Nitrogen availability and plant–plant interactions drive leaf silicon concentration in wheat genotypes

Abstract Estimating plasticity of leaf silicon (Si) in response to abiotic and biotic factors underpins our comprehension of plant defences and stress resistance in natural and agroecosystems. However, how nitrogen (N) addition and intraspecific plant–plant interactions affect Si concentration remains unclear. We grew 19 durum wheat genotypes (Triticum turgidum ssp. durum) in pots, either alone or in intra‐ or intergenotypic cultures of two individuals, and with or without N. Above‐ground biomass, plant height and leaf [Si] were quantified at the beginning of the flowering stage. Nitrogen addition decreased leaf [Si] for most genotypes, proportionally to the biomass increase. Si plasticity to plant–plant interactions varied significantly among genotypes, with both increases and decreases in leaf [Si] when mixed with a neighbour, regardless of the mixture type (intra‐/intergenotype). Besides, increased leaf [Si] in response to plant–plant interactions was associated with increased plant height. Our results suggest the occurrence of both facilitation and competition for Si uptake from the rhizosphere in wheat mixtures. Future research should identify which leaf and root traits characterise facilitating neighbours for Si acquisition. We also show that Si could be involved in height gain in response to intraspecific competition, possibly for increasing light capture. This important finding opens up new research directions on Si and plant–plant interactions in both natural ecosystems and agroecosystems. More generally, our results stress the need to explore leaf Si plasticity in responses to both abiotic and biotic factors to understand plant stress resistance. Read the free Plain Language Summary for this article on the Journal blog.

