Climate change is likely to perturb rainfall patterns in the humid tropics, with uncertain consequences for biogeochemical processes that control soil greenhouse gas fluxes. The response of soil-atmosphere fluxes of carbon dioxide (CO2), nitrous oxide (N2O), and methane (CH4) to increased soil moisture remains contested. We tested the hypothesis that increasingly intense rainfall events could affect greenhouse gas fluxes from an upland tropical forest soil by altering redox potential, using a field water addition experiment. We predicted that increased soil moisture would deplete soil oxygen (O2) and other terminal electron acceptors, decreasing CO2 and N2O fluxes while increasing CH4 fluxes. At our Puerto Rican field site, we measured soil-atmosphere fluxes of CO2, N2O, and CH4 from 18 replicate plots every 48 hours over a period of 25 days. We measured soil moisture and soil oxygen concentrations continuously and measured a broader suite of redox-active species (nitrate, manganese (II), and iron (II)) in soil extractions every eight days. A subset of plots received daily water additions over 24 days, or fluctuating water addition over eight-day intervals.
Results/Conclusions
Surprisingly, our watering treatments (60 mm/day) had little effect on soil moisture, which varied between 30 – 60 % among plots according to landscape position (scale of meters). Fluxes of CO2 and N2O displayed strongly significant negative relationships with soil moisture and soil oxygen concentrations, and 82% and 75% of the variance in these fluxes was explained by plot identity (random intercept). Concentrations of redox-active species co-varied according to thermodynamic predictions; oxygen and nitrate declined while reduced iron, manganese, and CH4 increased with soil moisture, with an apparent non-linear threshold response at > 55 % soil moisture. The first principal component of the correlation matrix of redox-active species explained 54 % of the total variance among these constituents. Using this principle component as an integrative multivariate measure of soil redox yielded significant relationships with CO2 and N2O fluxes, supporting our hypothesis that redox potential influences these fluxes in our upland soils. Several plots were consistent weak sources of CH4, but fluxes showed no relationships with moisture or redox. In sum, our data suggest that given high background soil moisture, increased short-term rainfall intensity (days to weeks) may have little additional impact on soil biogeochemical processes. Instead, subtle topographic variation appears to create constitutive differences in moisture that dramatically impact redox potential and greenhouse gas fluxes.