Beta-diversity of ectoparasites at two spatial scales: nested hierarchy,geography and habitat type

Beta-diversity of biological communities can be decomposed into (a) dissimilarity of communities among units of finer scale within units of broader scale and (b) dissimilarity of communities among units of broader scale. We investigated compositional, phylogenetic/taxonomic and functional beta-diversity of compound communities of fleas and gamasid mites parasitic on small Palearctic mammals in a nested hierarchy at two spatial scales: (a) continental scale (across the Palearctic) and (b) regional scale (across sites within Slovakia). At each scale, we analyzed beta-diversity among smaller units within larger units and among larger units with partitioning based on either geography or ecology. We asked (a) whether compositional, phylogenetic/taxonomic and functional dissimilarities of flea and mite assemblages are scale dependent; (b) how geographical (partitioning of sites according to geographic position) or ecological (partitioning of sites according to habitat type) characteristics affect phylogenetic/taxonomic and functional components of dissimilarity of ectoparasite assemblages and (c) whether assemblages of fleas and gamasid mites differ in their degree of dissimilarity, all else being equal. We found that compositional, phylogenetic/taxonomic, or functional beta-diversity was greater on a continental rather than a regional scale. Compositional and phylogenetic/taxonomic components of beta-diversity were greater among larger units than among smaller units within larger units, whereas functional beta-diversity did not exhibit any consistent trend regarding site partitioning. Geographic partitioning resulted in higher values of beta-diversity of ectoparasites than ecological partitioning. Compositional and phylogenetic components of beta-diversity were higher in fleas than mites but the opposite was true for functional beta-diversity in some, but not all, traits.     

Modified TWINSPAN classification in which the hierarchy respects cluster heterogeneity

Aim: To propose a modification of the TWINSPAN algorithm that enables production of divisive classifications that better respect the structure of the data. Methods: The proposed modification combines the classical TWINSPAN algorithm with analysis of heterogeneity of the clusters prior to each division. Four different heterogeneity measures are involved: Whittaker's beta, total inertia, average Sørensen dissimilarity and average Jaccard dissimilarity. Their performance was evaluated using empirical vegetation datasets with different numbers of plots and different levels of heterogeneity. Results: While the classical TWINSPAN algorithm divides each cluster coming from the previous division step, the modified algorithm divides only the most heterogeneous cluster in each step. The four tested heterogeneity measures may produce identical or very similar results. However, average Jaccard and Sørensen dissimilarities may reach extreme values in clusters of small size and may produce classifications with a highly unbalanced cluster size. Conclusions: The proposed modification does not alter the logic of the TWINSPAN classification, but it may change the hierarchy of divisions in the final classification. Thus, unsubstantiated divisions of homogeneous clusters are prevented, and classifications with any number of terminal clusters can be created, which increases the flexibility of TWINSPAN.     

A guide through a family of phylogenetic dissimilarity measures among sites

Ecological studies have now gone beyond measures of species turnover towards measures of phylogenetic and functional dissimilarity. This change of perspective has a main objective: disentangling the processes that drive species distributions from local to broad scales. A fundamental difference between phylogenetic and functional analyses is that phylogeny is intrinsically dependent on a tree‐like structure whereas functional data can, most of time, only be forced to adhere a tree structure, not without some loss of information. When the branches of a phylogenetic tree have lengths, then each evolutionary unit on these branches can be considered as a basic entity on which dissimilarities among sites should be measured. Several of the recent measures of phylogenetic dissimilarities among sites thus are traditional dissimilarity indices where species are replaced by evolutionary units. The resulting indices were named PD‐dissimilarity indices, in reference to early work on the phylogenetic diversity (PD) measure. Here I review and compare indices and ordination approaches that, although first developed to analyse the differences in the species compositions of sites, can be adapted to describe PD‐dissimilarities among sites. Using simulations of species distributions along environmental gradients, I compare indices, associated with permutation tests and null models, in their ability to reveal existing phylogenetic patterns along the gradients. As an illustration, I show that the amount of bat PD‐dissimilarities along a disturbance gradient in Selva Lacandona of Chiapas, Mexico is dependent on whether species' abundance is considered, and on the PD‐dissimilarity index used. Overall, the family of PD‐dissimilarity indices has a critical potential for future analyses of phylogenetic diversity as it benefits from decades of research on the measure of species dissimilarity. I provide clues to help to choose among many potential indices, identifying which indices satisfy minimal basic properties, and analysing their sensitivity to abundance, size, diversity and joint absences.     

