Importance of environmental and spatial ... - Wiley Online Library

0 downloads 0 Views 438KB Size Report
Groningen, The Netherlands. *Correspondence: Tina Astor, Department of. Ecology, Swedish University of Agricultural. Sciences, PO Box 7044, SE-75007 ...
Journal of Biogeography (J. Biogeogr.) (2017) 44, 1362–1372

ORIGINAL ARTICLE

Importance of environmental and spatial components for species and trait composition in terrestrial snail communities Tina Astor1* , Ted von Proschwitz2, Joachim Strengbom1, Matty P. Berg3,4 and Jan Bengtsson1

1

Department of Ecology, Swedish University of Agricultural Sciences, PO Box 7044, SE-75007 Uppsala, Sweden, 2G€oteborg Museum of Natural History, PO Box 7283, 402 35 G€oteborg, Sweden, 3Department of Ecology Science, Section Animal Ecology, Vrije Universiteit, 1081 HV Amsterdam, The Netherlands, 4Groningen Institute for Evolutionary Life Sciences, Community and Conservation Ecology Group, University of Groningen, Nijenborgh 7, 9747 AG Groningen, The Netherlands

ABSTRACT

Aim Despite the huge diversity of soil animals and their recognized contribution to many ecosystem functions, little is known about the relative importance of factors controlling their abundance and distribution. We examined the relative importance of environmental and spatial factors in explaining the species and functional trait composition of terrestrial snail communities at the level of metacommunities (spatial extent c. 100 9 100 km) in a heterogeneous, intensively used landscape. We hypothesized that both spatial and environmental factors contribute to the variation in community structure across the landscape, but expected environmental variables describing local habitat conditions to be most important. Location County of Sk ane, south Sweden. Methods We quantified community structure in terms of species composition and as functional trait composition, because functional traits directly link species performance to environmental conditions. To disentangle the unique and shared contribution of environmental and spatial factors to the variation in snail community structure (in terms of species and trait composition) we applied a partial redundancy analysis. Results Species traits explained more of the variance in community composition than species identity. Snail traits such as tolerance to environmental stress (related to soil moisture content) and niche width were correlated with the main environmental gradient. Environmental variables (i.e. soil moisture content, vegetation characteristics and soil pH) contributed considerably more to variation in community composition (species: 11.4%; traits: 24.9%) than the spatial variables (species: 6.5%; traits: 4.2%).

*Correspondence: Tina Astor, Department of Ecology, Swedish University of Agricultural Sciences, PO Box 7044, SE-75007 Uppsala, Sweden. E-mail: [email protected]

Main conclusions The results highlight that both environmental and spatial variables are required to understand the relative importance of niche-based and intrinsic population processes as drivers of terrestrial snail community structure. However, at the scale of our study niche-based community structuring explained by the trait–environment relationship is considerably more important than spatial patterning independent of the environment. Keywords community assembly, functional traits, Gastropoda, soil fauna, spatial heterogeneity, variance partitioning

INTRODUCTION Soil animals constitute a diverse and functionally important group of organisms in terrestrial ecosystems. However, our knowledge of the factors controlling community structure, 1362

http://wileyonlinelibrary.com/journal/jbi doi:10.1111/jbi.12946

that is, shifts in composition and diversity of soil fauna across time and space is relatively poor (Bardgett, 2002). Most studies explaining spatiotemporal shifts in community structure have used a taxonomic approach, and they have largely increased our knowledge on how abundances of ª 2017 John Wiley & Sons Ltd

Environmental and spatial components affect snail community composition species may vary along environmental gradients. However, they have not been very useful for unravelling the causal mechanisms behind changes in species composition in response to environmental variation across different spatial and temporal scales, often because of a strong context dependency. In contrast, trait-based approaches are regarded to be much more useful in this respect (McGill et al., 2006; Messier et al., 2010), as the functional traits of species directly link their performance to environmental conditions (Lavorel & Garnier, 2002; Violle et al., 2007; Suding et al., 2008; de Bello et al., 2010). Compared to plant ecology, trait-based approaches are still relatively new to animal ecology, partly because of a lack of reliable or relevant trait data in many groups (Moretti et al., 2016). The situation is, however, rapidly changing. Traits underlying the performance of animals are currently being measured (Moretti et al., 2016), and trait databases are becoming available for analyses (e.g. Makkonen et al., 2011; Burkhardt et al., 2012). Yet, few studies have shown consistent and general links between environmental factors and soil fauna traits that directly underly organisms’ growth, reproduction and survival, collectively determining a species performance. For instance, physiological traits underpinning desiccation resistance explained species distribution patterns of terrestrial isopods in relation to water availability across Hungary (Dias et al., 2013). Traits related to life-form and the vertical structuring of communities in soils can be used to predict the responses of springtails (Collembola) to environmental changes (Krab et al., 2010; Makkonen et al., 2011; Widenfalk et al., 2015). Moreover, traits related to moisture conditions can be important drivers of community composition of land snails on islands (Astor et al., 2014). Soil organisms often show a patchy distribution, across both horizontal and vertical space (reviewed by Berg, 2012), which may result from spatial heterogeneity in environmental conditions, or from intrinsic population processes, such as dispersal, reproduction and competition (e.g. Ettema & Wardle, 2002) that is independent of environmental heterogeneity. For instance, microhabitat complexity and landscape features, such as the amount of forest area and connectivity of forest patches, can explain differences in collembolan life-forms (soil-dwelling versus surface-dwelling), which can be explained by traits that are related to the animals’ dispersal abilities (Martins da Silva et al., 2012). Thus, in order to fully understand how communities are assembled, it is necessary to disentangle the trait–environment relationship from processes leading to spatial patterning which is independent of the environment, such as intrinsic population processes like dispersal limitation. Methods that partition variation in community structure into spatial and environmental components (Borcard et al., 1992; Legendre et al., 1994) have been used for a range of organisms (Labaune & Magnin, 2001; Benefer et al., 2010; Martins da Silva et al., 2012), both in studies based on functional groups (e.g. Viketoft, 2013) and across groups of organisms (Hajek et al., 2011; Chytry et al., 2012). These Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

