This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Acta Oecologica 36 (2010) 569e577 Contents lists available at ScienceDirect Acta Oecologica journal homepage: www.elsevier.com/locate/actoec Original article Contribution of molehill disturbances to grassland community composition along a productivity gradient Merav Seifan*, Katja Tielbörger, Daniela Schloz-Murer, Tal Seifan Plant Ecology Department, Institute for Evolution and Ecology, Tübingen University, Auf der Morgenstelle 3, 72076 Tübingen, Germany a r t i c l e i n f o a b s t r a c t Article history: Received 26 June 2009 Accepted 27 August 2010 Available online 20 September 2010 Site productivity and disturbances are among the main factors determining plant community composition. With increasing site productivity, plant diversity is often reduced and the importance of traits associated with light competition increases. Small-scale disturbances created by mole activity are assumed to affect this dynamics by two factors: increasing site heterogeneity and providing survival opportunity for weaker light competitors. These effects are expected to become stronger with increasing site productivity. To test this hypothesis we compared species composition on molehills and in undisturbed plots within grasslands and along a productivity gradient. The differences in species composition were evaluated both by species diversity and by traits representing trade-offs between species’ response to competition and disturbance. Species diversity generally decreased with site productivity, while disturbances had no significant contribution to the variation among samples. When analyzing shifts in species traits, we found a central role of site productivity, but also a significant, though secondary, effect of molehill disturbances. In particular, communities showed increased potential height and specific leaf area and decreased seed weight and Ellenberg’s light indicator values with increasing site productivity, supporting the assumption of increasing light limitation along the gradient. Species observed on molehills were generally less shade tolerant, especially in the more productive grasslands. We conclude that, although overall plant community composition is mainly shaped by large environmental gradients such as site productivity, small-scale disturbances may contribute to local environmental heterogeneity and trait variation by enabling species less adapted to light competition to survive. Ó 2010 Elsevier Masson SAS. All rights reserved. Keywords: Species traits Species richness Small-scale disturbances Plant ecology strategy scheme Seed mass SLA 1. Introduction Productivity and disturbances are two of the major factors shaping biodiversity in time and space (e.g., Hastings, 1980; Huston, 1999; Kondoh, 2001). While many studies have investigated productivity and disturbance effects separately, their interactive effects on plant communities are relatively less studied (Gross et al., 2005; Kadmon and Benjamini, 2006). This is surprising, because the existing studies, as well as general plant community theory, allow the generation of clear and testable hypotheses. Keddy (1992) suggested that plant communities are the product of hierarchical filtering mechanisms. The first filter is created by the habitat conditions, selecting species whose traits enable them to persist under the specific conditions (McGill et al., 2006). The second filter is a combination of biotic and abiotic factors determining which * Corresponding author. Tel.: þ49 7071 2978814; fax: þ49 7071 295356. E-mail addresses: merav.seifan@uni-tuebingen.de (M. Seifan), katja.tielboerger@ uni-tuebingen.de (K. Tielbörger), daniela.schloz-murer@uni-tuebingen.de (D. Schloz-Murer), tal.seifan@mail.huji.ac.il (T. Seifan). 1146-609X/$ e see front matter Ó 2010 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.actao.2010.08.005 of the potentially adapted species will indeed be observed in the community (Keddy, 1992; Diaz et al., 1998). Small-scale disturbances, such as fossorial animal activity, are a fundamental part of the second filtering process, locally changing soil properties and increasing the spatial and temporal environmental heterogeneity (Pickett and White, 1985; Eldridge and Whitford, 2009). The disturbed patches are usually enriched in nutrients because of the high animal activity and especially due to a constant mixture of soil layers (Huntly and Reichman, 1994; Kinlaw, 1999). In addition, they show lower soil moisture due to higher evaporation from the exposed soil (Huntly and Reichman, 1994; Canals and Sebastia, 2000; accompanied by lower infiltration rates in dry areas, e.g. Eldridge, 2009) and a higher light exposure because of the removal/burial of older vegetation layers (e.g. Canals and Sebastia, 2000). Because of their properties, small-scale disturbances by fossorial animals are expected to increase plant community diversity. In addition to the potential increase in suitable establishment habitats, as described above, the direct physical damage to the established vegetation may help to limit dominant species’ success and provide a refuge for weaker competitors on the newly disturbed mounds. Therefore, it is expected that the role of disturbances will Author's personal copy 570 M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 become more important when a community is dominated by a relatively low number of highly competitive species (e.g. the intermediate disturbance hypothesis, Connell, 1978). The probability of a few competitive species’ dominating is usually assumed to increase with productivity, because higher resource availability enhances plant growth and increases the magnitude of their interactions and in particularly the occurrence of competitive displacement (e.g. Bobbink et al., 1998; Huston, 1999; Grime, 2001). This lead Huston, as early as 1979, to suggest that a balance between increasing competitive exclusion at high productivity and decreasing population growth caused by disturbances, helps to maintain high diversity in relatively productive sites (Huston, 1979; Kondoh, 2001). While diversity indices and species composition help to link species distributions and environmental factors, the usage of trait analysis goes beyond the botanical complexity and gives insight to the potential mechanisms governing the observed patterns (Garnier et al., 2007; Violle et al., 2007; Fortunel et al., 2009). In this case, analysis of community functional composition, reflected in shifts in community trait attributes (sensu Violle et al., 2007), may help our understanding of the particular effects of small-scale disturbances and productivity in shaping plant communities (see also Diaz et al., 1999). One of the theoretical frameworks for such studies is the Leaf-Height-Seed (LHS) plant ecology strategy scheme (Westoby, 1998). Species’ potential height, their specific leaf area (SLA) and seed mass represent fundamental strategies related to their light capturing, longevity, leaf structure, reproduction and dispersal (Westoby, 1998; Westoby et al., 2002). In accordance with the hierarchical filtering theory (Keddy, 1992), the habitat factors should act as a first filter. In particular, we predict that biomass, growth rate and photosynthetic efficiency will increase with environmental productivity. Thus, with increasing productivity we expect to detect an increase in light competition, resulting in a trait convergence toward taller species with better abilities to exploit light (e.g. Westoby et al., 2002; De Bello et al., 2009). Fossorial animal activity, similar to other gap-opening disturbances, are expected to counteract this trait convergence by offering microhabitats with different