How well can we predict soil respiration with climate indicators, now and in the future?

Abstract. Soils contain the largest terrestrial store of carbon; three times greater than present atmospheric concentrations, whilst the annual soil-atmosphere exchange of carbon is an order of magnitude larger than all anthropogenic effluxes. Quantifying future pool sizes and fluxes is therefore sensitive to small methodological errors, yet unfortunately remains the second largest area of uncertainty in Intergovernmental Panel on Climate Change projections.

The flux of carbon from heterotrophic decomposition of soil organic matter is parameterized as a rate constant. This parameter is calculated from observed total soil carbon efflux and contemporaneously observed temperature and soil moisture. This metric is then used to simulate future rates of heterotrophic respiration, as driven by the projections of future climate- temperature and precipitation. We examine two underlying assumptions: how well current climate (mean temperature and precipitation) can account for contemporary soil respiration, and whether an observational parameter derived from this data will be valid in the future.

We find mean climate values to be of some use in capturing total soil respiration to the 95% confidence interval, but note an inability to distinguish between subtropical and Mediterranean fluxes, or wetland-grassland and wetland-forest fluxes.

Regarding the future, we present a collection of CO₂ enrichment studies demonstrating a strong agreement in soil respiration response (a 25% increase) independent of changes in temperature and moisture, however these data are spatially limited to the northern mid-latitudes.

In order to "future-proof" simple statistical parameters used to calculate the output from heterotrophic soil respiration, we propose a correction factor derived from empirical observations, but note the spatial and temporal limitations.

In conclusion, there seems to be no sound basis to assume that models with the best fit to contemporary data will produce the best estimates of future fluxes, given the methods, future dynamics and the nature of the observational constraints. Only through long-term field observations and appropriate, perhaps novel, data collection can we improve statistical respiration modelling, without adding mechanistic details at a computational cost.