studies often show the relative importance of environmental and spatial factors but do not directly provide a mechanistic understanding why the one factor dominates over the other. Functional trait approaches might provide this mechanistic understanding. However, studies on trait composition of communities in relation to environment-space partitioning are rare. There are numerous studies showing that occurrences of land snail species are limited by environmental conditions (e.g. G€ardenfors et al., 1995; Martin & Sommer, 2004), habitat and vegetation characteristics (e.g. Barker & Mayhill, 1999). In landscapes characterized by discontinuous habitats, occurrences of snail species might also be determined by their dispersal abilities (e.g. Hajek et al., 2011). Dispersal mechanisms in terrestrial snails comprise both active and passive dispersal, but active dispersal ability is generally believed to be poor (e.g. Baur & Baur, 1993). These aspects of terrestrial snails are shared with many other soil fauna groups and make snails a good model system to unravel underlying mechanisms behind spatiotemporal changes in community composition. The main objectives of the present study were to (1) identify the most important environmental and spatial variables related to community composition of soil fauna (both for species and traits), and (2) quantify the relative importance of environmental and spatial predictors for species and trait composition of soil animal communities on the scale of meta-communities (spatial extent c. 100 9 100 km) in a highly heterogeneous and fragmented landscape. At this scale, both local environmental conditions and dispersal ability could potentially play a role in community assembly. To study these questions, we selected land snails, a group of soil fauna of which the ecology and taxonomy is relatively wellknown and for which an extensive trait database is available (Falkner et al., 2001). MATERIALS AND METHODS Study area We used data from a land snail survey (Walden, 1965, 1986; von Proschwitz, 1996) conducted by the Gothenburg Natural History Museum within an area of 10.939 km2 in Sk ane Province, South Sweden from 1960 to 2012 (Fig. 1). Sk ane is in the nemoral and continental environmental zones (Metzger et al., 2005), with average annual precipitation ranging between 500 and 1000 mm, and average temperature ranging from 0 to 2 °C in January and from 15 to 17 °C in July. The area was completely covered by the latest Pleistocene ice-sheet, and hence land snails (as well as all other organisms) have colonized the area during the last c. 12 kyr. The landscape of Sk ane is strongly influenced by humans due to intensive agricultural practice and land use resulting in high small-scale heterogeneity and habitat fragmentation. In our study, we restricted the analyses to sites with available soil pH data, resulting in a total of 622 sites. For sites that had 1363

T. Astor et al.

Figure 1 Location of the province of Sk ane (grey; 55°500 2400 N/13°310 4800 E, WGS 84) in south Sweden (left), and a map of Sk ane with the location of the 622 sampling sites (right).

been repeatedly sampled, we only used the last of the sampling years. Snail sampling We restricted our analyses to shelled land snails because we lacked adequate trait information for slugs. Snails were sampled with a standard method (Walden, 1965; von Proschwitz, 1996) from April to October each year. A sample consisted of 15–20 L of fresh soil and litter, with the litter representative of the plant species composition of the site, and was sieved through an 8 9 8 mm mesh. The fraction left in the sieve was searched manually, and shells were either identified and counted directly, or brought to the laboratory for identification. The smaller snails accumulated in the portion passing through the sieve (c. 1–2 L). This fraction was air-dried, and then sieved in two fractions through a 4 9 4 mm mesh. The snails from both fractions were sorted out by hand under a 4–5 times magnifying glass. At some sites, snails were collected by manual search in the habitat by one person for 45 min. Our analyses were based on presence–absence data only and the manual search was conducted by expert malacologists. After sorting, all snails were identified to species under the microscope using 6–50 times magnification. Soil pH was measured calorimetrically (using S.K.P. Soil tester; Weibull Ltd) on the fresh litter sample, to the nearest 0.25 pH units. Detailed field notes on the geomorphology, structure (e.g. bare ground cover, presence of stones, signs of disturbances and human influence) and vegetation composition (tree cover and ground flora composition) were taken for each site. These notes were evaluated and categorized in order to be able to use the information for statistical analysis (see section ‘Explanatory environmental variables’). 1364

Trait information and calculation of community mean trait values Because we only had access to already collected data, it was not possible to measure key traits directly on snails present at the study sites. Instead, we utilized information from a trait database (Falkner et al., 2001), one of the most comprehensive databases currently available for soil fauna. We selected 12 traits that are related to life-history, reproduction, dispersal ability, habitat preferences and tolerance to abiotic conditions (Table 1), all linking land snails to spatial heterogeneity in environmental factors (Astor et al., 2014). The database has a fuzzy coding structure, meaning that each trait consists of several categories wherein the entries describe the species’ affinity to the respective categories, with 0 defined as no affinity, 1 as weak affinity, 2 as moderate affinity and 3 as strong affinity. To obtain one value for each trait and species, we calculated the sum of the relative (to the number of categories) affinities times the category number (see Appendix S1, Table S1 in Supporting Information for an example). For instance, the trait ‘humidity preference’ consisted of three categories (1: dry, 2: moist, 3: wet). If a species has a low affinity (value 1) for category 1 and a high affinity (value 3) for category 2 this means that one quarter of the population has a preference for dry conditions (category 1) and three quarters of the population has a preference for moist conditions (category 2). The average trait value was then calculated as ¼ 9 1 + ¾ 9 2, to weight the relative importance of the trait values for that particular species. The number of annual reproduction periods (how often snails reproduce per year) was established by counting the occurrences in the corresponding main reproduction period categories (see Appendix S1, Table S2 for an example). For instance, if a Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

Environmental and spatial components affect snail community composition Table 1 Snail traits selected and their original database values (category numbers in bold) from which an average trait value per species was calculated (see Appendix S1, Table S2). Details concerning the units of expression and the rationale behind these categories can be found in Falkner et al. (2001). These species-specific trait values were used to calculate community mean trait values for each site.

