Models of Joint Distribution of Species on the Example of Benthic Communities from Small Rivers of the Volga Basin

摘要

Abstract— The theoretical and practical aspects of building of joint species-distribution models, a modern tool for the analysis of ecological communities, are considered. It is shown that it is inappropriate to use the MaxEnt method or other methods based on the concept of “pseudo-absence” points when the observational data are quantitative indicators of the population density (in particular, the abundance of species in hydrobiological studies). Contemporary multidimensional models of the joint distribution of communities should include a set of parameters that assess the impact of the following groups of fixed and random factors on the species occurrence: (a) the covariates and categorical variables describing the environmental conditions and characteristics of biotopes, (b) the main indicators characterizing each species and the phylogenetic structure of communities, (c) the functions of spatial autocorrelation of the data at observation points, (d) the residual (i.e., not caused by external factors) associativity of species. Analysis of the published data and practical examples of implementation showed that the mentioned requirements, in general, are satisfied by the methodological platform and the R package Hierarchical Modeling of Species Communities (HMSC). They form the basis for the construction of multidimensional hierarchical generalized linear models with mixed parameters estimated by Bayesian procedure. The main concepts and blocks of the HMSC platform are described, and the results of models based on the authors’ data and long-term hydrobiological studies of benthic communities in 132 small and medium-sized rivers in the middle and lower Volga basin are discussed. The parameters of a set of one-dimensional candidate models for the abundance distribution of the subfamily Prodiamesinae (Diptera, Chironomidae) are analyzed, and a map of its forecast range within the region is constructed. To illustrate the multidimensional case, a model of the joint spatial distribution of 31 species of chironomids is constructed and its coefficients are analyzed. A residual correlation graph of statistically significant interspecies interactions has been built. It is concluded that the HMSC method and software package can be effectively used to solve fundamental problems of communities’ ecology: the ways in which the areas of individual populations, the structure of their communities, and the nature of interspecific interactions depend the on environmental conditions and methods to predict future trends of these processes in response to global changes.