Soil respiration is the largest flux of terrestrial carbon to the atmosphere.  Despite its importance, we lack a solid understanding of the factors affecting the temporal dynamics of soil respiration. This is especially true in arid ecosystems where soil respiration and productivity are highly variable and largely controlled by episodic water availability.  Towards improving our understanding of soil respiration in arid systems, this study employs a hierarchical Bayesian model to synthesize multiple datasets on soil respiration representing four major North American deserts (i.e., Chihuahuan, Sonoran, Mojave, and Great Basin).  We examine patterns of soil respiration and thresholds of limitation created by soil moisture, temperature, and available carbon.  Coupling of the large and diverse datasets with the Bayesian modeling framework allows us to explicitly quantify antecedent soil moisture and to estimate the effect of antecedent soil moisture on soil respiration.  Many previous studies have indicated the importance of antecedent conditions in understanding current ecological processes, but the quantification of antecedent conditions has been treated in a qualitative or ad hoc manner.

Results/Conclusions

Here, we provide a rigorous, quantitative method for quantifying the effects of antecedent conditions.  Using our model, we found that antecedent soil moisture can explain up to 55% of the observed variation between observed and predicted levels of soil respiration in these desert systems.  The importance of seasonal thresholds also emerged.  During the dry summer months, moisture received 25 days in the past significantly affected soil respiration, explaining 31% of the observed variation between observed and predicted soil respiration.  In the cooler winter months, the threshold was closer to 10 days.  This approach expands our understanding of soil respiration in arid ecosystems, and it can be easily generalized to quantify the impacts of antecedent conditions on other ecological processes and systems.