Skip to main content
NCBI home page
Proceedings of the Royal Society B: Biological Sciences logoLink to Proceedings of the Royal Society B: Biological Sciences
. 2019 Jul 24;286(1907):20190429. doi: 10.1098/rspb.2019.0429

Leaf trait variability between and within subalpine grassland species differs depending on site conditions and herbivory

Jennifer Firn 1,✉, Huong Nguyen 1, Martin Schütz 2, Anita C Risch 2
PMCID: PMC6661350  PMID: 31337314

Abstract

Plant traits are commonly used to predict ecosystem-level processes, but the validity of such predictions is dependent on the assumption that trait variability between species is greater than trait variability within a species—the robustness assumption. Here, we compare leaf trait intraspecific and interspecific variability depending on geographical differences between sites and 5 years of experimental herbivore exclusion in two vegetation types of subalpine grasslands in Switzerland. Four leaf traits were measured from eight herbaceous species common to all 18 sites. Intraspecific trait variability differed significantly depending on site and herbivory. However, the amount and structure of variability depended on the trait measured and whether considering leaf traits separately or multiple leaf traits simultaneously. Leaf phosphorus concentration showed the highest intraspecific variability, while specific leaf area showed the highest interspecific variability and displayed intraspecific variability only in response to herbivore exclusion. Species identity based on multiple traits was not predictable. We find intraspecific variability is an essential consideration when using plant functional traits as a common currency not just species mean traits. This is particularly true for leaf nutrient concentrations, which showed high intraspecific variability in response to site differences and herbivore exclusion, a finding which suggests that the robustness assumption does not always hold.

Keywords: robustness assumption, interspecific trait variability, plant–herbivore interactions, predicting species identity, progressive exclusion of herbivores, scale of environmental change

1. Introduction

Plant functional traits have increasingly become a common currency for predicting how communities and ecosystems respond to environmental change—ranging from broad-scale climatic gradients to finer-scale anthropogenic disturbance [1,2]. Specific leaf area (SLA) and leaf nutrient concentrations are commonly measured plant traits that have been used as surrogate measures for biological activities such as how plants acquire, process and then re-invest resources [3], even being referred to as part of a ‘holy grail’ of plant functional traits [1,2].

Originally, functional trait studies were focused on interspecific trait variability using species mean trait values, which are now available in plant trait databases [4]. An acknowledged limitation of this approach is, that in many cases, species mean trait values are derived from only a few individuals, and from singular or just a few populations [5]. Another limitation of sourcing mean trait values from databases concerns the validity of a fundamental assumption within trait-based ecology known as the ‘robustness assumption’ [1,6,7]. The robustness assumption assumes that interspecific variability is greater than intraspecific (within species) variability and therefore justifies the use of mean trait values for a species.

Seminal research showed considerable intraspecific trait variability particularly across indirect (e.g. altitude, latitude and longitude) or direct (e.g. temperature, rainfall) landscape gradients [5,7,8]. The amount of intraspecific trait variability displayed by species depends on genetic differences between individuals, subpopulations and populations, as well as phenotypic plasticity reflecting environmental conditions, ontogeny and ecological interactions including competition [5,9]. An extensive meta-analysis of more than 600 plant communities and 36 functional traits found that on average intraspecific trait variability explained 25% of the total trait variability within a population and 32% among populations [10]. Overall, the amount of trait variability explained by individuals of the same species has been found to depend on (i) the type and strength of the environmental gradient being tested, (ii) the identity of the species measured including life form (e.g. sedge, grass, tree) and (iii) whether the type of trait measured was physiological or morphological [5,10–13].

The majority of studies investigating intraspecific trait variability were conducted across environmental gradients, and have therefore been observational [14] and not experimental studies [13,15]. Yet mean trait values from databases are used to infer how plant communities respond functionally to short-term perturbations or alterations in disturbance regimes [1,16]. Observational studies represent plant trait responses to changes in abiotic conditions that regulate plant growth. By contrast, short-term experimental manipulations, such as the exclusion of herbivores, can alter induced defence mechanisms against tissue loss [17].

Leaf traits such as SLA and leaf nutrient concentrations are used as surrogate measures for how plants invest in photosynthate and mineral nutrients in the construction of their leaves. They have been found to correlate across a continuum of fast to slow returns on investment known as the leaf economic spectrum [16]. Species with higher SLAs and leaf nutrient contents are described as resource acquisition specialists that are able to grow faster [18], but tend to be more palatable experiencing higher rates of herbivory, while resource conservation specialists are slow growing, investing more in tissue construction and tending to be less palatable to herbivores [1]. It remains, however, unclear how plant traits vary within and between species when short-term perturbations occur. Although changes in plant traits have been used in ecology and evolutionary biology to infer function beyond species identity [1], assessing whether intraspecific trait variability is important is essential for developing a set of rules for ecologists to use when predicting and interpreting how plant communities ‘function’, particularly in regard to the common practice of sourcing trait values from databases [8].

