COMPARISON OF LABORATORY AND MODELING SIMULATION METHODS FOR ESTIMATING SOIL CARBON POOLS IN TROPICAL FOREST SOILS

Availability of methods to determine kinetically-defined soil carbon p ools may assist in better understanding soil organic matter turnover i n tropical soils and its relationship with soil mineral fractions and nutrient cycling. Our objective was to compare three methods of estima ting soil C pools for the top mineral soil horizon of 13 tropical fore st soils with a wide range of clay content and differing soil mineralo gies. The methods were: (i) regression analysis of CO2-C release from a 341 day incubation of unamended soils; (ii) results of C analysis pr ocedures including determinations of soluble C, microbial biomass C an d light fraction C; and (iii) CENTURY model simulations of equilibrium values of soil C pools. Soil mineralogy did not have a significant ef fect on CO2-C release, expressed as a proportion of total organic C. d uring incubation. However, allophanic soils had significantly higher t otal organic C, soluble C and light fraction C than soils of smectitic , kaolinitic or oxidic mineralogies. Clay and sand contents significan tly correlated with cumulative proportional CO2-C release. The active C pool, as determined by the CENTURY equilibrium method and measuremen ts of soluble plus microbial biomass C, were less than the active C po ol estimated by the incubation-regression method. Measured light fract ion e was smaller than estimates from the CENTURY equilibrium method a nd incubation-regression estimates. Total organic C, soluble plus micr obial biomass C and light fraction C had the highest correlations with cumulative incubation CO2-C release. Of the CENTURY model C pool esti mates, only the slow C pool estimate correlated with incubation CO2-C release. The use of C analyses as soil C pool estimates for model simu lations of the long-term incubation resulted in an underestimation of actual incubation C02-C release. This underestimate was caused by a sm aller slow pool estimated by light fraction analysis. In addition, str uctural and metabolic C pools were not measured and they have a large short-term effect on CO2-C release. Use of CENTURY equilibrium estimat es, including estimates of structural and metabolic C, resulted in sim ulated CO2-C release comparable to actual CO2-C release patterns. Howe ver, the use of the CENTURY equilibrium method may be limited by the d ifficulty of obtaining adequate soil, plant and climatic information t o run model simulations and by the validity of CENTURY model assumptio ns for factors controlling soil C pools under tropical climatic condit ions.