TY - JOUR
AU1 - Buchmann,, Tina
AU2 - Schumacher,, Jens
AU3 - Roscher,, Christiane
AB - Abstract Aims Studies along environmental gradients have shown that intraspecific trait variation (ITV) may contribute considerably to community-level trait variation. However, we lack knowledge about how the extent of ITV varies on a local scale and whether a varying extent of ITV is related to differences in local environmental site and plant community characteristics. Methods We investigated plant height, specific leaf area (SLA), leaf dry matter content (LDMC) and leaf greenness of three common grass species (Arrhenatherum elatius, Dactylis glomerata, Poa pratensis) in 12 mown grasslands in a local study area around Jena (Thuringia, Germany) across three spatially hierarchical sampling levels: between sites, between subplots within site and within subplots. Important Findings Arrhenatherum elatius and D. glomerata had higher means in plant height and a lower variation in leaf traits than P. pratensis. The major proportions of variation in leaf traits of P. pratensis and D. glomerata were found within subplots, while the traits of A. elatius varied mainly between sites. Trait correlations across the hierarchical sampling levels were highly consistent in A. elatius, but more variable in D. glomerata and P. pratensis. Environmental site and plant community characteristics mostly explained a larger proportion of variation in trait means in A. elatius than in D. glomerata and P. pratensis, while metrics of ITV were generally less predictable. Our results suggest that trait variation in P. pratensis and D. glomerata is more strongly related to within-site conditions (i.e. biotic interactions), while differences in local environmental conditions between sites have a strong impact on trait variation in A. elatius. Since our study was limited to three grass species, further studies with a greater number of species are required to make generalizations about the importance of biotic interactions and environmental conditions as drivers of ITV at local scale. grass species, intraspecific trait variation, leaf traits INTRODUCTION During recent years, the research focus in community ecology shifted from a species-based to a more trait-based view assuming that functional characteristics of species are the base to understand the mechanisms underlying community assembly (McGill et al. 2006). In this context, functional traits are morphological, physiological or phenological characteristics measurable at the plant individual level that affect species performance and fitness (Violle et al. 2007). Trait-based analyses of plant communities often focused mainly on mean trait values of species and interspecific trait differences. Indeed, interspecific trait variation represented a major proportion of community-level trait variation in many studies (Hulshof et al. 2010; Kichenin et al. 2013) supporting the assumption that intraspecific trait variation (ITV) is negligible. However, ignoring ITV may underestimate species’ abilities to adjust to changing environmental conditions or lead to misinterpretation of the mechanisms underlying community assembly (Violle et al. 2012). ITV due to genetic variation or phenotypic plasticity can contribute considerably to community-level trait variation (Albert et al. 2010a;,Jung et al. 2010;,Siefert 2012). Only recently, results of a meta-analysis on global scale determining the relative extent of ITV in plant communities across different traits and ecosystems showed that the relative importance of ITV increases with decreasing spatial scale emphasizing the need to incorporate ITV especially in studies on local scale and with short environmental gradients (Siefert et al. 2015). Species exist only under those environmental conditions which they are able to tolerate (environmental filtering) and this may lead to a convergence of possible trait values within a community (Grime 2006). Competition among coexisting species can be reduced due to a limited trait similarity between them (niche differentiation) causing a divergence of possible trait values in a community (Cornwell and Ackerly 2009; MacArthur and Levins 1967). However, it is also possible that competition may increase trait similarity among species that minimize competitive inequalities (Hille Ris Lambers et al. 2012). Community assembly processes such as environmental filtering or niche differentiation are barely directly measurable, but trait-based approaches are based on the assumption that assembly processes are expressed in predictable patterns of trait distribution (Cornwell and Ackerly 2009; Siefert 2012). Trait convergence can be detected by a reduced range and variance of trait values, whereas trait divergence would be expressed by an even spacing of trait values (Kraft and Ackerly 2010; Siefert 2012). According to the niche-based theory stating that functional differences among species promote species coexistence, ITV could be expected to decrease with increasing species richness of plant communities to reduce niche overlap among species (Violle et al. 2012). In line with this expectation, Siefert et al. (2015) found negative relationships between species richness and ITV, while other studies also observed positive relationships between species richness and ITV (Le Bagousse-Pinguet et al. 2014). Several studies revealed that the amount of ITV may differ among traits (Auger and Shipley 2013; Jung et al. 2010; Messier et al. 2010; Siefert et al. 2015). It is also possible that correlations or physiological or evolutionary trade-offs among traits constrain trait variation (Boucher et al. 2013; Valladares et al. 2007). Moreover, not all species and all traits exhibit the same amount of ITV (Albert et al. 2010b; Siebenkäs et al. 2015; Siefert et al. 2015). It is commonly assumed that species with a more variable trait expression should have a greater capacity to adjust to varying environmental conditions and match optimal trait values (Mitchell et al. 2017), but it is also possible that a low variation in observed trait values indicates that species may perform well under varying environmental conditions. The relative importance of ITV can differ among spatial scales because ITV is influenced by different mechanisms (Albert et al. 2010b; Lajoie and Vellend 2015; Messier et al. 2010). The signature of biotic interactions is more likely to be detectable at small spatial scales (e.g. within site), whereas environmental conditions are more important at larger spatial scales (e.g. between sites). Incorporating spatially hierarchical sampling levels in exploring the extent of ITV is therefore important to identify the different potential mechanisms affecting ITV. Easily measurable aboveground key plant traits are plant height, specific leaf area (SLA), leaf dry matter content (LDMC) and leaf greenness as a surrogate for leaf nitrogen concentration. These traits are expected to reflect species’ competitive and resource capture abilities, but are also known to respond highly plastically to variation in environmental conditions such as light and nutrient availability (e.g. Al Haj Khaled et al. 2005; Hodgson et al. 2011; Siebenkäs et al. 2015) and plant community diversity (e.g. Gubsch et al. 2011; Roscher et al. 2011). Plant height corresponds to species’ competitive ability in light acquisition (Cornelissen et al. 2003). Competition for light is asymmetric due to the unidirectional supply of solar radiation, i.e. taller individuals intercept disproportionally more light than smaller species (Weiner 1990). SLA and LDMC are important components of the leaf economic spectrum and are related to species’ resource use strategies (Reich et al. 2003; Wright et al. 2004). Species with fast tissue turnover and high relative growth rates are expected to have