Modelling the spatial distribution of beta diversity in Australian subtropical rainforest

Investigation of the spatial distribution of biodiversity among communities or across habitats (beta diversity) is often hampered by a scarcity of biological survey data. This is particularly the case in communities of high floristic diversity, such as the subtropical rainforests of eastern Australia. In contrast, there is excellent spatial coverage of environmental data for this region, such as geology, elevation and climate data. Generalized dissimilarity modelling was used in this study to combine biological survey data and environmental data grids for the investigation and prediction of floristic turnover among vegetation communities at a regional scale. Generalized dissimilarity modelling identified four environmental predictors of floristic turnover in the study region, all of which are linked with moisture stress: radiation of the driest quarter, precipitation of the driest period of the year, slope and aspect. Ten land classes representing largely homogeneous floristics and environment were identified and mapped for the region, allowing significantly greater discrimination than currently available mapping for this region. With increases in evapotranspiration and moisture stress predicted as a result of climate change, these results may allow future floristic shifts to be assessed in relation to regional‐scale gradients in floristic turnover.     

Phylogenetic generalised dissimilarity modelling: a new approach to analysing and predicting spatial turnover in the phylogenetic composition of communities

Compared to species turnover, patterns of phylogenetic turnover provide extra information about the spatial structure of biodiversity, for example providing more informative comparisons between the biota of sites which share no species. To harness this information for broad‐scale spatial analysis, we present phylo‐GDM, a technique for interpolating the spatial structure of phylogenetic turnover between sampled locations in relation to environment, based on generalised dissimilarity modelling (GDM). Using a database of over 150 000 location records for 114 myobatrachid frog species in Australia, linked to a species‐level phylogeny inferred from 2467 base pairs of mitochondrial DNA, we calculated species and phylogenetic turnover between pairs of sites. We show how phylogenetic turnover extended the range of informative comparison of compositional turnover to more biologically and environmentally dissimilar sites. We generated GDM models which predict species and phylogenetic turnover across Australia, and tested the fit of models for different ages within the phylogeny to find the phylogenetic tree depth at which the relationship to current day environment is greatest. We also incorporated explanatory variables based on biogeographic patterns, to represent broad‐scale turnover resulting from divergent evolutionary histories. We found that while the predictive power of our models was lower for full phylogenetic turnover than for species turnover, models based on the more recent components of the phylogeny (closer to the tips) outperformed species models and full phylogenetic models. Phylo‐GDM has considerable potential as a method for incorporating phylogenetic relationships into biodiversity analyses in ways not previously possible. Because phylogenies do not require named taxa, phylo‐GDM may also provide a means of including lineages with poorly resolved taxonomy (e.g. from metagenomic sequencing) into biodiversity planning and phylogeographic analysis.     

Spread of common species results in local‐scale floristic homogenization in grassland of Switzerland

Aim We assess changes in plant species richness and changes in species dissimilarity at local scale in Swiss grassland between the time periods 2001–2004 and 2006–2009. Further, we provide an ecological interpretation of the observed taxonomic homogenization of vascular plants. Location Switzerland. Methods Changes in species richness and changes in Simpson dissimilarity index of vascular plants in grassland (meadows and pastures) were examined. The analyses were based on species lists recorded on 339 10‐m² sample plots from a systematic sample covering the entire Switzerland. Each sample plot had been surveyed once in 2001–2004 and once in 2006–2009 with 5 years between the first and the second survey. Changes in species dissimilarity were interpreted by comparing the relative contribution of several indicator species groups. Results Mean species richness of vascular plants in grassland increased during the study period. In contrast, species dissimilarity of plants decreased, suggesting local‐scale floristic homogenization of grassland in Switzerland. It was mostly because of the spread of common species, namely the species that are tolerant to high nutrient levels, the species of low conservation value and the species adapted to moderate temperature levels that led to taxonomic homogenization. Target species for conservation did only marginally affect taxonomic homogenization. In contrast to the predictions from studies of taxonomic homogenization on larger scales, the taxonomic homogenization of grassland at local scale was not explained by the spread of neophytic species. Main conclusions The biotic diversity of grassland in Switzerland changed considerably between 2001–2004 and 2006–2009. The observed taxonomic homogenization was merely because of the spread of common species. Local‐scale changes in land use regimes implemented by agri‐environmental schemes and other conservation efforts on parts of the entire grassland area were, apparently, not enough to prevent the total grassland from recent taxonomic homogenization.     

