Variation partitioning in double-constrained multivariate analyses: linking communities, environment, space, functional traits, and ecological niches

Constrained multivariate analysis is a common tool for linking ecological communities to environment. The follow-up is the development of the double-constrained correspondence analysis (dc-CA), integrating traits as species-related predictors. Further, methods have been proposed to integrate information on phylogenetic relationships and space variability. We expand this framework, proposing a dc-CA-based algorithm for decomposing variation in community structure and testing the simple and conditional effects of four sets of predictors: environment characteristics and space configuration as predictors related to sites, while traits and niche (dis)similarities as species-related predictors. In our approach, ecological niches differ from traits in that the latter are distinguished by and characterize the individual level, while niches are measured on the species level, and when compared, they are characteristics of communities and should be used as separate predictors. The novelties of this approach are the introduction of new niche parameters, niche dissimilarities, synthetic niche-based diversity which we related to environmental features, the development of an algorithm for the full variation decomposition and testing of the community–environment–niche–traits–space (CENTS) space by dc-CAs with and without covariates, and new types of diagrams for the results. Applying these methods to a dataset on freshwater mollusks, we learned that niche predictors may be as important as traits in explaining community structure and are not redundant, overweighting the environmental and spatial predictors. Our algorithm opens new pathways for developing integrative methods linking life, environment, and other predictors, both in theoretical and practical applications, including assessment of human impact on habitats and ecological systems. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s00442-021-05006-6.