conditions (e.g. Kinlaw, 1999; Canals and Sebastia, 2000), and by providing refuge for species which are less adapted to strong light competition (e.g. Bakker and Olff, 2003). To study the potential effects of productivity and fossorial animal disturbances on plant communities, we focused on grasslands, one of the most intensively studied ecosystems for biodiversity research. Our disturbance agent was the European mole (Talpa europaea). Because the European mole feeds mainly on earthworms, we assumed that its effect on plant communities is mainly indirect, via the creation of molehills. Surprisingly, although the European mole is abundant throughout central Europe and occupies many different habitats, there are relatively few studies on its role as a plant community agent (but see Edwards et al., 1999; Canals and Sebastia, 2000, 2002). One possible explanation might be that in the European tradition, moles are treated as a pest rather than an integral part of natural and semi-natural ecosystems (Atkinson and MacDonald, 1994). We suggest that not only is the European mole an integral part of a natural ecosystem, but it may play an important role in maintaining plant species diversity. The goal of this study was to test for a consistent association between community characteristics, using a trait-based approach and diversity indices, and both productivity-related soil properties and European mole disturbances. Using the above consideration we hypothesized the following: (A) Species diversity: Since productivity is a significant environmental factor in the study area (i.e. part of the first filter), its role in explaining species diversity is expected to be larger than that of small-scale disturbances. Nevertheless, with increasing productivity the impact of European mole activity on species diversity is expected to increase, due to larger dissimilarities in light and moisture conditions between the disturbed and undisturbed microhabitats. (B) Community trait values: Trait attributes related to high light intensities and dry conditions will dominate on molehills while attributes related to increased light-utilizing efficiency and intense competition are expected to dominate the undisturbed vegetation. Because the two microhabitats are expected to increase their dissimilarity with increasing productivity, the differences in trait values are expected to increase with productivity. In order to test these hypotheses, we first compared the abiotic conditions on molehills and in the undisturbed vegetation near them and established the presence of a productivity related soil gradient between the studied grasslands. We then compared the diversity and traits of species growing in the disturbed and undisturbed vegetation along the productivity gradient. 2. Material and methods 2.1. Study area The study was conducted in the Swabian Alb, BadenWürttemberg, Germany. The Alb, a mountain range (600e1000 meters above sea level) situated north of the European Alps, is particularly suitable to study the above questions, since there is a clear negative correlation between altitude and resource levels in this region. Specifically, the area, which is built on limestone and dolomite bedrock (white and brown Jurassic limestone), is strongly influenced by erosion so that richer soils are found in crevices and valleys (Grees, 1993). Six grasslands, situated south of Tübingen, were chosen for repeated field surveys. The grasslands were located in close proximity to each other (2e10 km), but at different altitudes (Table 1), covering two landscape types (“Mittleres Albvorland” and “Mittlere Kuppenalb”; Breuning and Demuth, 1999). Both landscape types are characterized by similar temperatures, with an average January temperature of 2  C, and 16  C in July. The region’s annual precipitation is between 800e1000 mm and vegetation growth period is between 225 days in the low areas to 195e205 days in the higher areas (Renners, 1991). Therefore, the grasslands created an environmental gradient (mainly because of variation in soil properties) while maintaining similar climatic conditions. In addition, the relative short distance between grasslands supports the assumption that all grasslands originated from a common regional species pool. All the grasslands are situated in nature reserves or landscape conservation areas and are subjected to agricultural management (both mowing and sheep grazing), but not to fertilizer management. All grasslands exhibit activity by fossorial mammals (mainly European moles). An agreement was reached with the land-owners Table 1 Grassland information. Geological and geographical information concerning the six grasslands sampled in the study. Grassland location Coordinates Altitude Geology Krebsbachtal Olgahöhe Unter Lauhern Rossberg Urselhochberg Won 08 560 4100 E/48 240 2600 N 09 030 3200 E/48 230 2700 N 09 080 2900 E/48 250 2700 N 09 100 1800 E/48 250 2700 N 09 160 0700 E/48 270 0800 N 09 130 0400 E/48 250 2800 N 443 m 586 m 627 m 722 m 776 m 808 m Alluvium Brown Jurassic Brown Jurassic White Jurassic White Jurassic White Jurassic KB OH UL RB UH W Author's personal copy M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 to refrain from activity in the survey sites during spring and early summer of the study season (2005). 2.2. Data collection 2.2.1. Species survey Forty quadrats of 20  20 cm were marked in each grassland in March 2005 (a total of 240 quadrats). Half of the quadrats were positioned on randomly chosen molehills, and the rest were randomly positioned in the vegetation matrix between the molehills. All molehills used in the survey were newly created, indicating recent activity of moles in the grasslands. The size of the quadrats was determined to be slightly smaller than the average molehill, thus enabling us to compare species diversity at the same scale in both microhabitats (see also Canals and Sebastia, 2002; Schiffers et al., 2010). The quadrats were surveyed twice during the growing season (April and July), to enable detection of a maximum number of species. In each survey, all observed species within the quadrat were recorded. Because all the grasslands are privately owned, the study had a restricted duration which ended with the final mowing of all the grasslands in August 2005, and did not include autumn species. 2.2.2. Species diversity Analyses of change in species composition along the suggested productivity gradient and between disturbed and undisturbed microhabitats within grasslands were based on the species lists acquired in the two vegetation surveys. Three (related) variables were created for each combination of grassland and microhabitat: (i) number of species presented in each quadrat, which reflected richness at the quadrat scale, (ii) maximum species number observed in a microhabitat within each grassland (sum over all 20 quadrats), (iii) species diversity (ShannoneWiener index) based on species frequency of appearance in the relevant 20 sampling quadrats. Many studies emphasized the importance of abundance measurements and dominance in diversity analysis (see Cingolani et al., 2007). However, the nature of the plant species observed (e.g. perennial grasses, annuals, herbs with significant vegetative growth) caused difficulties in choosing a common abundance measurement. Therefore, we used the frequency of observed species (out of maximum 20 quadrats) as a surrogate for its abundance. 