In this study, we quantified the amount and structure of interspecific and intraspecific variability in response to different abiotic conditions and short-term changes in herbivory in grassland ecosystems. We tested four commonly used morphological and physiological leaf traits [16]: specific leaf area (SLA), and leaf carbon (C), leaf nitrogen (N) and leaf phosphorus (P) concentrations, which were sampled across 18 experimental sites and two subalpine vegetation types [17,19]. The two vegetation types are a result of different land-use histories, with short-grass vegetation today having higher soil phosphorus availability and grazing intensities than tall-grass vegetation [20]. At 18 sites, herbivores, ranging from ungulates to invertebrates, were excluded for 5 years [19]. We collected traits for eight herbaceous species common to all plots and sites representing four life forms, i.e. forbs, legumes, grasses and sedges. Based on past research [10,21], we expected to find that both morphological (SLA) and physiological traits (leaf nutrient concentrations) will show high intraspecific variance, but physiological traits will show higher intraspecific variance than SLA [18] because of high mobility and translocation of nutrients in plant tissue [22]. We also expected to find higher intraspecific variance in leaf nutrient traits than SLA between the two vegetation types because of differences in soil nutrient limitations and depending on grazing intensity, as herbivores are known to differ in their feeding behaviour [23]. We aimed to specifically answer the following four questions to address our hypotheses.

  • (1) How much variance is explained inter- and intraspecifically, and does this vary depending on the trait measured?

  • (2) How do inter- and intraspecific trait variance differ depending on vegetation type, site and herbivory?

  • (3) How do inter- and intraspecific trait variance change when multiple traits are simultaneously considered?

  • (4) Do combinations of leaf traits discriminate between plant species and consequently, does the robustness assumption hold for the traits measured (i.e. is interspecific variance higher than intraspecific variance).

2. Material and methods

(a). Study area

Our study was conducted in the Swiss National Park (SNP; 170 km2 in area) located at elevations ranging between 1350 and 3170 m.a.s.l. in south-eastern Swiss Alps (see also [19]). The SNP was founded in 1914 and has received minimal human disturbance for the past 100+ years (no hunting, fishing or camping, visitors are not allowed to leave the trails). The park's subalpine grasslands are characterized by large (greater than 1 ha) homogeneous patches of short- and tall-grass vegetation. These two vegetation types represent different historical land-use by domestic livestock as well as grazing regimes by wild ungulates. Short-grass vegetation developed in areas where cattle and sheep stayed overnight leading to high nutrient inputs via urine and faeces during agricultural land-use from the fourteenth century until 1914. After the SNP foundation and the removal of domestic livestock, these areas became favourite grazing sites of red deer (Cervus elaphus L.) and grazing intensity is quite high [19]. Tall-grass vegetation developed in areas where cattle and sheep used to graze, but did not rest overnight, which resulted in low overall nutrient input [20,24]. Consequently, today soil nutrient availability and grazing intensity are much lower. Domestic grazing no longer occurs in either vegetation type since 1914, but a diverse herbivore community that can be divided into four groups based on body size/weight is found: large (red deer and chamois, Rupicapra rupicapra L.; 30–150 kg), medium (marmot, Marmota marmota L.; and mountain hare, Lepus timidus L.; 3–6 kg), and small vertebrate herbivores (small rodents, e.g. Clethrionomys spp., Microtus spp., Apodemus spp.; 30–100 g), as well as invertebrates (e.g. grasshoppers, caterpillars < 5 g).

(b). Experimental design

We selected 18 subalpine locations, nine in short-grass and nine in tall-grass vegetation, across the entire SNP, hereafter sites, distributed over six grasslands at altitudes ranging from 1975 to 2300 m (electronic supplementary material, table S1). The sites were all located on dolomite parent material, but varied in both abiotic (electronic supplementary material, table S1) and biotic conditions, with aboveground productivity ranging from 176 to 1015 g dry weight m−2 yr−1. Average pre-treatment mineral soil C content in the short-grass vegetation was 11.3% (3.5%; standard deviation), N 0.55% (0.27%) and P 10.3 mg kg−1 (6.8 mg kg−1). For tall-grass vegetation, the values for C were 10.9% (3.6%), for N 0.54% (0.22%), and for P 8.0 mg kg−1 (4.2 mg kg−1). The minimum distance between two grassland sites was always greater than 1 km, and the maximum distance between sites was approximately 8 km. Most of the grasslands were also separated by mountain ranges.

Both short- and tall-grass vegetation as well as the full complement of the above-listed herbivores co-occurred on these six grasslands. Two exclosure networks, per vegetation type were erected in the three larger grasslands and one set-up per vegetation type in three smaller grasslands (18 in total, electronic supplementary material, figure S1). Each exclosure network progressively excluded herbivores by body size from ungulates to invertebrates using size-selective exclosures. Each exclosure network contained five plots, four herbivore exclusion plots and a control plot greater than 5 m from the exclusion plots that allowed access to all herbivores. Overall, this made for a total of 90 plots with area 2 × 3 m each. Details about the construction of the exclosures networks can be found in Risch et al. [19,25]. In this study, ‘between site’ intraspecific variability in leaf traits refers to the difference between sites (abiotic differences across sites), while ‘within-site’ intraspecific variability in leaf traits refers to differences in herbivory caused by the experimental exclusions.

