Monthly measurements of carbon dioxide ͑CO 2 ͒, methane ͑CH 4 ͒ and nitrous oxide ͑N 2 O͒ fluxes in peat soils were carried out and compared with groundwater level over a year at four sites ͑drained forest, upland cassava, upland and lowland paddy fields͒ located in Jambi province, Indonesia. Fluxes from swamp forest soils were also measured once per year as the native state of this investigated area. Land-use change from drained forest to lowland paddy field significantly decreased the CO 2 ͑from 266 to 30 mg C m -2 h -1 ͒ and N 2 O fluxes ͑from 25.4 to 3.8 g N m -2 h -1 ͒, but increased the CH 4 flux ͑from 0.1 to 4.2 mg C m -2 h -1 ͒ in the soils. Change from drained forest to cassava field significantly increased N 2 O flux ͑from 25.4 to 62.2 g N m -2 h -1 ͒, but had no significant influence on CO 2 ͑from 266 to 200 mg C m -2 h -1 ͒ and CH 4 fluxes ͑from 0.1 to 0.3 mg C m -2 h -1 ͒ in the soils. Averaged CO 2 fluxes in the swamp forests ͑94 mg C m -2 h -1 ͒ were estimated to be one-third of that in the drained forest. Groundwater levels of drained forest and upland crop fields had been lowered by drainage ditches while swamp forest and lowland paddy field were flooded, although groundwater levels were also affected by precipitation. Groundwater levels were negatively related to CO 2 flux but positively related to CH 4 flux at all investigation sites. The peak of the N 2 O flux was observed at Ϫ 20 cm of groundwater level. Lowering the groundwater level by 10 cm from the soil surface resulted in a 50% increase in CO 2 emission ͑from 109.1 to 162.4 mg C m -2 h -1 ͒ and a 25% decrease in CH 4 emission ͑from 0.440 to 0.325 mg C m -2 h -1 ͒ in this study. These results suggest that lowering of groundwater level by the drainage ditches in the peat lands contributes to global warming and devastation of fields. Swamp forest was probably the best land-use management in peat lands to suppress the carbon loss and greenhouse gas emission. Lowland paddy field was a better agricultural system in the peat lands in terms of C sequestration and greenhouse gas emission. Carbon loss from lowland paddy field was one-eighth of that of the other upland crop systems, although the Global Warming Potential was almost the same level as that of the other upland crop systems because of CH 4 emission through rice plants.

Methane ͑CH 4 ͒ emissions from peat soils in tropical and temperate wetlands were compared. Annual CH 4 emission rates in Ozegahara, the largest wetland on Honshu main island, Japan, were higher than in drained forest wetland areas examined in Indonesia. Methane emissions from the lowland paddy fields examined in Indonesia were higher than those of peaty paddy fields in Japan. There was generally a positive correlation ͑r 2 ϭ 0.09; P Ͻ 0.1͒ between CH 4 emissions and CH 4 production activities in wetland soils. In Ozegahara, there was a positive relationship ͑r 2 ϭ 0.80; P Ͻ 0.01͒ between CH 4 production activities and soil pH, but there was no such relationship in Indonesia. The range of soil pH in Ozegahara was 5-7, while pH values in the Indonesian sites were lower than 5. There was a positive response of CH 4 emission with respect to groundwater level in all of these areas. In Indonesia, land-use change from swamp and drained forest to cassava or coconut field lowered groundwater levels and decreased CH 4 emission, while change to lowland paddy raised the groundwater level and increased CH 4 emission. Addition of acetate generally inhibited CH 4 production during the early period ͑un-til 2 weeks͒ of incubation, then enhanced it afterward in both Ozegahara and Indonesian wetland soils. Addition of hydrogen mostly enhanced CH 4 production. From the results of this study, CH 4 fluxes from peat soil to the atmosphere were positively correlated with CH 4 production activities, and CH 4 production activity in peat soil was regulated by soil pH, while land-use change from wetland to upland crop lowered groundwater level and thus reduced CH 4 production and enhanced CH 4 oxidation.