2.2.3. Species traits In accordance with Westoby’s recommendations (1998), we used species height, specific leaf area (SLA) and seed weight as the main traits studied. Height (canopy height, in meters) and SLA (in mm2 mg 1) information for each listed species were obtained from the LEDA traitbase, an open northwest European trait database which provides species trait information based on standardized observations (Kleyer et al., 2008). Seed weight (in grams for 1000 seeds) was obtained from the SID database of the Royal Botanic Gardens, Kew, another open database with standardized seed data (Liu et al., 2008). In addition, we used Ellenberg’s light indicators (Ellenberg et al., 1992) for each species. Ellenberg’s indicators are commonly used for central European plant communities (e.g. Nygaard and Ejrnaes, 2004), especially for interpreting site conditions based on species composition (e.g. Ozinga et al., 2005). The values are based on an ordinal scale (usually between 1 and 9) representing the distribution (and thus suggesting the ecological and physiological adaptations) of each species in relation to an abiotic factor (here: light availability). Similar to the analysis of diversity, we created two related variables for the traits: (i) average trait value per quadrat, calculated from the values of the observed species in each quadrat; (ii) 571 weighted trait range within a microhabitat in each grassland, calculated by weighting the trait values of all the observed species by their frequency of appearance (out of the 20 relevant quadrats; similar to Garnier et al., 2007 “aggregated traits”). This variable studies possible effects of species abundance on the observed trends. 2.2.4. Explanatory variables Although productivity is central to numerous studies, different studies refer to different aspects of the term. For example, many studies focus on the direct response of plant communities, measured as a plants’ above ground biomass or correlated observations (e.g. Foster, 2001; Foster and Dickson, 2004), while others focus on habitat fertility (i.e. soil properties and resource availability) as an indicator for inherent differences which potentially influence plant community development (e.g. Campbell and Grime, 1992; Smit and Olff, 1998; Fynn et al., 2005). Although the different approaches are strongly connected and may replace each other under certain conditions (e.g. Grace, 1993; Huston and DeAngelis, 1994), it is exceptionally important to specify precisely the aspects of productivity studied. Here we defined the term ‘siteproductivity’ as describing the relative fertility of a certain location according to the availability of soil resources and potential plant growth. Therefore, we collected soil samples from all grasslands on sunny days in May 2005, at least 24 h after the last rain. At each site, five samples were taken from the vegetation matrix in random locations, and from five random molehills in the same local area. The soil samples were analyzed by standard methods (Egner et al., 1960) in the Geographical Institute, Tübingen University, for available phosphorus (extraction with acetateelactate, UV/Vis spectrometry), potassium and magnesium content (atomic absorption spectrometry). Total nitrogen availability could not be measured from the samples, but nitrite and nitrate content were determined by DX120 ion chromatography. Two microclimate variables were also collected: before taking the samples, the soil temperature (five centimeters below surface) was measured in the field. Soil moisture was estimated from the soil samples (as percent of soil weight lost after drying at 80  C for 24 h). Note, that because instantaneous soil measurements are relatively simple to collect in the field, they are advocated as an efficient surrogate for site productivity, although they may be considered as an imperfect representation of soil property dynamics (Garnier et al., 2007). As such, they should not be considered as absolute values, but as an estimation of the potential differences in soil properties along the studied gradient. Because many studies have shown that geographical location has a strong effect on plant community development and habitat conditions (e.g. Sebastia, 2004; Hofer et al., 2008), we added the grassland altitude as an additional variable. To represent relative site productivity according to plant composition, we used the aggregated value of the sampled species Ellenberg’s nitrogen indicators (Ellenberg et al., 1992). This approach is commonly used for describing relative productivity of central European plant communities when direct measurements are difficult to collect or unavailable (e.g. Ter Braak and Gremmen, 1987; Bakker et al., 2002; Ozinga et al., 2005; Wieger et al., 2005; Bernhardt-Romermann et al., 2008). 2.3. Data analysis Variation partitioning (Borcard et al., 1992) was applied to determine the relative influence of site productivity and disturbances on community properties (diversity and trait composition). To further analyze the relative importance of the variables composing site productivity in our study, we compared the performance of several models, selecting the most parsimonious Author's personal copy 572 M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 among them using information criteria (AICc, Burnham and Anderson, 2004). First, we tested for the existence of an environmental gradient and for differences in the abiotic conditions between molehills and undisturbed vegetation by analyzing potential shifts in soil properties (nutrient content and microclimate) between grasslands and among the microhabitats (molehills and undisturbed vegetation). We used generalized linear models (GLM; normal distribution with log link) in which grassland and microhabitat were fixed factors. Because many of the explanatory variables were not independent from each other, they may lead to biased model analysis, when testing for diversity and trait differences, due to multicollinearity (Draper and Smith, 1998). Therefore, we also calculated a correlation matrix for all the variables using Pearson’s correlations before choosing the site productivity variables for further analyses. We eliminated from the analyses variables which were highly correlated (r > 0.60) with other variables. The reduction in the number of explanatory variables was also important to reduce the chances of overfitting models due to the relative low number of samples (Draper and Smith, 1998). The contribution of the chosen site productivity variables and the specific microhabitat (a molehill or undisturbed vegetation) to the variation in community diversity and traits was analyzed using a series of partial redundancy analyses (pRDA; Borcard et al., 1992), in which the total explained variation between plant communities (either traits or diversity measurements) was separated into three components: the effect of site productivity, the effect of disturbance (represented as the microhabitat in which species were observed) and the joint effect of the two (see also Marini et al., 2008; Pakeman et al., 2009). The significant contribution of site productivity and disturbance was tested by Monte Carlo permutation tests (999 permutations). The data set was also analyzed using canonical correspondence analysis (CCA). Because the analyses yielded very similar results, it suggested that the assumptions of a relatively short gradient and linear effects of the variables on community properties was not violated (Jongman et al., 1995), and therefore only the pRDA results are presented. To further determine which of the chosen variables composing site productivity presented the strongest connection with community composition, we compared various GLMs which were a derivative of a dummy-variable regression model. In the basic model, the chosen site productivity variables were used as continuous variables and the two microhabitats (molehills and undisturbed vegetation) were used as a dummy variable. Among the potential combinations of explanatory variables, we selected the most parsimonious model which presented the lowest Akaike’s criterion values. As recommended by Burnham and Anderson (2004), we used the correction for small sample sizes (AICc). The coefficients of the selected models were then estimated and their significance was tested using Wald c2 tests. All models were tested for heteroscedasticity and residual distribution. In most cases, a normal distribution with identity link satisfied the model assumptions. In some cases, a log link was required. These cases are reported with the results. Because of the multiple uses of the same data for different comparisons, Benjamini and Hochberg correction for individual probabilities was used in the relevant places (false discovery rate; Benjamini and Hochberg, 1995; Verhoeven et al., 2005). Data analysis was conducted using SPSS version 17.0. Multivariate analyses were conducted using CANOCO version 4.5. 