(c). Measuring leaf traits

We measured leaf traits in year 5 of the exclosure experiment, which was 2013. In each of the 90 plots, we randomly collected five fully developed leaves with no signs of herbivore damage from five mature individuals of eight-plant species representing four life forms (forbs: Galium anisophyllon Vill., Crepis alpestris (Jacq.) Tausch; legumes: Lotus corniculatus L., Trifolium repens L.; grasses: Festuca rubra L., Briza media L.; sedges: Carex sempervirens Vill., Carex caryophyllea Latour). We chose these eight species because they were the most dominant within each life form across all sites in terms of both their frequency and cover as measured at the plot level. We collected leaves with the standardized protocols detailed by Cornelissen et al. [26]. Note that in this study the main focus is on inter- and intraspecific variability depending on both abiotic conditions and herbivory in general, not on specific impacts of the different-sized herbivores, which we have addressed previously [17].

All collected leaves for each species from each plot were combined, scanned for area using a flatbed scanner (Epson perfection V300) and image analysis software (ImageJ [27]). Thereafter the leaves were dried, weighed (dry weight) and SLA calculated as leaf area (mm2)/dry weight (g). An additional 10 leaves per species were collected (where available) in each plot, dried, combined with the leaves used for the SLA measurements, ground and analysed for total leaf nutrients. Leaf N and C were analysed with a LECO TruSpec analyser (combustion), and leaf P with a Varian Vista Pro ICPOES instrument on samples that were open digested with a 5 : 1 nitric and perchloric acids [28].

(d). Data analyses

(i). Quantification of inter- and intraspecific functional variance for individual and multiple traits (questions 1–3)

To describe the amounts of intraspecific variability in leaf traits (i.e. SLA, leaf N, C and P concentration), we created violin plots using the ggplot2 package [29]. Violin plots are both a boxplot and a kernel density plot that provide a non-parametric estimate of the probability density function of a random variable.

We used linear mixed effect models to represent the hierarchical sampling design of the experiment [30]. We developed these models to investigate trait variability for each of the two vegetation types, i.e. short-grass and tall-grass vegetation, separately. We used the lme4 package [31] to build models that included the fixed effect plant species (fixed ∼ species) and random intercept site effect (random ∼1 | site) to decompose the trait variability (SLA, leaf C, N, P). To quantify the amount of variability between sites (abiotic conditions) and within sites (grazing exclusion), we calculated measures of explained variability with random effects (R2c) and without random effects (R2m) in the models using the package MuMIN [32]. The decomposition of trait variance between the different sampling levels (i) interspecific, (ii) intraspecific between sites and (iii) intraspecific within sites was calculated as percentage of variance at the site level in the different models [33]. Variances were estimated using restricted maximum likelihood. To test the variability within each plant species, we built null models that contained only random effects of site for each functional trait and again used explained variability estimates (i.e. R2c and R2m) for the models [33].

We examined whether relationships between intraspecific and interspecific trait variability were robust when considering multiple leaf traits simultaneously. For this purpose, we used between-group principal component analyses (PCAs) of the four measured functional traits for each species combined. The PCAs used the correlation matrix based on species' means. Between-group PCA calculates linear combinations of variables maximizing the between-group (in this case between species) variance instead of the overall variance. The inertia calculated in a between-group PCA represents the part of the total variance due to the differences between species [34].

We also ran PCAs at species levels to observe whether the overall trends remained consistent. To do this, we calculated the correlations between traits at the interspecific (axes discriminating species in the trait space) and intraspecific (axes discriminating individuals within species in the trait space) levels using both between- and within-PCA analyses. Within-group PCAs breakdown the structure within a group, in this case the group we used was species, by considering the data centred on species means [34]. To observe trait correlation within species, we also ran a PCA on individual data for each species. To calculate these ordinations, we used the packages FactoMineR [35] and ade4 [36].

(ii). Species discrimination (question 4)

We investigated the probability of attributing a given leaf to the correct plant species based on the leaf traits measured—a test of the robustness assumption. If interspecific variability is higher than intraspecific variability, then we should be able to predict species identity based on the four leaf traits we measured. We used the Mahalanobis metric to calculate whether an individual leaf (based on all four traits measured) was closer to individual leaves from its own species than to others [37]. We also used linear discriminant analyses (LDA) to measure combinations of continuous explanatory variables (i.e. SLA, leaf C, N, P concentration) that best separate two or more classes of categorical variables (i.e. species [38]). LDAs, although similar to between-group PCAs, maximize the ratio of the between-groups variance to the total variance. LDA was used to predict the identity of a species a leaf comes from based on all four traits. This then allowed us to calculate the probability (Pij) of attributing a leaf from species i to species j. To conduct these analyses, we used the packages MASS [39,40]. All data analyses were conducted in R version 3.2.2 [41].

3. Results

(a). How much variance is explained inter- and intraspecifically, and does this vary depending on the trait measured?

Specific leaf area showed the highest interspecific variance and the largest discrepancy between interspecific and intraspecific variance, with more than 95% of the total variance being partitioned between species regardless of vegetation type (figure 1a,b). Both leaf N and C concentration also showed higher interspecific variance (N: >74%; C: > 65% in both vegetation types) than intraspecific variance (figure 1a,b), with leaf N concentration showing the highest overall variance of any of the traits measured, at 77% (figure 1c,d). Leaf P concentration was the only trait where intraspecific variance was higher than interspecific variance with 84% of the variance explained within a species in short- and 60% in tall-grass vegetation (figure 1a,b).

Figure 1.

Figure 1.