high SLA (low LDMC) according to their high resource capture abilities. On the opposite, species with slower tissue turnover and low relative growth rates have low SLA (high LDMC) due to high resource conservation (Reich et al. 2003). Leaf greenness estimates the chlorophyll content in a leaf and is related to leaf nitrogen concentration (e.g. Chapman and Barreto 1997; Errecart et al. 2012). Therefore, leaf greenness can also be considered as component of the leaf economic spectrum. In the present study, we investigated variation in key functional traits (i.e. plant height, SLA, LDMC, leaf greenness) of 3 common grass species in 12 semi-natural grasslands in a local study area with a minimum distance of 0.5 km and a maximum distance of 26.6 km between study sites. All studied grasslands were managed by mowing, but differed in the intensities of fertilisation to some degree. The investigated grass species varied in dominance (i.e. proportional contribution to community biomass) and represented one mostly dominant species (Arrhenatherum elatius (L.) P. Beauv. ex J. Presl & C. Presl), and two often subordinate species (Poa pratensis L., Dactylis glomerata L.). Traits were measured comparing three spatially hierarchical sampling levels: between sites, between subplots within sites and within subplots. We asked the following questions: (i) Do trait means and the extent of ITV differ among species and how does it relate to environmental site and plant community characteristics? We expected that trait means and the extent of ITV differ among species due to a limited trait similarity among coexisting species and different strategies of species to deal with varying abiotic and biotic conditions. We also expected that trait means and ITV show significant relationships to environmental site and plant community characteristics reflecting the potential of these common grass species to adjust to local-scale environmental conditions. Specifically, we hypothesized that the extent of ITV decreases with increasing species richness because of a reduced niche overlap among coexisting species. (ii) Do environmental filtering and niche differentiation influence the extent of ITV at the site- and subplot-level? We expected a stronger signal of environmental filtering (reduced range and variance of trait values) at the site-level due to differences in local environmental conditions, whereas signals of niche differentiation (even spacing of trait values) should be stronger at the subplot-level due to a predominant role of biotic interactions. (iii) How does the relative extent of ITV differ across the spatially hierarchical sampling levels? We expected that the relative extent of ITV should be greater among sites than among subplots within sites or within subplots for plant height because asymmetric competition for light strongly depends on environmental site characteristics (i.e. fertilisation, soil nutrients), while ITV at the subplot-level is more important in leaf traits and fine-scale adjustment to growth conditions. (iv) Is there a consistent correlation structure of traits across spatially hierarchical sampling levels? We hypothesized that patterns of trait correlations are different across spatially hierarchical sampling levels, because the influence of biotic interactions should be greatest at the within-subplot level, whereas the influence of environmental site characteristics should be greatest at the between-site level. MATERIALS AND METHODS Study sites and design The study sites were 12 semi-natural mesophilic grasslands (Arrhenatherion type according to Ellenberg 1988) located in the floodplain or adjacent slopes in the valley of the Central Saale and its tributaries around Jena (Thuringia, Germany; see online supplementary Table S1). Sites were chosen to represent a broad range of environmental site conditions for this grassland type in the local study region. Species richness and plant community structure of the studied grassland type is highly dependent on land-use intensity with respect to fertilisation (Socher et al. 2013). Therefore, the selected sites were managed by mowing usually twice per year as it is typical for meadows of the Arrhenatherion type, but differed in the amount of applied fertiliser (see online supplementary Table S1). Minimum distance between study sites was 0.5 km, and the maximum distance was 26.6 km. The region around Jena has a mean annual air temperature of 9.9°C and mean annual precipitation of about 610 mm (1980–2010; Hoffmann et al. 2014). In August 2012, an area of 7.3 × 7.3 m divided into a grid of 36 subplots of 0.8 × 0.8 m separated by 0.5 m distance from each other was established on each site. Vegetation surveys were conducted on each subplot to determine species occurrences and cover using a modified decimal scale (Londo 1976). Based on this survey, six subplots were chosen on each site, which covered the subplot-level variation of species richness, categorized into low, intermediate and high species richness. Each category contained two subplots which were randomly chosen out of the subplots representing the lowest and highest and the mean species richness specified for each site (see online supplementary Table S1). Soil characteristics and potential solar irradiation Five samples of the top soil (0–15 cm depth) were taken with a soil corer (1 cm diameter) on each of the six subplots and pooled per subplot in October 2012. Soil samples were analysed for total nitrogen concentrations as well as plant-available phosphorus and potassium concentrations (for methods of soil analyses see online supplementary Appendix S1). Exposition, inclination and the horizon using a horizontoscope (Tonne 1954) were measured on each site to calculate the mean potential solar irradiation during the growing season from March to October based on an algorithm by Volz (1958). Aboveground plant biomass Aboveground plant biomass was harvested before mowing in May and September 2013. Vegetation was cut 3 cm above ground in a 0.4 × 0.4 m quadrat in the centre of each subplot. Biomass data from both harvests were summed to get annual community biomass production. In May 2013, biomass samples were sorted by species to assess species biomass proportions. Samples were dried at 70°C for 48 h and weighed. Species selection for trait measurements Three plant species present on almost all subplots of each study site were selected for trait measurements. This applied to three common grass species: Arrhenatherum elatius, Dactylis glomerata and Poa pratensis. Arrhenatherum elatius occurred on 11 sites with a mean biomass proportion of 21.14 ± 17.69 % (see online supplementary Fig. S1). Dactylis glomerata and P. pratensis occurred on all sites with mean biomass proportions of 7.85 ± 5.76 % and 6.35 ± 8.70 %, respectively. The three selected grass species differ in their growth form, especially in their clonal growth organs. Arrhenatherum elatius and D. glomerata are tussock grasses forming short rhizomes aboveground, which allow for vegetative spread over short distances only (CLO-PLA data base; Klimešová and de Bello 2009). Poa pratensis is not a tussock grass, but builds fast-growing rhizomes belowground, which allow for vegetative spread over longer distances. In the vegetative stage, A. elatius forms leafy stems (Pfitzenmeyer 1962), whereas D. glomerata and P. pratensis have basally attached leaves. Trait measurements Between 5 and 13 August 2013, plant height and three leaf traits (SLA, LDMC and leaf greenness) were measured. Ten shoots per species were sampled on each subplot. Plant height was measured in the field as the distance between ground level and top of the upper leaf (Cornelissen et al. 2003). Then, the second fully expanded leaf of each shoot was cut and stored separately between moist paper towels