Abstract Background and Aims How functional traits vary with environmental conditions is of fundamental importance in trait-based community ecology. However, how intraspecific variability in functional traits is connected to species distribution is not well understood. This study investigated inter- and intraspecific variation of a key functional trait, i.e. specific leaf area (leaf area per unit dry mass; SLA), in relation to soil factors and tested if trait variation is more closely associated with specific environmental regimes for low-variability species than for high-variability species. Methods In a subtropical evergreen forest plot (50 ha, southern China), 106 700 leaves from 5335 individuals of 207 woody species were intensively collected, with 30 individuals sampled for most species to ensure a sufficient sample size representative of intraspecific variability. Soil conditions for each plant were estimated by kriging from more than 1700 observational soil locations across the plot. Intra- and interspecific variation in SLA were separately related to environmental factors. Based on the species-specific variation of SLA, species were categorized into three groups: low-, intermediate- and high-intraspecific variability. Intraspecific habitat ranges and the strength of SLA–habitat relationships were compared among these three groups. Key Results Interspecific variation in SLA overrides the intraspecific variation (77 % vs. 8 %). Total soil nitrogen (TN, positively) and total organic carbon (TOC, negatively) are the most important explanatory factors for SLA variation at both intra- and interspecific levels. SLA, both within and between species, decreases with decreasing soil nitrogen availability. As predicted, species with low intraspecific variability in SLA have narrower habitat ranges with respect to soil TOC and TN and show a stronger SLA–habitat association than high-variability species. Conclusions For woody plants low SLA is a phenotypic and probably adaptive response to nitrogen stress, which drives the predominance of species with ever-decreasing SLA towards less fertile habitats. Intraspecific variability in SLA is positively connected to species’ niche breadth, suggesting that low-variability species may play a more deterministic role in structuring plant assemblages than high-variability species. This study highlights the importance of quantifying intraspecific trait variation to improve our understanding of species distributions across a vegetated landscape. Environmental gradient, intraspecific trait variation, niche, specific leaf area, soil nitrogen limitation INTRODUCTION How functional traits vary across particular environmental gradients is a central question in trait-based community ecology (McGill et al., 2006; Shipley et al., 2016). It is commonly assumed that species-mean trait values represent the niche differentiation among species (Sterck et al., 2011; D’Andrea and Ostling, 2016). However, recent empirical studies have demonstrated that interspecific trait variation does not always override intraspecific variation and the latter may be important in community trait composition at fine scales (Messier et al., 2010; Albert et al., 2011; Auger and Shipley, 2013; Siefert et al., 2015). In addition, studies, empirically and theoretically, have increasingly shown that intraspecific variation contributes substantially to species coexistence and biodiversity maintenance (Violle et al., 2012; Adler et al., 2013; Siefert, 2014; Laughlin and Joshi, 2015). Thus, quantifying intraspecific trait variation along with the interspecific variation is crucial to understanding niche differentiation and species distribution at local scales. Specific leaf area (SLA), defined as the ratio of leaf surface area to dry mass, is a key functional trait for plants (Cornelissen et al., 2003; Pérez-Harguindeguy et al., 2013). SLA reflects the area available for solar irradiance interception per unit of assimilate investment (Milla and Reich, 2007). As a way of resource allocation within leaves, SLA profoundly influences the efficiency of assimilating photosynthetic active radiation and ultimately plant photosynthetic capacity (Green et al., 2003). In addition, SLA is closely associated with other biologically important traits including leaf nitrogen concentration, lifespan and plant relative growth rate (Lambers and Poorter, 1992; Reich et al., 1997, 2003; Wright et al., 2004). Representing one of the main ecological strategies of plants (Westoby, 1998; Wright et al., 2004), SLA is critical for modulating plant species distribution and community assembly (cf. Niinemets and Kull, 1994; Ackerly et al., 2002; Burns, 2004; Poorter et al., 2009; Hulshof et al., 2013). Interspecific differences in SLA have been closely linked to niche partitioning among species, with low-SLA species associated with a slow growth strategy and infertile soils (Poorter and De Jong, 1999; Baraloto et al., 2006; Sterck et al., 2006). However, the mean trait value of a species mainly reflects its niche positioning rather than niche breadth. Theoretical studies proposed that niche breadth would be defined by intraspecific trait variability, the latter being due to a combination of genotypic variation and trait plasticity (Violle and Jiang, 2009). Species with low trait variability may display a high level of habitat specialization (Sultan, 2000) and possess a relative fitness advantage within their preferred habitats (Caley and Munday, 2003). However, only sparse empirical evidence exists in support of this argument (Laurans et al., 2012; Sides et al., 2014), which calls for further tests. Moreover, the link between intraspecific trait variability and niche breadth, if any, could play a significant role in shaping community assembly. Intraspecific trait variability determines the probability that a species will pass through environmental filters (Jung et al., 2010; Violle et al., 2012). In a highly heterogeneous system, low-variability species can only establish in habitats close to their niche optimum (Umaña et al., 2015), but high-variability species could potentially colonize across the entire landscape due to their broad niche width. Thus, we predict that trait variation is more closely associated with specific environmental regimes for low-variability species than for high-variability species (Fig. 1). Fig. 1. View largeDownload slide