Functional dissimilarity across trophic levels as a driver of soil processes in a Mediterranean decomposer system exposed to two moisture levels

The role of biodiversity for soil processes remains poorly understood. Existing evidence suggests that functional diversity rather than species richness is relevant for soil functioning. However, the importance of functional diversity has rarely been assessed simultaneously at more than one trophic level, critically limiting the prediction of consequences of biodiversity loss for soil functioning. In a laboratory microcosm experiment, we tested the hypothesis that increasing functional dissimilarity of both litter‐feeding soil fauna and litter mixtures interactively affects the rates of five different soil processes related to litter decomposition. We created trait‐based functional dissimilarity gradients using five assemblages of two detritivore species and five mixtures of two plant litter species commonly found in Mediterranean shrubland ecosystems of southern France. With increasing drought periods predicted for Mediterranean ecosystems in the future, we additionally included two different watering frequencies to evaluate the impact of drought on soil processes and how drought interacts with functional dissimilarity. The different fauna assemblages and litter mixtures showed strong effects on litter mass loss, soil organic carbon and nitrogen leaching, as well as on soil microbial activities. Up to 20% of the variation in response variables was explained by functional dissimilarity, suggesting an ecologically relevant impact of functional diversity on soil process rates. Detritivore functional dissimilarity tended to have stronger effects when combined with increasingly dissimilar litter mixtures, suggesting that trait dissimilarity interacts across trophic levels. Drought affected several soil processes but did not modify the relationships between functional dissimilarity and process rates. Our results indicate that trait diversity of detritivore assemblages and litter mixtures is an important predictor of soil process rates. The common and easily measurable traits used in our study suggest straightforward application across different types of ecosystems and environmental conditions.     

Classification,composition, richness and diversity of British hedgerows

Abstract. A floristic classification of British hedgerows is constructed. Hedges (the woody, structural element) and hedge‐bottoms (the mainly herbaceous ground flora) were analysed separately because different natural and management processes were expected to affect them. The 11 hedge classes varied in size from five to over 500 plots. Over two‐thirds of all hedges were dominated by Crataegus spp. or Primus spinosa. Hedge‐bottoms were more evenly grouped into four classes whose species composition could be related to broad land use/habitat types: intensive arable, rotational/mixed farming, managed grasslands, and woodland. The woodland class was only 12% of the total, and species characteristic of true woodland habitats were uncommon even within that class, so our data do not support the hypothesis of hedgerows acting as‘woodland corridors’for plants. Richness/diversity scores differed markedly between different hedge and hedge‐bottom classes. The richest hedgerows, both in species and inferred habitats, were among the rarer types, while the most common classes (both hedge and hedge‐bottom) were consistently the poorest. Comparison of species composition and diversity in hedges and hedge‐bottoms, and associations between them, all supported the hypothesis that the relative importance of historical, management and ecological effects differed between the two parts of the hedgerow; hedges and hedge‐bottoms should be treated ecologically as largely independent units. The classifications and richness/diversity scores enable easy preliminary assessment of the conservation value of a hedgerow. They also enable individual hedgerows to be placed in a wider (national) context.     

Determinants of Plant Community Assembly in a Mosaic of Landscape Units in Central Amazonia: Ecological and Phylogenetic Perspectives

The Amazon harbours one of the richest ecosystems on Earth. Such diversity is likely to be promoted by plant specialization, associated with the occurrence of a mosaic of landscape units. Here, we integrate ecological and phylogenetic data at different spatial scales to assess the importance of habitat specialization in driving compositional and phylogenetic variation across the Amazonian forest. To do so, we evaluated patterns of floristic dissimilarity and phylogenetic turnover, habitat association and phylogenetic structure in three different landscape units occurring in terra firme (Hilly and Terrace) and flooded forests (Igapó). We established two 1-ha tree plots in each of these landscape units at the Caparú Biological Station, SW Colombia, and measured edaphic, topographic and light variables. At large spatial scales, terra firme forests exhibited higher levels of species diversity and phylodiversity than flooded forests. These two types of forests showed conspicuous differences in species and phylogenetic composition, suggesting that environmental sorting due to flood is important, and can go beyond the species level. At a local level, landscape units showed floristic divergence, driven both by geographical distance and by edaphic specialization. In terms of phylogenetic structure, Igapó forests showed phylogenetic clustering, whereas Hilly and Terrace forests showed phylogenetic evenness. Within plots, however, local communities did not show any particular trend. Overall, our findings suggest that flooded forests, characterized by stressful environments, impose limits to species occurrence, whereas terra firme forests, more environmentally heterogeneous, are likely to provide a wider range of ecological conditions and therefore to bear higher diversity. Thus, Amazonia should be considered as a mosaic of landscape units, where the strength of habitat association depends upon their environmental properties.     

