Modeling future changes in potential habitats of five alpine vegetation types on the Tibetan Plateau by incorporating snow depth and snow phenology.

Although snow cover is a major factor affecting vegetation in alpine regions, it is rarely introduced into ecological niche models in alpine regions. Snow phenology over the Tibetan Plateau (TP) was estimated using a daily passive microwave snow depth dataset, and future datasets of snow depth and snow phenology were projected based on their sensitivity to temperature and precipitation. Furthermore, the potential habitats of five alpine vegetation types on the TP were predicted under two future climate scenarios (SSP245 and SSP585) by using a model with incorporated snow variables, and the driving factors of habitat change were analyzed. The results showed that the inclusion of snow variables improved the prediction accuracy of MaxEnt model, particularly in alpine meadow habitats. By the end of the 21st century, the potential habitats of steppes, meadows, shrubs, deserts, and coniferous forests on the TP will migrate to higher latitudes and altitudes, in which the potential habitats of alpine desert will recede (replaced by alpine steppe), and the potential habitats of other four vegetation types will expand. The random forest importance analysis showed that the recession of potential habitat was mainly driven by the increase in average annual temperature, and the expansion of potential habitat was mainly driven by the increase in precipitation. With the gradual increase in temperature and precipitation in the future, the snow depth and snow cover duration days will decrease, which may further lead to the transition of vegetation types from cold-adapted to warm-adapted on the TP. Our study highlights both that the prediction accuracy of alpine vegetation was improved by incorporating snow variables into the species distribution model, and that a changing climate will likely have a powerful influence on the distribution of alpine vegetation across the TP.