Quantifying and mapping soil organic carbon in Mali, West Africa using spatiotemporal methods

The Kyoto protocol recognized the importance of the terrestrial sink of carbon and proposed schemes that allow countries to treat sequestered carbon as a commodity that can be traded for global environmental benefit. Carbon sequestration can be a win-win scenario because it also introduces a set of new benefits into dryland farming communities particularly in Sub-Saharan Africa. The possibility, however, for agricultural producers to participate in the emerging market for tradable carbon-credits requires a reliable verification mechanism. Soil carbon inventories of many developing nations rely on a broad scale assessment. These approaches do not account for the spatial and temporal variability of soil carbon nor do they provide a measure of uncertainty associated with these assessments. This study proposed the use of Bayesian Maximum Entropy (BME) to quantify and map soil organic carbon at field scale in four agroecological zones of Mali, Sub-Saharan Africa. The prediction model comparisons using the mean error (ME) indicated that BME performed better than did the kriging methods (0.033, 0.41, respectively). BME prediction also provided a lower MSE representing a 25% reduction compared with Kriging, and 10% compared with cokriging. This study also demonstrated potential use of space---time covariances as tools to improve our understanding of spatial and temporal variability of soil organic carbon. Based on the temporal and spatial models maps were generated to predict mean trends. The estimation of tree biomass in Sub-Saharan Africa is important for an accurate assessment of the potential of these systems to capture and store carbon. The results show that tree carbon represented as much as 34% of the amount of organic carbon stored in soil surface (0-20 cm). Data from 2000 to 2006 indicated a net increase of soil organic carbon, which varied between 2.6 to 13.9 Mg ha-1. Despite the complexities that characterize the spatial and temporal distribution of most environmental processes, BME provides a framework to analyze both space and time components.