Drought effects on soil carbon dioxide efflux in two ecosystems in Central Sulawesi, Indonesia

Abstract

Drought response on soil CO2 efflux and production dynamics were examined in two tropical ecosystems in central Sulawesi, Indonesia. Large-scale throughfall displacement roofs were built in a cacao (Theobroma cacao) / Gliricidia sepium agroforestry plantation (560 m.a.s.l.) and in an undisturbed sub-montane tropical rainforest (1,050 m.a.s.l.) to simulate drought conditions. At each site, ecosystem drought responses from three roof plots were compared to three undisturbed control plots. Soil CO2 production was measured spatially at the soil surface and vertically within the soil profile to 2.5 m depth every two weeks. 1. The simulated drought in the cacao / Gliricidia agroforestry ecosystem consisted of a one month baseline evaluation phase, a 13 months simulated drought and a five months rewetting phase. During the drought phase of the experiment soil CO2 efflux decreased by 13% in comparison to the control. The mild drought response is attributed to two reasons. First, soil CO2 efflux peaked at intermediate soil moisture contents, but was low when soil conditions became dry (in the induced drought plots) and when the soil became very wet (in the control plots). This means that respiration differences between control and roof plots may have been masked when soil moisture conditions were wet in the control and concurrently very dry in roof plots. Secondly, CO2 efflux drought response was localized. At some measurement chamber sites (n = 11) drought effects were very pronounced: soil CO2 efflux decreased as soil moisture levels decreased. At other chamber sites (n = 7) however, there was no evident drought effect on soil CO2 emissions. Soil CO2 efflux was highest at the base of tree stems and decreased with distance radiating outwards. Overall, the experiment had a CO2 neutral effect: the decreases in emissions during the induced drought period were compensated for during the five month rewetting phase when CO2 efflux rebounded and surpassed the control. 2. The simulated drought in the undisturbed sub-montane tropical rainforest consisted of a two and a half months baseline evaluation phase, a 24.5 months simulated drought and a four months rewetting phase. In contrast to the cacao agroforest, the sub montane tropical rainforest experienced a severe decrease in soil CO2 production. Here, soil CO2 efflux decreased by an average of 39 % in comparison to the control during the induced drought period. Soil moisture, the main variable controlling CO2, exhibited a strong positive linear relationship with soil CO2 production (R2 = 0.72). A two phase ecosystem drought response was observed. During the first phase, which lasted nine months, leaf litter respiration declined as did the CO2 production between 30 and 70 cm soil depth. During the second phase of the experiment (the next 16 months) drought conditions intensified further and belowground CO2 production from heterotrophic and autotrophic sources decreased at all soil depths. Leaf litter respiration remained negligible. Recuperation after the drought was slow in this ecosystem and did not rebound to control plot levels. In this ecosystem, the simulated drought resulted in a reduction in overall CO2 emission.