to promote rehydration in plastic boxes in a cooler. In total, 692 leaves of D. glomerata, 694 leaves of P. pratensis and 582 leaves of A. elatius were sampled. Leaf samples were kept overnight in a cooling chamber (4°C) under dark conditions before sample processing on the following day. After removing any water droplets from the leaf surface with tissue paper each leaf was weighed immediately to determine water-saturated fresh mass (Wilson et al. 1999). Afterwards, leaf area was measured with a leaf area meter (LI-3000C area meter equipped with LI3050C transparent belt conveyor accessory; LI-COR, Lincoln, USA) and leaf greenness was assessed with a portable chlorophyll meter (SPAD-502 Plus, Minolta Camera Co. Ltd., Osaka, Japan). Leaf greenness estimates the relative proportion of chlorophyll in a leaf, expressed as index from 1 to 100. It was assessed by measuring the absorption of two different wavelengths (650 nm and 940 nm). Measurements of leaf nitrogen concentrations on a subsample of leaves showed a strong positive correlation between LG and leaf nitrogen concentrations in the three studied grass species (Pearson’s correlation coefficients: A. elatius: r = 0.68 (N = 12), D. glomerata: 0.75 (N = 12), P. pratensis: 0.79 (N = 12); P < 0.05). Leaf samples were dried at 70°C for 48 h and weighed to determine leaf dry mass. SLA expressed in mm2 mg−1 was calculated by dividing the area of a fresh leaf by its dry mass. LDMC expressed in mg g−1 is the quotient of leaf dry mass and water-saturated fresh mass. Missing data of trait values in leaf greenness (N = 58) and LDMC (N = 1) were replaced by the median of the site. Statistical analyses Statistical analyses were performed using the software R 3.1.3 (R Core Team 2015). Trait means and extent of ITV (question 1) The mean and three different metrics (coefficient of variation [CV], range, standard deviation of the nearest neighbour distance) were calculated to quantify ITV at the subplot-level. The CV is the standard deviation divided by the mean and estimates the trait variation relative to the mean. Range is the difference between maximum and minimum value of a trait. A restricted range of trait values is regarded as a signal of environmental filtering (Siefert 2012). Standard deviation of the nearest neighbour distance divided by range (SDNDr) describes the evenness of spacing along a trait axis with low values indicating niche differentiation (Kraft and Ackerly 2010). Linear mixed-effects models were applied to test whether species differed with respect to their trait means and ITV metrics (CV, range and SDNDr) using the function lmer in the R package lme4 (Bates et al. 2015). Data were log-transformed to fulfil assumptions of normality. Modelling started with a null model with ‘site’ as random effect. In a second model, it was tested whether species identity as fixed effect improved the model using the maximum likelihood (ML) and likelihood ratio tests for model comparison. A post-hoc test (TukeyHSD) using the R package multcomp (Hothorn et al. 2008) revealed which species differed significantly (using restricting ML [REML]). Linear mixed-effects models were also used to test for significant effects of local environmental site characteristics (fertilisation, soil nitrogen, plant-available soil phosphorus and potassium, potential solar irradiation) and plant community characteristics (species richness, community biomass production, biomass proportions of the study species) on trait means and ITV. Trait means, ranges, SDNDr and all fixed explanatory variables except for fertilisation and potential solar irradiation were log-transformed to fulfil assumptions of normality. First, a minimal adequate model based on subplot-level environmental site characteristics as fixed effects and ‘site’ as random effect was obtained by applying the R package MuMIn (Barton 2015) and the AIC criterion for model selection to account for effects of environmental variation. In a next step, it was tested whether the addition of subplot-level species richness, community biomass production and species biomass proportion as fixed effects led to further significant model improvements. R2 statistics for the final minimal adequate model was calculated by applying likelihood ratio test statistics (Magee 1990) comparing the log-likelihood of the final model with the log-likelihood of the null model (intercept-only model). Null model approach to detect environmental filtering and niche differentiation (question 2) Null models can be applied to detect environmental filtering and niche differentiation by comparing observed trait distributions within communities with trait distributions expected by chance (Weiher and Keddy 1995; Cornwell et al. 2006). Therefore, permutation tests were conducted to estimate whether ITV metrics (CV, range, SDNDr) were significantly different from random expectations at the site- and subplot-level, respectively (for details see online supplementary Appendix S2). Variance decomposition of single traits (question 3) To analyse the relative extent of ITV across hierarchical sampling levels linear mixed-effects models were applied. Trait variance of each species was decomposed according to the hierarchical sampling design into the variance within subplots, between subplots within sites and between sites according to Albert et al. (2010b). Analyses were based on null models (m0) with ‘subplot’ as random effect and models (m1) with ‘site’ as fixed and ‘subplot’ as random effect for each trait and species. To obtain trait variance between sites, the difference of the random effect variances was extracted from (m0) and (m1). Trait variances between subplots within site and within subplots were determined by extracting the random effect variance and the variance of residuals, respectively, from (m1). Variances were estimated using REML. Plant height and SLA were log-transformed beforehand to achieve normal distributions. Analyses of multiple traits (question 4) Trait correlations at the three hierarchical sampling levels (within subplots, between subplots within site, between sites) for each species were examined in principal component analyses (PCAs) using the R package ade4 (Dray and Dufour 2007). Leaf-level data were averaged per site (=site mean) and subplot (=subplot mean). Following Albert et al. (2010a) between-site PCAs were run on the site mean data. The deviation from the site means between subplots (subplot mean − site mean) and within subplots (leaf-level data − site mean) were used for the between-subplot and within-subplot PCAs, respectively. Trait data were standardized (mean = 0, variance = 1) and the Euclidean distance was used to compute PCAs. RESULTS Trait means and extent of ITV (question 1) Trait means and the extent of ITV differed among species (Fig. 1). Poa pratensis showed lower means of plant height, SLA, leaf greenness and a higher mean of LDMC than the other two species (Fig. 1a–d). Subplot-level ranges and CV differed among species or at least P. pratensis differed from the other two species except for the ranges in SLA (Fig. 1e–l). The CVs in P. pratensis were larger than in A. elatius and D. glomerata. The CVs of SLA also differed between A. elatius and D. glomerata (larger in D. glomerata), while the CVs of other traits did not differ between the latter species. SDNDr calculated at the subplot level were similar across species for the studied traits (Fig. 1m–p). Figure 1: View largeDownload slide (a–d) means and ITV metrics (e–h) range, (i–l) coefficient of variation (CV) and (m–p) standard deviation of nearest neighbour distances divided by range (SDNDr) for plant height, leaf dry matter content (LDMC), leaf greenness (LG), and specific leaf area (SLA) of the studied grass species. Shown are species means (±1 SE) based on site