Disparate responses of soil-atmosphere CO2 exchange to biophysical and geochemical factors over a biocrust ecological succession in the Tabernas Desert

There is growing evidence that dryland soils could act as substantial but so far disregarded carbon sinks at the global scale. In drylands, potential abiotic processes of CO2 uptake are still debated while estimates of the biotic contribution of photosynthetizing biocrusts to the net carbon uptake remain uncertain. This uncertainty is partly attributable to a common neglect of the underlying soil and spatiotemporal variability of soil CO2 fluxes. Moreover, it is still unknown how those fluxes evolve during the ecological succession of biocrusts and which factors control them. Therefore, in this study, we aimed to (1) identify those factors and use them for predictions, and (2) explain the inter-annual variation of cumulative CO2 fluxes over the succession. We conducted 2 years of continuous measurements of the topsoil CO2 molar fraction (chi s) and microclimatic variables and estimated the soil-atmosphere CO2 flux using the gradient method. Statistical spatiotemporal models were developed of chi s dynamics and cumulative annual fluxes. A soil CO2 uptake potentially due to coupled gypsum dissolution-calcite precipitation was consistently detected at night, with cumulative values ranging from-4 to-65 gC m- 2 depending on succession stage. In comparison, cumulative soil CO2 emissions ranged from 101 to 307 gC m(-2) depending on succession stage. The succession stage, soil water content (theta w), and temperature (Ts) and the in-teractions of these variables explained and efficiently predicted the daily averaged chi s dynamics. Soil CO(2 )emissions were more sensitive to theta w and Ts in late successional stages, apparently because of organic carbon accumulation and higher porosity. Our measurements suggest that CO2 consumption processes were progressively counterbalanced by the increase in biological CO2 production during succession. That is probably why those processes could mainly be detected in early successional stages and more generally in drylands, as they sustain a low biological activity. However, such processes could be ubiquitous in ecosystems but difficult to detect. We discuss the implication of those results for the extensive dryland soils in the context of climate change.