Life-history

Morphology Environmental tolerance Specialization

Traits

Database categories

Maximum shell size Age at maturity Number of offspring (eggs) Reproduction mode Main reproduction periods Shell shape Humidity preference Dry period survival Inundation tolerance Ecosystem preferences

1: < 2.5 mm; 2: 2.5–5 mm; 3: 5–15 mm; 4: > 15 mm 1: < 1 year; 2: 1 year; 3: > 1 year 1: 1–10; 2: 11–100

Average Average average

1: Cross-fertilization, 2: self-fertilization 1: January–February; 2: March–April; 3: May–June; 4: July–August; 5: September–October; 6: November–December 1: Depressed; 2: globose/conical; 3: oblong 1: Dry (xerophilous); 2: moist (mesophilous); 3: wet (hygrophilous) 1: Days; 2: weeks; 3: months 1: Low; 2: moderate; 3: high Deciduous forest; scrub; mixed forest; coniferous forest; tall herb; thermophilous forest fringe; unimproved grassland; heathland; coastal dunes; inland dunes; cliff/rock; scree/walls; hedge; fen; reed; water edge Trees; shrubs/bushes/saplings; herbs; mosses; timber; forest litter; stones; strand debris; sand; soil; bare rock; root zone; crevices; caves Deciduous forest litter; fungi; lichens; mosses; algae, vascular plants; carnivorous or saprophagous

Percentage self-fertilization Annual number of reproduction periods Average Average Average Average Ecosystem niche width using Shannon Index

Micro-habitat preferences

Food preferences

Microhabitat niche width using Shannon Index Diet niche width using Shannon Index

Table 2 Full set of habitat variables used as explanatory variables in the global redundancy analysis of snail species and trait composition. Mean ( SD) and range are given for the numerical variables. All other variables are coded as dummy variables (1/0). For a more detailed description of habitat types and soil types, see Appendix S1 Table S1 and S4. Mean  SD

Variables

Description

Major habitat types

Coniferous forest, deciduous forest, mixed forest, forest edge, tree patch, shrub land, grassland with trees or shrubs, open grassland (without trees or shrubs), fen, bog, swamp, other wetland, park, stone wall, ruderal, open other, mixed Dry, intermediate, moist, wet, variable 1/0 1/0 1/0 1/0 1/0 Dystric cambisol (CMdy), Eutric Cambisol (CMeu), Dystric Regosol (RGdy), Haplic Podzol (PZha)

Soil moisture Stones Bare ground Grazed Inland shore Sea shore Soil type

pH Mean temperature vegetation period Mean precipitation vegetation period Maximum temperature summer Minimum temperature winter

Range

(°C)

5.86  0.873 12.2  0.498

4.24–8.50 11.2–13.5

(mm)

2.33  0.541

1.42–4.27

(°C)

21.4  1.43

17.7–24.6

(°C)

10.5  3.19

species has an entry in three categories, its annual reproduction period is set to 3. The niche width calculations (diet-, microhabitat- and ecosystem niche width) were based on the number of Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

19.4 to 1.00

respective diet-, microhabitat preference and ecosystem preference categories (Table 2) that a species occurred in, and were calculated using the Shannon index (Shannon, 1948). From the trait ‘reproduction mode’, which has two 1365

T. Astor et al. non-mutually exclusive categories (cross-fertilization and self-fertilization), we calculated the percentage of self-fertilization for each species (e.g. if a species had an affinity of 2 for self-fertilization and an affinity of 1 for cross-fertilization the percentage degree of self-fertilization was 66.7%). For each community we calculated community mean trait values by averaging the trait values of all species present in a community. Explanatory environmental variables The original field notes contain detailed, but not standardized descriptions of the site habitats, including dominant vegetation types, the presence of single plant species, signs of disturbance or human influence (e.g. heaps of swept leaves, brick stone fragments, rubble), and pH values. Based on this information we generated seven environmental variables (major habitat type, 15 levels, see also Appendix S1, Table S3), soil moisture (5 levels), bare ground (2 levels), stones (2 levels), livestock grazing (2 levels), inland shore (i.e. edges of rivers, streams, lakes and ponds; 2 levels), sea shore (2 levels) and soil type (4 levels); Table 2. Habitat classification was based on the dominant vegetation types, the degree of vegetation openness and the presence of plant species indicating specific conditions (dry/wet, nutrient-rich/poor presence/absence of grazing or disturbance), and was determined by expert knowledge of an independent plant ecologist. Climatic data (daily mean precipitation and air temperature) were obtained in the form of grid data, 4 9 4 km resolution, from the Swedish Meteorological and Hydrological Institute (SMHI, http://luftwebb.smhi.se/). From this we extracted mean precipitation and mean temperature over the vegetation growth period (daily mean temperature exceeding 5 °C), minimum winter temperature (January–February), and maximum summer temperature (June–August) (Table 1). Information on major soil types (Table 1) of the sites was obtained from the European Soil Database (ESDB) v2.0 (Panagos et al., 2012). The soil types varied in soil depth, fertility and acidity (Driessen et al., 2001) (see Appendix S1, Table S4). Explanatory spatial variables To generate spatial variables at different spatial scales we used PCNM (principal coordinate of neighbour matrices) analysis (Borcard & Legendre, 2002), which belongs to the wider family of methods called MEM (Moran’s Eigenvector Maps) (Dray et al., 2006). First, a Euclidean distance matrix was constructed from the geographical site coordinates. Then, the matrix was truncated to retain only the distances among close neighbours. Finally, a principal coordinate analysis was computed and eigenvalues with positive spatial correlation (Moran’s I) were retained and used as explanatory variables for variation partitioning. Lower numbered PCNM axes resemble broad-scale spatial structures, whereas higher numbered axes resemble fine scale spatial structures. 1366

