Modeling vertical distribution of carbon in oxisols of the western Brazilian Amazon (Rondonia)

Oxisols have great ecological significance in tropical soils because they are the dominant soil type of the Brazilian Amazon ecosystem, comprising more than 40% of its total. To estimate carbon (C) stocks and changes requires knowledge of the vertical distribution of C in profiles, The objective of this study was to determine if specific patterns occur in C profiles of Oxisols in the Western Brazilian Amazon so that total C storage can be assessed down to any given depth by simple models with low input data requirements, Two models, a power-based model and an exponential-based model, were tested using nonlinear regression analysis on a soil database made up of 129 Oxisol profiles corresponding to 519 soil horizons, These models, as judged by the coefficient of determination (R-2) value, explained more than 55% of the total variance for all of the horizons, whether or not segregated by taxonomic unit. The models were then tested with individual profiles. The power model exhibited a tendency to overestimate C stocks when integration was done for the 0-20-cm and the 0-100-cm layers, Results from the exponential model were better than those from the power model, The R-2 values were greater than 0.82, and the associated standard error was reduced, In a validation procedure, the mean error (ME) was close to zero for the exponential model, with a systematic ME of only 0.06 kg C.m(-2) for the 0-100-cm layer.