Background/Question/Methods: Terrestrial ecosystem carbon (C) storage in soils and vegetation exceeds that in the atmosphere by a factor of four, and represents a dynamic balance among C input, allocation, and loss. This balance is likely being altered by climate change, but the response of terrestrial C cycling to rising temperatures remains poorly quantified. Importantly, temperature-induced changes in ecosystem C flux and storage have the potential to feedback into atmospheric CO₂ levels and global climate. In this study we examined how tropical forest ecosystems will respond to rising temperature by examining ecosystem C input (gross primary production; GPP), C fluxes (aboveground litterfall and soil-surface CO₂ efflux) and C partitioning (fraction of GPP partitioned to aboveground vs. belowground) along a 5°C mean annual temperature (MAT) gradient on the Island of Hawaii. Along the MAT gradient, substrate age and type, dominant overstory vegetation, and plant available water are constant, allowing us to isolate the impacts of temperature on ecosystem C cycling. We used a state-of-the-science approach to select a network of nine permanent plots along a 5°C MAT gradient (13-18°C) with LiDAR-based information (1.2 m resolution) on forest structure.  Monthly litterfall collections were used to estimate GPP along the gradient from global relationships between litterfall and GPP (Litton et al. 2007). Soil-surface CO₂ efflux (‘soil respiration’) was measured monthly and, along with litterfall rates and a mass-balance approach, used to estimate total belowground C flux (TBCF). Finally, C partitioning to belowground (the fraction of GPP partitioned to belowground) was estimated as TBCF/GPP.

Results/Conclusions: Aboveground litterfall and GPP increased linearly and positively with temperature (r²=0.61), with GPP ranging from ~2,000 to 4,500 g C m^-2 yr^-1 at 13 and 18°C MAT, respectively. Soil-surface CO₂ efflux (r²=0.45) and TBCF (r²=0.35) also increased linearly and positively with temperature, increasing by ~100 and 80 g C m^-2 yr^-1 for every 1°C increase in MAT, respectively. The fraction of GPP that was partitioned belowground decreased linearly with increasing temperature (r²=0.46), which was opposite the pattern expected from a recent global analysis (Litton and Giardina 2008). However, belowground partitioning increased linearly and positively with GPP (r²=0.82), suggesting that changes in C partitioning with rising temperature are a result of alterations in belowground resource availability. The results of this study have important implications for understanding how tropical forest ecosystem C input, flux, partitioning, and, ultimately, storage will respond to rising temperatures in the coming decades.