Open in a new tab

Breakdown of leaf trait variance: contributions of interspecific, between-site intraspecific (differences between experimental sites; nine in short-grass and nine in tall-grass vegetation) and within-site intraspecific conditions (grazing exclusion treatments) variance (a,b) and overall total variance explained in the models (c,d). (a,c) Leaf trait variance within short-grass vegetation. (b,d) Leaf trait variance within tall-grass vegetation. SLA, specific leaf area; C, leaf carbon concentration; N, leaf nitrogen concentration; P, leaf phosphorus concentration.

(b). How do inter- and intraspecific trait variance differ depending on vegetation type, site and herbivory?

Trait variance among species from the same life form was lower than between life forms (e.g. grass, forb, sedge or legume), and these trends were consistent across the two vegetation types. Sedge species generally showed the lowest intraspecific trait variance, forbs and legumes the highest (electronic supplementary material, figure S2A,B). However, the intraspecific variance was similar between species within a life form, with the exception being the intraspecific variability in SLA between the two-grass species in the short-grass vegetation and the two-legume species in the tall-grass vegetation (electronic supplementary material, figure S2A,B).

Overall, intraspecific variance in SLA was low compared with the interspecific variance for each of the eight-plant species in both vegetation types (figure 2a,b). Intraspecific variance in leaf C and N concentration showed differences but no consistent patterns between species depending on site conditions, experimental exclusions and vegetation type (figure 2a,b). For example, in the short-grass vegetation the amount of intraspecific variance in leaf C in response to our exclusions (within site) was higher for Briza media (G2) than for Festuca rubra (G1), yet, a similar amount was found for the two forbs. Similarly, leaf N in the tall-grass vegetation showed a higher level of intraspecific variance in response to herbivore exclusion for Carex sempervirens (S1) compared to Carex caryophyllea (S2), which showed one of the lowest levels of intraspecific variance.

Figure 2.

Figure 2.

Open in a new tab

Breakdown of intraspecific between-site and within-site (herbivore exclusion treatments) variance in leaf traits for each of the eight plant species depending on vegetation type: (a) short-grass vegetation and (b) tall-grass vegetation. F1, Galium anisophyllon; F2, Crepis alpestris; L1, Lotus corniculatus; L2, Trifolium repens; G1, Festuca rubra; G2, Briza media; S1, Carex sempervirens; S2, Carex caryophyllea. SLA, specific leaf area; C, leaf carbon concentration; N, leaf nitrogen concentration; P, leaf phosphorus concentration.

Intraspecific variance was highest overall for leaf P concentration, yet there was little consistency within and between sites and the two vegetation types (figure 2a,b). For example, differences in leaf P concentration of Lotus corniculatus, Trifolium repens and Carex caryophyllea were higher than the overall pattern shown in the short-grass vegetation (figure 1a), and these differences were within sites and therefore explained by herbivory treatments (figure 2a). By contrast, leaf P concentration of the same three species in the tall-grass vegetation was mostly explained by differences between sites (figure 2b), which is similar to the overall pattern (figure 1b).

(c). How do inter- and intraspecific trait variance change when multiple traits are simultaneously considered?

When all plant species, all traits and both intraspecific and interspecific trait variability (only between sites) were combined in multivariate analyses, we found similar patterns (figure 3a,b,e,f). However, when assessing the intraspecific variability of multiple traits for each species separately, there were differences in trait variance between species (electronic supplementary material, figure S3) because of the highly variable responses of our eight species to the exclusion of herbivores (figure 2).

Figure 3.

Figure 3.

Open in a new tab

Multidimensional relationships of the four leaf traits measured in the short-grass and tall-grass vegetation, including differences in ordinations depending on whether intraspecific variance is also considered. (a,e) Total interspecific variation PCAs; (b,f) total interspecific and intraspecific variation PCAs; (c,g) species dispersion; (d,h) species dispersion and intraspecific variability. F1, Galium anisophyllon; F2, Crepis alpestris; L1, Lotus corniculatus; L2, Trifolium repens; G1, Festuca rubra; G2, Briza media; S1, Carex sempervirens; S2, Carex caryophyllea; SLA, specific leaf area; C, leaf carbon concentration; N, leaf nitrogen concentration; P, leaf phosphorus concentration.

(d). Do combinations of leaf traits discriminate between plant species and consequently, does the robustness assumption hold for the traits measured (i.e. is interspecific variance higher than intraspecific variance)?

We found that individuals of different species and even life forms showed overlapping distributions in their leaf traits based on calculated Mahalanobis distances, and these distributions were similar in the short- and tall-grass vegetation (figure 4a,b). Individuals of Carex sempervirens (sedge) and Festuca rubra (grass) overlapped considerably in their trait distribution distances (Mahalanobis distances); and Carex sempervirens individuals also showed some similarity to Galium anisophyllon (forb) and Briza media (grass) (figure 4). As expected, grass and sedge species and then forb and legume species were found to overlap in their Mahalanobis distances (electronic supplementary material, figures S4–S10).

Figure 4.

Figure 4.

Open in a new tab

Functional distance between species based on all four leaf traits. Densities of Mahalanobis distances are calculated between individuals from the species Carex sempervirens and individuals of all other focal species in (a) short-grass vegetation and (b) tall-grass vegetation. Small differences infer pairs of individuals show similar traits. Densities of Mahalanobis distances calculated for all species as the comparison are shown in the electronic supplementary material, figures S4–S10. (Online version in colour.)

The probability of attributing leaves to the correct species based on their traits was just over 50% in both vegetation types (i.e. mean = 57% for short-grass vegetation; mean = 55% for tall-grass vegetation). Sedge and forb species showed the lowest probability of being distinguished from other species based on the four leaf traits measured. In the short-grass vegetation, a leaf of Carex sempervirens had a 20% and 60% probability, respectively, of being wrongly attributed to Carex caryophyllea or Festuca rubra (figure 5a). In the tall-grass vegetation, there was an estimated 20–40% probability of a Carex caryophyllea leaf being mistaken for Carex sempervirens, Briza media or Festuca rubra leaves (figure 5b).

Figure 5.

Figure 5.

Open in a new tab

Probabilities of attributing a leaf of species A to species B based on LDA for all four traits. Probabilities of correctly predicting the origin of a leaf from its traits are given for the different observed and predicted species. Shaded circles shown outside the diagonal represent prediction errors. (a) Short-grass vegetation and (b) tall-grass vegetation. (Online version in colour.)

4. Discussion

We found accounting for intraspecific trait variance explains a higher proportion of variance when compared with interspecific variance and therefore, if measured, can improve predictions concerning how plant communities change functionally soil resource limitations and historical land use [8,21,42], but also short-term perturbations such as changes in herbivory. Other studies have found that intraspecific trait variability is high; however, they were largely observational and measured differences because of landscape gradients. Our study is unique in that we also found high intraspecific trait in response to a short-term herbivore exclosure experiment in two vegetation types with differing legacies of past land-use practices (greater than 100 years ago). We found considerable differences in the amount of intraspecific trait variance depending on the trait measured, plant life form and whether considering single or multiple traits of individuals [43]. When we considered all four traits simultaneously, we found just a 55% overall probability of successfully attributing a random plant leaf to the correct plant species. Similarly, the distributions of distances between the multivariate traits from two different species were sometimes closer than within a species, surprisingly even between the life forms of forbs and sedges.

Specific leaf area is a commonly used trait to characterize plant communities, as it has been found to correlate with investments plants make in capturing energy for growth versus plant defence strategies, particularly herbaceous species [44,45]. We found that SLA showed lower levels of intraspecific variability in comparison to the physiological traits. This suggests then that the robustness assumption held for our SLA measurements [21,46], and generally concurs with other studies that have shown SLA to have higher variability between than within a species [6,47]. Our findings, however, contrast with a study that reported higher intraspecific versus interspecific variability in SLA in response to local-scale environmental change [12]. These discrepancies may be explained by differences in trait manifestations between herbaceous and the woody species included in other studies, and the local-scale environmental change measured here is also of experimental and not only observational nature. Long-lived woody species can experience considerable ontogenetic changes and variability in irradiance on SLA development in complex forest canopies such as those of tropical forests [12,48,49]. Albert et al. [21] found higher intraspecific variability in SLA than our results here, but also included woody species (dwarf shrubs) in their study [21].

The responsiveness of leaf nutrients to herbivory and local site conditions in the same species is likely explained by the high mobility and translocation qualities of nutrients such as P and N [22]. Leaf P concentration showed a higher intraspecific than interspecific variability [50], whereas intraspecific and interspecific variabilities in leaf N and leaf C concentration were more evenly distributed [13]. The subalpine grasslands of the Swiss National Park are known to be limited by soil P availability, with the tall-grass vegetation displaying the lowest soil P levels [51]. In contrast to N that can be supplied to plants from bacteria, protozoa and fungi [52], plant available P is limited to soil stocks [53]. In addition, plant P uptake is dependent on herbivory levels, as herbivores are highly selective in the plant tissue they eat, choosing leaves with higher nutrient levels first [20,23,50]. Previous studies in this same experimental network of fences has shown that herbivores consume up to 60% of the biomass in the short-grass vegetation compared with less than 20% in the tall-grass vegetation [19,20].

Our findings raise important considerations for designing future trait-based studies. The results validate the use of mean values at the species level for SLA in herbaceous communities to assess differences in community level functional diversity and ecosystem functioning, while traits of leaf nutrient concentration show a high intraspecific variability in response to local-scale changes in site conditions such as soil fertility and changes in herbivory. Therefore, studies aimed at measuring short-term herbivore effects on leaf functional traits should focus on physiological traits such as leaf nutrients, rather than morphological traits such as SLA. It is important to note that a recent study found that across a diverse set of globally disturbed grasslands neither leaf nutrient concentrations nor SLA were consistent indicators of vertebrate consumer exclusion [18], although this study did not measure intraspecific trait variance.

Our results show the importance of matching the traits measured with the context of the ecosystems being studied. In our case, we found leaf P concentration could be used to detect within species responses to herbivory, because the subalpine grasslands we were working in are P-limited [54,55]. This also suggests that in some cases measuring just one trait may be sufficient, depending on the research questions and the environmental context. When we used multiple traits to understand responses to environmental change, patterns were largely consistent across our vegetation types, but not at the level of life form or species. This finding suggests that single-trait approaches, especially leaf nutrient concentrations, may produce misleading results for understanding functional responses at the community level because interactions between the growth trade-off in plants, herbivores and environmental context, and that evaluations of plant species functional responses to environmental change may be more reliably detected using multiple traits from multiple individuals within a species [56,57].

Overall, our results show the importance of measuring intraspecific trait variability, especially when using physiological traits, as intraspecific variability can be high when compared to interspecific variability. Consequently, our findings suggest that the robustness assumption does not always hold. This confirms what was suggested by another study, which concluded that ‘species tell more about traits than traits about species’ [58].

Supplementary Material

Additional Figures and Table
rspb20190429supp1.docx (3.5MB, docx)
Reviewer comments
rspb20190429_review_history.pdf (954.2KB, pdf)

Acknowledgements

We thank various employees and volunteers of the Swiss Federal Institute for Forest, Snow, and Landscape Research and the Swiss National Park for assistance with fence construction and maintenance. We thank Constant Signarbieux and Felix Hernandez for their invaluable help in the field and Emma LaDouceur and Max Rosenthal for leaf grinding and weighing in the laboratory.

Data accessibility

Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.n41d2dv [59].

Authors' contributions

J.F., A.C.R. and M.S. carried out field work, participated in data analysis, participated in the design of the study and drafted the manuscript; H.N. participated in data analyses and assisted by editing the manuscript.

Competing interests

The authors do not have any conflicts of interest to declare.

Funding

This study was funded by the Swiss National Science Foundation (grant nos. 31003A_122009/1 and 31003A_140939/1).

References

  • 1.Funk JL, et al. 2017. Revisiting the Holy Grail: using plant functional traits to understand ecological processes. Biol. Rev. 92, 1156–1173. [DOI] [PubMed] [Google Scholar]
  • 2.Lavorel S, Garnier E. 2002. Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Funct. Ecol 16, 545–556. ( 10.1046/j.1365-2435.2002.00664.x) [DOI] [Google Scholar]
  • 3.Diaz S, et al. 2004. The plant traits that drive ecosystems: evidence from three continents. J. Veg. Sci. 15, 295–304. ( 10.1111/j.1654-1103.2004.tb02266.x) [DOI] [Google Scholar]
  • 4.Kattge J, et al. 2011. TRY—a global database of plant traits. Glob. Change Biol. 17, 2905–2935. [Google Scholar]
  • 5.Albert CH, Thuiller W, Yoccoz NG, Soudant A, Boucher F, Saccone P, Lavorel S. 2010. Intraspecific functional variability: extent, structure and sources of variation. J. Ecol. 98, 604–613. ( 10.1111/j.1365-2745.2010.01651.x) [DOI] [Google Scholar]
  • 6.McGill BJ, Enquist BJ, Weiher E, Westoby M. 2006. Rebuilding community ecology from functional traits. Trends Ecol. Evol. 21, 178–185. ( 10.1016/j.tree.2006.02.002) [DOI] [PubMed] [Google Scholar]
  • 7.Violle C, Enquist BJ, McGill BJ, Jiang L, Albert CH, Hulshof C, Jung V, Messier J. 2012. The return of the variance: intraspecific variability in community ecology. Trends Ecol. Evol. 27, 244–252. ( 10.1016/j.tree.2011.11.014) [DOI] [PubMed] [Google Scholar]
  • 8.Albert CH, Grassein F, Schurr FM, Vieilledent G, Violle C. 2011. When and how should intraspecific variability be considered in trait-based plant ecology? Persp. Plant Ecol. Evol. Syst. 13, 217–225. ( 10.1016/j.ppees.2011.04.003) [DOI] [Google Scholar]
  • 9.Bolnick DI, Amarasekare P, Araujo MS, Burger R, Levine JM, Volker HWR, Schreiber SJ, Urban MC, Vasseur DA. 2011. Why intraspecific trait variability matters in community ecology. Trends Ecol. Evol. 26, 183–192. ( 10.1016/j.tree.2011.01.009) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Siefert A, et al. 2015. A global meta-analysis of the relative extent of intraspecific trait variation in plant communities. Ecol. Lett. 18, 1406–1419. ( 10.1111/ele/12508) [DOI] [PubMed] [Google Scholar]
  • 11.Messier J, McGill BJ, Lechowicz MJ. 2010. How do traits vary across ecological scales? A case for trait-base ecology. Ecol. Lett. 13, 838–848. ( 10.1111/j.1461-0248.2010.01476.x) [DOI] [PubMed] [Google Scholar]
  • 12.Messier J, McGill BJ, Enquist BJ, Lechowicz MJ. 2016. Trait variation and integration across scales: is the leaf economic spectrum present at local scales? Ecography 39, 1–13. ( 10.1111/ecog.01369) [DOI] [Google Scholar]
  • 13.Kinchenin E, Wardle DA, Morse CW, Freschet GT. 2013. Contrasting effects of plant inter- and intraspecific variation on community-level trait measures along an environmental gradient. Funct. Ecol. 27, 1254–1261. ( 10.1111/1365-2435.12116) [DOI] [Google Scholar]
  • 14.Midolo G, De Frenne P, Holzel N, Wellstein C.. 2019. Global patterns of instraspecific leaf trait responses to elevation. Glob. Change Biol. 25, 2485–2498. [DOI] [PubMed] [Google Scholar]
