Carbon dioxide exchange over multiple temporal scales in an arid shrub ecosystem near La Paz, Baja California Sur, Mexico

Arid environments represent 30% of the global terrestrial surface, but are largely under-represented in studies of ecosystem carbon flux. Less than 2% of all FLUXNET eddy covariance sites exist in a hot desert climate. Long-term datasets of these regions are vital for capturing the seasonal and interannual variability that occur due to episodic precipitation events and climate change, which drive fluctuations in soil moisture and temperature patterns. The objectives of this study were to determine the meteorological variables that drive carbon flux on diel, seasonal, and annual scales and to determine how precipitation events control annual net ecosystem exchange (NEE). Patterns of NEE from 2002 to 2008 were investigated, providing a record with multiple replicates of seasons and conditions. Precipitation was extremely variable (55339 similar to mm) during the study period, and reduced precipitation in later years (20042008) appears to have resulted in annual moderate to large carbon sources (62258 similar to g similar to C similar to m-2 similar to yr-1) in contrast to the previously reported sink (20022003). Variations in photosynthetically active radiation were found to principally drive variations in carbon uptake during the wet growing season while increased soil temperatures at a 5 similar to cm depth stimulated carbon loss during the dry dormant season. Monthly NEE was primarily driven by soil moisture at a 5 similar to cm depth, and years with a higher magnitude of precipitation events showed a longer growing season with annual net carbon uptake, whereas years with lower magnitude had drier soils and displayed short growing seasons with annual net carbon loss. Increased precipitation frequency was associated with increased annual NEE, which may be a function of increased microbial respiration to more small precipitation events. Annual precipitation frequency and magnitude were found to have effects on the interannual variability of NEE for up to 2 similar to years.