Forests store carbon in two distinct pools—the living biomass and the detrital mass. Biomass is a recognized carbon sink while the detritus is not. The various carbon fluxes in forests—growth, carbon exchange, decomposition—are commonly studied as disparate parts and not as a whole system. There are relatively few whole-system internal carbon budgets for forests, though many for streams. The disparate data for forests can be assembled into system budgets, but they often won’t balance. We developed balanced, whole-system carbon budgets for four forests in northern California as a way to predict carbon storage in detrital pools and test hypotheses about carbon partitioning and storage and loss potentials. Field methods were tested and modified to reduce sampling bias and variation. We tested the accuracy of carbon calculations based on individual soil profiles versus those based on site means, use of small soil pits versus cores, efficient root separations techniques, and tests of chamber-based soil CO2 effluxes. Resultant estimates of carbon pools and fluxes were combined with literature values into a bookkeeping model that balances carbon in a reasonable manner. Gaming the model is a way to test hypotheses about carbon cycling.
Soil carbon contents were 50 % lower when calculated as the product of bulk density and organic matter concentration from the same soil sample versus those calculated using mean site values. The product of mean values was biased high because soil bulk density and organic matter concentrations were not independent but negatively correlated. Bulk densities were 25 % higher from soil pits higher versus soil cores. Roots separated from moist soil using multiple meshes and winnowing allowed rapid processing of soil profiles and this allowed increased larger samples sizes (36 profiles in each forest) and reduced error (95% C.I. of +/- 30 %). Soil CO2 efflux using dynamic chambers increased with fan speeds. July effluxes averaged 6 µmol C m-2 s-1; annual effluxes were 10 Mg C ha-1 yr-1 and 40 % higher than modeled rates. The model worked well to constrain fluxes. Carbon in these forests range from 250 Mg C ha-1 at lower elevations to over 600 Mg C ha-1 for an undisturbed fir forests at high elevation and appear to be small sinks of carbon. Our model is most responsive to gross primary production, tree mortality, decomposition, and a humification coefficient.