Geospatial and Environmental Variable Dataset for Predicting Potential Habitats of Mammals in the South Korea

Recent climate change and land-use alteration are weakening the structure and connectivity of mammal habitats, leading to range contractions and population isolation. However, habitat and distribution forecasting for Korean mammals has largely been confined to specific species or local areas, leaving a lack of quantitative assessments at the national scale. This study compiled nationwide occurrence records and environmental spatial data to predict potential habitats for 21 major mammal species in the South Korea using species distribution models. Occurrence points were integrated from survey datasets provided by the National Institute of Ecology (EcoBank) and the Korea National Park Service. Environmental predictors included climate variables: topographic factors (altitude, slope, and topographic wetness index), land cover-based distance variables (to forest, cultivated land, transportation areas, residential areas, and water bodies), and a population-based human activity index. To reduce multicollinearity (|r|≥0.7) and prioritize discriminatory power, independent-sample t-tests were conducted between presence and pseudo-absence points, and 6-9 variables were finally selected for each species based on p-value ranking and ecological relevance. Potential habitats were modeled using an ensemble approach that combined ten algorithms (GLM, GBM, GAM, CTA, ANN, SRE, FDA, RF, MARS, and MaxEnt), weighted by true skill statistics (TSS). Model performance ranged from 0.255 to 0.991 in TSS (mean=0.588), indicating generally moderate to high predictive accuracy, although some widely distributed rodents showed relatively low performance. The predicted maps distinguished species preferring high-elevation mountainous regions from those favoring lowland and coastal areas, and identified high suitability for the invasive nutria in the Nakdong River Basin. The data provided in this study are expected to be useful for a range of national-scale conservation and management applications, including habitat status assessment, priority conservation area selection, ecological network and corridor planning, and invasive species control.