The essentiality of Si for plants remains challenging to assess (Coskun et al., 2019;Epstein, 1994) but increased resistance to herbivores and stress alleviation following Si fertilisation can lead to increased plant primary productivity and crop yields (Liang et al., 2015;Savant et al., 1999;Tubana et al., 2016;Xu et al., 2020). Because graminoid crop species can exhibit very high Si concentrations ([Si]) (e.g. up to 20% of SiO 2 in rice; Klotzbücher et al., 2018), the beneficial role of Si in agriculture is well recognised, and Si is routinely applied to croplands in many countries (e.g. China, Japan, USA, Brazil) (Datnoff et al., 2001;Yan et al., 2018). Thus, it is important to understand the factors affecting plant Si nutrition but, to date, we still have limited knowledge of how soil nutrient availability and interactions between plants affect Si concentration.
Increasing evidence suggests that plant Si concentration depends on soil nutrient status (de Tombeur, Laliberté, et al., 2021;Johnson et al., 2021;Minden et al., 2021;Quigley et al., 2020). In particular, decreases in Si concentration and resulting Si-based defences following nitrogen (N) fertilisation have recently been reported for different grassland/pasture species (Johnson et al., 2021;Minden et al., 2021;Quigley et al., 2020) (but see Moise et al., 2019). This has been attributed to the investment in 'cheap' Si versus relatively 'more expensive' carbon (C) (Raven, 1983) during N stress (Johnson et al., 2021;Minden et al., 2021) and reflects trade-offs between plant growth rate and carbon-or Si-based defences within Poaceae family (Massey et al., 2007). However, past studies have generally focused only on a single, non-cultivated genotype (Johnson et al., 2021;Minden et al., 2021). Significant genotypic variation in Si concentration has been reported in rice and wheat (Ma et al., 2007;Merah et al., 1999;Talukdar et al., 2019), so the plasticity (i.e. production of multiple phenotypes from a single genotype depending on environmental conditions; Miner et al., 2005) of leaf [Si] in response to N fertilisation might differ among genotypes, but this has not yet been tested.
So far, the influence of plant-plant interactions on plant Si nutrition has received surprisingly little attention in the literature (but see Garbuzov et al., 2011;Ning et al., 2017Ning et al., , 2021, especially compared with other nutrients (Li et al., 2014). At the interspecific level, Ning et al. (2021) showed that rice accumulates significantly more Si when grown with water spinach (Ipomoea aquatica Forsk)-a low Si-accumulating species-compared with a rice monoculture, possibly through the effect of root exudates on soil Si mobilisation (de Tombeur, Ning et al., 2021). However, when two grasses with high Si-concentration (Poa annua and Lolium perenne) were investigated, such interspecific facilitation on Si concentration was not observed (Garbuzov et al., 2011). The influence of plant-plant interactions on Si concentration at the intraspecific level, to our knowledge, has received no attention, either in intragenotypic cultures or intergenotypic mixtures. It is important to consider both intra-and intergenotypic cultures because facilitation for Si uptake in the rhizosphere might prevail over competition when genotypes are functionally different (e.g. they contrast in nutrientacquisition strategies and/or Si demand). Furthermore, both types of genotypic cultures should be considered because intragenotypic stands are typical of modern agriculture, but there is increasing interest in the role of genetic diversity in increasing the sustainability of agriculture as greater intraspecific diversity may increase productivity and resistance to pests and pathogens (Barot et al., 2017;Hajjar et al., 2008;Litrico & Violle, 2015;Montazeaud et al., 2022).
Finally, leaf Si has been linked to different plant architecture traits that could in turn influence competition for light capture, including decreasing leaf insertion angle and leaf arc/straightness (Ando et al., 2002;de Tombeur, Cooke, et al., 2021;Yamamoto et al., 2012;Zanão Júnior et al., 2010), and increasing plant height (Gong et al., 2003;Ma et al., 1989;Zanão Júnior et al., 2010). As such, we might expect some relationships between the Si concentration of a genotype and the outcomes of plant-plant interactions (i.e. in this case, biomass loss or gain when mixed with a neighbour).
It remains challenging to predict potential links between Si and competition outcomes, since greater plant height might increase competition intensity (Falster & Westoby, 2003;Violle et al., 2009), but decreasing leaf insertion angle and arc reduces the light extinction coefficient inside the canopy and may thus decrease competition intensity (Ando et al., 2002). Nevertheless, studies on Si benefits against biotic and abiotic stresses have greatly expanded during the last 10 years (Coskun et al., 2019), and investigating previously overlooked functions of silicification, such as its influence on plant architecture and potential impact on plant-plant interactions, is thus needed.
Here, we studied 19 genotypes of durum wheat (Triticum turgidum ssp. durum), a major staple crop, which we grew in pots, either alone, in intragenotypic culture or in intergenotypic culture, at two levels of N availability. We quantified plant above-ground biomass,  (David et al., 2014). The 19 genotypes represented a large phenotypic diversity on below-and above-ground traits.
The 19 genotypes were grown either alone in single (alone in the pot), in intragenotypic culture (two plants of the same genotype in the same pot) or in intergenotypic culture (two plants from different genotypes in the same pot), hereafter growth modalities, with two levels of N (treatment N + and N − ), and in triplicate. We randomly assembled 26 intergenotypic mixtures among the 171 possibilities. The modality single thus represents 114 individuals (19 genotypes × 2 N levels × 3 replicates), intragenotypic culture 228 individuals (2 plants × 19 genotypes × 2 N levels × 3 replicates) and intergenotypic culture 312 individuals (26 mixtures of 2 plants × 2 N levels × 3 replicates). In total, 384 pots and 654 wheat individuals were considered.

| Growth conditions
The experiment was conducted at the CEFE experimental field (Montpellier, France) from January to May 2021, in outdoor conditions. We used a randomised complete block design using three blocks (one replicate in each block). Plants were grown in 4-L plastic pots (18.5 cm diameter; 21.5 cm depth) filled with approximately 4.5 kg of local soil (52% sand, 27% silt and 21% clay; 6.9% CaCO 3 ; 4.1% organic carbon; 0.21% total N; pH 8.0), and amended with PK fertiliser (0.38 g per pot; P 2 O 5 and K 2 O). The effect of plant-plant interactions on plant Si uptake might be influenced by soil Si availability (Ning et al., 2021). Here, although not quantified, we expect Si availability to be rather high in this young, high-pH and clay + silt-rich soil (Cornelis & Delvaux, 2016). Indeed, a recent analysis of soil Si availability in French soils shows that this soil type exhibits the highest Si concentrations extracted with CaCl 2 and is unlikely to be Si limited (Caubet et al., 2020). Two seeds per plant were sown in each pot and the largest plant was kept after germination. Pots of the N + treatment received N four times during the experiment, for a total input of 0.94 g N per pot, whereas pots of the N − treatment did not receive any N fertilisation. Plants were not protected from the rain and were watered with amounts to avoid water excess or deficit.

