Abstract

Soils are the largest organic carbon terrestrial reservoir (COS), compared to the oceans, terrestrial vegetation, and the atmosphere. Soil organic carbon (COS) is dynamic and, due to anthropogenic action, it can become a sink or a net source of greenhouse gases (GHG). For this reason, Ecuador joined efforts and contributed to the construction of the World Map of Organic Soil Carbon (GSOCmap), with which it was possible to estimate, by means of digital soil mapping, the COS content within the 30 cm depth level. national as well as its spatial variability by identifying the environmental factors involved in its storage and those covariates that influence its uncertainty regarding its content. To achieve this, a geostatistical model (Regression - Kriging) was used, which combined 12 924 data from profiles of Heritage soils at 1: 25 000 scale along with 140 edaphic and environmental covariates. The model obtained with a spatial resolution of 1 km, estimated the COS reserve between 10 to 297.8 t had, (mean of 55.69 t. ha(-1)), and suggests a COS reservoir in the surface 30 cm of 1.37 Pg. However, in order to strengthen the confidence of the reported value, it was essential to calculate the uncertainty through this study through ordinary kriging between the differences in the COS values estimated by the regression equation and the real values, through cross-validation (5%), and later by external validation, from which an uncertainty value of COS stored in the soils of Ecuador of 1.63 +/- 0.38 Pg of COS was obtained. It was obtained from this model that 40% of the spatial variation of COS presented a root of the root mean square error (RMSE) of 0.52 t ha(-1) and a correlation (R-2) of 0.41. The covariates with the greatest weight and which directly influence the carbon stock are temperature, soil type, altitude, and the topographic humidity index. The highest concentration of COS is present in the Andes region (> 120 t.ha(-1)) while in the Coast region and Insular region the values fluctuate between medium to low and in the Amazon region the values are low (<40 t. ha(-1)). The analysis allowed to know the areas that present greater uncertainty and less precision, with which it will be possible to strengthen monitoring in areas that currently have information gaps in future studies.