Predictor variables for soil organic carbon contents in the Miombo woodlands ecosystem of Kitonga forest reserve, Tanzania

Abstract

Few studies have established the variables which adequately predict SOC storage in the Miombo woodlands. Multiple regression analysis was used to establish the variables which could predict SOC contents in dominant soils of the Miombo woodlands of Kitonga Forest Reserve, Tanzania. Thirty soil minipits located at different elevations across a topographical gradient were selected, geo-referenced, excavated and samples from the natural horizons were collected for physico-chemical analysis. A total of 85 samples were collected, each representing a natural soil horizon. The results indicated that total nitrogen (TN), (P< 0.001, R2 = 0.97) and TN in combination with calcium (Ca) (P< 0.001, R2 = 0.99) were important predictor variables of SOC contents. The combination of cation exchange capacity, Zinc, Copper, clay and iron together with TN and Ca predicted well the SOC contents (P< 0.001, R2 = 0.999). Considering time and cost implications for field and laboratory analysis in predicting SOC stocks, the combination of TN and Ca that predicted the SOC contents by 99% provided equally strong prediction when compared to the combination of all the variables. Thus, proper land management strategies which enhance conservation of TN and Ca in concert would provide adequate prediction of SOC contents in soils.