Conceptual links between intraspecific trait variability and niche occupancy. Species with low intraspecific trait variability are expected to occupy narrow niches and have more predictable positioning along a trait spectrum in terms of their environmental niche (left panels), whereas species with high intraspecific trait variability are expected to display the opposite (right panels). Fig. 1. View largeDownload slide Conceptual links between intraspecific trait variability and niche occupancy. Species with low intraspecific trait variability are expected to occupy narrow niches and have more predictable positioning along a trait spectrum in terms of their environmental niche (left panels), whereas species with high intraspecific trait variability are expected to display the opposite (right panels). In this study, we aimed to quantify the intra- and interspecific variations of SLA and reveal the effects of soil conditions on the SLA variations to understand the plant species assemblage in a stem-mapped subtropical forest (50 ha) in China. SLA data were intensively sampled and soil properties were fully mapped across the plot. To empirically test the hypothesis, we addressed the following three important questions: (1) How does SLA vary with edaphic factors both within and between species? (2) Do species with low SLA variability show a stronger relationship between SLA and habitats than high-variability species? (3) Do species with low SLA variability have narrower habitat ranges than high-variability species? Answers to these questions would contribute substantially to understanding the functional significance of SLA for species distribution and community assembly. MATERIALS AND METHODS Study site The study site is located in the Heishiding Provincial Natural Reserve (23°27′N, 111°52′E), Guangdong Province, southern China. The region has a humid subtropical monsoon climate with mean annual temperature of 19.6 °C and mean annual precipitation of 1740 mm. Relief is predominantly of low hills, with elevation ranging from 150 to 1000 m above sea level. The vegetation is characterized by evergreen broadleaved forests dominated by Lauraceae, Fagaceae and Theaceae species. Leaf flushing occurs from March to May for most woody species in the region (Li and Wang, 1984). Following the field protocol of the CTFS-forestGEO network (http://www.forestgeo.si.edu/), we established a 50-ha (500 × 1000 m) forest plot in the reserve to monitor long-term forest dynamics. The plot roughly covers a local catchment, with runoffs generally flowing north (Fig. 2). The bedrocks are granitic, on which ferrasol develops as the dominant soil type. Elevation ranges from 430 to 700 m. Slope inclination varies greatly between 10° and 70°. Fig. 2. View largeDownload slide The geographical location of Heishiding forest plot and the distribution of 5335 sampled plants within the plot. Values are in metres and colours represent altitude above sea level. Fig. 2. View largeDownload slide The geographical location of Heishiding forest plot and the distribution of 5335 sampled plants within the plot. Values are in metres and colours represent altitude above sea level. All free-standing woody stems with diameter at breast height (dbh) ≥ 1 cm in the plot were tagged, measured, identified to species and mapped within a geographical coordinate frame. A total of 264 394 stems, belonging to 214 woody species, 129 genera and 60 families, were recorded. Most of the 214 woody species were evergreen tree species, along with 32 shrubs and 25 deciduous tree species (Supplementary Data Table S1). The two most abundant species were Cryptocarya concinna Hance (Lauraceae) and Neolitsea phanerophlebia Merrill (Lauraceae), with abundances of 24 302 and 16 594, respectively. Leaf sampling and measurement In 2013 and 2014, leaf sampling was conducted within the 50-ha forest plot during the summer (from June to August) after the leaf flushing season. To capture the community-wide leaf trait variation, we adopted a stratified random sampling strategy with particular reference to species. For species with abundance ≥ 50 individuals across the plot, 30 individuals were randomly selected for leaf sampling. For species with abundance < 50 individuals, three to 30 individuals were sampled depending on the abundance of those species and the availability of satisfactory leaves. We first sampled leaves along random routes, with nearly 60 % of samples representing common species collected in this step. Subsequently, we sampled rare species according to their locations and in the meantime covered a few common species yet to reach 30 samples. Seven extremely rare species were not sampled due to either tree mortality or the lack of satisfactory leaves (see below). In total, we sampled 5335 individuals across 207 species. All sampled individuals were well interspersed across the 50-ha plot and a wide variety of microhabitat types was thus encompassed (Fig. 2). The mean dbh of all sampled individuals was very close to that for the whole plot (5.01 ± 0.09 vs. 5.55 ± 0.02 cm, mean ± s.e.), which indicates the size structure of the forest was representatively sampled. Twenty healthy-looking, fully developed current-year leaves were detached from each plant, with the assistance of tree pruners where necessary. We removed leaves from one or two twigs about 1–3 cm in diameter from the outer canopy. Note that the vertical range of leaf sampling was limited to the lowest 6.8 m (height of the person sampling plus pruner length). Our sampling was therefore biased towards collecting leaves from the understorey and may be ontogenetically biased towards early stages for canopy species. On the other hand, it facilitated SLA comparison of understorey plants and saplings of overstorey species in relation to soil parameters at similar (shaded) light regimes. In total, we collected 106 700 leaves from 5335 individuals for subsequent analyses. All sampled leaves were wrapped in moist filter paper and packed into resealable plastic bags before laboratory processing. Within 24 h of field sampling, leaf area was measured with an LI-3000C leaf area meter (LI-COR) to the nearest 0.01 cm2. For some exceptionally large or non-flat leathery leaves, a digital scanner plus with the software ImageJ (v1.48, http://imagej.nih.gov/ij/) was used to measure leaf area. Dry mass was measured to the nearest 0.0001 g after 48 h of