Global patterns of terrestrial assemblage turnover within and among land uses

Land use has large effects on the diversity of ecological assemblages. Differences among land uses in the diversity of local assemblages (alpha diversity) have been quantified at a global scale. Effects on the turnover of species composition between locations (beta diversity) are less clear, with previous studies focusing on particular regions or groups of species. Using a global database on the composition of ecological assemblages in different land uses, we test for differences in the between‐site turnover of species composition, within and among land‐use types. Overall, we show a strong impact of land use on assemblage composition. While we find that compositional turnover within land uses does not differ strongly among land uses, human land uses and secondary vegetation in an early stage of recovery are poor at retaining the species that characterise primary vegetation. The dissimilarity of assemblages in human‐impacted habitats compared with primary vegetation was more pronounced in the tropical than temperate realm. An exploratory analysis suggests that this geographic difference might be caused primarily by differences in climate seasonality and in the numbers of species sampled. Taken together the results suggest that, while small‐scale beta diversity within land uses is not strongly impacted by land‐use type, compositional turnover between land uses is substantial. Therefore, land‐use change will lead to profound changes in the structure of ecological assemblages.     

Primary and secondary vegetation patches as contributors to floristic diversity in a tropical deciduous forest landscape

The floristic diversity of Mexican tropical deciduous forests (TDF) is of critical importance given the high species richness (alpha diversity), species turnover (beta diversity), and the intense deforestation rates. Currently, most TDF landscapes are mosaics of agricultural land, secondary vegetation, and patches of relatively undisturbed primary vegetation. Here we illustrate how both primary forest remnants and secondary vegetation patches contribute to the floristic diversity of TDF in a landscape of volcanic origin in central Veracruz, Mexico. Our objectives were to assess sampling efficiency and inventory completeness, to compare mean and cumulative species richness between primary forest and secondary vegetation sites, and to analyze beta diversity between vegetation types. In an area of 12,300 m2 we recorded 105 families, 390 genera, and 682 species. Species inventories for both vegetation types were about 80% complete. Secondary vegetation is more alpha diverse than primary forest, both in terms of cumulative and mean species richness. We found a remarkably high beta diversity between vegetation types (75% of complementarity, 91.60% of mean dissimilarity). We also identified the species that contribute the most to similarity within vegetation types and to dissimilarity between vegetation types. Our results support the idea that assessing biodiversity on the landscape scale is an appropriate way to ascertain the impact of human activities. For this land mosaic, conservation of the flora would not be possible by focusing solely on primary forest remnants. We propose the implementation of a network of small conservation areas with a flexible structure, following the “archipelago reserve” model.     

Whole‐landscape modelling of compositional turnover in aquatic invertebrates informs conservation gap analysis: An example from south‐western Australia

相似文献   

Evaluation of floristic diversity in urban areas as a basis for habitat management

Questions: How can floristic diversity be evaluated in conser‐vation plans to identify sites of highest interest for biodiversity? What are the mechanisms influencing the distribution of species in human‐dominated environments? What are the best criteria to identify sites where active urban management is most likely to enhance floristic diversity? Location: The Hauts‐de‐Seine district bordering Paris, France. Methods: We described the floristic diversity in one of the most urbanized French districts through the inventory of ca. 1000 sites located in 23 habitats. We built a new index of floristic interest (IFI), integrating information on richness, indigeneity, typicality and rarity of species, to identify sites and habitats of highest interest for conservation. Finally, we explored the relationship between site IFI and land use patterns (LUP). Results: We observed a total of 626 vascular plant species. Habitats with highest IFI were typically situated in seminatural environments or environments with moderate human impact. We also showed that neighbouring (urban) structures had a significant influence on the floristic interest of sites: for example, the presence of collective dwellings around a site had a strong negative impact on IFI. Conclusions: Our approach can be used to optimize management in urban zones; we illustrate such possibilities by defining a ‘Site Potential Value’, which was then compared with the observed IFI, to identify areas (e.g. river banks) where better management could improve the district's biodiversity.     

