On the importance of soil texture for predicting future global soil respiration

Soil respiration, the flux of CO₂ from soils to the atmosphere, is an important component of the global carbon budget, but our understanding of its future responses and patterns remains highly uncertain. Global soil respiration has usually been modelled as a function of climate variables (e.g., temperature and precipitation), but rarely explored soil properties such as soil texture may be highly influential. Here, we model and test for the influence of soil texture on global soil respiration under future climate change scenarios until the year 2100 at 1 km resolution using a machine learning approach. Total soil respiration (Rs) is projected to increase from 91.9 to 97.6 PgC y^-1 (∼6% change) under the SSP5-8.5 scenario, with consistent rising heterotrophic respiration (Rh) at high latitudes. Not accounting for soil texture may lead to higher estimates of future Rs (+7.8%) and Rh (+27.8%), indicating that soil carbon-climate feedbacks solely based on climate variables-the dominant approach up to now-may be overestimated. Using the best available proxies for climate, vegetation and soil, our study provides spatial and temporal insights about future soil carbon dynamics and highlights the urgent need to consider more soil properties that influence soil respiration at local scale to accurately project the specific responses of global Rs and Rh to the changing climate of the 21st century.