oven-drying at 80 °C. Soil survey Soil cores were extracted to a depth of 10 cm from 1708 sites across the 50-ha plot. The sampling sites were spatially outlined on a regular grid of points every 30 m, with each grid intersection accompanied by additional sample points at 2, 4 or 12 m in a random compass direction (N, E, S, W, NE, NW, SE or SW) from the grid. All soil cores were extracted within 5 h around noon on clear days in autumn 2013. Six soil physical and chemical properties were measured for each core: moisture (MST), total organic carbon (TOC), total nitrogen (TN), available phosphorus (AP), available potassium (AK) and available aluminium (AAl). These factors have been shown to be relevant for shaping species composition in this forest (Zhu, 2015); and AAl was considered because of its toxic effects on plants especially in acid soils (Kochian et al., 2004) that are known to occur in this region (pH = 4.38 ± 0.26, mean ± s.d.). Soil moisture was gravimetrically determined. A TOC analyser (Shimadzu, TOC-Control V 2.00) was used to determine the TOC concentration using the acid sparing (total carbon minus inorganic carbon) method. TN was determined colorimetrically by using a SmartChem-200 automated discrete analyser (AMS westco, Smartchem 3.1.30) after Kjeldahl digestion. AP, AK and AAl were extracted in Mehlich III solution and measured with an ICP-OES 2100DV atomic emission spectrometer (PerkinElmer, Winlab32 4.0). A summary of these edaphic properties is presented in Supplementary Data Table S2. Kriging interpolation Kriging maps were created for each of the soil properties for the 50-ha plot so that the soil data can be obtained at any location in the plot, including the locations of the 5335 sampled trees. First, observed soil properties were Box–Cox transformed and detrended by quadratic trend-surface regressions. Empirical variograms were computed based on the residuals after detrending and fitting against spherical or exponential models. Models with the lowest residual sums of squares were selected for ordinary kriging. The removed trend was then added back to the kriged values, and the values were back-transformed to the original scale. Leave-one-out cross-validations showed that the coefficient of variation (CV) of the root-mean-square deviation was <16 % in most cases except that AP had a root-mean-square deviation of = 25 % (see details in Table S2). Geostatistical modelling was performed using the R package ‘geoR’(Ribeiro and Diggle, 2001). Data analysis Pairwise Pearson’s correlations for the six edaphic variables were performed to examine collinearity between variables. The strongest correlation was found between TOC and TN with a coefficient of 0.46, followed by that between MST and TN with a coefficient of 0.34. The highest variance inflation factor (VIF) in the dataset was 1.38, much less than the commonly recommended VIF = threshold of 10, and therefore no factor was considered redundant and all were retained. All pairwise correlation coefficients for the six variables are shown in Supplementary Data Table S3. We used a linear random effect model (LRE) to decompose the hierarchical variance (Messier et al., 2010; Auger and Shipley, 2013), with leaves nested within individuals and with individuals nested within species. Sampling dates were also incorporated as a random factor in the model. The nested LRE was fit using restricted maximum likelihood (REML). Intra- and interspecific variation in SLA were separately modelled in relation to environmental factors. The following linear mixed effects model (LME) was applied at the intraspecific level: yij=β0+∑6k=1βkeijk+ui+εij (1) where yij is the SLA (averaged across 20 leaves and then log-transformed) of the jth individual plant of the ith species, eijk is the kth environmental variable (centred and scaled by the standard deviation), βk is the fixed effect of each of the k = 1, 2, …, 6 environmental variables, β0 is the fixed intercept, ui is the random variance term to account for the species-level SLA variation, and εij is the error. This model evaluated how SLA of individual plants varies with proximate environmental variables. Marginal and conditional R2 values (for the variance explained by the fixed and random effects, respectively) were obtained following the method proposed by Nakagawa and Schielzeth (2013). We also incorporated a random slope term (i.e. interspecific differences in response to soil variables) into this model but found the coefficient estimates were minimally influenced. In addition, we also considered a spherical spatial correlation structure into this model but found that the results were well preserved (Supplementary Data Table S4). Here we only presented the results from models without considering random slope or spatial autocorrelation. At the interspecific level, SLA was modelled using a simple linear regression: ui=α0+∑k=16αk∑j=1neijkn+εi (2) where ui is the random effect (i.e. averaged SLA) of the ith species in Model (1), α0 is the intercept, α1, α2, α3 and up to αk are the effects of species-averaged environmental variables (centred and scaled by the standard deviation) and n is the number of sampled plants of the ith species. This model evaluates how species with different SLA are distributed along environmental gradients. The most parsimonious model was selected using backward selection based on the Akaike (1974) information criterion (AIC) (Burnham and Anderson, 2003). The R packages ‘stats’ and ‘lmerTest’ (Kuznetsova et al., 2014) were used in model fitting and model selection. The CV of SLA across conspecific individuals was assumed to represent a species’ capacity to adjust its traits according to the local environment. No evident association was detected between species-specific CV and sample size (r = 0.12, P = 0.08, n = 207). The hypervolume of edaphic factors within a species was assumed to reflect the species’ niche breadth. We used the algorithm proposed by Blonder et al. (2014) to quantify each species’ niche space. This algorithm integrated multivariate kernel density estimation and importance-sampling Monte Carlo methods to define a hypervolume. Bandwidth was specified with a Silverman estimator across 264 394 individuals and then uniformly used for each species. In addition to high-dimensional hypervolumes defined by all six edaphic factors, we considered low-dimensional ones defined by edaphic