| Plant height, biomass and leaf [Si] measurements
Vegetative plant height, plant above-ground biomass and leaf [Si] were quantified at the beginning of the flowering stage.
Vegetative height was measured as the distance between the soil surface and the tallest leaf without stretching the plant leaf. The leaf [Si] was quantified with an X-ray fluorescence spectrometer (Reidinger et al., 2012). Briefly, three most recent ligulate adult leaves were sampled on each individual, dried at 60°C for 72 h and ball-milled (Retsch MM400 Mixer mill) for 3 min at a frequency of 20 Hz. Ground samples were pressed at 10 tons into pellets using a manual hydraulic press (Specac). Si analyses were performed using a Nitron XL3t900 GOLDD XRF analyser (Thermo Scientific).
Silicon-spiked synthetic cellulose was used for calibration, and analyses were performed under helium atmosphere to avoid signal loss by air absorption (Reidinger et al., 2012). A reading was taken of each side of the pellet, approximately 1 h apart, to account for u-drift in the instrument (Johnson, 2014). The concentration of Si in these three most recent ligulate adult leaves (in % of Si by dry weight) was considered to capture the intraspecific variation in leaf [Si] among the genotypes, the response to N fertilisation and plant-plant interactions, and potential relations between leaf [Si] and competition outcomes. Finally, all plant materials were harvested, dried at 60°C for 72 h and weighed to obtain aboveground biomass.

| Variation in leaf [Si] among genotypes and response to N fertilisation
Variation in leaf [Si] among the 19 wheat genotypes and their plasticity to N fertilisation were assessed only for the single plants to discriminate it from the neighbour effect. For both N treatments, differences in leaf Si across the 19 genotypes were tested by a oneway analysis of variance (ANOVA). To quantify the plasticity of leaf [Si] in response to N fertilisation among the 19 genotypes, we calculated log response-ratios (hereafter logRR) as the logarithm of ratios between individual trait values and corresponding genotype-mean values in N − , as follows: Differences in logRR among genotypes were tested by ANOVA, and genotype-mean logRR significantly different from zero were assessed with Student's t-tests. A logRR below zero means that the treatment significantly decreased the trait values, while the opposite is true for logRR above zero.

| Plasticity to plant-plant interactions
We first tested differences in leaf [Si] among the treatments single, intra-and intergenotypic cultures by ANOVA followed by post hoc tests using the 'multcomp' package (Hothorn et al., 2008) for both  (Zuur et al., 2009). All analyses were conducted in the R environment (R Core Team, 2021).

| Intraspecific variation in leaf [Si] and plasticity to N fertilisation
Without N fertilisation, genotype-mean leaf [Si] ranged from 1.0% to 2.9%, but did not significantly differ among genotypes (p = 0.09, Figure 1a). N fertilisation resulted in an overall decrease in leaf [Si] of 42%, with genotype-mean ranging from 0.7% to 1.9% and that differed significantly among genotypes (Figure 1a). The response of leaf [Si] to N fertilisation (logRR) varied significantly among genotypes, and N fertilisation significantly decreased leaf [Si] for 12 out of the 19 genotypes (logRR < 0) (Figure 2a).

