Improving the prediction of Phragmites australis marsh distribution by adding edaphic properties to climatic...

Soil properties have been widely shown to influence wetland vegetation distribution. However, the degree to which adding edaphic variables can improve climate-only models remains unclear. Here, we examine whether incorporating soil properties into climate-only Phragmites australis marsh distribution models (SDMs) can improve statistical performance and predictive power. The results show that including soil predictors significantly improved SDMs' performance compared to climate-only models (TSS: + 21.07%, Student's t-tests: t = 24.364, P < 0.01; AUC: + 8.51%, t = 27.209, P < 0.01). On average, soil predictors contributed more than climate predictors (57.9% vs 42.1%). Notably, the topographic wetness index, soil pH, and soil organic carbon were crucial predictors. Climate + soil models reduced the predicted suitable areas by 23.94%-55.13% relative to those from climate-only models. Our findings suggest that soil properties may constrain the potential range expansion of wetland plants, highlighting the necessity of incorporating these properties into models for predicting wetland plant distributions under environmental change.