The analysis was performed with the ‘PCNM’ package in R version 3.1.0 (R Core Team 2014). Data analysis All analyses were done in R (R Core Team 2014). A detrended correspondence analysis carried out prior to analysis suggested that a linear response can be assumed both for the species and the trait data (see Appendix S2, Table S5). To test the relative importance of environmental and spatial variables for determining snail community composition we carried out two analyses, one based on species occurrence and one on community mean trait values. Assuming a linear response, we used partial redundancy analysis with unbiased coefficients of determination using adjusted R2 (Peres-Neto et al., 2006) to partition the variation in community composition (species or traits) among sites into the following fractions: environment (E), space (S), pure environment after accounting for spatial variables (E|S), pure space after accounting for environmental variables (S|E), a fraction shared by environment and space (SSE, spatially structured environment) and residual variation. To ensure that only significant variables entered the variation partitioning procedure, a forward selection of environmental variables and spatial variables was carried out prior to the variation partitioning. To prevent the risk of including too many variables in the model, and to reduce the risk of type-I error, we used the double stopping criterion proposed by Blanchet et al. (2008). In this procedure the variables were selected in two steps. First, a global model with all variables was computed. If this model was significant, the adjusted coefficient of multiple determination (adj. R2) of the global model was used as a second stopping criterion in the forward selection procedure (Blanchet et al., 2008). All significance tests were done by permutation tests for constrained analysis with 999 permutations using the ‘vegan’ package in R. Canonical ordination in combination with PCNM analysis is commonly used to partition variation in community structure into environmental and spatial variables. However, recent studies comparing commonly used methods for variation partitioning suggest that raw data approaches, such as canonical ordinations with spatial explanatory variables (e.g. PCNM eigenvectors), tend to underestimate the contribution of environmental variables, whereas distance-based methods, such as multiple regression on distance matrices, tend to underestimate the effect of spatial variables (Gilbert & Bennett, 2010). Therefore, we also performed multiple regressions on distance matrices (MRM) (Lichstein, 2007). For this purpose the community, environmental and the geographical coordinate matrices were converted to Euclidean distance matrices. The community distance matrix served as a response variable for three different models, that is, one with both the environmental and spatial distance matrix (representing the total explained variation in trait composition), one with only the environmental distance matrix, and a third Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

Environmental and spatial components affect snail community composition with only the spatial distance matrix. Significance of regression coefficients, and R2 values were tested using permutation tests with 1000 permutations. The R2 values were used to estimate the variation in community composition explained by pure environmental, and spatial components, as well as their shared contribution. Potential limitations of the dataset The information from Falkner et al. (2001) is based largely on literature sources supplemented by expert knowledge. In order to account for local adaptation it would be preferable to measure traits that directly underlie the performance of species in their habitat. This was unfortunately not possible in the present study because such data were not gathered in the surveys from which we derived our data. The Falkner database not only contains valuable information on trait differences among snail species but also information on the potential range of trait variation within species. A prerequisite for a meaningful analysis of among-species average trait variation is that among-species trait variation is larger than within-species variation. This prerequisite turned out to hold for all traits included in our study, which support the validity of using the Falkner trait database. A further test for the validity of the trait information of the database can be found in Appendix S2). Since we used only the last sampling year in case a site was sampled multiple times, the sampling year varied among sites in our dataset. To test if ‘sampling year’ had an effect on the results, we carried out an additional variation partitioning analysis with ‘sampling year’ as third component (besides environmental and spatial variables). Sampling year had no effect on our general conclusions (see Appendix S2, Fig. S1). To test if the snail sampling method (sorting litter samples versus manual search at the sites) influenced the results we carried out an additional variation partitioning analysis with ‘sampling method’ as a third component (besides environmental and spatial variables). Differences in sampling methods did not affect our general conclusions (see Appendix S2, Fig. S2). RESULTS Species–environment relationships Four environmental variables (i.e. deciduous forest, soil pH, fen and wet sites) were retained in the forward selection procedure of the RDA (based on species occurrences), and explained 13.5% of the total variance in the RDA. The first RDA axis accounted for 8.9% of the total variation in species composition. The variables ‘wet’, and ‘fen’ had the highest positive scores on the first axis, followed by soil pH, whereas deciduous forest was negatively associated with the first axis. The second RDA axis accounted for 2.6% of the total variation and was mainly represented Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

by soil pH, which had a negative association with this axis. Trait–environment relationships Eight local environmental variables were retained in the forward selection procedure based on traits (i.e. wetland, fen, deciduous forest, soil pH, haplic Podzol soil type, presence of stones, wet sites and grazed sites), whereas none of the climatic variables was retained. The environmental variables explained 33.9% of the total variance in the RDA. The first RDA axis accounted for 23.5% of the total variation in trait composition, and represents mainly a soil moisture and vegetation coverage gradient, with more dry, deciduous forest having positive scores and wet, open habitats having negative scores. The second axis accounted for 8.2% of the total variation in the RDA, and represents a soil pH gradient, with pH having high negative scores and Podzol (PZha, representing an acidic soil) having positive scores (Fig. 2). Community mean traits with high scores on the first axis were mainly represented by tolerance traits (e.g. humidity preference, inundation tolerance and dry period survival) and community mean traits related to specialization (e.g. diet niche width and microhabitat niche width). The community mean values for humidity preference and inundation tolerance were positively correlated with soil moisture, whereas dry period survival showed a negative correlation with soil moisture. All measures of niche-width, as well as community mean for maximum shell size, were greater in deciduous forests. The community mean values for shell size, number of offspring and inundation tolerance were strongly related to the second axis, and their scores were positively correlated with soil pH (negative values along RDA axis 2). Variation partitioning of species occurrences Twenty-seven of 186 spatial PCNM variables were chosen by forward selection (PCNM 1-18, 20, 21, 26, 28, 42, 56, 68, 105, 148 and 156), most of them representing large-scale spatial variables. Variation partitioning showed that across sites, all environmental and spatial variables together (TOTsp) explained 25.3% of the variation in species composition (Fig. 3a). The environment (Esp) explained 18.8% of the variation, of which 11.4% represented the pure environmental factors (E|Ssp) and 7.4% represented spatially structured environmental factors (SSEsp, variation shared by environment and space). The lowest amount of variation (6.5%) was explained by space alone (S|Esp). Permutation tests showed that all testable fractions were significant (see Appendix S3, Table S6). In MRM (see Appendix S3, Table S7), environment and space together explained 17.5% of the total variation in species composition, of which pure environment made up the largest part (16.2%). Pure space (0.5%) and the shared spaceenvironment component (0.8%) had a similar contribution to the explained variation in species composition. 1367