| Plasticity to plant-plant interactions
We found no overall effect of growth modality (single, intra-or intergenotypic culture) on leaf [Si], whether plants were N-fertilised or not ( Figure 1b). However, plasticity in leaf [Si] in response to plantplant interactions (logRR) varied significantly among genotypes for both N treatments (Figure 2b). The presence of a neighbour significantly decreased leaf [Si] for seven genotypes in the N − and for five genotypes in the N + treatments (logRR < 0), and significantly increased leaf [Si] for three genotypes in the N − and for seven genotypes in the N + treatments (logRR > 0) (Figure 2b).
Genotypes varied significantly in their responses to plantplant interactions also within the intra-and intergenotypic culture treatments and for both N treatments (see Figure S1). Genotypemean responses were not consistent between N treatments

| Plant height, above-ground biomass and responses to N fertilisation and plantplant interactions
Overall, N fertilisation increased plant biomass and height, while the presence of a neighbour decreased biomass for both N treatments but had no significant effect on plant height (Figure 1c,d).
We found a strong significant negative relationship between leaf [Si] and above-ground biomass, but not plant height, for single plants (model including both N treatments) ( Table 1). We found a strong negative relationship between the plasticity (logRR) in leaf [Si] and that of biomass in response to N fertilisation (Figure 4a), suggesting that larger increase in plant biomass following N fertilisation implied a stronger decrease in leaf [Si], and confirming the strong dependency between these two traits when N was manipulated. In contrast, plasticity of plant height to N fertilisation was not related to that of leaf [Si] (Figure 4a).
In the models considering plants with a neighbour, we also observed a slight negative relationship between leaf [Si] and aboveground biomass, but only in the N − treatment ( Table 1). When intra-and intergenotypic culture were considered separately, none of the biomass-Si relationships were significant (Table S1). However, significant positive relationships between plant height and leaf [Si] were identified for both N treatments (Table 1), and within the intraand intergenotypic culture (Table S1).  Figure S3).

F I G U R E 1 Leaf silicon concentrations ([Si]) of 19 durum wheat genotypes grown alone (single) and for two levels of N availability (means ± SE; n = 3) in (a). Boxplots showing the effects of plant growth modalities (single, intra-and intergenotypic culture) on leaf [Si] in (b)
, plant above-ground biomass in (c) and plant height in (d), for each N treatment. In (a), data are ranked by increasing genotype-mean leaf [Si] in the N − treatment for both plots, and results of ANOVA (F-values) conducted between the genotypes are given. In (b)-(d), the central horizontal bar in each box shows the median, the box represents the interquartile range (IQR) and the whiskers show the location of the most extreme data points that are still within a factor of 1.5 of the upper or lower quartiles. Each point indicates one individual, and the y-axis for leaf [Si] in (b) is on a logarithmic scale to improve visualisation. Different letters indicate significant differences (p < 0.05) between single, intra-and intergenotypic culture within an N treatment. ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant.

(a) (b)
F I G U R E 2 Variation in log response ratios (logRR) of leaf silicon (Si) concentrations to nitrogen (N) fertilisation for the single plants in (a) and to plant-plant interactions for both N treatments in (b) among 19 wheat genotypes. Both intra-and intergenotypic culture were considered together in the analysis in (b) (see Figure S1 for separate analyses). Data are ranked by increasing genotype-mean logRR. The central horizontal bar in each box shows the median, the box represents the interquartile range (IQR), the whiskers show the location of the most extreme data points that are still within a factor of 1.5 of the upper or lower quartiles, and black points are values that fall outside the whiskers. Results of ANOVA (F-values) conducted between the genotypes are given. LogRR significantly different from zero following student t-tests are indicated with stars. ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant.

Genotypes
Genotypes Genotypes * ** ** * * * * * * ** * * ****** ** ** ** * * * ** *** ** * ** ** * * * ** ** * ** ****** The strong decrease in Si concentrations following N fertilisation confirms our hypothesis and previous studies using natural grassland/pasture species (Johnson et al., 2021;Massey et al., 2007;Minden et al., 2021;Quigley et al., 2020). The results are also in line with the resource availability hypothesis, which proposes higher levels of defence in resource-limited environments (Coley et al., 1985;Endara & Coley, 2011). Since Si is thought to incur lower C costs than C-based structural/defensive compounds (Raven, 1983), this might also reflect a selective advantage of plants reducing leaf construction/defence costs when resources are limiting (Minden et al., 2021), but the underlying mechanism remains unclear (Hodson & Guppy, 2022). Although N deficiency might directly increase the expression of Si transporters (Wu et al., 2017), our results suggest a N-driven 'dilution effect' on leaf [Si] (Hodson & Guppy, 2022;Jarrell & Beverly, 1981) since we found a strong negative relation between biomass and leaf [Si]. This likely explains the strong negative relationship between the plasticity of biomass and that of leaf [Si] to N fertilisation. The significant interactions between wheat Si concentrations, total above-ground biomass and responses to N fertilisation stress the need to combine data on total Si content and total dry matter content, wherever possible (Jarrell & Beverly, 1981).