3. Results 3.1. Site productivity and identification of relevant explanatory variables The soil property analyses revealed significant differences between grasslands. Generally, molehills showed lower soil moisture and potassium content and higher soil temperature and nitrate content. The differences in nitrate content and soil temperature between microhabitats were the smallest in the lowest altitude grassland (KB) and the largest in the highest grassland (UH), while for potassium content, the trend was opposite. For further information concerning site productivity properties see supplementary material 1. The general correlations among explanatory variables are summarized in Table 2. As expected, site productivity as measured by plant composition (using Ellenberg’s nitrogen indicator values), was strongly correlated with magnesium, phosphorus, nitrite, soil moisture content and altitude. Potassium and nitrate contents were each correlated with nitrite content (though not with each other). The microclimate measurements (soil temperature and moisture) were correlated with each other and with altitude. According to these results, we chose altitude, Ellenberg’s nitrogen indicator values (hereafter Ellenberg’s N), potassium and nitrate contents and temperature to represent site productivity in the following analyses. 3.2. Species diversity A list of the species observed in each microhabitat and grassland is presented in supplementary material 2. Overall, the analyses of species richness and diversity at grassland level presented very similar results, with the effect of site productivity and disturbances explaining 59% and 56% of the variation, respectively. Therefore, the diversity results are presented Table 2 Pearson correlation coefficients between the nine explanatory variables: Altitude e Average grassland location in meters above sea level; K e average potassium content (mg kg 1); Mg e average magnesium content (mg kg 1); Ps e average available phosphorus content (mg kg 1); Nitrite e average nitrite content (mg kg 1); Nitrate e average nitrate content (mg kg 1); Temp e average soil surface temperature ( C); Moist e average soil moisture (% of sample weight); EllenN e Average Ellenberg’s nitrogen indicator value calculated by a weighted appearance of the species in the samples; * 0.01 P < 0.05; **0.001 < P < 0.01; *** P < 0.001. Parameter Altitude K Mg Ps Nitrite Nitrate 0.50 0.10 0.25 Temp 0.62* 0.60* 0.67* 0.30 0.34 0.88*** 0.66* 0.36 0.53 0.12 0.01 0.36 0.59* 0.31 0.38 0.13 0.23 0.52 0.46 0.64* 0.36 0.10 0.58* 0.50 0.85*** 0.78** 0.57* 0.13 0.75** 0.54 0.61* Altitude Soil variables K Mg Ps Nitrite Nitrate 0.02 0.77*** 0.50 0.10 0.25 0.30 0.34 0.66* 0.12 0.88*** 0.36 0.01 0.53 0.36 0.59* Temp Moist 0.62* 0.60* 0.31 0.46 0.38 0.64* 0.13 0.36 0.23 0.10 0.52 0.58* 0.75** EllenN 0.67* 0.50 0.85*** 0.78** 0.57* 0.13 0.54 Microclimate Plant productivity 0.02 0.77*** Topography Moist 0.61* EllenN Author's personal copy M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 only in supplementary material 3. At the quadrat level, the site productivity explained a much smaller amount of the general variation among grasslands (17%; Fig. 1). Despite these differences, the general trends were very similar: site productivity always explained a larger part of the variation between communities relative to molehill disturbances. The contribution of molehill disturbances was not only much smaller but also not significant (Monte Carlo permutation tests; Fig. 1). Model selection analyses showed that community diversity and richness were significantly lower on molehills than in the undisturbed vegetation at quadrat and grassland levels (quadrat level: c21,235 ¼ 512.42, P < 0.001; grassland level: c21,235 ¼ 77.63, P < 0.001). At both levels, altitude (quadrat level: c21,235 ¼ 26.08, P < 0.001; grassland level: c21,235 ¼ 43.86, P < 0.001), Ellenberg’s N (quadrat level: c21,235 ¼ 36.82, P < 0.001; grassland level: c21,235 ¼ 11.54, P ¼ 0.001) and potassium content (quadrat level: c21,235 ¼ 4.65, P ¼ 0.031; grassland level: c21,235 ¼ 20.26, P < 0.001) best described community composition. However, the site productivity effect showed opposite trends for the two sampling levels: while diversity and richness decreased with increasing site productivity at grassland level, richness increased at the quadrat level (Fig. 1; supplementary material 3). In addition, at the grassland level, although the interaction between microhabitat and Ellenberg’s N was only marginally significant (c21,235 ¼ 4.59, P ¼ 0.032), its effect on undisturbed vegetation was more negative than on molehill vegetation. At the quadrat level, the interaction 573 between microhabitat and Ellenberg’s N was not significant and negatively affected the model (i.e. significantly increased AICc values), and therefore cannot be seen as a significant trend. 3.3. Species traits Similar to the diversity and richness analyses, overall variation explained by site productivity and disturbances at the quadrat level was much lower than at the grassland level (w16% vs. w57% respectively, for all traits). However, since the general trends of the pRDA and the model selection analyses were similar at both levels, we present here only the results at the grassland level. The results calculated based on the quadrat level may be found in the supplementary material (supplementary material 4). Canopy height and Ellenberg’s light indicator values for the studied communities increased with altitude (canopy height: c21,234 ¼ 13.82, P < 0.001; Ellenberg’s L: c21,233 ¼ 298.38, P < 0.001) and decreased with Ellenberg’s N (canopy height: c21,234 ¼ 51.34, P < 0.001; Ellenberg’s L: c21,233 ¼ 257.60, P < 0.001) and potassium content (canopy height: c21,234 ¼ 18.62, P < 0.001; Ellenberg’s L: c21,233 ¼ 9.34, P ¼ 0.002, Fig. 2). For both traits, disturbance contributed significantly to the explained variation between communities (9% and 6% respectively; supported by the GLM results: canopy height: c21,234 ¼ 31.00, P < 0.001; Ellenberg’s L: c21,233 ¼ 209.35, P < 0.001). In addition, a significant interaction between microhabitat and site productivity was found. These Fig. 1. Changes in species richness with site productivity at (a) the grassland and (b) quadrat scale. Scatter plots present change in species richness along site productivity variables which were chosen as the most important by model fitting (see Material and Methods). - e richness in the undisturbed samples; , e richness on molehills (SE, when appropriate); continuous lines represent the linear regression trend for undisturbed vegetation and dashed lines for molehills and are for trend demonstration only. The pie chart describes the general percentage of species matrix variation explained by site productivity and disturbances derived