means (6 subplots per site) for A. elatius (AE), D. glomerata (DG) and P. pratensis (PP). P values and letters above the bar plots indicate significant differences between species in trait means and ITV metrics (calculated at the subplot level) tested in linear mixed-effects models. Figure 1: View largeDownload slide (a–d) means and ITV metrics (e–h) range, (i–l) coefficient of variation (CV) and (m–p) standard deviation of nearest neighbour distances divided by range (SDNDr) for plant height, leaf dry matter content (LDMC), leaf greenness (LG), and specific leaf area (SLA) of the studied grass species. Shown are species means (±1 SE) based on site means (6 subplots per site) for A. elatius (AE), D. glomerata (DG) and P. pratensis (PP). P values and letters above the bar plots indicate significant differences between species in trait means and ITV metrics (calculated at the subplot level) tested in linear mixed-effects models. The means showed in nearly all traits and species (with exception of means of LDMC in D. glomerata) significant relationships with measured local environmental site characteristics (Table 1; see online supplementary Table S2). In A. elatius, means of LDMC decreased and leaf greenness increased with fertilisation, while D. glomerata and P. pratensis had larger means of SLA with increasing levels of fertilisation. Potential solar radiation had negative effects on plant height and positive effects on SLA in all species, as well as negative effects on LDMC in A. elatius and leaf greenness in P. pratensis. In addition, different combinations of nutrient-related soil variables (N, plant-available P and K) were included in the models explaining best differences in trait means, with exception of the model for leaf greenness in D. glomerata (Table 1). After accounting for the effects of environmental site characteristics, community biomass production was always included in the models best explaining differences in plant height (positive effects). In addition, community biomass production was also incorporated in the models best explaining differences in mean values of SLA in D. glomerata (negative effects) and in P. pratensis (positive effects). Species biomass proportions were weakly related to traits means, with the exception of positive relationships to leaf greenness in D. glomerata. Species richness of the studied subplots showed positive relationships with the means of LDMC in P. pratensis. Table 1: summary of linear mixed-effects model analyses testing the influence of environmental site and plant community characteristics on trait means and metrics of intraspecific trait variability (ITV) (range, coefficient of variation [CV], standard deviation of nearest neighbour distances divided by range [SDNDr]) for three studied grass species Response variable Arrhenatherum elatius Dactylis glomerata Poa pratensis Minimal adequate model R2 Minimal adequate model R2 Minimal adequate model R2 Plant height  Mean ↓N + ↑P + ↓K + ↓Sun + ↑BMCom 0.469*** ↑P + ↓Sun + ↑BMCom 0.283*** ↑P + ↓K + ↑BMCom 0.209***  Range ↓N + ↑P + ↓K + ↓Sun 0.310*** ↓Sun 0.148*** ↓N + ↓K + ↓Sun 0.070*  CV ↓BMCom 0.090* ↓BMCom 0.139** ↓BMCom 0.063*  SDNDr Specific leaf area (SLA)  Mean ↑N + ↓P + ↑Sun 0.216** ↑Fert + ↓P + ↑Sun + ↓BMCom 0.161* ↑N + ↑Fert + ↑Sun + ↑BMCom 0.204**  Range ↑N + ↑Sun 0.177** ↓BMCom + ↓Sr 0.074*  CV ↑Fert + ↑Sun 0.102* ↑K + ↓BMCom 0.148**  SDNDr Leaf dry matter content (LDMC)  Mean ↓Fert + ↓N + ↓Sun + ↑BMCom 0.331*** ↓N + ↓K + ↑Sr 0.239***  Range ↑N + ↓P + ↑K + ↓BMCom 0.181**  CV ↑Fert + ↑Sun 0.187** ↑N + ↓P + ↑K + ↓BMCom 0.268***  SDNDr ↓N 0.107** Leaf greenness  Mean ↑Fert + ↓N 0.208** ↑BMSp 0.081* ↓N + ↑P + ↑K + ↓Sun 0.237***  Range ↑N + ↑K 0.086* ↓Sun 0.070*  CV ↑Sun 0.081* ↑K + ↓BMSp 0.137** ↑BMCom + ↓BMSp 0.125**  SDNDr ↓BMCom 0.066* Response variable Arrhenatherum elatius Dactylis glomerata Poa pratensis Minimal adequate model R2 Minimal adequate model R2 Minimal adequate model R2 Plant height  Mean ↓N + ↑P + ↓K + ↓Sun + ↑BMCom 0.469*** ↑P + ↓Sun + ↑BMCom 0.283*** ↑P + ↓K + ↑BMCom 0.209***  Range ↓N + ↑P + ↓K + ↓Sun 0.310*** ↓Sun 0.148*** ↓N + ↓K + ↓Sun 0.070*  CV ↓BMCom 0.090* ↓BMCom 0.139** ↓BMCom 0.063*  SDNDr Specific leaf area (SLA)  Mean ↑N + ↓P + ↑Sun 0.216** ↑Fert + ↓P + ↑Sun + ↓BMCom 0.161* ↑N + ↑Fert + ↑Sun + ↑BMCom 0.204**  Range ↑N + ↑Sun 0.177** ↓BMCom + ↓Sr 0.074*  CV ↑Fert + ↑Sun 0.102* ↑K + ↓BMCom 0.148**  SDNDr Leaf dry matter content (LDMC)  Mean ↓Fert + ↓N + ↓Sun + ↑BMCom 0.331*** ↓N + ↓K + ↑Sr 0.239***  Range ↑N + ↓P + ↑K + ↓BMCom 0.181**  CV ↑Fert + ↑Sun 0.187** ↑N + ↓P + ↑K + ↓BMCom 0.268***  SDNDr ↓N 0.107** Leaf greenness  Mean ↑Fert + ↓N 0.208** ↑BMSp 0.081* ↓N + ↑P + ↑K + ↓Sun 0.237***  Range ↑N + ↑K 0.086* ↓Sun 0.070*  CV ↑Sun 0.081* ↑K + ↓BMSp 0.137** ↑BMCom + ↓BMSp 0.125**  SDNDr ↓BMCom 0.066* Shown are the predictor variable combinations included in the minimal adequate model. Arrows indicate a positive (↑) or negative (↓) influence of the predictor variable on the ITV metric. R2 statistics were assessed by likelihood ratio tests. Significance levels are indicated with * P ≤ 0.05 ** P ≤ 0.01 *** P ≤ 0.001 Abbreviations of predictor variables: BMCom = community biomass production, BMSp = species biomass proportion, Fert = fertilisation, K = plant-available soil potassium, N = total soil nitrogen, P = plant-available soil phosphorus, Sr = species richness, Sun = potential solar irradiation. View Large Table 1: summary of linear mixed-effects model analyses testing the influence of environmental site and plant community characteristics on trait means and metrics of intraspecific trait variability (ITV) (range, coefficient of variation [CV], standard deviation of nearest neighbour distances divided by range [SDNDr]) for three studied grass species Response variable Arrhenatherum elatius Dactylis glomerata Poa pratensis Minimal adequate model R2 Minimal adequate model R2 Minimal adequate model R2 Plant height  Mean ↓N + ↑P + ↓K + ↓Sun + ↑BMCom 0.469*** ↑P + ↓Sun + ↑BMCom 0.283*** ↑P + ↓K + ↑BMCom 0.209***  Range ↓N + ↑P + ↓K + ↓Sun 0.310*** ↓Sun 0.148*** ↓N + ↓K + ↓Sun 0.070*  CV ↓BMCom 0.090* ↓BMCom 0.139** ↓BMCom 0.063*  SDNDr Specific leaf area (SLA)  Mean ↑N + ↓P + ↑Sun 0.216** ↑Fert + ↓P + ↑Sun + ↓BMCom 0.161* ↑N + ↑Fert + ↑Sun + ↑BMCom 0.204**  Range ↑N + ↑Sun 0.177** ↓BMCom + ↓Sr 0.074*  CV ↑Fert + ↑Sun 0.102* ↑K + ↓BMCom 0.148**  SDNDr Leaf dry matter content (LDMC)  Mean ↓Fert + ↓N + ↓Sun + ↑BMCom 0.331*** ↓N + ↓K + ↑Sr 0.239***  Range ↑N + ↓P + ↑K + ↓BMCom 0.181**  CV ↑Fert + ↑Sun 0.187** ↑N + ↓P + ↑K + ↓BMCom 0.268***  SDNDr ↓N 0.107** Leaf greenness  Mean ↑Fert + ↓N 0.208** ↑BMSp 0.081* ↓N + ↑P + ↑K + ↓Sun 0.237***  Range ↑N + ↑K 0.086* ↓Sun 0.070*  CV ↑Sun 0.081* ↑K + ↓BMSp 0.137** ↑BMCom + ↓BMSp 0.125**  SDNDr ↓BMCom 0.066* Response variable Arrhenatherum elatius Dactylis glomerata Poa pratensis Minimal adequate model R2 Minimal adequate model R2 Minimal adequate model R2 Plant height  Mean ↓N + ↑P + ↓K + ↓Sun + ↑BMCom 0.469*** ↑P + ↓Sun + ↑BMCom 0.283*** ↑P + ↓K + ↑BMCom 0.209***  Range ↓N + ↑P + ↓K + ↓Sun 0.310*** ↓Sun 0.148*** ↓N + ↓K + ↓Sun 0.070*  CV ↓BMCom 0.090* ↓BMCom 0.139** ↓BMCom 0.063*  SDNDr Specific leaf area (SLA)  Mean ↑N + ↓P + ↑Sun 0.216** ↑Fert + ↓P + ↑Sun + ↓BMCom 0.161* ↑N + ↑Fert + ↑Sun + ↑BMCom 0.204**  Range ↑N + ↑Sun 0.177** ↓BMCom + ↓Sr 0.074*  CV ↑Fert + ↑Sun 0.102* ↑K + ↓BMCom 0.148**  SDNDr Leaf dry matter content (LDMC)  Mean ↓Fert + ↓N + ↓Sun + ↑BMCom 0.331*** ↓N + ↓K + ↑Sr 0.239***  Range ↑N + ↓P + ↑K + ↓BMCom 0.181**  CV ↑Fert + ↑Sun 0.187** ↑N + ↓P + ↑K + ↓BMCom 0.268***  SDNDr ↓N 0.107** Leaf greenness  Mean ↑Fert + ↓N 0.208** ↑BMSp 0.081* ↓N + ↑P + ↑K + ↓Sun 0.237***  Range ↑N + ↑K 0.086* ↓Sun 0.070*  CV ↑Sun 0.081* ↑K + ↓BMSp 0.137** ↑BMCom + ↓BMSp 0.125**  SDNDr ↓BMCom 0.066* Shown are the predictor variable combinations