Biodiversity model projections are becoming increasingly important, helping to improve understanding of ecological systems, highlighting threats, and informing investment in conservation actions. Most commonly applied are ‘bottom-up’ models focussed on individual species (e.g. SDMs), which can perform well in predicting individual species distributions, but are less accurate in predicting community-level diversity and composition. In contrast, ‘top-down’ community-level approaches more accurately predict species richness, compositional turnover, or other community metrics, but they do not provide information at the species-level, where conservation interest and investment often focusses. Recently however, a new generation of modelling techniques have emerged, that harness data from both the species- and community-level to predict the composition of assemblages over space and time.
Here we examine a semi-mechanistic modelling framework that harnesses data and makes predictions at both the species- and community-level (‘DynamicFOAM’ linked with the ‘M-SET’ metacommunity model). Using case-studies from a number of diverse taxonomic groups across large regions, we assess how well these linked modelling techniques predict current patterns in species assemblages, and their capacity to incorporate a range of key mechanisms in predicting changes in species assemblages over time (dispersal, community assembly, trophic interactions, genetic adaptation).
Results/Conclusions
Our analyses highlight some of the key challenges for emerging modelling techniques in making accurate predictions of both the distributions of individual species and patterns in community diversity. The quantitative model assessments we present also demonstrate how incorporating ecological processes into diversity models can substantially alter predicted changes in species assemblages over time under global change, and alter our understanding of how beneficial different management actions are likely to be in retaining biodiversity.
While there remains a number of important methodological challenges in the ongoing development of modelling techniques that simultaneously consider both the species- and community-level, we highlight the potential benefits in continued development of these approaches. We also demonstrate that although ecological interactions and processes can be complex and difficult to model, accounting for such complexities is vital if we are to improve our capacity to understand and respond to the influence of global change on biodiversity.