from the variance partitioning analysis (- e site productivity; e disturbance; e shared effect of site productivity and disturbance; , e unexplained variation). The numbers represent the percentage of variation explained by the factor. * 0.01 < P < 0.05; **0.001 < P < 0.01;*** P < 0.001 of Monte Carlo permutation tests. The shared effect was obtained by subtraction and could not be tested for significance. Results for species diversity at the grassland scale were similar (supplementary material 2). Author's personal copy 574 M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 Fig. 2. Change in (a) potential canopy height and (b) in Ellenberg’s light indicator values of the communities between grasslands and microhabitats. The pie chart represents the general percentage of variation explained by site productivity and disturbance and the scatter plots the relationship between canopy height/Ellenberg’s light indicator and the site productivity variables chosen by the model fitting process. Symbols and shading are as in Fig. 1. significant interaction terms are supported by the pRDA which showed that the combined effect of disturbance and site productivity explained 5% of the variation in Ellenberg’s light indicator and 17% in canopy height. In both cases, traits were more positively affected by site productivity on molehills relative to the undisturbed vegetation (Ellenberg’s light indicator: interactions with Ellenberg’s N: c21,233 ¼ 183.96, P < 0.001 and potassium: c21,233 ¼ 21.04, P < 0.001; canopy height: interaction with potassium content: c21,234 ¼ 21.00, P < 0.001). Variation in seed weight among communities were best explained by soil temperature, showing an increasing seed weight with increasing temperature (c21,238 ¼ 8.43, P ¼ 0.004). Although disturbance explained 9% of the variation between communities, the contribution of disturbance was not significant, as indicated also by the GLM results. However, the results indicate a weaker effect of temperature on seed weight on molehills relative to the undisturbed vegetation, though the results were significant only at the quadrat level (Fig. 3a; supplementary material 4). Similar to the results for seed weight, SLA values were significantly affected only by site productivity (c21,238 ¼ 45.08, P < 0.001). At the grassland level, a positive effect of Ellenberg’s N on SLA was detected. At the quadrat level, an additional similar effect of potassium was detected, as well as a significant decrease in SLA with increasing altitude and soil temperature (Fig. 3b, supplementary material 4). 4. Discussion The results confirmed our prediction that site productivity serves as the main factor determining species composition in the study area. As predicted, plant communities became less diverse with increasing site productivity, selecting for species with potentially higher canopy and better light intercepting abilities. Species diversity was generally lower on molehills, and its composition a subset of the surrounding vegetation. Nevertheless, species growing on molehills, although present also in the undisturbed vegetation, were generally less shade tolerant relative to the undisturbed vegetation, especially in the more productive sites. This suggests that at a local scale, molehills contribute to the maintenance of a richer community by acting as potential colonization sites where species which are less adapted to strong light competition can survive. 4.1. Site productivity and the analysis of soil properties The analyses of soil properties supported our working assumption that molehills and undisturbed vegetation provided significantly different microhabitats, as found in previous studies concerning animal created gaps (e.g. Kinlaw, 1999; Canals and Sebastia, 2000; Suding and Goldberg, 2001; Rebollo et al., 2003). In particular, on molehills, soil temperature was always higher and soil moisture lower than in the surrounding undisturbed vegetation. In addition, nitrate levels were much higher on molehills, especially in less productive grasslands. Similar findings were recorded for molehills in other geographical locations (e.g. Spanish calcareous grasslands; Canals and Sebastia, 2000) and for other fossorial animals (e.g. vole activity in Californian grasslands; Canals et al., 2003; several examples in arid areas; Kinlaw, 1999), suggesting that these differences are common and are not limited to certain ecosystems. The large differences in nitrogen content between the two microhabitats may be a product of animal activity, which mixed soil profiles (Canals et al., 2003), or due to nitrogen-loss processes (e.g. leaching) which were magnified by the disturbances (e.g. Canals and Sebastia, 2000; Eldridge and Whitford, 2009). Author's personal copy M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 575 Fig. 3. change in (a) seed weight and (b) in SLA values between grasslands and microhabitats. The pie chart represents the general percentage of variation explained by site productivity and disturbance and the scatter plots the relationship between seed weight/SLA and the site productivity variables chosen by the model fitting process. Symbols and shading are as in Fig. 1. As expected, many of the site productivity variables were intercorrelated and therefore were removed from further analyses to prevent multicollinearity. In particular, Ellenberg’s N values were positively correlated with phosphorus, magnesium and nitrite contents and with soil moisture. Ellenberg’s N and the aforementioned soil variables were all negatively correlated with altitude, confirming the assumption of decreasing site productivity with increasing altitude in the studied area. Because of these correlations, our site productivity estimator was constructed of five variables: Ellenberg’s N, potassium and nitrate contents, soil temperature and altitude. Among these variables three were found to be of great importance in explaining variation in species composition: Ellenberg’s N, thus supporting the usage of Ellenberg’s N as a simple estimator for site productivity in middle European ecosystems (e.g. Ozinga et al., 2005; Bernhardt-Romermann et al., 2008); potassium content, which is known to be one of the major factors determining plant growth and photosynthesis (e.g. Hogh-Jensen, 2003; Jordan-Meille and Pellerin, 2004) and altitude, which probably represented other unmeasured geographical and climatic factors (e.g. Sebastia, 2004; Hofer et al., 2008). Interestingly, nitrate content had no significant contribution to any of the tested models. These results support the assumption that the observed higher nitrate content on mounds is a short-lived event (Canals et al., 2003; Eldridge and Whitford, 2009). 