factors showing significant relationships with SLA. We tested the correlations between species-specific CV in SLA and edaphic hypervolumes, with sample size (species abundance for niche hypervolumes) as the controlling factor. We also quantified edaphic hypervolume based on a reduced dataset (i.e. the 5335 plants with SLA measured) and checked its correlation with intraspecific SLA variability (i.e. CV). The 33th and 66th percentiles of species-specific CV were taken as the cut-points to categorize 207 species into three equal-size subsets, referring to low-, intermediate- and high-variability species, respectively. Equation (2) was applied separately to each subset of species so that we could examine the differences in environmental constraints on the three subsets. Specifically, we were concerned with whether the coefficient of determination (R2) was lower for high- than for low-variability species. Equation (1) was also applied separately to three subsets with different degree of SLA variability but at the individual level. For this, determined whether intraspecific SLA variation originating from environmental forces (marginal R2) differed between the contrasting subsets. RESULTS SLA ranges from 55.4 to 652.4 cm2 g−1 across all sampled individuals with a mean of 146.8 cm2 g−1. The two most abundant species have similar SLA, both of which are close to the average value (Cryptocarya concinna with 143.9 cm2 g−1 and Neolitsea phanerophlebia with 135.7 cm2 g−1). The species with the lowest and highest mean SLA are Camellia semiserrata C.W. Chi (76.9 cm2 g−1) and Alangium chinense (Lour.) Harms (461.8 cm2 g−1), respectively. Variance partitioning demonstrates that 77.1, 8.3 and 14.6 % of the total variation in SLA are attributable to species, individual and unexplained origins, respectively (Fig. 3). The contribution of leaves-within-individual and sampling dates to total variation is negligible (<<0.001 and 0.019 %, respectively) when these two factors were incorporated into the nested model structure. Fig. 3. View largeDownload slide Hierarchical variance partitioning in specific leaf area as revealed by the linear random effects model. Fig. 3. View largeDownload slide Hierarchical variance partitioning in specific leaf area as revealed by the linear random effects model. SLA varies positively with soil TN but negatively with soil TOC, and these trends are consistent at both intraspecific and interspecific levels (Fig. 4). Environmental factors account for 1.2 % of intraspecific SLA variation (marginal R2 = 0.001 divided by intraspecifically originating variance 0.083) and 8.5 % of interspecific SLA variation, respectively. Intraspecific variability in SLA is positively correlated with the soil niche hypervolume defined by TOC and TN but not with that defined by all six factors (Fig. 5). When the soil niche hypervolume was quantified for the 5335 plants with SLA measured, these relationships were qualitatively similar (results not shown). Fig. 4. View largeDownload slide Effect sizes of environmental variables on specific leaf area at the intraspecific (A) and interspecific (B) levels. Only the most parsimonious models based on the Akaike information criterion are shown (i.e. predictors with zero effect size were excluded). The magnitude (± s.e.) and direction of the effects are shown by slope estimates determined from the linear models. The overall strength of the effects is shown by the coefficient of determination (R2). The variations explained by environmental variables (R2m) and by species (R2c) are also shown. Abbreviations on the y-axis: MST = soil moisture, TOC = soil total organic carbon, TN = total nitrogen, AP = available phosphorus, AK = available potassium, AAl = available aluminium. Fig. 4. View largeDownload slide Effect sizes of environmental variables on specific leaf area at the intraspecific (A) and interspecific (B) levels. Only the most parsimonious models based on the Akaike information criterion are shown (i.e. predictors with zero effect size were excluded). The magnitude (± s.e.) and direction of the effects are shown by slope estimates determined from the linear models. The overall strength of the effects is shown by the coefficient of determination (R2). The variations explained by environmental variables (R2m) and by species (R2c) are also shown. Abbreviations on the y-axis: MST = soil moisture, TOC = soil total organic carbon, TN = total nitrogen, AP = available phosphorus, AK = available potassium, AAl = available aluminium. Fig. 5. View largeDownload slide Correlations between intraspecific variability (i.e. CV) in SLA and soil-defined hypervolume with sample size as the controlling variable. (A) Six-dimensional hypervolume defined by all soil factors. (B) Two-dimensional hypervolume defined by TOC and TN (i.e. two significant factors). Fig. 5. View largeDownload slide Correlations between intraspecific variability (i.e. CV) in SLA and soil-defined hypervolume with sample size as the controlling variable. (A) Six-dimensional hypervolume defined by all soil factors. (B) Two-dimensional hypervolume defined by TOC and TN (i.e. two significant factors). SLA is not related to any environmental factor for low-variability species (Fig. 6A). For high-variability species, only soil TN was strongly positively associated with SLA (Fig. 6E). However, for intermediate-variability species, three environmental factors, i.e. soil TN, TOC and AAl, significantly influenced SLA (Fig. 6C). Intraspecific SLA variation originating from environmental factors for each of the three groups (low-, intermediate- and high-variability species) is consistently low (marginal R2 ≤ 0.004, Fig. 6A, C, E). Fig. 6. View largeDownload slide Effect sizes of environmental variables on specific leaf area (SLA) for low-, intermediate- and high-SLA-variability species. Only the most parsimonious models based on the Akaike information criterion are shown (i.e. excluded predictors with zero effect size). The effects for the low-, intermediate- and high-SLA-variability species are successively shown from top to bottom. Left (A, C and E) and right (B, D and F) panels represent the intraspecific and interspecific level, respectively. See Fig. 4 for the interpretation of R2 and abbreviation notation on the y-axis. Fig. 6. View largeDownload slide Effect sizes of environmental variables on specific leaf area (SLA) for low-, intermediate- and