included in the minimal adequate model. Arrows indicate a positive (↑) or negative (↓) influence of the predictor variable on the ITV metric. R2 statistics were assessed by likelihood ratio tests. Significance levels are indicated with * P ≤ 0.05 ** P ≤ 0.01 *** P ≤ 0.001 Abbreviations of predictor variables: BMCom = community biomass production, BMSp = species biomass proportion, Fert = fertilisation, K = plant-available soil potassium, N = total soil nitrogen, P = plant-available soil phosphorus, Sr = species richness, Sun = potential solar irradiation. View Large ITV metrics (range, CV and SDNDr) showed less consistent relationships to environmental site characteristics across species and among traits. Only the range of plant height was always negatively related to potential solar radiation, and lower concentrations of soil nitrogen and potassium also negatively affected the range of plant height in A. elatius and P. pratensis (Table 1). Apart from environmental site characteristics, the CV in plant height in all grass species as well as the ranges and CVs in SLA and LDMC of D. glomerata were lower on more productive sites (negative relationships to community biomass production). The CV in leaf greenness of P. pratensis was higher on more productive sites. Dactylis glomerata and P. pratensis had lower CVs in leaf greenness on sites, where their biomass proportions were lower. Species richness of the studied subplots did not explain variation in ITV metrics after accounting for environmental site characteristics except for negative relationships to the range of SLA in D. glomerata. While models for trait means explained 21–48% of variation in A. elatius, the proportions of explained variation were 8–28% in D. glomerata and 20–24% in P. pratensis. Proportions of explained variation in other trait metrics were generally smaller or environmental site and plant community characteristics even did not explain a significant proportion of variation. Null model approach to detect environmental filtering and niche differentiation (question 2) All observed CV and ranges of traits were significantly smaller than null distributions indicating that ITV was lower than expected by chance at the site-level (see online supplementary Table S3). Observed SDNDr were in most cases not different from random expectations across sites (see online supplementary Table S3). Only SDNDr of SLA and leaf greenness in A. elatius as well as SDNDr of SLA and LDMC in D. glomerata were significantly greater than null distributions indicating a less evenly spacing of these trait values than expected by chance. Permutation tests at the subplot-level within site revealed that the observed CV and ranges of plant height were significantly smaller than null distributions within most of the sites indicating that ITV of plant height was also lower than expected by chance at the subplot-level (see online supplementary Fig. S2). The majority of CV and ranges of leaf traits showed no differences from random expectations at the subplot-level (see online supplementary Fig. S2). Observed SDNDr also did not differ from random expectations at the subplot-level (see online supplementary Fig. S2). Permutation tests with between-sample distances at the site-level showed that samples of the same site were more similar in the observed traits than samples of different sites (see online supplementary Table S4). Permutation tests with between-sample distances at the subplot-level within site revealed that samples of the same subplot were more similar in plant height and also partly in LDMC and SLA than samples of different subplots within site (see online supplementary Fig. S3). The samples showed no subplot-specific differences in leaf greenness (see online supplementary Fig. S3). Variance decomposition of single traits (question 3) In all species, the major proportion of variance in plant height was attributable to between-site variation (55–75%; Fig. 2a), whereas most of the variance in leaf greenness was within subplots (60–85%; Fig. 2c). The results for SLA and LDMC depended on species identity: A. elatius showed the largest proportion of variance at the between-site level but P. pratensis and D. glomerata rather within subplots (Fig. 2b and d). A small proportion of variance was attributable to differences between subplots within site in all studied traits. Figure 2: View largeDownload slide variance decomposition at the within-subplot, between-subplot and between-site level for (a) plant height, (b) leaf dry matter content, (c) leaf greenness and (d) specific leaf area of Arrhenatherum elatius (AE), Dactylis glomerata (DG), and Poa pratensis (PP). The 50 % threshold is indicated by a dashed line. Figure 2: View largeDownload slide variance decomposition at the within-subplot, between-subplot and between-site level for (a) plant height, (b) leaf dry matter content, (c) leaf greenness and (d) specific leaf area of Arrhenatherum elatius (AE), Dactylis glomerata (DG), and Poa pratensis (PP). The 50 % threshold is indicated by a dashed line. Analyses of multiple traits (question 4) PCAs showed that the first axis was mainly explained by LDMC and SLA (43–71% of variance) for all species and across sampling levels (Fig. 3). Differences among species were due to plant height and leaf greenness and the correlations between these traits and LDMC and SLA. Arrhenatherum elatius showed the most preserved correlation structure across sampling levels with LDMC and SLA determining together with plant height the first axis and leaf greenness the second axis (Fig. 3a–c). Dactylis glomerata showed at the within- and between-subplot level similar correlation structures with plant height explaining the second axis mainly independent from LDMC and SLA (Fig. 3d and e). At the between-site level, plant height was closer related to LDMC on the first axis and D. glomerata had a similar correlation structure as A. elatius at this level (Fig. 3f). Poa pratensis showed the least consistent correlation structures across sampling levels. At the within-subplot level, the correlation structure was similar to the PCAs of A. elatius with a strong positive correlation between LDMC and plant height on the first axis (Fig. 3g). This correlation weakened at the between-subplot level (Fig. 3h) and changed at the between-site level where plant height was closely related to leaf greenness (Fig. 3i). Figure 3: View largeDownload slide principal component analysis (PCA) of the measured plant traits at the within-subplot, between-subplot and between-site levels for (a–c) Arrhenatherum elatius, (d–f) Dactylis glomerata and (g–i) Poa pratensis. Analyses were based on plant height (H), specific leaf area (SLA), leaf dry matter content (LDMC) and leaf greenness (LG). Correlation circles and the first two principal component axes are shown. Proportions of explained variation are given for the first two principal component axes. Figure 3: View largeDownload slide principal component analysis (PCA) of the measured plant traits at the within-subplot, between-subplot and between-site levels for (a–c) Arrhenatherum elatius, (d–f) Dactylis glomerata and (g–i) Poa pratensis. Analyses were based on plant height (H), specific leaf area (SLA), leaf dry matter content (LDMC) and leaf greenness (LG). Correlation circles and the first two principal component axes are shown. Proportions of explained variation are given for the first two principal component axes. DISCUSSION Previous studies indicated that the recognition of ITV is particularly important in studies at a local scale (Siefert et al. 2015), but studies focusing on ITV in ecosystems without remarkable environmental gradients are scarce. Therefore, the aim of this study was to evaluate how consistent patterns of ITV are in three common grass species