4.2. The combined effect of productivity and disturbances on biodiversity As expected, site productivity had a stronger and more significant effect on community diversity than disturbance. Overall, grassland diversity decreased with increasing site productivity, as found in many previous studies (e.g. Foster, 2001; Grime, 2001) and as predicted by theoretical models (e.g. Huston, 1979; Kondoh, 2001). Although molehills provided a significantly different microhabitat, plant species observed on mounds were a subset of the species pool in the surrounding undisturbed vegetation. Hence, molehills did not act as refuge sites where a distinctively different species assemblage may take advantage of the different conditions. Similar results were found in previous studies of gap-opening disturbances (Guo, 1996; Zobel et al., 2000; Gross et al., 2005; Sebastia and Puig, 2008). The low species diversity on molehills was probably a result of a relatively slow recovery rate from disturbances, which may be caused by limited colonization rates (e.g. Kovar et al., 2001; Rebollo et al., 2003). When studying species richness at the quadrat level, the amount of unexplained variation was especially large. In addition, closer inspection revealed a positive effect of site productivity on richness, contradicting the results at the grassland level and the general expectations. These results emphasize the significant role of scale in predicting changes in biodiversity (Palmer and White, 1994; Stoms, 1994; Aarssen, 2004). Following Huston’s (1999) advice, we conducted our survey at the same scale at which the studied disturbances were operating, which dictated a focus on very small scale diversity. However, as indicated by the large unexplained variation between grasslands, studying species richness at such a small scale might create a bias toward other local effects that override our ability to analyze large scale shifts in species composition (Stoms, 1994; Kalkhan and Stohlgren, 2000; Aarssen, 2004). Since our diversity indices were also based on richness, they might suffer from similar disadvantages, and only partially reflect dominant species (Cingolani et al., 2007). 4.3. The combined effect of productivity and disturbances on trait composition The analyses of the shift in trait composition between grasslands and microhabitats helped revealing the nature of the interaction between site productivity and disturbances. Overall, site productivity had a larger contribution to the explained trait and diversity variation. These results confirm our working assumption that the productivity gradient is the main factor shaping the observed communities, and support the findings of other studies, that environmental gradients are central filtering factors on plant traits (e.g. Diaz et al., 1999; Pakeman et al., 2009). In our grasslands, communities showed increasing investment in trait attributes associated with increased light limitation and biotic competition (e.g. increasing canopy height and SLA values) with increased site productivity (see also Bullock et al., 2001; Louault et al., 2005). Interestingly, for SLA and seed weight, site productivity was the only parameter explaining variation among grasslands, while disturbances had no additional significant effect. SLA represents the light intercepting area of a leaf per dry mass unit, and it is closely related to the plant net assimilation rate (Reich et al., 1992) and photosynthetic capacity (e.g. Wright et al., 2001). Changes in SLA values were previously shown to correlate not only with increased light availability but also with reduction in nutrient and water availability (Wright et al., 2001; Westoby et al., 2002). High SLA is associated with rapid production of new leaves, especially at young age, suggesting faster growth rate in relatively benign conditions (Westoby, 1998). Lower SLA values are favored with a reduction in environmental fertility (Garnier et al., 2007), because they are associated with longer residence time of nutrients in the leaves and longer life span, which are related to increased nutrient conservation strategies (Reich et al., 1992; Aerts and van Author's personal copy 576 M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 der Peijl, 1993). The decrease in seed weight with increasing site productivity is more difficult to explain, because analyses of seed weight trends along abiotic gradients and especially the effect of temperature, the main parameter explaining seed weight variation in our study, are equivocal (Baskin and Baskin, 2001; Westoby et al., 2002). A possible explanation for the observed increase in seed weight with increasing soil temperature may be related to higher germination costs in higher temperatures, which selects for larger seeds (Leishman et al., 2000). Another possibility is that, since seeds tend to increase in size and weight with environmental harshness (Leishman et al., 2000; Westoby et al., 2002), the positive effect of temperature on seed size reflects also the decrease in soil moisture along the studied gradient (e.g. Pakeman et al., 2009). This explanation is supported by Kahmen and Poschlod (2008) who found that smaller seeds were more successful in less disturbed, shady and dense grassland communities. With the decrease in site productivity and the increased importance of nutrient and microclimate limitation (as indicated by an increase in soil temperature), species with higher seed weight and lower SLA may be favored because of their potential ability to buffer resource limitations (e.g. Westoby et al., 2002; Wright et al., 2002). Mole activity had a secondary (as shown by the lower proportion of explained variation) but significant effect in shaping community trait attributes. Species assemblages observed on molehills were generally shorter and more light-demanding relative to the undisturbed vegetation surrounding them. These differences support the assumption that grasslands are mainly governed by above-ground competition, and that the creation of disturbances allows weaker competitors to survive in the community (e.g. Rajaniemi, 2002). In addition, it supports the findings of Violle et al. (2009) that plant height is especially sensitive to light depletion and is, therefore, a preferable trait for estimating asymmetric light competition. As predicted, the relative contribution of mole activity to the community trait values changed with site productivity. At relatively productive sites, the undisturbed vegetation was dense and tall, creating a shaded habitat which favored establishment of species adapted to light limitation (e.g. Seidlova et al., 2009). In these grasslands, molehills may provide establishment opportunities for species with inferior light interception ability, expressed by their shorter structure and lower Ellenberg’s light values (e.g. Wright et al., 2001). In less productive sites, differences in species assemblages’ trait values were weaker, reflecting, perhaps, the general resource limitation in these grasslands. In accordance with this explanation, results may indicate interplay between strong biotic interactions and habitat filtering which affects, and may limit, trait convergence (see De Bello et al., 2009). At a local scale, when environment is benign, biotic interactions are expected to play a significant role in shaping communities, and disturbance is expected to counteract their effects. When habitat filtering becomes the main limiting factor (i.e. when site productivity is low), disturbances are expected to have less dramatic effect on functional diversity (Diaz et al., 1999). However, field manipulations are required to confirm this assumption. 5. Conclusions Although information on animal activity in grasslands is starting to accumulate (Canals and Sebastia, 2000; Kovar et al., 2001; Canals et al., 2003; Rebollo et al., 2003; Sebastia and Puig, 2008), our knowledge about the effects of gap-openings in grasslands is still scarce relative to the information found on woody vegetation (Seidlova et al., 2009). Our results suggest that small-scale disturbances play an important role in increasing environmental heterogeneity and with it increasing community functional diversity, supporting previous studies’ results (e.g., Sebastia and Puig, 2008). In particular, since species occupying a site are drawn selectively from the regional species pool (e.g. Diaz et al., 1998; Westoby et al., 2002), the increased heterogeneity enables the survival of less adapted species within the local species pool and with it the maintenance of a relatively higher community diversity. Furthermore, since the results from the trait composition analyses were more sensitive to the combined effects of disturbance and site productivity than were the results from species diversity, we advocate the use of trait-based analyses for the detection of local patterns of adaptation. Acknowledgements We would like to thank M. Koltzenburg for guidance in the species surveys, H. Horstmann, T. Obergfell, D. Schloz and I. SchlozMurer for help in the field, and P. Kühn and T. Scholten of the Geographical institute, Tübingen for the soil analysis. We acknowledge funding by the MINERVA foundation to MS, and by the DFG (FK Ti 338/4-3) to KT and TS. Appendix. Supplementary material Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.actao.2010.08.005. References Aarssen, L.W., 2004. Interpreting co-variation in species richness and productivity in terrestrial vegetation: making sense of causations and correlations at multiple scales. Folia Geobot. 