The magnitude and variability of soil-surface CO2 efflux increase with mean annual temperature in Hawaiian tropical montane wet forests

Soil-surface CO ₂ efflux (F _S; 'soil respiration') accounts for ≥50% of the CO ₂ released annually by the terrestrial biosphere to the atmosphere, and the magnitude and variability of this flux are likely to be sensitive to climate change. We measured F _S in nine permanent plots along a 5.2 °C mean annual temperature (MAT) gradient (13-18.2 °C) in Hawaiian tropical montane wet forests where substrate type and age, soil type, soil water balance, disturbance history, and canopy vegetation are constant. The objectives of this study were to quantify how the (i) magnitude, (ii) plot-level spatial variability, and (iii) plot-level diel variability of F _S vary with MAT. To address the first objective, annual F _S budgets were constructed by measuring instantaneous F _S monthly in all plots for one year. For the second objective, we compared plot-level mean instantaneous F _S in six plots derived from 8 versus 16 measurements, and conducted a power analysis to determine adequate sample sizes. For the third objective, we measured instantaneous F _S hourly for 24 h in three plots (cool, intermediate and warm MATs). The magnitude of annual F _S and the spatial variability of plot-level instantaneous F _S increased linearly with MAT, likely due to concomitant increases in stand productivity. Mean plot-level instantaneous F _S from 8 versus 16 measurements per plot yielded statistically similar patterns. The number of samples required to estimate plot-level instantaneous F _S within 10% and 20% of the actual mean increased with MAT. In two of three plots examined, diel variability in instantaneous F _S was significantly correlated with soil temperature but minimal diel fluctuations in soil temperature (<0.6 °C) resulted in minimal diel variability in F _S. Our results suggest that as MAT increases in tropical montane wet forests, F _S will increase and become more spatially variable if ecosystem characteristics and functioning undergo concurrent changes as measured along this gradient. However, diel variation in F _S will remain a minor component of overall plot-level variation.