Professor Northern Arizona University, United States
Abstract: Soil respiration, the cumulative efflux of CO2 from plant root metabolism and microbial decomposition (hereafter soil CO2 flux), is an essential ecosystem process and a major flux in the terrestrial carbon cycle. Despite its importance, accurate modeling of soil CO2 fluxes remains a challenge in arid ecosystems. While respiration in mesic, temperate ecosystems is primarily controlled by temperature, both temperature and water availability are thought to be limiting factors in arid systems. Yet, models that include only these abiotic factors perform relatively poorly. We hypothesize soil CO2 fluxes in arid ecosystems are driven by interactions among (a) temperature, (b) soil moisture, and (c) photosynthetic activity. The main objective of this study is to devise a new model formulation that better represents soil CO2 fluxes in arid systems across all seasons. Here, we combined soil temperature, soil volumetric water content, and a proxy for plant photosynthetic activity (PhenoCam Green Chromatic Coordinate or “Gcc”) into a better soil respiration model. We measured soil CO2 fluxes in an arid grassland near Flagstaff, AZ using the Eosense FD forced diffusion system, January 2021 through June 2022. We used simulated annealing to optimize parameters in a suite of nonlinear regression models. A combination of classic model fit metrics (AIC, MAPE, RMSE, etc.) were used to assess and compare model performance. Our final model, three multiplicative sigmoidal curves with a scalar parameter, performed very well (marginal R2 = 0.85) and the prediction timeseries mirrored the pulse-like respiration dynamics characteristic of drylands. Modeling these pulses has been challenging due to the influence of antecedent climatic conditions and the responsiveness of microbes. Our proxy for plant activity (PhenoCam Gcc) contributed the most to the success of our final model. We think this is because highly active vegetation has more active root systems, which in turn stimulate microbial activity. Surprisingly, soil temperature had the least influence on model fit, which supports our hypothesis that factors other than temperature and water drive soil CO2 fluxes in arid ecosystems. Our three key drivers (temperature, water content, and plant activity) trade off as the dominant control throughout the year. More accurate predictions of soil CO2 fluxes are essential for understanding the future of these vulnerable, yet critical systems in the context of climate change.