39, 385e403. Aerts, R., van der Peijl, J., 1993. A simple model to explain the dominance of low productive perennials in nutrient poor habitats. Oikos 66, 144e147. Atkinson, R.P.D., MacDonald, D.W., 1994. Can repellents function as a non-lethal means of controlling moles (Talpa europaea)? J. Appl. Ecol. 31, 731e736. Bakker, J.P., Elzinga, J.A., de Vries, Y., 2002. Effects of long-term cutting in a grassland system: perspectives for restoration of plant communities on nutrient-poor soils. Appl. Veg. Sci. 5, 107e120. Bakker, E.S., Olff, H., 2003. Impact of different-sized herbivores on recruitment opportunities for subordinate herbs in grasslands. J. Veg. Sci. 14, 465e475. Baskin, C.C., Baskin, J.M., 2001. Seeds: Ecology, Biogeography and Evolution of Dormancy and Germination. Academic Press, New York. Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J.R. Stat. Soc. Series B (Methodological) 57, 289e300. Bernhardt-Romermann, M., Romermann, C., Nuske, R., Parth, A., Klotz, S., Schmidt, W., Stadler, J., 2008. On the identification of the most suitable traits for plant functional trait analyses. Oikos 117, 1533e1541. Bobbink, R., Hornung, M., Roelofs, J.G.M., 1998. The effects of air-borne nitrogen pollutants on species diversity in natural and semi natural European vegetation. J. Ecol. 86, 717e738. Borcard, D., Legendre, P., Drapeau, P., 1992. Partialling out the spatial component of ecological variation. Ecology 73, 1045e1055. Breuning, T., Demuth, S., 1999. Rote Liste der Farn und Samenpflanzen BadenWürttembergs. Landesanstalt für Umweltschutz Baden-Württemberg, Mannheim. Bullock, J.M., Franklin, J., Stevenson, M.J., Silvertown, J., Coulson, S.J., Gregory, S.J., Tofts, R., 2001. A plant trait analysis of responses to grazing in a long-term experiment. J. Appl. Ecol. 38, 253e267. Burnham, K.P., Anderson, D.R., 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociol. Method. Res. 33, 261e304. Campbell, B.D., Grime, J.P., 1992. An experimental test of plant strategy theory. Ecology 73, 15e29. Canals, R.M., Sebastia, M.T., 2000. Soil nutrient fluxes and vegetation changes on molehills. J. Veg. Sci. 11, 23e30. Canals, R.M., Sebastia, M.T., 2002. Heathland dynamics in biotically disturbed areas: on the role of some features enhancing heath success. Acta Oecol. 23, 303e312. Canals, R.S., Herman, D.J., Fierestone, M.K., 2003. How disturbance by fossorial mammals alters N cycling in a California annual grassland. Ecology 84, 875e881. Cingolani, A.M., Cabido, M., Gurvich, D.E., Renison, D., Diaz, S., 2007. Filtering processes in the assembly of plant communities: are species presence and abundance driven by the same traits? J. Veg. Sci. 18, 911e920. Connell, J.H., 1978. Diversity in tropical rain forests and coral reefs. Science 199, 1302e1310. De Bello, F., Thuiller, W., Leps, J., Choler, P., Clement, J.C., Macek, P., Sebastia, M.T., Lavorel, S., 2009. Partitioning of functional diversity reveals the scale and extent of trait convergence and divergence. J. Veg. Sci. 20, 475e486. Author's personal copy M. Seifan et al. / Acta Oecologica 36 (2010) 569e577 Diaz, S., Cabido, M., Casanoves, F., 1998. Plant functional traits and environmental filters at a regional scale. J. Veg. Sci. 9, 113e122. Diaz, S., Cabido, M., Zak, M., Carretero, E.M., Aranibar, J., 1999. Plant functional traits, ecosystem structure and land-use history along a climatic gradient in centralwestern Argentina. J. Veg. Sci. 10, 651e660. Draper, N.R., Smith, H., 1998. Applied Regression Analysis. John Wiley & Sons. Edwards, G.R., Crawley, M.J., Heard, M.S., 1999. Factors influencing molehill distribution in grassland: implications for controlling the damage caused by molehills. J. Appl. Ecol. 36, 434e442. Egner, H., Riehm, H., Domingo, W.R., 1960. Untersuchingen über die chemische Bodenanalyse als Grundlage für die Beurteilung des Nährstoffzustandes der Böden. Ann. Royal Agric. Coll. Sweden Ann. 26. Eldridge, D.J., 2009. Badger (Taxidea taxus) mounds affect soil hydrological properties in a degraded shrub steppe. Am. Midl. Nat. 161, 350e358. Eldridge, D.J., Whitford, W.G., 2009. Badger (Taxidea taxus) disturbances increase soil heterogeneity in a degraded shrub-steppe ecosystem. J. Arid Environ. 73, 66e73. Ellenberg, H., Weber, H.E., Düll, R., Wirth, V., Werner, W., Paulißen, D., 1992. Zeigerwerte von Pflanzen in Mitteleuropa. Erich Goltze, Göttingen. Fortunel, C., Granier, E., Joffre, R., Kazakou, E., Quested, H., Grigulis, K., Lavorel, S., Ansquer, P., Castro, H., Cruz, P., Dolezal, J., Eriksson, O., Freitas, H., Golodets, C., Jouany, C., Kigel, J., Kleyer, M., Lehsten, V., Leps, J., Meier, T., Pakeman, R., Papadimiriou, M., Papanastasis, V.P., Quetier, F., Robson, M., Sternberg, M., Theau, J.P., Thebault, A., Zarovali, M., 2009. Leaf traits capture the effects of land use changes and climate on litter decomposability of grasslands across Europe. Ecology 90, 598e611. Foster, B.L., 2001. Constraints on colonization and species richness along a grassland productivity gradient: the role of propagule availability. Ecol. Lett. 4, 530e535. Foster, B.L., Dickson, T.L., 2004. Grassland diversity and productivity: the interplay of resource availability and propagule pools. Ecology 85, 1541e1547. Fynn, R.W.S., Morris, C.D., Kirkman, K.P., 2005. Plant strategies and trait trade-offs influence trends in competitive ability along gradients of soil fertility and disturbance. J. Ecol. 93, 384e394. Garnier, E., Lavorel, S., Ansquer, P., Castro, H., Cruz, P., Dolezal, J., Eriksson, O., Fortunel, C., Freitas, H., Golodets, C., Grigulis, K., Jouany, C., Kazakou, E., Kigel, J., Kleyer, M., Lehsten, V., Leps, J., Meier, T., Pakeman, R., Papadimitriou, M., Papanastasis, V.P., Quested, H., Quetier, F., Robson, M., Roumet, C., Rusch, G., Skarpe, C., Sternberg, M., Theau, J.P., Thebault, A., Vile, D., Zarovali, M.P., 2007. Assessing the effects of land-use change on plant traits, communities and ecosystem functioning in grasslands: a standardized methodology and lessons from an application to 11 European sites. Ann. Bot. 99, 967e985. Grace, J.B., 1993. The effects of habitat productivity on competition intensity. Trend. Ecol. Evol. 8, 228e230. Grees, H., 1993. Die Schwäbische Alb. In: Borcherdt, C. (Ed.), Geographische Landeskunde von Baden-Württemberg, third ed. W. Kohlhammer, pp. 332e362. Grime, J.P., 2001. Plant Strategies, Vegetation Processes and Ecosystem Properties, second ed. John Wiley & Sons, Chichester. Gross, K.L., Mittelbach, G.G., Reynolds, H.L., 2005. Grassland invasibility and diversity: responses to nutrients, seed input, and disturbance. Ecology 86, 476e486. Guo, Q., 1996. Effects of bannertail kangaroo rat mounds on small-scale plant community structure. Oecologia 106, 247e256. Hastings, A., 1980. Disturbance, coexistence, history, and competition for space. Theor. Popul. Biol. 18, 363e373. Hogh-Jensen, H., 2003. The effect of potassium deficiency on growth and N2 fixation in Trifolium repens. Physiol. Plantarum 3, 440e449. Hofer, G., Wagner, H.H., Herzog, F., Edwards, P.J., 2008. Effects of topographic variability on the scaling of plant species richness in gradient dominated landscapes. Ecography 31, 131e139. Huntly, N., Reichman, O.J., 1994. Effects of subterranean mammalian herbivores on vegetation. J. Mammal 75, 852e859. Huston, M.A., 1979. A general hypothesis of species diversity. Am. Nat. 113, 81e101. Huston, M.A., 1999. Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. Oikos 86, 393e401. Huston, M.A., DeAngelis, D.L., 1994. Competition and Coexistence e the effect of resource transport and supply rates. Am. Nat. 144, 954e977. Jongman, R.H.G., Ter Braak, C.J.F., Van Tongeren, O.F.R., 1995. Data Analysis in Community and Landscape Ecology. Cambridge University Press, Cambridge. Jordan-Meille, L., Pellerin, S., 2004. Leaf area establishment of a maize (Zea mays L.) field crop under potassium deficiency. Plant Soil 265, 75e92. Kadmon, R., Benjamini, Y., 2006. Effects of productivity and disturbance on species richness: a neutral model. Am. Nat. 167, 939e946. Kahmen, S., Poschlod, P., 2008. Does germination success differ with respect to seed mass and germination season? Experimental testing of plant functional trait responses to grassland management. Ann. Bot. 101, 541e548. Kalkhan, M.A., Stohlgren, T.J., 2000. Using multi-scale sampling and spatial crosscorrelation to investigate patterns of plant species richness. Environ. Monit. Assess. 64, 591e605. Keddy, P.A., 1992. Assembly and response rules: two goals for predictive community ecology. J. Veg. Sci. 3, 157e164. Kinlaw, A., 1999. A review of burrowing by semi-fossorial vertebrates in arid environments. J. Arid Environ. 41, 127e145. Kleyer, M., Bekker, R.M., Knevel, I.C., Bakker, J.P., Thompson, K., Sonnenschein, M., Poschlod, P., van Groenendael, J.M., Klimes, L., Klimesová, J., Klotz, S., Rusch, G.M., Hermy, M., Adriaens, D., Boedeltje, G., Bossuy, B., Endels, P., Götzenberger, L., Hodgson, J.G., Jackel, A.K., Dannemann, A., Kühn, I., 577 Kunzmann, D., Ozinga, W.A., Römermann, C., Stadler, M., Schlegelmilch, J., Steendam, H.J., Tackenberg, O., Wilmann, B., Cornelissen, J.H.C., Eriksson, O., Garnier, E., Fitter, A., Peco, B., 2008. The LEDA Traitbase: a database of lifehistory traits of the Northwest European flora. J. Ecol 96, 1266e1274. Kondoh, M., 2001. Unifying the relationships of species richness to productivity and disturbance. Proc. R. Soc. Lon. Ser. B-Biol. Sci. 268, 269e271. Kovar, P., Kovarova, M., Dostal, P., Herben, T., 2001. Vegetation of ant-hills in a mountain grassland: effects of mound history and of dominant ant species. Plant Ecol. 156, 215e227. Leishman, M.R., Wright, I.J., Moles, A.T., Westoby, M., 2000. The evolutionary ecology of seed size. In: Fenner, M. (Ed.), Seeds: The Ecology of Regeneration in Plant Communities. CABI International, New York, pp. 31e57. Liu, K., Eastwood, R.J., Flynn, S., Turner, R.M., Stuppy, W.H., 2008. Seed information database (release 7.1, May 2008). http://www.kew.org/data/sid. Louault, F., Pillar, V.D., Aufrere, J., Garnier, E., Soussana, J.F., 2005. Plant traits and functional types in response to reduced disturbance in a semi-natural grassland. J. Veg. Sci. 16, 151e160. Marini, L., Scotton, M., Klimek, S., Pecile, A., 2008. Patterns of plant species richness in Alpine hay meadows: local vs. landscape controls. Basic Appl. Ecol. 9, 365e372. McGill, B.J., Enquist, B.J., Weiher, E., Westoby, M., 2006. Rebuilding community ecology from functional traits. Trends Ecol. Evol. 21, 178e185. Nygaard, B., Ejrnaes, R., 2004. A new approach to functional interpretation of vegetation data. J. Veg. Sci. 15, 49e56. Ozinga, W.A., Schaminee, J.H.J., Bekker, R.M., Bonn, S., Poschlod, P., Tackenberg, O., Bakker, J., van Groenendael, J.M., 2005. Predictability of plant species composition from environmental conditions is constrained by dispersal limitation. Oikos 108, 555e561. Pakeman, R.J., Leps, J., Kleyer, M., Lavorel, S., Granier, E., the VISTA consortium, 2009. Relative climatic, edaphic and management controls of plant functional trait signatures. J. Veg. Sci. 20, 148e159. Palmer, M.W., White, P.S., 1994. Scale dependence and the species-area relationship. Am. Nat. 144, 717e740. Pickett, S.T.A., White, R.S., 1985. The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, London. Rajaniemi, T.K., 2002. Why does fertilization reduce plant species diversity? Testing three competition-based hypotheses. J. Ecol. 90, 316e324. Reich, P.B., Walters, M.B., Ellsworth, D.S., 1992. Leaf life span in relation to leaf, plant and stand characteristics among diverse ecosystems. Ecol. Monog. 62, 365e392. Renners, M., 1991. Geoökologische Raumgliederung der Bundesrepublik Deutschland. Forschung zur deutschen Landeskunde Band 235, Trier. Rebollo, S., Perez-Camacho, L., Valencia, J., Gomez-Sal, A., 2003. Vole mound effects and disturbance rate in a Mediterranean plant community under different grazing and irrigation regimes. Plant Ecol. 169, 227e243. Schiffers, K., Tielbörger, K., Jeltsch, F., 2010. Changing importance of environmental factors driving secondary succession on molehills. J. Veg. Sci. 21, 500e506. Sebastia, M.T., 2004. Role of topography and soils in grassland structuring at the landscape and community scales. Basic Appl. Ecol. 5, 331e346. Sebastia, M.T., Puig, L., 2008. Complex vegetation responses to soil disturbances in mountain grassland. Plant Ecol. 199, 77e88. Seidlova, L., Verlinden, M., Gloser, J., Milbau, A., Nijs, I., 2009. Which plant traits promote growth in the low-light regimes of vegetation gaps? Plant Ecol. 200, 303e318. Smit, R., Olff, H., 1998. Woody species colonization in relation to habitat productivity. Plant Ecol. 139, 203e209. Stoms, D.M., 1994. Scale dependence of species richness maps. Professional Geographer 46, 346e358. Suding, K.N., Goldberg, D., 2001. Do disturbances alter competitive hierarchies? Mechanisms of change following gap creation. Ecology 82, 2133e2149. Ter Braak, C.J.F., Gremmen, N.J.M., 1987. Ecological amplitudes of plant species and the internal consistency of Ellenberg’s indicator values for moisture. Vegetatio 69, 79e87. Verhoeven, K.J.F., Simonsen, K.L., McIntyre, L.M., 2005. Implementing false discovery rate control: increasing your power. Oikos 108, 643e647. Violle, C., Navas, M.L., Vile, D., Kazakou, E., Fortunel, C., Hummel, I., Garnier, E., 2007. Let the concept of trait be functional! Oikos 116, 882e892. Violle, C., Garnier, E., Lecoeur, J., Roumet, C., Podeur, C., Blanchard, A., Navas, M.L., 2009. Competition, traits and resource depletion in plant communities. Oecologia 160, 747e755. Westoby, M., 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant Soil 199, 213e227. Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A., Wright, I.J., 2002. Plant ecological strategies: some leading dimensions of variation between species. Ann. Rev. Ecol. Evol. Syst. 33, 125e159. Wieger, W.G.W., Goedhart, P.W., Van Dobben, H.F., Berendse, F., 2005. Plant species as predictors of soil pH: replacing expert judgment with measurements. J. Veg. Sci. 16, 461e470. Wright, I.J., Reich, P.B., Westoby, M., 2001. Strategy shifts in leaf physiology, structure and nutrient content between species of high- and low-rainfall and highand low-nutrient habitats. Funct. Ecol. 15, 423e434. Wright, I.J., Westoby, M., Reich, P.B., 2002. Convergence towards higher leaf mass per area in dry and nutrient poor habitats had different consequences for leaf life span. J. Ecol. 90, 534e543. Zobel, M., Otsus, M., Liira, J., Moora, M., Mols, T., 2000. Is small-scale species richness limited by seed availability or microsite availability? Ecology 81, 3274e3282.