high-SLA-variability species. Only the most parsimonious models based on the Akaike information criterion are shown (i.e. excluded predictors with zero effect size). The effects for the low-, intermediate- and high-SLA-variability species are successively shown from top to bottom. Left (A, C and E) and right (B, D and F) panels represent the intraspecific and interspecific level, respectively. See Fig. 4 for the interpretation of R2 and abbreviation notation on the y-axis. Low-variability species with high SLA tend to be distributed at locations with high soil nitrogen but low organic carbon (Fig. 6B). This pattern holds for the intermediate- and high-variability species with high SLA (Fig. 6). The proportion of interspecific SLA variation explained by environmental factors declines from the low- to high-variability species (21.0, 14.4 and 7.2 %; Fig. 6B, D, F). DISCUSSION SLA variation and edaphic factors Experimental evidence has demonstrated that nitrogen fertilization can increase SLA (Knops and Reinhart, 2000; Meziane and Shipley, 2001; Al Haj Khaled et al., 2005) and high nitrogen availability favours the dominance of high-SLA species (Craine et al., 2001). Likewise, our analysis showed that SLA is positively related to soil nitrogen concentration (TN) both within and across species (Fig. 4). In contrast, the relationship between SLA and soil organic carbon (TOC) displays an opposite pattern (Fig. 4). This is probably because more nitrogen was locked up in relatively recalcitrant soil organic matter and therefore less available for uptake by plants, and also because of stoichiometric constraints (Janssen, 1996; Ordoñez et al., 2009). In fact, a negative relationship between SLA and soil C/N ratio has been reported on a global scale (Ordoñez et al., 2009). In the present study, we found a similar negative association between these two variables (Supplementary Data Fig. S1a,f). This lends further support for a ubiquitous role of soil nitrogen limitation in driving SLA variability, especially at a given (understorey) light regime (Hodgson et al., 2011). For species with intermediate variability in SLA, a negative relationship between SLA and available aluminium content in soil (AAl) was detected (Fig. 6C). It is widely accepted that a high concentration of aluminium in soil would limit water and mineral nutrient uptake, comprising a stress for plants (Kochian et al., 2004). In fact, acid soils rich in soluble aluminium in cerrado vegetation were found to favour leaf scleromorphy (Delgado et al., 2013), a feature characterized by low SLA. After analysing species with intermediate intraspecific variability, we found that many of them were acid-sensitive, such as Michelia spp., Eurya spp., Manglietia moto, Rhodomyrtus tomentosa and Camellia sinensis (Hou, 1954). This is partly responsible for the significant SLA–AAl relationship found in this group. Although phosphorus is a limiting nutrient for plant development and growth throughout the world (Vance et al., 2003), SLA and soil AP were not related in the present study. Interestingly, at the global scale no effect of soil phosphorus on SLA was detected either (Ordoñez et al., 2009). This may intricately involve the interaction between soil phosphorus and radiation. Generally, for plant growth phosphorus deficiency discourages the interception of photosynthetically active radiation (PAR) rather than the efficiency of conversion of PAR into biomass (Colomb et al., 2000). The function of SLA is, however, largely linked to PAR conversion efficiency (Green et al., 2003), and this function seems more important in less well-exposed environments (Mitchell and Woodward, 1988) just as for the situation in the present study. Nitrogen limitation appeared to have prevalent impacts on SLA, as demonstrated at both the intra- and interspecific levels (Fig. 4), suggesting this factor represents a relevant environmental gradient in this subtropical forest plot. For the intermediate-variability species, besides nitrogen availability, available aluminum (AAl) strongly influenced SLA as well, but negatively (Fig. 6C). These results are in line with patterns observed in the leaf economic spectrum (Reich et al., 2003; Wright et al., 2004), which shows that more stressed habitats (particularly low nitrogen availability here) select against resource-acquisitive, weak and disposable leaves (with high SLA) but favour resource-retentive, sturdy and enduring leaves (with low SLA). Inter- vs. intraspecific SLA variation A meta-analysis of recent studies reported that intraspecific variation in SLA accounts for on average 30 % of the total variation within plant communities (Siefert et al., 2015). We found that interspecific SLA variation in a subtropical forest in South China quantitatively overrides its intraspecific counterpart (77 % vs. 8 %, Fig. 5). This result is not necessarily at odds with the expectation of more pronounced intraspecific trait variation at local scales from previous work (Albert et al., 2011; Auger and Shipley, 2013). In our study the rather low intraspecific SLA variability may to some extent be attributed to sampling leaves from the understorey. If we had collected overstorey leaves as well, the overall variability would probably have been greater (Keenan and Niinemets, 2016). Furthermore, the relative weight of interspecific/intraspecific trait variation could be context-dependent. Critically, defining how fine a local scale is should consider the dimensions of the community (e.g. forest vs. herbaceous vegetation) and the level of environmental heterogeneity. In a speciose and heterogeneous assemblage such as that in the present study, it is not surprising that interspecific trait variation would dominate the variability of the trait of interest (Siefert et al., 2015). It is notable that environmental heterogeneity in this forest is more salient in soil nutrient concentration, but gentle in moisture (Table S2). The locally narrow gradient of moisture may partly explain the lack of its relationship with SLA (Figs 4 and 6). Interspecific SLA variation, based at least partly on the evolutionary adaptation of the species involved, apparently has profound implications for species distribution across communities. Species with low SLA are associated with soils poor in total nitrogen but rich in total organic carbon (Fig. 4B), suggesting that low SLA or the underlying