Increasing Soil Organic Carbon (SOC), the largest terrestrial carbon pool, through proper land management has been suggested as a nature-based solution to mitigate climate change. In this context, it is important to understand the impacts of land transformations on regional SOC stocks. The study spatially analyzed the tree plantation expansion in Kerala, India, along with other land transformations in the last five decades and its effect on surface (0–30 cm) Soil Organic Carbon (SOC) density and stocks. This study adopted a machine learning-based predictive modelling approach by combining: (1) a detailed two-time period land use map separating major plantation types; (2) legacy soil data representing ground SOC measurements for each land use category; (3) other climatic, topographic and soil variables that affect the spatial variation of SOC, in order to spatially assess the changes in SOC stocks in Kerala due to land use changes over the last five decades (1972–2020). The study highlighted significant local hotspots of losses and gains that the traditional area-based methods do not fully capture. Interestingly, although there was a large increase in the area under tree cover in the last five decades, SOC gains in certain regions were compensated by losses in other regions leading to a very small change (~ 2%) in the overall SOC pool size. Land use and soil type were the most important predictors of SOC based on the developed Random Forest model. The findings highlighted that afforestation with tree plantations might not always lead to an increase in SOC stocks at regional scales. Its effect on SOC stocks varied by plantation type and previous land use. These implications must be considered while adopting climate mitigation strategies. Also, spatially explicit evaluation of various plantation types improves SOC source sink modelling and should be considered for preparing more accurate regional & national SOC inventories.