T. Astor et al.

Figure 2 RDA bi-plot with the location of southern Swedish terrestrial snail community mean trait values (indicated by the tips of the arrows) in relation with the significant environmental variables (grey triangles) after forward selection. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3 Unique (pure environment: E|S, pure space: S|E) and shared (spatially structured environment: SSE) contribution of environmental and spatial variables to (a) species composition and (b) trait composition of terrestrial snail communities in the province of Sk ane, south Sweden (n = 622). Level of explained variance is expressed as the percentage of total variation in community mean trait values. Subscript sp: species, subscript tr: trait.

Variation partitioning of community mean trait values Out of 186 spatial PCNM eigenvectors, 17 were chosen by forward selection (PCNM 1-9, 11, 12, 15, 16, 20, 21 and 65), most of these representing large-scale spatial structures. Variation partitioning revealed that all environmental and spatial variables together (TOTtr) explained 41.7% of the variation in trait composition across sites (Fig. 3b). The component environment (Etr) explained 37.6% of the variation, of which 24.9% represented the pure environmental factors (E| Str), describing the true trait–environment relationship, and 12.6% represented spatially structured environmental factors (SSE, variation shared by environment and space). Only 4.2% 1368

of the total variation in trait composition across sites was explained by space alone (S|Etr). Permutation tests showed that all testable fractions were significant (see Appendix S3, Table S8). In MRM (see Appendix S3, Table S9), environment and space together explained 14.7% of the total variation in trait composition, of which pure environment explained 14.1%. Pure space explained 0.1%, and the shared component explained 0.5% of the total variation. DISCUSSION We have assessed the relative importance of environmental and spatial factors for variation in land snail community composition across landscapes, by adopting a trait-based Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

Environmental and spatial components affect snail community composition approach. Our study showed that species traits underlying species performance explained more of the variance in community composition than species identity, which agrees with studies of other groups of organisms (among others, Ackerly & Cornwell, 2007; Martins da Silva et al., 2012). This is not surprising because functional traits that determine how species cope with their environment provide a direct link between species performance or fitness and environmental conditions (Violle et al., 2007). Thus, a trait-based approach should provide a more mechanistic understanding of processes structuring communities, such as niche-based and intrinsic population processes. Moreover, we found that local environmental conditions were more important than spatial variables in explaining community composition of terrestrial snails, supporting our prediction and indicating the predominance of niche-based community structuring. Of the environmental variables, soil moisture and vegetation characteristics explained most of the variation both in species and in trait composition. We found strong correlations between these environmental variables and the community mean of traits. This indicates that tolerance to environmental stress (related to moisture) and niche width (related to vegetation characteristics) are important factors affecting community composition among land snails. These results reinforce the observations of earlier studies that have demonstrated the importance of local environmental conditions, such as humidity and pH (e.g. G€ardenfors et al., 1995; Martin & Sommer, 2004), habitat and vegetation characteristics (e.g. W€areborn, 1970; Van Es & Boag, 1981; Ports, 1996; Barker & Mayhill, 1999; Ondina & Mato, 2001), or soil architecture (Nekola, 2003) for the distribution of terrestrial snails in many parts of the world. Our results are in accordance with other studies on soil fauna groups with a supposedly limited dispersal ability, such as Collembola (Martins da Silva et al., 2012), suggesting for soil fauna niche-based community structuring might be more important than spatial patterning at the landscape scale, independent of the environment. Because spatial variables explained substantially less of the variance than did environmental variables, the poor active dispersal ability of snails (e.g. Baur & Baur, 1993; Schilthuizen & Lombaerts, 1994) does not appear to be a major force for determining their community composition. Especially small snail species can exhibit extensive distribution ranges (Nekola et al., 2009; Nekola, 2014), suggesting that passive long-distance dispersal of snails by, for instance, wind (Rees, 1965; Vagvolgyi, 1975) or assisted dispersal by birds (Gittenberger et al., 2006; Kawakami et al., 2008) may be of importance for their large-scale distributions. Sk ane lies within one biogeographical zone. All snail species have colonized the region since the last ice-age, and hence are unlikely to be poor dispersers over longer geological time periods. However, the landscape is strongly influenced by humans due to intensive agricultural practice and land use. Thus, the area exhibits high small-scale heterogeneity, even within biotopes. On the spatial scale we address in this study – the meta-community scale – landscape configuration and fragmentation can Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