traits (e.g. leaf tissue density; Witkowski and Lamont, 1991) is beneficial in N-limited habitats. This highlights the functional significance of SLA in shaping species distribution along environmental gradients (Burns, 2004; Pollock et al., 2012) and again substantiates that SLA plays a crucial part in plant overall adaptive strategies (Westoby, 1998). For example, evergreen species on average have lower SLA than deciduous species (158 vs. 257 cm2 g−1, Table S1), and thus tend to specialize their niche on poor soils. At the intraspecific level, SLA also tends to decrease with stronger nitrogen limitation (Fig. 4A). This corroborated that there is a fundamental underlying association between SLA and nitrogen availability. Moreover, this finding indicates that a species could decrease its SLA in response to reduced nitrogen availability, and will be replaced by species with lower SLA if nitrogen supply is below some threshold. Low- vs. high-variability species Intraspecific variability in SLA is positively correlated with its habitat variability with respect to soil TOC and TN (Fig. 5). Similarly, Laurans et al. (2012) noted that SLA plasticity was lower in more specialized species along a light gradient in a tropical forest, while others documented that higher variability in SLA was related to broader species distribution across plant communities (Burns, 2004; Sides et al., 2014), which implies that intraspecific trait variability has a role in determining a species’ niche breadth (Laughlin and Joshi, 2015). Environmental variables explained more interspecific SLA variation where low-variability species were exclusively considered (Fig. 6B). Indeed, there was a more stringent SLA shift with environmental variations for low-variability species. This is in agreement with the result from Nordic lake islands, where species with lower intraspecific trait variability were more responsive to a composite environmental gradient linked to island size, soil fertility and spatial heterogeneity (Kumordzi et al., 2014). SLA of individual plants, from within either high- or low-variability species, showed similarly weak relationships with proximate habitats (Fig. 6A, C, E), which indicated that other abiotic and/or biotic constraints (e.g. light competition) not measured in this work could have operated on individual plants. Also, a plant’s SLA itself can influence the competitive effect it has on its neighbours (Kunstler et al., 2016), thereby indirectly affecting these neighbours’ SLA as well irrespective of any influence of soil regime. While individual plants were grouped into species, SLA of high-variability species were less associated with the species’ average habitats (Fig. 6F). This suggests that the SLA variation within each high-variability species can more flexibly adjust the species’ traits to improve overall fitness in given environmental conditions. Umaña et al. (2015) noted that species in homogeneous habitats were subject to stabilizing selection and consequently showed low intraspecific trait variability. According to Umaña et al. (2015), how intraspecific trait variability shapes fitness should be dependent on the magnitude of habitat heterogeneity. Our result clearly demonstrates that species with low intraspecific trait variability are associated with limited habitat ranges covering relatively homogeneous environmental conditions. Intraspecific variability in SLA is positively associated with species’ niche breadth, which implies that low-variability species operate as a more deterministic force in structuring communities compared with high-variability species. As a result, given a particular combination of habitat and trait, we are able to predict more reliably the occurrence of species with lower trait variability across the landscape, which helps us to forecast species distributions in a changing world and offer more reliable information for next-generation dynamic global vegetation models (Scheiter et al., 2013). CONCLUSIONS This study, to our knowledge, represents the most extensive mapping of SLA variation and its environmental determinants in forests, based on over 100 000 leaves across plants in a large forest plot (50 ha). In particular, we have explicitly demonstrated how intraspecific variability in SLA is connected to species distribution across the plot. We found that species with high SLA preferentially colonize relatively N-rich habitats and this pattern is more predominant in low-variability species. In addition, species with low variability in SLA show narrow habitat ranges with respect to soil TN and TOC. These findings have important implications for linking trait patterns to niche occupancy, and in turn to species distribution. A caveat should be made explicitly that soil conditions could also be modified by plants, which in turn involves how to define niche itself. We generally follow the Grinnellian definition of niche (i.e. species’ positions along particular environmental gradients) in the present work. Thus, how to link SLA and other important functional traits to the Eltonian niche (i.e. species’ role in an interactive system) remains poorly resolved (D’Andrea and Ostling, 2016; Rosado et al., 2016). SUPPLEMENTARY DATA Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Table S1: Summary of species information. Table S2: Summary of observed environmental variables and Kriging modelling parameters. Table S3: Pairwise correlations between environmental variables. Table S4: Summary of linear mixed effect models with and without spatial autocorrelation. Fig. S1: Model diagnostics for assumptions of normality, homoscedasticity and linearity. ACKNOWLEDGEMENTS This work is a part of Dong He’s PhD programme, directed by Fangliang He who provided great financial and intellectual support. We are grateful to Wensheng Shu and his research group for the soil survey. David Deane, Bill Shipley, Ruben Milla and an anonymous reviewer offered valuable comments on an early version of the manuscript. We would like to thank Mingshan Xu and Huiling Zhu for the help to prepare the Figure 2. A number of colleagues at Sun Yat-sen University and volunteers from around South China are thanked for assistance with fieldwork. This study was financially supported by the National Key R & D Program of China (2017YFC0506100, 2017YFC0506101) and the National Natural Science Foundation of China (31570426 and 31622014) to