixer for 16 h. Finally, the extracted solutions were centrifuged at 4,000 rpm for 30 min and filtrated using 0.45 μm polyamide membrane filters (Whatman NL 17).

Si concentrations in the HF acid, Na₂CO₃, Tiron, and CaCl₂ extracts were measured at the ZALF Central Laboratory via ICP-OES (ICP-iCAP 6300 Duo, Thermo Fisher Scientific Inc.) using internal calibration standards made from a certified reference material (i.e., Certipur Si ICP Standard, Merck). Accuracy and long-term repeatability of Si concentration measurements via ICP-OES were systematically monitored. The detection limit was 0.6 × 10⁻³ ppb, the analytical measurement precision was ±1.3%. All analyses were performed in two lab replicates and three single ICP-OES measurements were performed per replicate resulting in six single values (n = 6) for every tested sample. Blank samples (one blank sample per 20 plant/soil samples) were used to analyze Si concentrations in the extraction/digestion chemicals. Blank sample Si concentrations were subtracted from corresponding plant/soil sample Si concentrations before further calculations. To avoid any potential Si contamination only plastic equipment was used during the entire laboratory work.

Phytoliths were extracted from husk and straw samples by ashing following a protocol modified from Puppe and Leue (2018): i) 2.5 g of dry plant material was weighed, ii) the plant samples were heated in a muffle furnace for 6 h at 450°C to combust organic matter, iii) residual organic matter was oxidized using H₂O₂ (30%) and HNO₃ (65%) at 80°C until the reaction subsided, iv) carbonates were dissolved by the use of hot HCl (10%, 30 min, 80°C), and v) the obtained phytoliths were washed with distilled water, dried at 105°C, and weighed.

Linear and monotonic relationships in the data set were analyzed via Pearson’s (r) and Spearman’s rank (r _s ) correlations (α level of 0.05), respectively, using the software package SPSS Statistics (version 22.0.0.0, IBM Corp.).

In general, our findings clearly show that alkaline extractions with Tiron are more efficient to determine Si contents in plants compared to Na₂CO₃. This corroborates the results of previous studies. Guntzer et al. (2010), for example, performed Tiron extractions using plant samples of wheat (Triticum durum), horsetail (Equisetum arvense), fern (Dicksonia squarrosa), elm (Ulmus laevis), and larch (Larix gmelinii) and compared their results to the corresponding results of electrothermal vaporization and lithium-metaborate digestion, which were used as reference methods to determine total Si contents. They found the results of electrothermal vaporization to be about 1.3 times higher than the ones obtained from Tiron extraction. This is generally comparable to the results of our study. We found that the results of HF digestion (used as reference for total Si contents) were about 1.04 (husk samples) and 1.4 (straw samples) times higher than the corresponding ones of Tiron extraction. In relation to the results of Na₂CO₃ extractions the corresponding results of HF digestion were about 1.4 and 1.8 times higher for husk and straw samples, respectively. The results of Tiron extraction were about 1.3 (husk) and 1.2 (straw) times higher than the corresponding ones of Na₂CO₃ extraction. How can these discrepancies between the results [(i) alkaline extractions vs. HF digestion, (ii) Na₂CO₃ vs. Tiron extraction, and (iii) husk vs. straw samples] be explained?

The discrepancies between single-step alkaline extractions and multistep HF digestions can be explained best by differences in the methods itself and the corresponding extractant strength. While the HF acid microwave digestion with its strong acids and relatively high temperatures can completely decompose siliceous (including crystalline silica such as quartz) and organic materials, alkaline extractions are comparatively weak. Na₂CO₃ is assumed to be unable to dissolve crystalline silica quickly, and thus a short-time Na₂CO₃ extraction is used for the specific extraction of amorphous silica (Meunier et al., 2014; Clymans et al., 2016; Schaller et al., 2022). Nakamura et al. (2020) compared the Na₂CO₃ method with the borate fusion method (which was used as reference) and found that the Na₂CO₃ method resulted in 16%–32% lower Si contents than the borate fusion method in five plant species. Based on their results these authors recommended to apply a correction factor to calculate ‘total’ Si contents of plants. However, ideally, this correction factor needs to be considered including a larger number of plant species. Moreover, the findings of Nakamura et al. (2020) are in line with our finding that extraction with Na₂CO₃ systematically underestimates Si contents in plant materials. Contrary, Tiron is assumed to dissolve noncrystalline (amorphous) silica quicker than Na₂CO₃ and to slightly attack crystalline silica (see Kodama & Ross, 1991). Thus, differences in the Si contents might also reflect a certain amount of amorphous silica with a very high condensation state in the plant samples, which has been observed in several previous studies (e.g., Dietrich et al., 2003; Schulz-Kornas et al., 2017).

