Soil carbon stock assessment using depth and spatial models on afforested arable lands

The change from grasslands and natural shrubs to afforested arable land has a major impact on soil organic carbon (SOC) stocks. Grasslands are known to be SOC sinks as seen in the Chernozems of North America, Eurasia and South Africa. However, determining the SOC stocks of soils can be financially costly as each location must be sampled in depth increments. This study aimed to estimate the SOC stocks for the Mvoti catchment (30 degrees 48 ' E and 29 degrees 18 ' S) in Kwa-Zulu Natal, South Africa by developing depth functions on a limited number of soil samples and expanding these functions to known land-uses and soil types. The results showed that splines captured the short-term vertical distribution of SOC better than exponential decay functions, which has major implications on arable lands. Long-term forest plantations showed a positive correlation with SOC stocks (32.7 kg m(-2)), while annual crop cultivation (27.0 kg m(-2)) showed a negative correlation when compared to natural grasslands (28.8 kg m(-2)). The Cubist algorithm predicted the total SOC stock of the catchment area at between 12 248 and 17 624 Mg depending on the depth function used. Soils with yellow-brown subsoils tend to have higher SOC stocks and the lowest degree of uncertainty.