Frontiers of Biogeography

Ecological niche models (ENMs) can project changes in species’ distributions under climate change and thus inform conservation efforts and further our understanding of patterns of change. Predictions of species’ distribution shifts under climate change in topographically and geologically complex landscapes, such as karst landforms, should be improved by better integration of non-climate abiotic variables, such as karst geology or habitat structure, into model projections. We built ENMs for one of the limestone langurs, a group of leaf monkeys adapted to forests on the Sino-Vietnamese limestone karst landform. We collected occurrence localities for François’ leaf monkeys (Trachypithecus francoisi) and thinned them to avoid sampling bias. We included as environmental parameters a global dataset for karst geology and 19 bioclimatic variables derived from monthly temperature and precipitation at 30 arc-second resolution. ENMs including karst geology and climatic variables outperformed and differed spatially from climate-only models. Across six future-climate scenario projections, the optimal karst+climate model differed from the best climate-only model and predicted more spatial overlap with karst in the future, a contraction in total area of suitable habitat by the 2070s, and a small loss in the amount of suitable habitat in existing conservation areas. This study shows the importance of considering other abiotic factors beyond climate in projections of suitable habitat under climate change for species in complex landscapes. Because our results show that karst and climate interact to explain the distribution of a karst-adapted species, the results also suggest that, under climate change, these interactions are likely to produce altered networks of species into novel biological communities. Finally, our results support the need for conservation of limestone habitats and cross-border collaboration to maintain refuges and movement connectivity for endangered species in the face of climate change.