There is an increasing demand for cost-effective and accurate methods to create digital maps of soil carbon (C). The objective of our study was to create maps of six different soil carbon pools, based on their interaction with ancillary landscape environmental data. Sequestration of carbon into long-lived soil carbon pools is an alternative to offset the increasing concentration of CO2 in the atmosphere, and typically involves movement of fixed carbon into the stable recalcitrant pool from the smaller labile pool which has greater exchange with the atmosphere. We hypothesized that human factors like land use are more influential in the concentration of carbon in the soil than natural factors, such as topography or climate.

We used 141 soil samples collected at four depths in the Santa Fe River watershed (3,585 km2), Florida across soil-landuse trajectories. Soil samples were analyzed for total C, hydrolysable C (6 N HCl), mineralizable C, and dissolved organic C at two fraction sizes (< 0.2 µm and < 0.7 µm). Recalcitrant C was calculated as the difference between total C and hydrolysable C. We based our soil landscape model in well-known conceptual models of soil formation and used a wide variety of environmental explanatory data including: digital elevation model and derived topographic properties, Landsat ETM+ imagery and derived transformations, soil, climate, land use, and geology data. We used regression kriging to account for local random variability and tested different parametric and non-parametric multivariate regression methods to model the global spatial trend.

Our approach provides a comprehensive assessment of how different biogeochemically active carbon pools interact with the landscape through its environmental attributes in a wide variety of soils. The identification of the environmental factors that influence the concentration of carbon in the soil is important to support sustainable land management.