C.C., and the 1000 Plan Grant to Fangliang He. LITERATURE CITED Ackerly D, Knight C, Weiss S, Barton K, Starmer K. 2002. Leaf size, specific leaf area and microhabitat distribution of chaparral woody plants: contrasting patterns in species level and community level analyses. Oecologia 130: 449– 457. Google Scholar CrossRef Search ADS Adler PB, Fajardo A, Kleinhesselink AR, Kraft NJ. 2013. Trait-based tests of coexistence mechanisms. Ecology Letters 16: 1294– 1306. Google Scholar CrossRef Search ADS Akaike H. 1974. A new look at the statistical model identification. IEEE Transactions on Automatic Control 19: 716– 723. Google Scholar CrossRef Search ADS Al Haj Khaled R, Duru M, Theau JP, Plantureux S, Cruz P. 2005. Variation in leaf traits through seasons and N-availability levels and its consequences for ranking grassland species. Journal of Vegetation Science 16: 391– 398. Google Scholar CrossRef Search ADS Albert CH, Grassein F, Schurr FM, Vieilledent G, Violle C. 2011. When and how should intraspecific variability be considered in trait-based plant ecology? Perspectives in Plant Ecology, Evolution and Systematics 13: 217– 225. Google Scholar CrossRef Search ADS Auger S, Shipley B. 2013. Inter-specific and intra-specific trait variation along short environmental gradients in an old-growth temperate forest. Journal of Vegetation Science 24: 419– 428. Google Scholar CrossRef Search ADS Baraloto C, Bonal D, Goldberg DE. 2006. Differential seedling growth response to soil resource availability among nine neotropical tree species. Journal of Tropical Ecology 22: 487– 497. Google Scholar CrossRef Search ADS Blonder B, Lamanna C, Violle C, Enquist BJ. 2014. The n-dimensional hypervolume. Global Ecology and Biogeography 23: 595– 609. Google Scholar CrossRef Search ADS Burnham KP, Anderson D. 2003. Model selection and multi-model inference: A practical information-theoretic approach . Berlin: Springer. Burns K. 2004. Patterns in specific leaf area and the structure of a temperate heath community. Diversity and Distributions 10: 105– 112. Google Scholar CrossRef Search ADS Caley MJ, Munday PL. 2003. Growth trades off with habitat specialization. Proceedings of the Royal Society of London. Series B: Biological Sciences 270: S175– S177. Google Scholar CrossRef Search ADS Colomb B, Kiniry JR, Debaeke P. 2000. Effect of soil phosphorus on leaf development and senescence dynamics of field-grown maize. Agronomy Journal 92: 428– 435. Google Scholar CrossRef Search ADS Cornelissen J, Lavorel S, Garnier Eet al. 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany 51: 335– 380. Google Scholar CrossRef Search ADS Craine J, Froehle J, Tilman D, Wedin D, ChapinIII F. 2001. The relationships among root and leaf traits of 76 grassland species and relative abundance along fertility and disturbance gradients. Oikos 93: 274– 285. Google Scholar CrossRef Search ADS D’Andrea R, Ostling A. 2016. Challenges in linking trait patterns to niche differentiation. Oikos 125: 1369– 1385. Google Scholar CrossRef Search ADS Delgado MN, de Aguiar Gomes MR, Báo SN, Rossatto DR. 2013. Fertilisation residues alter leaf scleromorphy in an evergreen savannah shrub (Maprounea brasiliensis, Euphorbiaceae). Australian Journal of Botany 61: 266– 273. Google Scholar CrossRef Search ADS Green DS, Erickson JE, Kruger EL. 2003. Foliar morphology and canopy nitrogen as predictors of light-use efficiency in terrestrial vegetation. Agricultural and Forest Meteorology 115: 163– 171. Google Scholar CrossRef Search ADS Hodgson JG, Montserrat-Martí G, Charles Met al. 2011. Is leaf dry matter content a better predictor of soil fertility than specific leaf area? Annals of Botany 108: 1337– 1345. Google Scholar CrossRef Search ADS Hou H-Y. 1954. Phytoindicators of acid, alkali and saline soils in China . Beijing: Academia Sinica. Hulshof CM, Violle C, Spasojevic MJet al. 2013. Intra-specific and inter-specific variation in specific leaf area reveal the importance of abiotic and biotic drivers of species diversity across elevation and latitude. Journal of Vegetation Science 24: 921– 931. Google Scholar CrossRef Search ADS Janssen B. 1996. Nitrogen mineralization in relation to C: N ratio and decomposability of organic materials. Plant and Soil 181: 39– 45. Google Scholar CrossRef Search ADS Jung V, Violle C, Mondy C, Hoffmann L, Muller S. 2010. Intraspecific variability and trait-based community assembly. Journal of Ecology 98: 1134– 1140. Google Scholar CrossRef Search ADS Keenan TF, Niinemets Ü. 2016. Global leaf trait estimates biased due to plasticity in the shade. Nature Plants 3: No.16201. Knops JM, Reinhart K. 2000. Specific leaf area along a nitrogen fertilization gradient. The American Midland Naturalist 144: 265– 272. Google Scholar CrossRef Search ADS Kochian LV, Hoekenga OA, Piñeros MA. 2004. How do crop plants tolerate acid soils? Mechanisms of aluminum tolerance and phosphorous efficiency. Annual Review of Plant Biology 55: 459– 493. Google Scholar CrossRef Search ADS Kumordzi BB, Nilsson M-C, Gundale MJ, Wardle DA. 2014. Changes in local-scale intraspecific trait variability of dominant species across contrasting island ecosystems. Ecosphere 5: art26. ES13-00339.1. Kunstler G, Falster D, Coomes DAet al. 2016. Plant functional traits have globally consistent effects on competition. Nature 529: 204– 207. Google Scholar CrossRef Search ADS Kuznetsova A, Brockhoff PB, Christensen RHB. 2014. lmerTest: Tests in Linear Mixed Effects Models . R package version 2.0–20. https://cran.r-project.org/web/packages/lmerTest/index.html Lambers H, Poorter H. 1992. Inherent variation in growth rate between higher plants: a search for physiological causes and ecological consequences. Advances in Ecological Research 23: 187– 261. Google Scholar CrossRef Search ADS Laughlin DC, Joshi C. 2015. Theoretical consequences of trait-based environmental filtering for the breadth and shape of the niche: New testable hypotheses generated by the Traitspace model. Ecological Modelling 307: 10– 21. Google Scholar CrossRef Search ADS Laurans M, Martin O, Nicolini E, Vincent G. 2012. Functional traits and their plasticity predict tropical trees regeneration niche even among species with intermediate light requirements. Journal of Ecology 100: 1440– 1452. Google Scholar CrossRef Search ADS Li M-J, Wang Z-H. 1984. The phenology of common plant species on Mount Dinghushan. Tropical and Subtropical