Understanding and modelling biotic interactions and their dynamics are prerequisite for predicting community and biodiversity response to climate change and, thus, for designing strategies and policies to halt the loss of biodiversity and associated ecosystem services. We propose to investigate uncertainties in biodiversity response to climate change by improving and integrating existing approaches for modelling biotic interactions in large-scale multispecies assemblages, taking into account reciprocal effects between species and their abiotic and biotic environment as well as variations in biotic interactions across space and time, and considering macroecological constraints on species assemblages. Using a hierarchical Bayesian approach, we will implement algorithms for describing biotic interactions based on simple qualitative links between species or functional groups, on quantitative interaction coefficients, and on interaction currencies, for example a common resource, that may mediate interactions. Algorithms will be benchmarked using species assemblage data on birds, butterflies and plants, mainly from Switzerland but also from other parts of Europe. Effects of different interaction mechanisms and data limitations on model performance will be tested using data from simulated communities. We will further assess the magnitude of variation in biotic associations throughout species’ environmental niches and geographic ranges. In a multi-scale approach, our multispecies interaction models will be combined with macroecological models to predict species richness and species compositions for an ensemble of climate change scenarios. By combining extensive empirical and theoretical analyses as well as expertise and datasets of researchers from four European countries, the proposed research is at the cutting edge, will provide theoretical and conceptual advancements in biodiversity science, and will help operationalizing community predictions under climate change.