Moreover, differences in phytolith composition likely influence extraction efficacy. Several studies have shown that phytolith surfaces are often associated with cell wall components like (hemi-) cellulose, cutin, or lignin (e.g., Law & Exley, 2011; Soukup et al., 2017; Zancajo et al., 2019). Additionally, organic compounds like proteins or glycoproteins can be found occluded in phytoliths (e.g., Elbaum et al., 2009; Alexandre et al., 2015; Kumar et al., 2020). These differences in phytolith composition are directly linked to the origin of phytoliths: While cell wall phytoliths are associated with a carbohydrate matrix, lumen phytoliths seem to contain more proteins and glycoproteins than cell wall phytoliths with consequences for phytolith dissolution kinetics and carbon sequestration, which is still discussed controversially [see the authoritative reviews of Hodson (2016; 2019) and references therein]. In a previous study Puppe et al. (2022b) analyzed leaf samples from harvest-fresh specimens of the identical winter wheat plants we used in the current study and found that about 62% of winter wheat phytoliths were lumen phytoliths, while 38% were cell wall phytoliths dividable equally in recognizable (in terms of phytolith morphology) cell wall phytoliths (i.e., short cells) and silicified tissue fragments. It can be assumed that the proportion of cell wall and lumen phytoliths in turn is directly linked to Si extraction efficacy as cell wall phytoliths are potentially more stable than lumen phytoliths. Our results on hand indicate that the proportion of cell wall and lumen phytoliths might differ between straw and husk samples resulting in differences of Si extraction efficacy. However, this has to be confirmed in future examinations. In this context, phytolith size and condensation state (potentially influenced by the location of phytolith formation and water evaporation, cf. Kumar et al., 2017) have to be considered as well (Schaller et al., 2021). In a further step we will analyze the morphology and elemental composition of phytoliths extracted from straw and husk samples to underline our assumption (cf. Puppe et al., 2022b).

In general, plant-Si-soil-Si relationships are mainly related to plant-specific Si uptake rates and Si availability in soils, factors that can be highly variable. This is why different studies often show inconsistent results. Our results, for example, show that i) phytolith contents in plant samples are not correlated to Si availability in soils at all (discussed in the last paragraph of this subsection) and ii) there are only weak relationships between Si contents in plant samples and plant available Si in corresponding soil samples. In fact, we only found statistically significant (moderate) correlations for husk samples using Spearman’s rank correlation (see Table 1). For straw samples, which showed lower Si contents than husk samples, no statistically significant correlations were detectable at all in our study. Contrary, Wu et al. (2020) found rice straw to better reflect plant available Si concentrations in soils, although rice straw also showed lower Si contents compared to rice husk. Wu and colleagues (2020) assumed the lower Si contents in rice straw to be less subject to variation—caused by fluctuations in concentrations of plant available Si in soils—than the higher Si contents in rice husk. Altogether, these results point to the need of further systematic analyses of plant-specific Si accumulation in different plant organs/materials. In this context, dynamics in plant Si uptake can be covered by analyses of plant samples taken at different plant growth stages (e.g., Schaller et al., 2022).

Regarding Si availability in soils our results partly (straw samples) corroborate the findings of Keeping (2017), who found that the uptake of Si by sugarcane in a shade house pot experiment did neither reflect the concentration of plant available Si in soils nor the Si content of used Si sources (calcium silicate slag, fused magnesium (thermo) phosphate, volcanic rock dust, magnesium silicate, and granular potassium silicate). This is in contrast to the studies of Miles et al. (2014) and Korndörfer et al. (2001), who found close correlations between plant available Si concentrations in soils and corresponding Si contents in sugarcane leaves and rice straw, respectively. These discrepancies between different studies can be explained by i) differences in the study design and ii) the fact that there is no standard extraction method for the determination of Si availability in soils yet. While studies that cover different soil types with widely varying physicochemical properties (e.g., pH, texture, or adsorption capacity) report close relationships between plant Si contents and Si concentrations in soils (e.g., Korndörfer et al., 2001; Miles et al., 2014), studies limited on few or only one study site can show ambiguous results (e.g., Klotzbücher et al., 2017; this study). Regarding Si availability in soils different procedures have been developed for specific plants grown under specific climates, i.e., mainly sugarcane and rice in (sub)tropical zones. In this context, potential correlations between plant available Si concentrations in soils and Si contents in plant materials are not only depending on soil texture, but also on the used extractant (cf. Crusciol et al., 2018) and how this extractant predicts Si availability in soils under a specific crop (potential mobilization of Si by, e.g., root exudates, see Wu et al., 2020). In future studies, measurements of the dynamics of plant available Si concentrations in soils will provide deeper insights into the complex interactions in the soil environment controlling Si availability in soils and corresponding Si uptake by plants (cf. de Tombeur et al., 2021). This knowledge is crucial to better assess (seasonal) human impacts (e.g., harvesting, fertilization, liming) on Si budgets of agricultural plant-soil systems and to implement sustainable strategies (e.g., straw recycling) in existing farming systems to prevent Si losses. In fact, a Si-based sustainable agriculture might be a promising approach to combine the beneficial effects of Si on plant performance with the demand for food security under global change (cf. Keeping and Reynolds, 2009; Van Bockhaven et al., 2013).

The detection of relationships between phytolith contents in plant samples and concentrations of plant available Si in corresponding soil samples is hampered by the fact that Si in plants is not only represented by extractable phytoliths, but also by fragile silica structures (Meunier et al., 2017; Puppe et al., 2017), which are not covered by phytolith extraction techniques like ashing performed in this study. As the majority of phytogenic silica might be stored in these fragile structures, current phytolith extraction techniques (ashing, acid digestion) thus might strongly underestimate the ‘total’ Si content of plants (cf. Puppe et al., 2017). Moreover, such methodological shortcomings might hamper an appropriate interpretation of the role of phytoliths in plant-soil systems (Kaczorek et al., 2019). Thus, we also need detailed analyses of phytogenic silica (extractable phytoliths vs. fragile structures) in different plant species as well as plant organs (e.g., leaves, stems, roots), which will help us to unravel the complex relationships between phytogenic silica in plant/soil samples and Si availability in soils.