Forest Ecosystems 2: 1– 10. McGill BJ, Enquist BJ, Weiher E, Westoby M. 2006. Rebuilding community ecology from functional traits. Trends in Ecology & Evolution 21: 178– 185. Google Scholar CrossRef Search ADS Messier J, McGill BJ, Lechowicz MJ. 2010. How do traits vary across ecological scales? A case for trait-based ecology. Ecology Letters 13: 838– 848. Google Scholar CrossRef Search ADS Meziane D, Shipley B. 2001. Direct and indirect relationships between specific leaf area, leaf nitrogen and leaf gas exchange. Effects of irradiance and nutrient supply. Annals of Botany 88: 915– 927. Google Scholar CrossRef Search ADS Milla R, Reich PB. 2007. The scaling of leaf area and mass: the cost of light interception increases with leaf size. Proceedings of the Royal Society of London B: Biological Sciences 274: 2109– 2115. Google Scholar CrossRef Search ADS Mitchell PL, Woodward FI. 1988. Responses of three woodland herbs to reduced photosynthetically active radiation and low red to far-red ratio in shade. Journal of Ecology 76: 807– 825. Google Scholar CrossRef Search ADS Nakagawa S, Schielzeth H. 2013. A general and simple method for obtaining R2 from generalized linear mixed‐effects models. Methods in Ecology and Evolution 4: 133– 142. Google Scholar CrossRef Search ADS Niinemets Ü, Kull K. 1994. Leaf weight per area and leaf size of 85 Estonian woody species in relation to shade tolerance and light availability. Forest Ecology and Management 70: 1– 10. Google Scholar CrossRef Search ADS Ordoñez JC, Van Bodegom PM, Witte JPM, Wright IJ, Reich PB, Aerts R. 2009. A global study of relationships between leaf traits, climate and soil measures of nutrient fertility. Global Ecology and Biogeography 18: 137– 149. Google Scholar CrossRef Search ADS Pérez-Harguindeguy N, Díaz S, Garnier Eet al. 2013. New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany 61: 167– 234. Google Scholar CrossRef Search ADS Pollock LJ, Morris WK, Vesk PA. 2012. The role of functional traits in species distributions revealed through a hierarchical model. Ecography 35: 716– 725. Google Scholar CrossRef Search ADS Poorter H, De Jong R. 1999. A comparison of specific leaf area, chemical composition and leaf construction costs of field plants from 15 habitats differing in productivity. New Phytologist 143: 163– 176. Google Scholar CrossRef Search ADS Poorter H, Niinemets Ü, Poorter L, Wright IJ, Villar R. 2009. Causes and consequences of variation in leaf mass per area (LMA): a meta-analysis. New Phytologist 182: 565– 588. Google Scholar CrossRef Search ADS Reich PB, Walters MB, Ellsworth DS. 1997. From tropics to tundra: global convergence in plant functioning. Proceedings of the National Academy of Sciences, USA 94: 13730– 13734. Google Scholar CrossRef Search ADS Reich P, Wright I, Cavender-Bares Jet al. 2003. The evolution of plant functional variation: traits, spectra, and strategies. International Journal of Plant Sciences 164: S143– S164. Google Scholar CrossRef Search ADS RibeiroJr PJ, Diggle PJ. 2001. geoR: A package for geostatistical analysis. R News 1: 14– 18. Rosado BHP, Figueiredo MSL, De Mattos EA, Grelle CEV. 2016. Eltonian shortfall due to the Grinnellian view: functional ecology between the mismatch of niche concepts. Ecography 39: 1034– 1041. Google Scholar CrossRef Search ADS Scheiter S, Langan L, Higgins SI. 2013. Next-generation dynamic global vegetation models: learning from community ecology. New Phytologist 198: 957– 969. Google Scholar CrossRef Search ADS Shipley B, De Bello F, Cornelissen JHC, Laliberté E, Laughlin DC, Reich PB. 2016. Reinforcing loose foundation stones in trait-based plant ecology. Oecologia 180: 923– 931. Google Scholar CrossRef Search ADS Sides CB, Enquist BJ, Ebersole JJ, Smith MN, Henderson AN, Sloat LL. 2014. Revisiting Darwin’s hypothesis: Does greater intraspecific variability increase species’ ecological breadth? American Journal of Botany 101: 56– 62. Google Scholar CrossRef Search ADS Siefert A. 2014. Intraspecific trait variation and community assembly in old-field grasslands . PhD Thesis, Syracuse University, Syracuse, NY. Siefert A, Violle C, Chalmandrier Let al. 2015. A global meta-analysis of the relative extent of intraspecific trait variation in plant communities. Ecology Letters 18: 1406– 1419. Google Scholar CrossRef Search ADS Sterck F, Poorter L, Schieving F. 2006. Leaf traits determine the growth-survival trade-off across rain forest tree species. The American Naturalist 167: 758– 765. Sterck F, Markesteijn L, Schieving F, Poorter L. 2011. Functional traits determine trade-offs and niches in a tropical forest community. Proceedings of the National Academy of Sciences, USA 108: 20627– 20632. Google Scholar CrossRef Search ADS Sultan SE. 2000. Phenotypic plasticity for plant development, function and life history. Trends in Plant Science 5: 537– 542. Google Scholar CrossRef Search ADS Umaña MN, Zhang C, Cao M, Lin L, Swenson NG. 2015. Commonness, rarity, and intraspecific variation in traits and performance in tropical tree seedlings. Ecology Letters 18: 1329– 1337. Google Scholar CrossRef Search ADS Vance CP, Uhde-Stone C, Allan DL. 2003. Phosphorus acquisition and use: critical adaptations by plants for securing a nonrenewable resource. New Phytologist 157: 423– 447. Google Scholar CrossRef Search ADS Violle C, Jiang L. 2009. Towards a trait-based quantification of species niche. Journal of Plant Ecology 2: 87– 93. Google Scholar CrossRef Search ADS Violle C, Enquist BJ, McGill BJet al. 2012. The return of the variance: intraspecific variability in community ecology. Trends in Ecology & Evolution 27: 244– 252. Google Scholar CrossRef Search ADS Westoby M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil 199: 213– 227. Google Scholar CrossRef Search ADS Witkowski ETF, Lamont BB. 1991. Leaf specific mass confounds leaf density and thickness. Oecologia 88: 486– 493. Google Scholar CrossRef Search ADS Wright IJ, Reich PB, Westoby Met al. 2004. The worldwide leaf economics spectrum. Nature 428: 821– 827. Google Scholar CrossRef Search ADS Zhu H-L. 2015. The contributions of seedlings to tree diversity and habitat effects on tree assemblages across life stages . Master’s thesis, Sun Yat-Sen University, Guangzhou. © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Annals of Botany – Oxford University Press
Published: Feb 3, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera