Scale dependence of vegetation-environment relationships: a meta-analysis of multivariate data

Questions How does spatial scale (extent and grain) influence the relative importance of different environmental factors as determinants of plant community composition? Are there general scale thresholds that mark the transition from primarily edaphic to primarily climatic control of plant communities? Location Global. Methods We surveyed the empirical literature and identified 89 analyses from 63 published studies that analysed vegetation–environment relationships involving at least two categories of predictor variables (edaphic, climatic, topographic, biotic, spatial or disturbance-related). For each analysis, we identified the primary predictor variable (i.e. the variable that explained the most variation in community composition) and the relative effect size of the best predictor variable from each category. We defined ‘primacy’ as the proportion of times a variable category was primary when it was measured, and analysed primacy and the relative effect size of each category as a function of spatial extent and grain. We also analysed the subset of studies that measured both edaphic and climatic variables to identify spatial extent and grain thresholds for the primacy of these factors. We surveyed the empirical literature and identified 89 analyses from 63 published studies that analysed vegetation–environment relationships involving at least two categories of predictor variables (edaphic, climatic, topographic, biotic, spatial or disturbance-related). For each analysis, we identified the primary predictor variable (i.e. the variable that explained the most variation in community composition) and the relative effect size of the best predictor variable from each category. We defined ‘primacy’ as the proportion of times a variable category was primary when it was measured, and analysed primacy and the relative effect size of each category as a function of spatial extent and grain. We also analysed the subset of studies that measured both edaphic and climatic variables to identify spatial extent and grain thresholds for the primacy of these factors. Results Edaphic variables had the highest primacy in the overall data set and at fine grain sizes (