F I G U R E 3
Variation in log response ratios (logRR) of leaf silicon (Si) concentrations to intergenotypic culture for both N treatments as a function of neighbour identity. Data are ranked by increasing neighbour identity-mean logRR for both plots. The central horizontal bar in each box shows the median, the box represents the interquartile range (IQR), the whiskers show the location of the most extreme data points that are still within a factor of 1.5 of the upper or lower quartiles, and black points are values that fall outside the whiskers. Results of ANOVA (F-values) conducted between the neighbour identity are given. LogRR significantly different from zero following student t-tests are indicated with stars for the N − treatment. ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant.

TA B L E 1
Results of the mixed-effect models (genotype as random factor) testing the effects of leaf silicon (Si) concentrations on aboveground plant biomass and height for single plants (both nitrogen (N) levels in the analyses), and for plants in interactions for both N levels separately (intra-and intergenotypic treatments combined; see Table S1 for separated analyses) and/or nutrient-acquisition strategies; Ning et al., 2021). Our results suggest that both competition and facilitation for Si uptake might exist at the intraspecific level with durum wheat genotypes.

Inter-and intragenotypic culture
Our comprehension of root-related processes influencing Si mobilisation in the rhizosphere is still limited, despite some progress in recent years (de Tombeur, Frew et al., 2017;Gattullo et al., 2016). Grasses release siderophores (i.e. lowmolecular weight chelators) in the soil solution to acquire limited nutrients (Ma, 2005;Oburger et al., 2014;Römheld, 1991), which also increase Si availability (Gattullo et al., 2016). This mechanism could possibly explain the increases of leaf  Cooke, et al., 2021;Yamamoto et al., 2012;Zanão Júnior et al., 2010). However, despite a slightly positive relationship between genotype-mean leaf [Si] in single and the response of aboveground biomass to competition in the N − treatment, genotype leaf [Si] did not appear to play a major role in intra-or intergenotypic competition outcomes. Nevertheless, increased leaf [Si] in response to competition was associated with increased plant height, and this was the case for both N and mixture treatments. Si might play an indirect role in intraspecific competition through its influence on plant height, given that this trait is often associated with a strong competitive ability in wheat (Thomas et al., 1993;Yenish & Young, 2004) and more generally with light capture (Falster &

F I G U R E 4
Relationships between the log response ratio (logRR) of leaf silicon (Si) concentrations and those of biomass and height to nitrogen (N) fertilisation for the single in (a) and to plant-plant interactions for both N treatments in (b). Both intra-and intergenotypic culture were considered together as 'plant-plant interactions' in the analyses (see Figure S3 for separate analyses). Red lines indicate regression lines between variables, and multiple R-squared are given. ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant.  Westoby, 2003;Violle et al., 2009). Height gain following Si fertilisation is, however, also associated with straighter leaves with lower leaf insertion angle (Zanão Júnior et al., 2010), which might in turn reduce the light extinction coefficient inside the canopy (Ando et al., 2002). In any case, this finding opens up new research directions on Si and plant-plant interactions in both natural and agroecosystems which remain strikingly scarce to date (but see Garbuzov et al., 2011;Ning et al., 2017Ning et al., , 2021.
Several perspectives arise from the results discussed above.

CO N FLI C T O F I NTE R E S T
C.V. is an Associate Editor of Functional Ecology but took no part in the peer review and decision-making processes for this paper. The authors declare that they have no competing interests.