Fluxes of N2O and CH4 from soils of savannas and seasonally-dry ecosystems

Aim: Savannas and seasonally-dry ecosystems cover a significant part of the world's land surface. If undisturbed, these ecosystems might be expected to show a net uptake of methane (CH4) and a limited emission of nitrous oxide (N2O). Land management has the potential to change dramatically the characteristics and gas exchange of ecosystems. The present work investigates the contribution of warm climate seasonally-dry ecosystems to the atmospheric concentration of nitrous oxide and methane, and analyses the impact of land-use change on N2O and CH4 fluxes from the ecosystems in question. Location: Flux data reviewed here were collected from the literature; they come from savannas and seasonally-dry ecosystems in warm climatic regions, including South America, India, Australasia and Mediterranean areas. Methods: Data on gas fluxes were collected from the literature. Two factors were considered as determinants of the variation in gas fluxes: land management and season. Land management was grouped into: (1) control, (2) 'burned only' and (3) managed ecosystems. The season was categorized as dry or wet. In order to avoid the possibility that the influence of soil properties on gas fluxes might confound any differences caused by land management, sites were grouped in homogeneous clusters on the basis of soil properties, using multivariate analyses. Inter- and intra-cluster analysis of gas fluxes were performed, taking into account the effects of season, land management and main vegetation types. Results: Soils were often acid and nutrient-poor, with low water retention. N2O emissions were generally very low (median flux 0.32 mg N2O m-2 day-1), and no significant differences were observed between woodland savannas and managed savannas. The highest fluxes (up to 12.9 mg N2O m-2 day-1) were those on relatively fertile soils with high air-filled porosity and water retention. The effect of season on N2O production was evident only when sites were separated in homogeneous groups on the basis of soil properties. CH4 fluxes varied over a wide range (-22.9 to 3.15 mg CH4 m-2 day-1, where the negative sign denotes removal of gas from the atmosphere), with an annual average daily flux of -0.48 ± 0.96 (SD) mg CH4 m-2 day-1 in undisturbed (control) sites. Land-use change dramatically reduced this CH4 sink. Managed sites were weak sinks of CH4 in the dry season and became sources of CH4 in the wet season. This was particularly evident for pastures. Burning alone did not reduce soil net CH4 oxidation, but decreased N2O production. Main conclusions: Despite the low potential for N2O production, both in natural and managed conditions, tropical seasonally-dry ecosystems represent a significant source of N2O (4.4 Tg N2O year-1) on a global scale, as a consequence of the large area they occupy. The same environments represent a potential CH 4 sink of 5.17 Tg CH4 year-1. However, assuming that c. 30% of the tropical land is converted to different uses, the sink would be reduced to 3.2 Tg CH4 year-1. The limited information on fluxes from Mediterranean ecosystems does not allow a meaningful scaling up. © 2006 Blackwell Publishing Ltd.