Abstract

Soil carbon fluxes are highly variable in space and time under tree crops such as oil palm, and attempts to model such fluxes must incorporate an understanding of this variability. In this work, we measured soil CO2 emission, root biomass and pruned frond deposition rates and calculated carbon fluxes into and out of the soil in a mature (20-year-old, second planting cycle) oil palm plantation in Papua New Guinea. Tree-scale spatial variability in CO2 emission and root biomass was quantified by making measurements on a 35-point trapezoid grid covering the 38.5-m(2) repeating unit of the plantation (n = 4 grids). In order to obtain an overall mean soil CO2 emission rate within 5% of the most accurate estimate, >= 24 measurement points were required. Soil CO2 emissions were spatially correlated with calculated carbon inputs (r(2) = 0.605, slope 1 : 1), but not with soil water content or temperature. However, outputs were higher than inputs at all locations, with a mean overall output of 7.24 mu mol m(-2) s(-1) and input of 3.02 mu mol m(-2) s(-1). Inputs related to fronds, roots and groundcover constituted 60%, 36% and 4% of estimated inputs, respectively. The spatial correlation of carbon inputs and outputs indicates that mineralisation rate is controlled mostly by the amount rather than the nature or input depth of the additions. The spatially uniform net carbon emission from soil may be due to inaccuracies in calculated fluxes (especially root-related inputs) or to non-biological emissions.