Soil organic carbon stocks and their determining factors in the Dano catchment (Southwest Burkina Faso)

• Published in: Catena (Giessen) 2018-07, Vol.166, p.298-309
• Main Authors: Hounkpatin, Ozias K.L., Op de Hipt, Felix, Bossa, Aymar Yaovi, Welp, Gerhard, Amelung, Wulf
• Format: Article
• Language: English
• Subjects: Carbon stock; Multiple linear regression land use; Random forest; Reference soil group; Soil organic carbon
• ISSN: 0341-8162; 1872-6887
• DOI: 10.1016/j.catena.2018.04.013

Although the evaluation of soil organic carbon (SOC) stocks across different types of land use and major reference soil groups is essential for mitigating climate change, there remains, to date, limited comprehensive understanding of whole tropical soil profiles. Therefore, this study aimed to explain the amount of SOC stocks in different land-use systems and across various soil groups, as well as its spatial pattern in the topsoil (0–30 cm) and subsoil (30–100 cm) within the savannah zone of Burkina Faso. Roughly 70 soil profiles were considered along with additional auger sampling to account for spatial variation in both cropland (CR) and savannah (SA). The machine learning technique random forest regression (RFR) and multiple linear regression (MLR) were used for modeling the surface and subsurface SOC stocks. For model calibration, covariates including land use, topographic, texture, and climatic data were considered as surrogate for soil forming factors. The prediction maps produced by the calibrated models were validated by an independent dataset. The results indicated that about 53% of the SOC stock over 1 m depth was held in the upper 30 cm. Only a marginal difference was recorded between the topsoil SOC stock in SA (41.4 t C ha−1) and CR (39.1 t C ha−1) soils. For the subsoil, a significant difference (p