Modelling and projecting the response of local assemblage composition to land use change across Colombia

Abstract Aim Understanding the impact of land use change within assemblages is fundamental to mitigation policies at local and regional scale. Here, we aim to quantify how site‐level terrestrial assemblages are responding to land use change in Colombia a mega‐diverse country and to project future biodiversity under different scenarios of land use change associated with climate change policies. Location Colombia (northern South America). Methods We collated original biodiversity data from 17 publications (285 sites) that examined how human impact affects terrestrial biodiversity in Colombia. From each site we estimated compositional intactness (i.e. compositional similarity to undisturbed sites). We fitted generalized linear mixed‐effects models to estimate how these measures of local biodiversity vary across land use habitats. Using space‐for‐time substitution, we applied our estimates to hindcast biodiversity changes since 1500 and project future changes under climate change policies of the four representative concentration pathways ( RCP s). Results Assemblages in urban, cropland and pasture sites were compositionally very different from those in primary vegetation. We infer that average compositional intactness has been reduced by 18% across Colombia to date, with strong regional variation. The best RCP scenario for future biodiversity is GCAM ‐ RCP 4.5, a path that favours the expansion of secondary forests under a strong carbon market; while the worst is MESSAGE ‐ RCP 8.5, ‘the business‐as‐usual’ scenario. Main conclusions Land use change has driven an increasing change in the composition of ecological assemblages in Colombia. By 2095, the implementation of carbon markets policy of climate change from GCAM ‐ RCP 4.5 could mitigate these changes in community composition. In contrast, the business‐as‐usual scenario MESSAGE ‐ RCP 8.5 predicts a steep community change placing the quality of ecosystems at risk.