Assessing the effectiveness of remote sensing variables for coarse-scale amphibian distribution modeling: A case study of Fejervarya multistriata in China

Amphibians are highly threatened and environmentally sensitive, which necessitates timely and precise habitat distribution information for effective conservation. Remote sensing provides a promising solution by offering direct, spatially explicit, and repeatable measurements of environmental conditions for habitat modeling and monitoring. In this study, we assessed the independent predictive ability of remote sensing data in large-scale amphibian habitat modeling. We constructed an ensemble species distribution model for Fejervarya multistriata across mainland China using 11 remote sensing variables, including remotely sensed ecosystem functioning attributes (RS-EFAs) and human impact factors, along with multi-temporal occurrence records from public databases and field surveys, and compared it with a climate-based model. The remote sensing-based model demonstrated high effectiveness and reliability in simulating the ecological niche of F. multistriata, as defined under the current taxonomic guidelines. The predicted habitat suitability pattern was largely consistent with that of the climate-based model, identifying highly suitable regions in the Middle and Lower Reaches of the Yangtze River Plain, the southeastern coastal region, the Pearl River Basin, and Hainan Island. Compared to the climate-based model, the remote sensing-based model revealed smaller fragmented habitat patches and excluded less suitable areas within continuous suitable regions, providing more detailed spatial information. Key factors influencing the distribution of F. multistriata included annual LST Std Dev, land cover type, and annual mean NDWI. This study demonstrates that remote sensing data can support accurate large-scale amphibian habitat modeling and offers a framework for future dynamic habitat monitoring and conservation planning.