Partitioning soil surface CO2 efflux into autotrophic and heterotrophic components, using natural gradients in soil δ13C in an undisturbed savannah soil

We used natural gradients in soil and vegetation δ13C signatures in a savannah ecosystem in Texas to partition soil respiration into the autotrophic (Ra) and heterotrophic (Rh) components. We measured soil respiration along short transects from under clusters of C3 trees into the C4 dominated grassland. The site chosen for the study was experiencing a prolonged drought, so an irrigation treatment was applied at two positions of each transect. Soil surface CO2 efflux was measured along transects and CO2 collected for analysis of the δ13C signature in order to: (i) determine how soil respiration rates varied along transects and were affected by localised change in soil moisture and (ii) partition the soil surface CO2 efflux into Ra and Rh, which required measurement of the δ13C signature of root- and soil-derived CO2 for use in a mass balance model. The soil at the site was unusually dry, with mean volumetric soil water content of 8.2%. Soil respiration rates were fastest in the centre of the tree cluster (1.5 ± 0.18 μmol m-2 s-1; mean ± SE) and slowest at the cluster-grassland transition (0.6 ± 0.12 μmol m-2 s-1). Irrigation produced a 7-11 fold increase in the soil respiration rate. There were no significant differences (p > 0.5) between the δ13C signature of root biomass and respired CO2, but differences (p < 0.01) were observed between the respired CO2 and soil when sampled at the edge of the clusters and in the grassland. Therefore, end member values were measured by root and soil incubations, with times kept constant at 30 min for roots and 2 h for soils. The δ13C signature of the soil surface CO2 efflux and the two end member values were used to calculate that, in the irrigated soils, Rh comprised 51 ± 13.5% of the soil surface CO2 efflux at the mid canopy position and 57 ± 7.4% at the drip line. In non-irrigated soil it was not possible to partition soil respiration, because the δ13C signature of the soil surface CO2 efflux was enriched compared to both the end member values. This was probably due to a combination of the very dry porous soils at our study site (which may have been particularly susceptible to ingress of atmospheric CO2) and the very slow respiration rates of the non-irrigated soils.