Nitrous oxide fluxes from savanna (miombo) woodlands in Zimbabwe

Aim: We test the hypothesis that land use and climate are important controls of nitrous oxide (N2O) emissions from savanna ecosystems, and that these emissions can be represented by a mechanistic model of carbon (C) and nitrogen (N) transformations. Location: Miombo woodlands in Zimbabwe are part of widespread woody savanna formations in southern and central Africa that cover more than 2.7 million km(2). The rainfall in this region is around 800 mm and is concentrated in the period between November and March. Methods: Losses of N2O were measured along transects in two field areas using static chambers over a period of 1 year. The vegetation in both areas was dominated by Julbernardia globiflora and Brachystegia spiciformis, but had differing management systems (burned and unburned), soil properties and site characteristics (slope and drainage). The effects of simulated rainfall and fertilizer additions were studied in laboratory incubations. Results: Patterns of N2O emissions were strongly linked to rainfall. The highest fluxes at both sites were measured within 18 days of the onset of the first rains in November, with fluxes of up to 42 mu g N m(-2) h(-1). During the dry season, fluxes were lower, but a large proportion (R-2 values between 0.8 and 0.95, P < 0.001) of the N2O flux could be predicted by variations in soil moisture. Soil columns were set up in the laboratory to which simulated rainwater was added, and the amounts and timing of rainwater addition were varied. Losses of N2O were highest within the first week of the laboratory study. Altering the amount of rainwater addition did not significantly affect N2O loss; however, a continuous addition of water resulted in higher losses of N2O (up to 79 mu g N m(-2) h(-1)) than periodic addition of the same amount. A model (denitrification-decomposition) was used to simulate N2O release over a 12 month period, using meteorological data recorded in the vicinity of the field site. The simulations and field data suggest that nitrification was the main process responsible for N2O release during the dry season but that denitrification was more important during the wet season. Main conclusions The release of N2O from dryland savannas was shown to constitute an important nutrient flux, and emissions were strongly linked to patterns of rainfall; however, there was evidence to suggest that the magnitude of fluxes is also influenced locally by differences in soil organic matter concentration and drainage.