  • 15.Albert CH, de Bello F, Boulangeat I, Pellet G, Lavorel S, Thuiller W.. 2012. On the importance of intraspecific variability for the quantification of functional diversity. Oikos 121, 116–126. ( 10.1111/j.1600-0706.2011.19672.x) [DOI] [Google Scholar]
  • 16.Wright IJ, et al. 2004. The worldwide leaf economics spectrum. Nature 428, 821–827. ( 10.1038/nature02403) [DOI] [PubMed] [Google Scholar]
  • 17.Firn J, Schuetz M, Nguyen H, Risch AC. 2017. Herbivores sculpt leaf traits differently in grasslands depending on life form and land-use histories. Ecology 98, 239–252. ( 10.1002/ecy.1637) [DOI] [PubMed] [Google Scholar]
  • 18.Firn J, et al. 2019. Leaf nutrients, not specific leaf area, are consistent indicators of elevated nutrient inputs. Nat. Ecol. Evol. 3, 400 ( 10.1038/s41559-018-0790-1) [DOI] [PubMed] [Google Scholar]
  • 19.Risch AC, Haynes AG, Busse MD, Filli F, Schüetz M. 2013. The response of soil CO2 fluxes to progressively excluding vertebrate and invertebrate herbivores depends on ecosystem type. Ecosystems 16, 1192–1202. ( 10.1007/s10021-013-9676-x) [DOI] [Google Scholar]
  • 20.Schütz M, Risch AC, Achermann G, Thiel-Egenter C, Page-Dumroese DS, Jurgensen MF, Edwards PJ. 2006. Phosphorus translocation by red deer on a subalpine grassland in the Central European Alps. Ecosystems 9, 624–633. ( 10.1007/s10021-006-0091-4) [DOI] [Google Scholar]
  • 21.Albert CH, Thuiller W, Yoccoz NG, Douzet R, Aubert S, Lavorel S. 2010. A multi-trait approach reveals the structure and the relative importance of intra- vs. interspecific variability in plant traits. Funct. Ecol. 24, 1192–1201. ( 10.1111/j.1365-2435.2010.01727.x) [DOI] [Google Scholar]
  • 22.Chapin FS., III 1980. The mineral nutrition of wild plants. Annu. Rev. Ecol. Syst. 11, 233–260. ( 10.1146/annurev.es.11.110180.001313) [DOI] [Google Scholar]
  • 23.Firn J, Price JN, Whalley RDB. 2013. Using strategically applied grazing to manage invasive alien plants in novel grasslands. Ecol. Process. 26 (http://www.ecologicalprocesses.com/content/2/1/26). [Google Scholar]
  • 24.Schütz M, Risch AC, Leuzinger E, Kruesi BO, Achermann G. 2003. Impact of herbivory by red deer (Cervus elaphus L.) on patterns and processes in subalpine grasslands in the Swiss National Park. For. Ecol. Manag. 181, 177–188. ( 10.1016/S0378-1127(03)00131-2) [DOI] [Google Scholar]
  • 25.Risch AC, et al. 2018. Size-dependent loss of aboveground animals differentially affects grassland ecosystem coupling and functions. Nat. Commun. 9, 1–11. ( 10.1038/s41467-018-06105-4) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Cornelissen JHC, et al. 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335–380. ( 10.1071/BT02124) [DOI] [Google Scholar]
  • 27.Abramoff MD, Magalhaes PJ, Ram SJ. 2004. Image processing with ImageJ. Biophoton. Int. 11, 36–42. [Google Scholar]
  • 28.Rayment GE, Higginson FR. 1992. The Australian handbook of soil and water chemical methods. Melbourne, Australia: Inkata Press. [Google Scholar]
  • 29.Wickham H. 2009. Ggplot2: elegant graphics for data analysis. New York, NY: Springer. [Google Scholar]
  • 30.Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, White J-SS. 2008. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol. Evol. 24, 127–135. ( 10.1016/j.tree.2008.10.008) [DOI] [PubMed] [Google Scholar]
  • 31.Bates D, Maechler M, Bolker B.. 2012. Package ‘Lme4’. CRAN.
  • 32.Barton K. 2012. Multi-model inference. R package version 172 See http://CRAN.R-project.org/package=MuMIn.
  • 33.Xu R. 2003. Measuring explained variation in linear mixed effects models. Stat. Med. 22, 3527–3541. ( 10.1002/sim.1572) [DOI] [PubMed] [Google Scholar]
  • 34.Dodelec S, Chesser D. 1991. Recent developments in linear ordination methods for environmental sciences. Adv. Ecol. 1, 133–155. [Google Scholar]
  • 35.Husson F, Josse J, Le S, Mazet J.. 2017. FactoMineR: Mulitvariate exploratory data analysis and data mining. CRAN.
  • 36.Dray S, Dufour A-B, Thioulouse J.. 2016. ade4, Analysis of ecological data: exploratory and Euclidean methods. CRAN.
  • 37.Mahalanobis PC. 1936. On the generalised distance in statistics. Proc. Natl Inst. Sci. India 2, 49–50. [Google Scholar]
  • 38.Venables WN, Ripley BD. 2002. Modern applied statistics with S, 4th edn New York, NY: Springer. [Google Scholar]
  • 39.Ripley B, Venables B, Bates D, Hornik K, Gebhardt A, Firth D. 2016. MASS. CRAN.
  • 40.Wei T, Simko V. 2016. corrplot: Visualization of a Correlation Matrix. CRAN.
  • 41.R Development Core Team. 2013. R: language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; (http://www.R-project.org/). [Google Scholar]
