Decomposing dark diversity affinities of species and sites using Bayesian method: What accounts for absences of species at suitable sites?

Abstract

Abstract Locally observed biodiversity always consists of only a fraction of its site‐specific species pool. Why some suitable species are absent, shaping dark diversity of that site, is a basic yet increasingly crucial question in the face of global biodiversity degradation. The ultimate processes underlying dark diversity associate with either dispersal or persistence limitations, or both. These two limitations in turn link to several characteristics of individual species and sites, making it challenging to detect the exact factors contributing to dark diversity in a particular metacommunity. Here, we propose a metric, dark diversity affinity (DDA), which measures the tendencies of individual species to be absent from suitable sites and of individual sites to miss suitable species. We developed a Bayesian model interrelating four types of datasets: metacommunity matrix of species presences in sites, species‐sites suitability matrix, species functional traits and site characteristics. In the model, DDA operates as an adjustment bridging the disparity between site‐specific suitability and observed presence/absence of each species at each site. Furthermore, DDA can be related to individual properties of species and sites through logistic regression sub‐models. We demonstrated our framework using nine empirical datasets of vertebrate, invertebrate and vascular plant metacommunities. We show the decomposed roles of species traits and site characteristics in defining DDA and, therefore, dark diversity in metacommunities. In the empirical datasets, various functional traits, which related to morphology, reproduction, dispersal ability, population attributes, resource specificity and life history, significantly affected species‐level DDA, while site characteristics regarding habitat types and attributes, resource availability, pollution, and edaphic and water conditions influenced DDA at the site level. Our framework provides a concept and methodological toolbox that allows identification of the processes underlying dark diversity and advances both the theory of community ecology and biodiversity conservation. Conservation actions can be more successful by knowing whether species loss in a particular metacommunity is associated to some species traits or site characteristics and what their relative contributions are.