Using multiple linear regression techniques to quantify carbon stocks of fallow vegetation in the tropics

Fallow ecosystems provide a significant carbon stock that can be quantified for inclusion in the accounts of global carbon budgets. Process and statistical models of productivity, though useful, are often technically rigid as the conditions for their application are not easy to satisfy. Multiple regression techniques have been applied to study some biophysical phenomena but yet to be applied to carbon stock estimation. Using ecological data from 28 sampling locations, the study applied the stepwise multiple regression technique to identify ecological variables that would explain carbon stock of fallow vegetation, aged between 3 and 8 years. The procedure generated three predictive regression models. The full model, could explain nearly 98% of variability of carbon stock (R 2 = .979), using cationexchange capacity and total nitrogen content of soil and leaf area index as the three predictor variables. Sampling inaccuracies could have contributed to the error component of models and sample size increase has been suggested for reduction of such errors. The advantage of the method is its simplicity. The paper suggests that the derived models be validated before broad application. Also, the cost-effectiveness of the approach should be tested against other approaches.