influence the occurrence of species in communities and for very specialized snail species, such as, for example, Vertigo geyeri, local extinction and limited recolonization ability may well play a role in such fragmented landscapes. Spatial patterning independent of the environment has been found on similar scales, for instance for arboreal and terrestrial mite communities (Lindo & Winchester, 2009) and snails (Hajek et al., 2011). A reason for the comparatively low importance of the purely spatial variables in our study could be that we analysed a wide range of habitats and environmental conditions (see above) within the same biogeographical region and that the full extent of dispersal limitation predominates at much larger scales (e.g. across biogeographical regions). Hence, we expect that our results may not hold in areas that have not been subjected to large-scale glaciation, such as central and southern Europe, and the Alps where many valleys contain unique species (Kerney & Cameron, 1979). Besides environmental and spatial variables, additional factors might influence community structure which is indicated by the amount of unexplained variation. The unexplained variation in our study was rather large (species: 74.7%; traits: 58.3%) despite the fact the PCNM analysis is regarded to capture both broad and fine scale spatial structures. However, it is not uncommon to find large amounts of unexplained variation in the type of analyses we have performed. In a recent meta-analysis of the usefulness of the PCNM method to capture the relative importance of space versus environment for a broad range of terrestrial and aquatic organisms, on average 50% of the variation was unexplained (Cottenie, 2005). High residual variation can originate from historic factors (Cameron et al., 1980; Magnin et al., 1995), effects of land use and landscape composition (G€ otmark et al., 2008), from environmental variables not included in the studies, or be a result of spatial processes operating at smaller or larger spatial scales than captured by our study. This study, using a trait-based approach, confirms the importance of local habitat conditions for regional distribution of terrestrial snails across landscapes, whereas climatic variables had a subordinate importance. Explicit inclusion of spatial components further enabled us to disentangle the true trait–environment relationship (niche-based process) from spatial patterns that are independent of the environment. The environmental component explained the largest amount of variation in our study. However, omitting the spatial component would have made it impossible to detect spatial pattern resulting from intrinsic population processes, such as dispersal limitation. Our findings highlight the need to include both environmental and spatial variables into studies on drivers of community structure and assembly as it increases the understanding of the relative importance of niche-based versus intrinsic population processes. ACKNOWLEDGEMENTS We thank the Swedish Research Council (grant to Jan Bengtsson) for funding Tina Astor’s work and the 1369

T. Astor et al. Gothenburg Natural History Museum for providing the snail occurrence data and habitat descriptions. Assistance with habitat classification by Aina Pihlgren is gratefully acknowledged. The study would not have been possible without the decades of snail inventories started by H. Lohmander, and continued by H.W. Walden and T. von Proschwitz. REFERENCES Ackerly, D.D. & Cornwell, W.K. (2007) A trait-based approach to community assembly: partitioning of species trait values into within- and among-community components. Ecology Letters, 10, 135–145. doi:10.1111/j.14610248.2006.01006.x. Astor, T., Strengbom, J., Berg, M.P., Lenoir, L., Marteinsd ottir, B. & Bengtsson, J. (2014) Underdispersion and overdispersion of traits in terrestrial snail communities on islands. Ecology and Evolution, 4, 2090–2102. doi:10.1002/ ece3.1084. Bardgett, R.D. (2002) Causes and consequences of biological diversity in soil. Zoology, 105, 367–374. doi:10.1078/09442006-00072. Barker, G.M. & Mayhill, P.C. (1999) Patterns of diversity and habitat relationships in terrestrial mollusc communities of the Pukeamaru Ecological District, northeastern New Zealand. Journal of Biogeography, 26, 215–238. doi:10. 1046/j.1365-2699.1999.00267.x. Baur, A. & Baur, B. (1993) Daily movement patterns and dispersal in the land snail Arianta arbustorum. Malacologia, 35, 89–98. Benefer, C., Andrew, P., Blackshaw, R., Ellis, J. & Knight, M. (2010) The spatial distribution of phytophagous insect larvae in grassland soils. Applied Soil Ecology, 45, 269–274. Berg, M.P. (2012) Patterns of biodiversity at fine and small spatial scales. Soil ecology and ecosystem services (ed. by D.H. Wall, R.D. Bardgett, V. Behan-Pelletier, J.E. Herrick, T. Hefin Jones, K. Ritz, J. Six, D.R. Strong and W.H. Van Der Putten), pp. 136–152. Oxford University Press, Oxford. Blanchet, F.G., Legendre, P. & Borcard, D. (2008) Forward selection of explanatory variables. Ecology, 89, 2623–2632. doi:10.1890/07-0986.1. Borcard, D. & Legendre, P. (2002) All-scale spatial analysis of ecological data by means of principal coordinates of neighbour matrices. Ecological Modelling, 153, 51–68. Borcard, D., Legendre, P. & Drapeau, P. (1992) Partialling out the spatial component of ecological variation. Ecology, 73, 1045–1055. doi:10.2307/1940179. Burkhardt, U., Russell, D.J., Decker, P., Do¨hler, M., Ho¨fer, H., Lesch, S., Rick, S., Ro¨mbke, J., Trog, C., Vorwald, J., Wurst, E., Voigtla¨nder, K. & Xylander, W.E.R. (2011) The Edaphobase project of GBIF-Germany - A new online soilzoological data warehouse. Applied Soil Ecology, 83, 3–12. http://dx.doi.org/10.1016/j.apsoil.2014.03.021. Cameron, R.A.D., Down, K. & Pannett, D.J. (1980) Historical and environmental influences on hedgerow snail 1370

