LAUSR

This study investigates how precipitation, temperature and seasonality (as a proxy of plant productivity) affect the temporal and spatial variability of soil CO2 efflux in two dry semiarid grasslands with different degrees of land degradation. We measured soil CO2 efflux over four years under plant, biological soil crust and bare soil patches and estimated annual soil carbon losses in both, a natural and a degraded grassland, by means of generalised additive mixed models considering temporal autocorrelation in the data. Soil CO2 efflux ranged from 0.08 to 3.70 and from 0.10 to 3.01 μmol CΟ2 m−2 s−1 in the natural and degraded grasslands, respectively. Daily soil CO2 efflux was mostly affected by moisture in the degraded grassland (25.4%), while in the natural grassland was affected jointly by seasonality, temperature and moisture (27.5%). Overall, the highest soil carbon fluxes were measured in soils covered by biological soil crusts (1.24 ± 0.02 and 1.10 ± 0.02) and the lowest in bare soils (1.11 ± 0.02 and 0.82 ± 0.02 μmol CΟ2 −2 s−1) in the natural and degraded sites, respectively. Cumulative soil carbon fluxes were mainly driven by temperature and previous precipitation (over three months). The highest soil carbon losses were estimated in the driest year (2009) and the lowest in the wettest (2010) with almost twice the amount of rainfall. The main difference between these years was the timing of the events that mostly occurred in the moments of maximum plant activity with optimum temperatures in spring in the dry year. Changes in precipitation patterns will affect soil carbon fluxes more than rainfall amount, particularly in degraded grasslands. Therefore, considering all climate drivers together with plant activity is essential to predict how climate change will affect soil biological processes in drylands.