Driving variables to explain soil organic carbon dynamics: páramo highlands of the Ecuadorian Real mountain range.

Purpose: Large soil organic carbon (SOC) reserves and a high soil capacity for SOC storage within an ecosystem contribute to mitigating the release of carbon into the atmosphere. Developing new spatially-explicit SOC estimation methods at local and micro-watershed scales is essential for gaining landscape understanding of SOC variability. Methods: This study provides new insights into the spatial variability of SOC in the Andean páramo soils. A range of variables from different sources (i.e., geophysical, meteorological, topographic, and spectral) were analyzed to identify driving variables to explain the SOC dynamic in the Andean páramo highlands of the Real range in the central region of Ecuador. This information was used to calibrate a SOC prediction model using Classification and Regression Trees (CART) and soil data samples from the 0-30 cm soil horizon. Results: Eight key variables linking with the SOC storage were used to calibrate the model for SOC estimation with an accuracy of 67% with an RMSE value of 2.17%. Results reveal that sand content emerged as the most significant variable, while taxonomic suborder and protected area variables provided crucial supplementary information. This study improves the ability to detect changes in SOC, particularly in smaller areas where traditional predictors, often more suitable for regional or national assessments, may exhibit insufficient explanatory power. Conclusion: The Andean páramo highlands of the Real range show high capacity for storing SOC, with values ranging from 3.5% to 19%. This variability highlights the ecosystem's importance as a globally relevant carbon reservoir.