Physicochemical Prediction of Tropical Agricultural Soil Carbon and Kinetic Soil Carbon Fractions

Abstract:

Improved understanding of key physicochemical soil properties that influence C storage, and how those controls can be generalized, will create better constraints for soil C models and help to reduce high uncertainty in current models of terrestrial C cycle. We investigated relationships between soil physicochemical properties, C concentration, and kinetic C fractions across a tropical agricultural landscape to improve mechanistic understanding of these systems. The study area was a 30,000-acre sugarcane plantation that was in operation for over 100 years. The 20 NRCS map units that we sampled to one meter represent a heavily disturbed model system of soil C across several important environmental gradients. The system was further constrained by consistent mono-cropping, irrigation, harvest burning, residual removal, and deep tillage for over a century. These constraints allowed for focused investigation into tropical soil C dynamics and the physicochemical properties that preferentially retained C despite constant disturbance. Relationships between physicochemical properties and C were developed using a linear mixed model (LMM) and model selection framework for each of three response variables: C %, the decay rate of the most labile C (k1), and the decay rate of slow C (k2). We found that the amount of information explained was reduced as we looked at faster moving soil C. Models of C % could explain up to 88% of variance, while models of k2 could explain up to 73% of variance, and models of k1 could only explain up to 14% of variance. The LMM and model selected results are not yet causal relationships but do suggest that we can gain insight and possible predictive power from these relationships. The data-driven models also present an opportunity to move past clay and depth as key model parameters so that variance within soil C models can be partitioned into more mechanistic metrics like mineralogy, aggregates, and surface charge.