Land-use change dominates amphibian biodiversity loss, with antagonistic interactions with climate change in the karst region of Guizhou, China.

Karst ecosystems support exceptionally high amphibian diversity but are increasingly threatened by the combined pressures of climate change and land-use change. To date, the independent and interactive effects of these two drivers remain poorly quantified in these highly heterogeneous karst landscapes. Focusing on Guizhou Province, a global karst biodiversity hotspot in southwestern China, this study quantifies the relative contributions of climate and land-use change to amphibian richness loss and disentangles their additive versus non-additive interaction patterns. Based on 2636 occurrence records of 112 species compiled from a decade of systematic field surveys and verified literature records, we constructed ensemble species distribution models (biomod2) and phylogenetically constrained ENphylo models. We projected species distributions to the 2070s under low- and strong-emission scenarios, contrasting land-use-only, climate-only, and coupled change scenarios. All scenarios consistently projected net declines in species richness, with losses ranging from 1.15 to 2.88 species per km2. Land-use change alone was associated with larger richness declines than climate change alone, while combined climate-land-use changes resulted in the smallest net losses. The smaller net loss under coupled scenarios reflected antagonistic non-additive patterns, which accounted for 77.7-79.7% of grid cells for species loss; in contrast, species gains were predominantly additive. Our findings highlight that regulating land-use intensity represents a more immediate and effective conservation lever than standalone climate mitigation in karst regions. We recommend that regional land-use planning prioritize habitat connectivity and constrain agricultural expansion, and that adaptive management strategies account for the spatial overlap of both stressors and the higher uncertainty associated with land-use projections to ensure robustness across development pathways.