Extended dissimilarity: a method of robust estimation of ecological distances from high beta diversity data

It is widely accepted that reliable ordination of ecological data requires a strong linear or ordinal relationship between the dissimilarity of sites, based on species composition, and the ecological distance between them. Certain dissimilarity measures, having the property that they take a fixed maximum value when sites have no species in common, have been shown to be strongly correlated with ecological distance. For ecological gradients of moderate length (moderate beta diversity), such measures, in conjunction with non-metric multidimensional scaling, will reliably yield successful ordinations. However, as beta diversity increases, more sites have no species in common, and such measures invariably under-estimate ecological distance for such sites. Thus ordinations of data with high species turnover (high beta diversity) may fail.Extended dissimilarities are defined using an iterative adaptation of flexible shortest path adjustment applied to the matrix of dissimilarities with fixed maximum values. By means of theoretical argument and simulations, this is shown to lead to far stronger correlations between the adjusted site dissimilarity and ecological distance for ecological gradients of greater length than previously considered. Hence ordinations of extended dissimilarities, by means of either metric or non-metric scaling techniques, are shown to outperform corresponding ordinations of unadjusted dissimilarities, with the difference increasing with increasing beta diversity.     

Modelling niche and neutral dynamics: on the ecological interpretation of variation partitioning results

Many studies have attempted to disentangle the effects of neutral and niche‐mediated processes on community composition using partial Mantel tests and variance partitioning on dissimilarity matrices. Recently, doubts about the reliability of these methods have emerged. Here we explore how the results are affected by three confounding factors that may affect ecological data to different degrees: spatial autocorrelation of the environmental variables, length of the compositional gradient, and sampling noise. We document that the statistical hypotheses tested in these methods may or may not correspond to the ecological hypotheses of interest. A major discrepancy emerges if a large proportion of sampling units in the analysed dataset share no species, in which case compositional dissimilarities saturate to a fixed maximum value although explanatory dissimilarities do not. With increasing dissimilarity saturation, the explanatory power of regression models decrease, which may lead to the erroneous conclusion that the ecological processes represented by the explanatory variables are not operating. A survey of recent literature suggests that there is a general lack of awareness of this problem, although it appears to affect > 10% of relevant studies. Our simulations show that if dissimilarity saturation is due to a long ecological gradient, extended dissimilarities essentially solve the problem for any degree of saturation. Using distances from a hybrid multidimensional scaling alleviates the saturation problem when the degree of saturation is < 60%. However, neither correction method can provide a solution to problems caused by insufficient sampling. How the presence of multiple explanatory gradients in combination with sampling noise affects overall analysis performance remains to be clarified.     

Relative importance of abiotic and land use factors in explaining variation in woody vegetation in a French rural landscape

Abstract. In a rural landscape, scale vegetation patterns of woody species are controlled by both abiotic and land use factors. The woody species composition in 126 sample plots was analysed and land use factors and some abiotic parameters were quantified using land register data. The relative importance of land use and abiotic factors was differentiated using a partial Canonical Correspondence Analysis (CCA); the influence of land use in neighbouring areas was explored by changing the scale of land use sampling. The woody species composition appears to be controlled equally by land use and abiotic factors. The fraction of floristic variation in tree species composition explained by land use variables was 33.2 % in 1980 and 30.8 % in 1992, while abiotic variables accounted for 31.2 %. Part of the 17 % of the floristic variance explained is related to the surrounding land use. Thus, when the land use of the sampled plots and the surrounding land use are considered simultaneously, up to 36.9 % of the species variation may be explained. Partial CCA enabled us to quantify the respective proportion of floristic variance which could be explained by land use (36.9 %), abiotic variables alone (20.2 %), shared variance (12.0 %) and unexplained variance (31.2 %). Our results indicated that a delayed effect of variation in land use on plant populations may exist. This delay may result either from population characteristics or from inadequate land use assessment. This study indicates the need for simultaneously examining land use and abiotic patterns in ecological studies, as many Mediterranean-type ecosystems have been shaped by these patterns.     