faunas. Biological Journal of the Linnean Society, 13, 75–87. doi:10.1111/j.1095-8312.1980.tb00071.x.  Chytry, M., Lososova, Z., Horsak, M., Uher, B., Cejka, T., Danihelka, J., Fajmon, K., Hajek, O., Juri^ckova, L., Kin trova, K., Lanıkova, D., Oty pkova, Z., Reho rek, V. & Tichy , L. (2012) Dispersal limitation is stronger in communities of microorganisms than macroorganisms across Central European cities. Journal of Biogeography, 39, 1101– 1111. doi:10.1111/j.1365-2699.2011.02664.x. Cottenie, K. (2005) Integrating environmental and spatial processes in ecological community dynamics. Ecology Letters, 8, 1175–1182. doi:10.1111/j.1461-0248.2005.00820.x. de Bello, F., Lavorel, S., Dıaz, S., Harrington, R., Cornelissen, J.H.C., Bardgett, R.D., Berg, M.P., Cipriotti, P., Feld, C.K., Hering, D., Martins da Silva, P., Potts, S.G., Sandin, L., Sousa, J.P., Storkey, J., Wardle, D.A. & Harrison, P.A. (2010) Towards an assessment of multiple ecosystem processes and services via functional traits. Biodiversity and Conservation, 19, 2873–2893. doi:10.1007/s10531-0109850-9. Dias, A.T.C., Krab, E.J., Marien, J., Zimmer, M., Cornelissen, J.H.C., Ellers, J., Wardle, D.A. & Berg, M.P. (2013) Traits underpinning desiccation resistance explain distribution patterns of terrestrial isopods. Oecologia, 172, 667–677. doi:10.1007/s00442-012-2541-3. Dray, S., Legendre, P. & Peres-Neto, P.R. (2006) Spatial modelling: a comprehensive framework for principal coordinate analysis of neighbour matrices (PCNM). Ecological Modelling, 196, 483–493. doi:10.1016/j.ecolmodel.2006.02. 015. Driessen, P.M., Deckers, J.A., Spaargaren, O.C. & Nachtergaele, F.O. (2001) World Soil Resources Report 94 – Lecture notes on the major soils of the world. 307 p. Food and Agriculture Organization of the United Nations, Rome. Ettema, C.H. & Wardle, D.A. (2002) Spatial soil ecology. Trends in Ecology and Evolution, 17, 177–183. doi:10.1016/ s0169-5347(02)02496-5. Falkner, G., Obrdlık, P., Castella, E. & Speight, M.C.D. (2001) Shelled Gastropoda of Western Europe. Verlag der Friedrich-Held-Gesellschaft, Munich. G€ardenfors, U., Walden, H.W. & W€areborn, I. (1995) Effects of soil acidification on forest land snails. Ecological Bulletins, 44, 259–270. Gilbert, B. & Bennett, J.R. (2010) Partitioning variation in ecological communities: do the numbers add up? Journal of Applied Ecology, 47, 1071–1082. doi:10.1111/j.1365-2664. 2010.01861.x. Gittenberger, E., Groenenberg, D.S.J., Kokshoorn, B. & Preece, R.C. (2006) Molecular trails from hitch-hiking snails. Nature, 439, 409–409. doi:10.1038/439409a. G€ otmark, F., von Proschwitz, T. & Franc, N. (2008) Are small sedentary species affected by habitat fragmentation? Local vs. landscape factors predicting species richness and composition of land molluscs in Swedish conservation forests. Journal of Biogeography, 35, 1062–1076. doi:10.1111/j. 1365-2699.2008.01882.x. Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

Environmental and spatial components affect snail community composition Hajek, M., Rolecek, J., Cottenie, K., Kintrova, K., Horsak, M., Poulıckova, A., Hajkova, P., Frankova, M. & Dıte, D. (2011) Environmental and spatial controls of biotic assemblages in a discrete semi-terrestrial habitat: comparison of organisms with different dispersal abilities sampled in the same plots. Journal of Biogeography, 38, 1683–1693. doi:10. 1111/j.1365-2699.2011.02503.x. Kawakami, K., Wada, S. & Chiba, S. (2008) Possible dispersal of land snails by birds. Ornithological Science, 7, 167–171. doi:10.2326/1347-0558-7.2.167. Kerney, M.P. & Cameron, R.A.D. (1979) A field guide to the land snails of Britain and North-West Europe (Collins Field Guide). HarperCollins Publisher, Hong-Kong. Krab, E.J., Oorsprong, H., Berg, M.P. & Cornelissen, J.H.C. (2010) Turning northern peatlands upside down: disentangling microclimate and substrate quality effects on vertical distribution of Collembola. Functional Ecology, 24, 1362– 1369. doi:10.1111/j.1365-2435.2010.01754.x. Labaune, C. & Magnin, F. (2001) Land snail communities in Mediterranean upland grasslands: the relative importance of four sets of environmental and spatial variables. Journal of Molluscan Studies, 67, 463–474. doi:10.1093/mollus/67.4.463. Lavorel, S. & Garnier, E. (2002) Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Functional Ecology, 16, 545–556. doi:10.1046/j.1365-2435.2002.00664.x. Legendre, P., Lapointe, F.-J. & Casgrain, P. (1994) Modelling brain evolution from behavior – a permutational regression approach. Evolution, 48, 1487–1499. doi:10.2307/ 2410243. Lichstein, J.W. (2007) Multiple regression on distance matrices: a multivariate spatial analysis tool. Plant Ecology, 188, 117–131. doi:10.1007/s11258-006-9126-3. Lindo, Z. & Winchester, N.N. (2009) Spatial and environmental factors contributing to patterns in arboreal and terrestrial oribatid mite diversity across spatial scales. Oecologia, 160, 817–825. doi:10.1007/s00442-0091348-3. Magnin, F., Tatoni, T., Roche, P. & Baudry, J. (1995) Gastropod communities, vegetation dynamics and landscape changes along an old-field succession in Provence, France. Landscape and Urban Planning, 31, 249–257. doi:10.1016/ 0169-2046(94)01049-E. Makkonen, M., Berg, M.P., van Hal, J.R., Callaghan, T.V., Press, M.C. & Aerts, R. (2011) Traits explain the responses of a sub-arctic Collembola community to climate manipulation. Soil Biology & Biochemistry, 43, 377–384. doi:10. 1016/j.soilbio.2010.11.004. Martin, K. & Sommer, M. (2004) Relationships between land snail assemblage patterns and soil properties in temperatehumid forest ecosystems. Journal of Biogeography, 31, 531– 545. doi:10.1046/j.1365-2699.2003.01005.x. Martins da Silva, P., Berg, M.P., Serrano, A.R.M., Dubs, F. & Sousa, J.P. (2012) Environmental factors at different spatial scales governing soil fauna community patterns in

Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd

fragmented forests. Landscape Ecology, 27, 1337–1349. doi:10.1007/s10980-012-9788-2. McGill, B.J., Enquist, B.J., Weiher, E. & Westoby, M. (2006) Rebuilding community ecology from functional traits. Trends in Ecology and Evolution, 21, 178–185. doi:10.1016/ j.tree.2006.02.002. Messier, J., McGill, B.J. & Lechowicz, M.J. (2010) How do traits vary across ecological scales? A case for trait-based ecology. Ecology Letters, 13, 838–848. doi:10.1111/j.14610248.2010.01476.x. Metzger, M.J., Bunce, R.G.H., Jongman, R.H.G., M€ ucher, C.A. & Watkins, J.W. (2005) A climatic stratification of the environment of Europe. Global Ecology and Biogeography, 14, 549–563. doi:10.1111/j.1466-822X.2005.00190.x. Moretti, M., Dias, A.T.C., de Bello, F., Altermatt, F., Chown, S.L., Azcarate, F.M., Bell, J.R., Fournier, B., Hedde, M., € Hortal, J., Ibanez, S., Ockninger, E., Sousa, J.P., Ellers, J. & Berg, M.P. (2016) Handbook of protocols for standardized measurement of terrestrial invertebrate functional traits. Functional Ecology, online, doi:10.1111/1365-2435.12776. Nekola, J.C. (2003) Large-scale terrestrial gastropod community composition patterns in the Great Lakes region of North America. Diversity and Distributions, 9, 55–71. doi:10.1046/j.1472-4642.2003.00165.x. Nekola, J.C. (2014) North American terrestrial gastropods through each end of a spyglass. Journal of Molluscan Studies, 80, 238–248. doi:10.1093/mollus/eyu028. Nekola, J.C., Coles, B.F. & Bergthorsson, U. (2009) Evolutionary pattern and process within the Vertigo gouldii (Mollusca: Pulmonata, Pupillidae) group of minute North American land snails. Molecular Phylogenetics and Evolution, 53, 1010–1024. doi:10.1016/j.ympev.2009.09.012. Ondina, P. & Mato, S. (2001) Influence of vegetation type on the constitution of terrestrial gastropod communities in northwest Spain. Veliger, 44, 8–19. doi:10.1016/j.apsoil. 2003.10.008. Panagos, P., Van Liedekerke, M., Jones, A. & Montanarella, L. (2012) European Soil Data Centre: response to European policy support and public data requirements. Land Use Policy, 29, 329–338. doi:10.1016/j.landusepol.2011.07. 003. Peres-Neto, P.R., Legendre, P., Dray, S. & Borcard, D. (2006) Variation partitioning of species data matrices: estimation and comparison of fractions. Ecology, 87, 2614–2625. doi:10.1890/0012-9658(2006) 87[2614:vposdm]2.0.co;2. Ports, M.A. (1996) Habitat affinities and distributions of land gastropods from the Ruby Mountains and East Humboldt Range of northeastern Nevada. Veliger, 39, 335–341. R Core Team (2014) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available at: http://www.R-project.org/ [Accessed 9 September 2016]. Rees, W.J. (1965) The aerial dispersal of Mollusca. Journal of Molluscan Studies, 36, 269–282.

1371

T. Astor et al. Schilthuizen, M. & Lombaerts, M. (1994) Population structure and levels of gene flow in the mediterranean land snail Albinaria corrugata (Pulmonata, Clausiliidae). Evolution, 48, 577–586. Shannon, C.E. (1948) A mathematical theory of communication. Bell System Technical Journal, 27, 379–423. doi:10. 1002/j.1538-7305.1948.tb01338.x. Suding, K.N., Lavorel, S., Chapin, F.S., Cornelissen, J.H.C., Dıaz, S., Garnier, E., Goldberg, D., Hooper, D.U., Jackson, S.T. & Navas, M.L. (2008). Scaling environmental change through the community-level: a trait-based response-andeffect framework for plants. Global Change Biology, 14, 1125–1140. doi:10.1111/j.1469-8137.2008.02650.x. Vagvolgyi, J. (1975) Body size, aerial dispersal, and origin of Pacific land snail fauna. Systematic Zoology, 24, 465–488. doi:10.2307/2412906. Van Es, J. & Boag, D.A. (1981) Terrestrial molluscs of central Alberta. Canadian Field-Naturalist, 95, 76–79. Viketoft, M. (2013) Determinants of small-scale spatial patterns: importansce of space, plants and abiotics for soil nematodes. Soil Biology and Biochemistry, 62, 92–98. doi:10.1016/j.soilbio.2013.03.012. 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, 882–892. doi:10.1111/j.2007. 0030-1299.15559.x. von Proschwitz, T. (1996) Survey of terrestrial fauna. National atlas of Sweden: geography of plants and animals. New knowledge (ed. by L. Gustafsson and I. Ahlen), pp. 150–151. Bra B€ ocker, H€ ogan€as. Walden, H.W. (1965) Terrestrial faunistic studies in Sweden. Proceedings of the 1st European Malacological Congress, pp. 95–109. London. Walden, H.W. (1986) The 1921-1981 survey of the distribution and ecology of land molluscs in southern and central Sweden. Proceedings of the 8th International Malacological Congress, pp. 329–336. Budapest. W€areborn, I. (1970) Environmental factors influencing the distribution of land molluscs of an oligotrophic area in southern Sweden. Oikos, 21, 285–291. doi:10.2307/3543685.

1372

Widenfalk, L.A., Bengtsson, J., Berggren,  A., Zwiggelaar, K., Spijkman, E., Huyer-Brugman, F. & Berg, M.P. (2015) Spatially structured environmental filtering of collembolan traits in late successional salt marsh vegetation. Oecologia, 179, 537–549. doi: 10.1007/s00442-015-3345-z. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Trait calculations and habitat description. Appendix S2 Potential biases. Appendix S3 Detailed variation partitioning results. DATA ACCESSIBILITY The species observation data are available after registration at Artportalen (https://www.artportalen.se/), a Swedish Species Observation System provided by the Swedish Species Information Centre (ArtDatabanken).

BIOSKETCH Tina Astor is broadly interested in trait-based approaches to explain community assembly, species’ responses to their biotic and abiotic environment and their effects on ecosystem functioning. Author contributions: T.A., M.P.B., J.S. and J.B. developed the idea, discussed the concept and analyses and contributed to the writing; T.v.P. collected the data and helped preparing the manuscript; T.A. analysed the data; T.A. led the writing.

Editor: Dr Richard Ladle

Journal of Biogeography 44, 1362–1372 ª 2017 John Wiley & Sons Ltd