Stress‐Testing Model‐Derived Biodiversity Indicators Under Different Assumptions of Land‐Use Change Impacts

ABSTRACT Aim There is an urgent need to identify sustainable development pathways promoting biodiversity recovery. However, biodiversity is multi‐dimensional, with individual facets only capturing part of the whole. Biodiversity indicators summarising this complexity vary in their sensitivity to different forms of biodiversity change, but the extent to which indicator choice affects estimated responses to drivers such as land‐use change is currently unclear. We evaluate the sensitivity and predictability of 31 indicators across several scenarios of ecological change. Location Simulated, global. Methods We simulate local species communities under a range of initial community properties and 11 scenarios representing varying magnitudes of land‐use change impacts on abundance, richness and evenness. Across these simulations, we model a suite of within‐ and between‐site biodiversity indicators, assessing their capacity to detect and predict land‐use‐driven ecological change. Results We show that model‐derived indicator responses form clusters representing distinct facets of biodiversity (e.g., abundance, richness), emphasising the need to use multiple indicators to capture the various dimensions of biodiversity change. Further, most indicators only perform strongly, in terms of response size and predictability, in a few scenarios of ecological change. Our simulation‐based assessment reveals important patterns in indicator responsiveness and predictability, that are likely applicable for real‐world ecological data. Main Conclusions Our stress tests suggest that at least three indicators, total abundance, geometric mean abundance and community dissimilarity (Canberra or Clark distance), should be used to provide a comprehensive perspective of land‐use impacts. Species richness and evenness/composition (e.g., Hill‐Simpson, Hill‐Shannon) offer additional informative context, but are less sensitive to ecological change and less conservation relevant, respectively. Using a collection of complementary indicators to reliably detect and predict anthropogenically driven ecological change enables more effective conservation monitoring and action.