studied in 12 semi-natural grasslands across three spatially hierarchical sampling levels (between sites, between subplots within sites and within subplots) on local scale. Trait means and extent of ITV (question 1) In line with our expectation, we found differences in trait means and ITV among the study species. Local environmental site characteristics and productivity mostly explained a larger proportion of variation in trait means of A. elatius than in D. glomerata or P. pratensis. The higher amounts of explained variation in mean and range of plant height in A. elatius indicated a plastic adjustment of height growth in response to resource availability which would be in line with the dominant plasticity hypothesis (Ashton et al. 2010). Studies supporting this hypothesis investigated plant growth in response to soil nutrient (nitrogen) availability (Ashton et al. 2010; Siefert and Ritchie 2016). They found that the most competitive species with the greatest access to light were more variable in response to soil nitrogen availability, probably due to a more efficient use of resources. This explanation is also supported by the high SLA values of A. elatius. SLA can be positively correlated with the potential relative growth rate of a species (Cornelissen et al. 2003; Weiher et al. 1999). Hence, A. elatius had likely the highest relative growth rate among the studied species and can therefore be considered as competitively dominant species. Poa pratensis with lower plant height and SLA showed the greatest extent of ITV in leaf traits quantified as CVs and ranges of trait values in the studied subplots. Aan et al. (2006) found a higher plasticity in leaf traits in subordinate species indicating their greater ability to adjust the expression of traits related to carbon acquisition in response to competition. However, the number of studied species in our analyses is too low to figure out whether the observed patterns in trait means and ITV depend on species relative abundances. Models best explaining trait means and ITV metrics were often based on a combination of environmental site characteristics and productivity. Site productivity generally increased plant height while the relationships between productivity and ITV in plant height and leaf traits were negative (Table 1). Interestingly, site productivity did not correlate with the studied soil variables or fertilisation (see online supplementary Table S5) indicating that productivity was likely dependent on multiple factors. Nevertheless, at more productive sites size-asymmetric light competition was increased and filtered for taller growth (Grime 2006). Reduced ITV at higher productivity, however, would support theory on equalizing mechanisms suggesting that similar trait values minimize average fitness differences among individuals in more competitive environments (Hille Ris Lambers et al. 2012). In contrast to our expectations, species richness was rarely included in the combination of predictor variables best explaining variation in trait means and ITV metrics. Significant effects of species richness on trait means and ITV metrics in linear mixed-effects models with species richness as single fixed explanatory variable (see online supplementary Table S2) disappeared after accounting for environmental site characteristics. Several studies in Central European temperate grasslands have shown that plant species richness declined in response to increasing land-use intensity, especially fertilisation (e.g. Kleijn et al. 2009; Socher et al. 2013). Dietrich et al. (2017) investigated how fertilisation, soil and plant community characteristics determine soil microbial activity in the same grasslands as used in this study and showed negative effects of fertilisation on species richness in structural equation models (see also negative correlations in online supplementary Table S5). Thus, the observed relationships between traits means and ITV metrics in separate models were mainly due to differences in site characteristics, which co-varied with species richness of the studied grasslands. Overall, the mostly low portions of explained variation in subplot-level trait means and ITV metrics suggest that possibly other sources of variation, e.g. biotic interactions, are more important for patterns of trait plasticity in the absence of strong environmental gradients. Additionally, we cannot exclude that other environmental factors not measured in our study (e.g. soil moisture, soil depth) account for some of the unexplained variation. Null model approach to detect environmental filtering and niche differentiation (question 2) In line with our expectations, we found evidence for trait convergence in all four traits for all three studied species at the site-level indicating a strong influence of environmental filtering on the extent of ITV. Contrary to our expectation, we found no indication of niche differentiation at the subplot-level. Instead, plant height even converged at the subplot-level. The important role of environmental filtering in influencing ITV in our study corresponds to findings of previous studies (Jung et al. 2010; Siefert 2012). For example, a previous study in old-field plant communities also reported on a stronger influence of environmental filtering than of niche differentiation, with a stronger signal of trait convergence in plant height than in SLA and LDMC (Siefert 2012). One possible explanation for the convergence in plant height in our study is that differences within subplots were reduced by competitive exclusion of smaller shoots which were inferior in the asymmetric competition for light. This would also explain why we observed a convergence in plant height within subplots, while this was not the case for leaf traits, which are known to be extremely plastic and respond rapidly to various environmental drivers and changes in the biotic environment (e.g. Hodgson et al. 2011; Roscher et al. 2011; Siebenkäs et al. 2017). However, it is also possible that samples within subplots were genetically closer related than at greater spatial distance, and the lower genetic variation at the subplot-level could also explain lower small-scale variation in plant height. Siefert (2012) discussed that clonal growth of plant species and multiple samples from the same clone (or genet) may affect trait distribution at small spatial scale. In our study, we do not have information if all sampled shoots on the subplot-scale were genetically distinct, but it is unlikely that effects of low genetic variation are expressed across species in one particular trait (i.e. plant height). Variance decomposition of single traits (question 3) In spite of the strong signal of trait convergence in all traits among sites, the relative extent of ITV across spatially hierarchical sampling levels differed among traits. In line with our expectations, the major portion of variation in plant height was due to between-site differences. The low proportion of within-subplot variation in plant height is consistent with the observed subplot-level convergence in this trait and suggests that local-scale environmental differences were drivers of variation in plant height. In contrast, sources of variation were more variable for leaf traits. Leaf traits of D. glomerata and P. pratensis varied to a large extent within subplots suggesting a fine-tuned adjustment of leaf traits at small scale. Bachmann et al. (2017) showed that LDMC and SLA of multiple grassland species vary with measured light at leaf height supporting our argumentation that leaf traits are highly plastic in response to the actual growth conditions. In contrast, SLA and LDMC of A. elatius were more variable between sites. Arrhenatherum elatius was the tallest species in our study and often reached dominance in our grasslands (see online supplementary Fig. S1). Sampled leaves were often exposed to full-light conditions in the top canopy, and therefore between-site variation due to local-scale environmental variation (see question 1) was more important for leaf trait expression than variation within subplots. Analyses of multiple traits (question 4) The negatively correlated leaf traits LDMC and SLA explained most of the variation in the data for all three studied species across the spatially hierarchical sampling levels. This correlation of trait values is well known from studies at global scale (e.g. Reich et al. 1997), but has also been reported from other studies at more local environmental gradients (Albert et al. 2010a; Mitchell et al. 2017). In contrast, species differed with respect to the correlations between plant height and leaf traits. These results correspond to findings of Albert et al. (2010a), who also observed in multi-trait analyses that the trait correlation structures were not completely consistent across species. While the trait correlation structure of A. elatius was consistent across spatially hierarchical sampling levels, in line with our expectations correlation patterns for D. glomerata and P. pratensis were more variable. Poa pratensis showed strong correlations between plant height and SLA (negative) and LDMC (positive) at the within-subplot level suggesting that light competition and a trade-off between tall growth/low SLA sun leaves and small growth/high SLA shade leaves was important at small spatial scale. The strong correlation between plant height and leaf greenness suggested that variation in nutrient availability was more important for multi-trait patterns of P. pratensis at the between-site level. Since A. elatius as a tall-growing species in our study usually reaches the upper canopy levels, it is unlikely that the negative correlation between plant height and SLA was due to plasticity in response to light availability. Possibly, biotic interactions and local environmental site conditions had no major impact on multi-trait patterns of A. elatius and patterns were therefore consistent across hierarchically sampling levels. In summary, trait-based ecology has gained increasing popularity, but the importance of ITV is among the “loose foundation stones” in the application of trait-based approaches (Shipley et al. 2016). Our empirical study showed that even on the local scale the extent of ITV can vary greatly among species and traits. Although our study only involved a small number of species, our results suggest that the impact of local-scale environmental site characteristics and biotic interactions within sites on ITV is highly variable among species. Therefore, further studies based on more species are needed to draw general conclusions about the importance of considering ITV at different spatial scales. ACKNOWLEDGEMENTS This work was supported by the German Research Foundation (RO2397/6). We thank the farmers and land owners for providing the study sites and information about management as well as S. Hahne, S. Schier, and B. Sawall for help during trait measurements. We especially acknowledge N. Hacker and Y. Oelmann for help with soil phosphorus analyses. We thank the associate editor Christa Mulder and the anonymous reviewer for their helpful comments on the manuscript. Conflict of interest statement. None declared. REFERENCES Aan A , Hallik L , Kull O ( 2006 ) Photon flux partitioning among species along a productivity gradient of an herbaceous plant community . J Ecol 94 : 1143 – 55 . Google Scholar Crossref Search ADS Albert CH , Thuiller W , Yoccoz NG , et al. . ( 2010a ) A multi-trait approach reveals the structure and the relative importance of intra- vs. interspecific variability in plant traits . Funct Ecol 24 : 1192 – 201 . Google Scholar Crossref Search ADS Albert CH , Thuiller W , Yoccoz NG , et al. . ( 2010b ) Intraspecific functional variability: extent, structure and sources of variation . J Ecol 98 : 604 – 13 . Google Scholar Crossref Search ADS Al Haj Khaled R , Duru M , Theau JP , et al. . ( 2005 ) Variation in leaf traits through seasons and N-availability levels and its consequences for ranking grassland species . J Veg Sci 16 : 391 – 8 . Google Scholar Crossref Search ADS Ashton IW , Miller AE , Bowman WD , et al. . ( 2010 ) Niche complementarity due to plasticity in resource use: plant partitioning of chemical N forms . Ecology 91 : 3252 – 60 . Google Scholar Crossref Search ADS PubMed Auger S , Shipley B ( 2013 ) Inter-specific and intra-specific trait variation along short environmental gradients in an old-growth temperate forest . J Veg Sci 24 : 419 – 28 . Google Scholar Crossref Search ADS Bachmann D , Roscher C , Buchmann N ( 2017 ) How do leaf trait values change spatially and temporally with light availability in a grassland diversity experiment ? Oikos . doi:10.1111/oik.04533 Barton K ( 2015 ) MuMIn: Multi-model inference. R package version 1.15.1 . https://CRAN.R-project.org/package=MuMIn ( 8 November 2017, date last accessed ) Bates D , Mächler M , Bolker B , et al. . ( 2015 ) Fitting linear mixed-effects models using lme4 . J Stat Softw 67 : 1 – 48 . Google Scholar Crossref Search ADS Boucher FC , Thuiller W , Arnoldi C , et al. . ( 2013 ) Unravelling the architecture of functional variability in wild populations of Polygonum viviparum L . Funct Ecol 27 : 382 – 91 . Google Scholar Crossref Search ADS PubMed Chapman SC , Barreto HJ ( 1997 ) Using a chlorophyll meter to estimate specific leaf nitrogen of tropical maize during vegetative growth . Agron J 89 : 557 – 62 . Google Scholar Crossref Search ADS Cornelissen JHC , Lavorel S , Garnier E , et al. . ( 2003 ) A handbook of protocols for standardised and easy measurement of plant functional traits worldwide . Aust J Bot 51 : 335 – 80 . Google Scholar Crossref Search ADS Cornwell WK , Ackerly DD ( 2009 ) Community assembly and shifts in plant trait distributions across an environmental gradient in coastal California . Ecol Monogr 79 : 109 – 26 . Google Scholar Crossref Search ADS Cornwell WK , Schwilk LD , Ackerly DD ( 2006 ) A trait-based test for habitat filtering: convex hull volume . Ecology 87 : 1465 – 71 . Google Scholar Crossref Search ADS PubMed Dietrich P , Buchmann T , Cesarz S , et al. . ( 2017 ) Fertilization, soil and plant community characteristics determine soil microbial activity in managed temperate grasslands . Plant Soil 419 : 189 – 99 . Google Scholar Crossref Search ADS Dray S , Dufour AB ( 2007 ) The ade4 package: implementing the duality diagram for ecologists . J Stat Softw 22 : 1 – 20 . Google Scholar Crossref Search ADS Ellenberg H ( 1988 ) Vegetation Ecology of Central Europe . Cambridge : Cambridge University Press . Errecart PM , Agnusdei MG , Lattanzi FA , et al. . ( 2012 ) Leaf nitrogen concentration and chlorophyll meter readings as predictors of tall fescue nitrogen nutrition status . Field Crops Res 129 : 46 – 58 . Google Scholar Crossref Search ADS Grime JP ( 2006 ) Trait convergence and trait divergence in herbaceous plant communities: mechanisms and consequences . J Veg Sci 17 : 255 – 60 . Google Scholar Crossref Search ADS Gubsch M , Buchmann N , Schmid B , et al. . ( 2011 ) Differential effects of plant diversity on functional trait variation of grass species . Ann Bot 107 : 157 – 69 . Google Scholar Crossref Search ADS PubMed Hille Ris Lambers J , Adler PB , Harpole WS , et al. . ( 2012 ) Rethinking community assembly through the lens of coexistence theory . Annu Rev Ecol Evol Syst 43 : 227 – 48 . Google Scholar Crossref Search ADS Hoffmann K , Bivour W , Früh B , et al. . ( 2014 ) Klimauntersuchungen in Jena für die Anpassung an den Klimawandel und seine erwarteten Folgen . Offenbach/Main, Germany : Deutscher Wetterdienst . Hodgson JG , Montserrat-Martí G , Charles