How do butterflies define ecosystems? A comparison of ecological regionalization schemes

Ecological regionalizations, such as ecoregions or environmental clusters, are often used as coarse filters for conservation. To be effective biodiversity surrogates, regionalizations should contain distinct species assemblages. This condition is not frequently evaluated and regionalizations are rarely assessed comparatively. We used a national dataset of Canadian butterfly collections to evaluate four regionalizations (ecoregions, land cover and productivity regime classifications, and a spatial grid) at two thematic resolutions using analysis of similarity (ANOSIM) and species indicator values. Overall, the spatially constrained schemes (ecoregions and grids) best captured patterns of butterfly community composition and species affinities, indicating that butterfly communities are strongly structured by space at the continent scale. In contrast, when comparing regions only within spatial or environmental neighbourhoods (i.e., comparing between regions that are adjacent along geographic or environmental gradients), all regionalizations performed similarly. Adjacency in environmental space is thus as important as physical adjacency at determining community dissimilarity. Productivity regimes and land cover will be useful biodiversity surrogates when considered in conjunction with space or within a spatially constrained area. This finding was confirmed with two ecoregional case studies (of the Algonquin-Lake Nipissing and Thompson–Okanagan Plateau ecoregions), which also revealed that the relative performance of regionalizations depends upon the context of the study area. We conclude that including species data can improve the efficiency of environmental surrogates for systematic conservation planning.