Seasonal variation in carbon dioxide exchange over a Mediterranean annual grassland in California

Understanding how environmental variables affect the processes that regulate the carbon flux over grassland is critical for large-scale modeling research, since grasslands comprise almost one-third of the earth's natural vegetation. To address this issue, fluxes of CO2 (F-c, flux toward the surface is negative) were measured over a Mediterranean, annual grassland in California, USA for 2 years with the eddy covariance method. To interpret the biotic and abiotic factors that modulate F-c over the course of a year we decomposed net ecosystem CO2 exchange into its constituent components, ecosystem respiration (R-eco) and gross primary production (GPP). Daytime R-eco was extrapolated from the relationship between temperature and nighttime F-c under high turbulent conditions. Then, GPP was estimated by subtracting daytime values of F-c from daytime estimates of R-eco. Results show that most of carbon exchange, both photosynthesis and respiration, was limited to the wet season (typically from October to mid-May). Seasonal variations in GPP followed closely to changes in leaf area index, which in turn was governed by soil moisture, available sunlight and the timing of the last frost. In general, R-eco was an exponential function of soil temperature, but with season-dependent values of Q(10). The temperature-dependent respiration model failed immediately after rain events, when large pulses of R-eco were observed. Respiration pulses were especially notable during the dry season when the grass was dead and were the consequence of quickly stimulated microbial activity. Integrated values of GPP, R-eco, and net ecosystem exchange (NEE) were 867, 735, and -132 g C m(-2), respectively, for the 2000-2001 season, and 729, 758, and 29 g C m(-2) for the 2001-2002 season. Thus, the grassland was a moderate carbon sink during the first season and a weak carbon source during the second season. In contrast to a well-accepted view that annual production of grass is linearly correlated to precipitation, the large difference in GPP between the two seasons were not caused by the annual precipitation. Instead, a shorter growing season, due to late start of the rainy season, was mainly responsible for the lower GPP in the second season. Furthermore, relatively higher R-eco during the non-growing season occurred after a late spring rain. Thus, for this Mediterranean grassland, the timing of rain events had more impact than the total amount of precipitation on ecosystem GPP and NEE. This is because its growing season is in the cool and wet season when carbon uptake and respiration are usually limited by low temperature and sometimes frost, not by soil moisture. (C) 2003 Elsevier B.V. All rights reserved.