  • 42.Jung V, Albert CH, Violle C, Kunstler G, Gregory L, Speigelberger T. 2014. Intraspecific trait variability mediates the response of subalpine grassland communities to extreme drought events. J. Ecol. 102, 45–53. ( 10.1111/1365-2745.12177) [DOI] [Google Scholar]
  • 43.Suding KN, Goldberg DE, Hartman KM. 2003. Relationships among species traits: separating levels of response and identifying linkages to abundance. Ecology 84, 1–16. ( 10.1890/0012-9658(2003)084[0001:RASTSL]2.0.CO;2) [DOI] [Google Scholar]
  • 44.Perez-Harguindeguy N, et al. 2013. New handbook for standardised measurement of plant functional traits worldwide. Aust. J. Bot. 61, 167–234. ( 10.1071/BT12225) [DOI] [Google Scholar]
  • 45.Garnier E, et al. 2004. Plant functional markers capture ecosystem properties during secondary succession. Ecology 85, 2630–2637. ( 10.1890/03-0799) [DOI] [Google Scholar]
  • 46.Garnier E, Shipley B, Roumet C, Laurent G. 2001. A standardized protocol for the determination of specific leaf area and leaf dry matter content. Funct. Ecol. 15, 688–695. ( 10.1046/j.0269-8463.2001.00563.x) [DOI] [Google Scholar]
  • 47.Roche P, Diaz-Burlinson N, Gachet S. 2004. Congruency analysis of species ranking based on leaf traits: which traits are the more reliable? Plant Ecol. 174, 37–48. ( 10.1023/B:VEGE.0000046056.94523.57) [DOI] [Google Scholar]
  • 48.Rozendaal DMA, Hurtado VH, Poorter L. 2006. Plasticity in leaf traits of 38 tropical tree species in response to light; relationships with light demand and adult stature. Funct. Ecol. 20, 207–216. ( 10.1111/j.1365-2435.2006.01105.x) [DOI] [Google Scholar]
  • 49.Poorter L, et al. 2008. Are functional traits good predictors of demographic rates? Evidence from five Neotropical forests. Ecology 89, 1908–1920. ( 10.1890/07-0207.1) [DOI] [PubMed] [Google Scholar]
  • 50.Niu K, Messier J, He J-S, Lechowicz MJ. 2015. The effects of grazing on foliar trait diversity and niche differentiation in the Tibetan alpine meadows. Ecosphere 6, 1–15. ( 10.1890/ES14-00174.1) [DOI] [Google Scholar]
  • 51.Schütz M, Wohlgemuth T, Krüesi BO, Achermann G, Gramiger H. 2000. Influence of grazing pressure on species richness in subalpine grassland in the Swiss National Park. Nationalpark-Forschung in der Schweiz 89, 39–65. [Google Scholar]
  • 52.Clarholm M. 1985. Possible roles for roots, bacteria, protozoa and fungi in supplying nitrogen to plants. In Ecological interactions in soil (ed. AH Fitter), pp. 255–265. Oxford, UK: Blackwell. [Google Scholar]
  • 53.Holford ICR. 1997. Soil phosphorus: its measurement, and its uptake by plants. Aust. J. Soil Res. 35, 227–240. ( 10.1071/S96047) [DOI] [Google Scholar]
  • 54.Loranger J, Meyer ST, Shipley B, Kattge J, Loranger H, Roscher C, Weisser WW. 2012. Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology 93, 2674–2682. ( 10.1890/12-0328.1) [DOI] [PubMed] [Google Scholar]
  • 55.Lind EM, Myron EP, Giaccai J, Parker JD. 2012. White-tailed deer alters specialist and generalist insect herbivory through plant traits. Environ. Entomol. 41, 1409–1416. ( 10.1603/EN12094) [DOI] [PubMed] [Google Scholar]
  • 56.Pigliucci M. 2003. Phenotypic integration: studying the ecology and evolution of complex phenotypes. Ecol. Lett. 6, 265–272. ( 10.1046/j.1461-0248.2003.00428.x) [DOI] [Google Scholar]
  • 57.Messier J, Lechowicz MJ, McGill BJ, Violle C, Enquist BJ. 2017. Interspecific integration of trait dimensions at local scales: the plant phenotype as an integrated network. J. Ecol. 105, 1775–1790. ( 10.1111/1365-2745.12755) [DOI] [Google Scholar]
  • 58.Clark JS. 2016. Why species tell more about traits than traits about species: predictive analysis. Ecology 97, 1979–1993. ( 10.1002/ecy.1453) [DOI] [PubMed] [Google Scholar]
  • 59.Firn J, Nguyen H, Schütz M, Risch AC. 2019. Data from: Leaf trait variability between and within subalpine grassland species differs depending on site conditions and herbivory Dryad Digital Respository. ( 10.5061/dryad.n41d2dv) [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

  1. Firn J, Nguyen H, Schütz M, Risch AC. 2019. Data from: Leaf trait variability between and within subalpine grassland species differs depending on site conditions and herbivory Dryad Digital Respository. ( 10.5061/dryad.n41d2dv) [DOI] [PMC free article] [PubMed]

Supplementary Materials

Additional Figures and Table
rspb20190429supp1.docx (3.5MB, docx)
Reviewer comments
rspb20190429_review_history.pdf (954.2KB, pdf)

Data Availability Statement

Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.n41d2dv [59].

Articles from Proceedings of the Royal Society B: Biological Sciences are provided here courtesy of The Royal Society

RESOURCES