Patterns of parasite community dissimilarity: the significant role of land use and lack of distance‐decay in a bat–helminth system

Increasing community dissimilarity across geographic distance has been described for a wide variety of organisms and understanding its underlying causes is key to understanding mechanisms driving patterns of biodiversity. Both niche‐based and neutral processes may produce a distance decay relationship; however, disentangling their relative influence requires simultaneous examination of multiple potential drivers. Parasites represent a unique opportunity in which to study distance decay because community dissimilarity may depend on environmental requirements and dispersal capability of parasites as well also those of their hosts. We used big brown bats Eptesicus fuscus and their intestinal helminths to investigate: 1) independent contributions of geographic and environmental distances on dissimilarity of intestinal helminth component communities between populations of big brown bats; 2) which environmental variables best explained variation in community dissimilarity; and 3) whether similar patterns of decay with geographic or environmental distance were observed for within‐host population and within‐individual host parasite communities. We used compositional measures of community dissimilarity to examine how parasite communities may change with geographic distance and varying environmental conditions. Non‐spatial variables strongly influenced compositional parasite community dissimilarity over multiple community scales, and we observed little evidence for spatial processes such as distance decay. Environment surrounding roost sites better predicted helminth community dissimilarity than any other class of variables and landcover classes representing anthropogenic modification consistently explained variation in community structure. Our results indicate that human disturbance drives significant patterns of parasite community dissimilarity, most likely by changing the presence or abundance of intermediate hosts in an area.     

Environment versus dispersal in the assembly of western Amazonian palm communities

Aim It is a central issue in ecology and biogeography to understand what governs community assembly and the maintenance of biodiversity in tropical rain forest ecosystems. A key question is the relative importance of environmental species sorting (niche assembly) and dispersal limitation (dispersal assembly), which we investigate using a large dataset from diverse palm communities. Location Lowland rain forest, western Amazon River Basin, Peru. Methods We inventoried palm communities, registering all palm individuals and recording environmental conditions in 149 transects of 5 m × 500 m. We used ordination, Mantel tests and indicator species analysis (ISA) to assess compositional patterns, species responses to geographical location and environmental factors. Mantel tests were used to assess the relative importance of geographical distance (as a proxy for dispersal limitation) and environmental differences as possible drivers of dissimilarity in palm species composition. We repeated the Mantel tests for subsets of species that differ in traits of likely importance for habitat specialization and dispersal (height and range size). Results We found a strong relationship between compositional dissimilarity and environmental distance and a weaker but also significant relationship between compositional dissimilarity and geographical distance. Consistent with expectations, relationships with environmental and geographical distance were stronger for understorey species than for canopy species. Geographical distance had a higher correlation with compositional dissimilarity for small‐ranged species compared with large‐ranged species, whereas the opposite was true for environmental distance. The main environmental correlates were inundation and soil nutrient levels. Main conclusions The assembly of palm communities in the western Amazon appears to be driven primarily by species sorting according to hydrology and soil, but with dispersal limitation also playing an important role. The importance of environmental characteristics and geographical distance varies depending on plant height and geographical range size in agreement with functional predictions, increasing our confidence in the inferred assembly mechanisms.     

Broad Scale Distribution of Ferns and Lycophytes along Environmental Gradients in Central and Northern Amazonia,Brazil

Establishing which factors determine species distributions is of major relevance for practical applications such as conservation planning. The Amazonian lowlands exhibit considerable internal heterogeneity that is not apparent in existing vegetation maps. We used ferns as a model group to study patterns in plant species distributions and community composition at regional and landscape scales. Fern species composition and environmental data were collected in 109 plots of 250 × 2 m distributed among four sites in Brazilian Amazonia. Interplot distances varied from 1 to ca 670 km. When floristically heterogeneous datasets were analyzed, the use of an extended Sørensen dissimilarity index rather than the traditional Sørensen index improved model fit and made interpretation of the results easier. Major factors associated with species composition varied among sites, difference in cation concentration was a strong predictor of floristic dissimilarity in those sites with pronounced heterogeneity in cation concentration. Difference in clay content was the most relevant variable in sites with uniform cation concentrations. In every case, environmental differences were invariably better than geographic distances in predicting species compositional differences. Our results are consistent with the ideas that: (1) the relative predictive capacity of the explanatory variables depend on the relative lengths of the observed gradients; and (2) environmental gradients can be hierarchically structured such that gradients occur inside gradients. Therefore, site‐specific relationships among variables can mask the bigger picture and make it more difficult to unravel the factors structuring plant communities in Amazonia.