M , et al. . ( 2011 ) Is leaf dry matter content a better predictor of soil fertility than specific leaf area ? Ann Bot 108 : 1337 – 45 . Google Scholar Crossref Search ADS PubMed Hothorn T , Bretz F , Westfall P ( 2008 ) Simultaneous inference in general parametric models . Biom J 50 : 346 – 63 . Google Scholar Crossref Search ADS PubMed Hulshof CM , Swenson NG ( 2010 ) Variation in leaf functional trait values within and across individuals and species: an example from a Costa Rican dry forest . Funct Ecol 24 : 217 – 23 . Google Scholar Crossref Search ADS Jung V , Violle C , Mondy C , et al. . ( 2010 ) Intraspecific variability and trait-based community-assembly . J Ecol 98 : 1134 – 40 . Google Scholar Crossref Search ADS Kichenin E , Wardle DA , Peltzer DA , et al. . ( 2013 ) Contrasting effects of plant inter- and intraspecific variation on community-level trait measures along an environmental gradient . Funct Ecol 27 : 1254 – 61 . Google Scholar Crossref Search ADS Kleijn D , Kohler F , Báldi A , et al. . ( 2009 ) On the relationship between farmland biodiversity and land-use intensity in Europe . Proc R Soc B 276 : 903 – 9 . Google Scholar Crossref Search ADS PubMed Klimešová J , de Bello F ( 2009 ) CLO-PLA: the database of clonal and bud bank traits of Central European flora . J Veg Sci 20 : 511 – 6 . Google Scholar Crossref Search ADS Kraft NJB , Ackerly DD ( 2010 ) Functional trait and phylogenetic tests of community assembly across spatial scales in an Amazonian forest . Ecol Monograph 80 : 401 – 22 . Google Scholar Crossref Search ADS Lajoie G , Vellend M ( 2015 ) Understanding context dependence in the contribution of intraspecific variation to community trait-environment matching . Ecology 96 : 2912 – 22 . Google Scholar Crossref Search ADS PubMed Le Bagousse-Pinguet Y , de Bello F , Vandewalle M , et al. . ( 2014 ) Species richness of limestone grasslands increases with trait overlap: evidence from within- and between-species functional diversity partitioning . J Ecol 102 : 466 – 74 . Google Scholar Crossref Search ADS Londo G ( 1976 ) Decimal scale for releves of permanent quadrats . Vegetatio 33 : 61 – 4 . Google Scholar Crossref Search ADS MacArthur R , Levins R ( 1967 ) The limiting similarity, convergence, and divergence of coexisting species . Am Nat 101 : 377 – 85 . Google Scholar Crossref Search ADS Magee L ( 1990 ) R2 measures based on Wald and likelihood joint significance tests . Am Stat 44 : 250 – 3 . McGill BJ , Enquist BJ , Weiher E , et al. . ( 2006 ) Rebuilding community ecology from functional traits . Trends Ecol Evol 21 : 178 – 85 . Google Scholar Crossref Search ADS PubMed Messier J , McGill BJ , Lechowicz MJ ( 2010 ) How do traits vary across ecological scales? A case for trait-based ecology . Ecol Lett 13 : 838 – 48 . Google Scholar Crossref Search ADS PubMed Mitchell RM , Wright JP , Ames GM ( 2017 ) Intraspecific variability improves environmental matching, but does not increase ecological breadth along a wet-to-dry ecotone . Oikos 126 : 988 – 95 . Google Scholar Crossref Search ADS Pfitzenmeyer CDC ( 1962 ) Biological flora of the British Isles. Arrhenatherum elatius (L.) J. and C. Presl (A. avenaceum Beauv.) . J Ecol 50 : 235 – 45 . Google Scholar Crossref Search ADS R Core Team ( 2015 ) R: a language and environment for statistical computing . Viena, Austria : R Foundation for statistical computing . http://www.r-project.org ( 8 November 2017, date last accessed ) Reich PB , Walters MB , Ellsworth DS ( 1997 ) From tropics to tundra: global convergence in plant functioning . Proc Natl Acad Sci USA 94 : 13730 – 4 . Google Scholar Crossref Search ADS PubMed Reich PB , Wright IJ , Cavender-Barres J , et al. . ( 2003 ) The evolution of plant functional variation: traits, spectra, and strategies . Int J Plant Sci 164 : S143 – 64 . Google Scholar Crossref Search ADS Roscher C , Kutsch WL , Kolle O , et al. . ( 2011 ) Adjustment to the light environment in small-statured forbs as a strategy for complementary resource use in mixtures of grassland species . Ann Bot 107 : 965 – 79 . Google Scholar Crossref Search ADS PubMed Shipley B , De Bello F , Cornelissen JH , et al. . ( 2016 ) Reinforcing loose foundation stones in trait-based plant ecology . Oecologia 180 : 923 – 31 . Google Scholar Crossref Search ADS PubMed Siebenkäs A , Schumacher J , Roscher C ( 2015 ) Phenotypic plasticity to light and nutrient availability alters functional trait ranking across eight perennial grassland species . AOB Plants 7 : plv029 . Google Scholar Crossref Search ADS PubMed Siebenkäs , A , Schumacher J , Roscher C ( 2017 ) Trait variation in response to resource availability and plant diversity modulates functional dissimilarity among species in experimental grasslands . J Plant Ecol 10 : 981 – 93 . Siefert A ( 2012 ) Incorporating intraspecific variation in tests of trait-based community assembly . Oecologia 170 : 767 – 75 . Google Scholar Crossref Search ADS PubMed Siefert A , Ritchie ME ( 2016 ) Intraspecific trait variation drives functional responses of old-field plant communities to nutrient enrichment . Oecologia 181 : 245 – 55 . Google Scholar Crossref Search ADS PubMed Siefert A , Violle C , Chalmandrier L , et al. . ( 2015 ) A global meta-analysis of the relative extent of intraspecific trait variation in plant communities . Ecol Lett 18 : 1406 – 19 . Google Scholar Crossref Search ADS PubMed Socher SA , Prati D , Boch S , et al. . ( 2013 ) Interacting effects of fertilization, mowing and grazing on plant species diversity of 1500 grasslands in Germany differ between regions . Basic Appl Ecol 14 : 126 – 36 . Google Scholar Crossref Search ADS Tonne F ( 1954 ) Besser Bauen mit Besonnungs- und Tageslicht-Planung . Schorndorf, Germany : Hoffmann . Valladares F , Gianoli E , Gómez JM ( 2007 ) Ecological limits to plant phenotypic plasticity . New Phytol 176 : 749 – 63 . Google Scholar Crossref Search ADS PubMed Violle C , Enquist BJ , McGill BJ , et al. . ( 2012 ) The return of the variance: intraspecific variability in community ecology . Trends Ecol Evol 27 : 244 – 52 . Google Scholar Crossref Search ADS PubMed Violle C , Navas ML , Vile D , et al. . ( 2007 ) Let the concept of trait be functional ! Oikos 116 : 882 – 92 . Google Scholar Crossref Search ADS Volz F ( 1958 ) Globalstrahlung auf geneigte Hänge . Meteorol Rundsch 11 : 132 – 5 . Weiher E , Keddy PA ( 1995 ) Assembly rules, null models, and trait dispersion - new questions front old patterns . Oikos 74 : 159 – 64 . Google Scholar Crossref Search ADS Weiher E , van der Werf A , Thompson K , et al. . ( 1999 ) Challenging Theophrastus: a common core list of plant traits for functional ecology . J Veg Sci 10 : 609 – 20 . Google Scholar Crossref Search ADS Weiner J ( 1990 ) Asymmetric competition in plant populations . Trends Ecol Evol 5 : 360 – 4 . Google Scholar Crossref Search ADS PubMed Wilson PJ , Thompson K , Hodgson JG ( 1999 ) Specific leaf area and leaf dry matter content as alternative predictors of plant strategies . New Phytologist 143 : 155 – 62 . Google Scholar Crossref Search ADS Wright IJ , Reich PB , Westoby M , et al. . ( 2004 ) The worldwide leaf economics spectrum . Nature 428 : 821 – 7 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2017. Published by Oxford University Press on behalf of the Institute of Botany, Chinese Academy of Sciences and the Botanical Society of China. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Intraspecific trait variation in three common grass species reveals fine-scale species adjustment to local environmental conditions
JF - Journal of Plant Ecology
DO - 10.1093/jpe/rtx068
DA - 2018-12-22
UR - https://www.deepdyve.com/lp/oxford-university-press/intraspecific-trait-variation-in-three-common-grass-species-reveals-eibPCppnrh
SP - 887
VL - 11
IS - 6
DP - DeepDyve
ER -