Downscaling Amphibian Species Richness Maps to Explore the Role of Spatial Scale in Conservation
摘要
Mapping species richness is a key goal of conservation research, but low data resolution and limited survey data make it challenging to accurately assess distribution patterns. In this study, the random forest (RF) and geographical random forest (GRF) models were used to construct a model of relationships between environmental factors and species richness, and high-resolution environmental data was used to downscale amphibian species distribution maps. The derived multi-scale species richness maps of 10 km, 5 km, and 1 km, revealed that the factors influencing the distribution of species richness and the locations of species richness hotspots vary with spatial scale. GRF outperformed GF in species richness map downscaling, with R2 above 97% and RMSE between 0.98 and 1.29. GRF analysis shows that the spatial distribution of environmental factors affecting species distribution varies greatly, and precipitation dominates the distribution of most regions. This study suggests that machine learning algorithms can be used to downscale species richness maps. The multiscale species richness distribution map demonstrates the sensitivity of species richness patterns to spatial scales, which is crucial for macro-ecological analysis and identifying priority conservation areas. This information should be taken into account in future conservation planning.