Deep Unsaturated Zone Contributions to Carbon Cycling in Semiarid Environments

Understanding terrestrial carbon cycling has relied primarily on studies of topsoils that are typically characterized to depths shallower than 0.5m. At a semiarid site instrumented down to 7m, we measured seasonal- and depth-resolved carbon inventories and fluxes and groundwater and unsaturated zone flow rates. Measurements showed that similar to 30% of the CO2 efflux to the atmosphere (60% in winter) originates from below 1m, contrary to predictions of less than 1% by Earth System Model land modules. Respiration from deeper roots and deeper microbial communities is supported by favorable subsurface temperatures, moisture, and oxygen availability. Below 1m, dissolved organic carbon fluxes from the overlying soil and C from deep roots and exudates are expected to be important in sustaining microbial respiration. Because these conditions are characteristic of semiarid climate regions, we contend that Earth System Model land modules should incorporate such deeper soil processes to improve CO2 flux predictions. Current understanding and prediction of terrestrial carbon cycling is primarily based on studies of soils shallower than 1m. It is extremely challenging to obtain quantitative understanding of carbon fluxes in deep subsurface needed to close the terrestrial carbon cycle. Our team conducted a field-based study in a semiarid region, spanning a depth of 7m through the unsaturated zone into groundwater. Through unique approaches including long-term sampling of depth-resolved gas and pore waters, we discovered that 30% of the CO2 efflux to the atmosphere (60% in winter) originates from below 1m. This result is contrary to the Earth System Model land model CLM4.5 that predicts less than 1% of the surface CO2 flux originates below 1-m depth. Moreover, we discovered an unexpectedly high dissolved organic carbon flux from rhizosphere into underlying unsaturated zone, which has not been previously recognized despite many decades of research on terrestrial carbon dynamics. Such fluxes are not detectable based on conventional bulk soil sample analyses. This dissolved organic carbon flux together with the other mechanisms discussed in the manuscript are characteristic of arid- and semiarid climate regions, leading us to contend that Earth System Model land modules should incorporate these deeper soil processes to improve carbon flux predictions.