Ecosystem-atmosphere carbon and water exchanges of subtropical evergreen and deciduous forests in India

Abstract

We studied the ecosystem-atmosphere carbon and water fluxes and their coupling at a broadleaf deciduous forest (Kaziranga National Park, KNP) and a coniferous evergreen forest (Kosi-Katarmal watershed, KKM) in subtropical India using the eddy covariance measurements. We analysed the data statistically in a bigleaf framework to compare the annual and seasonal carbon and water exchanges and their drivers at the two sites. The broadleaf deciduous forest at KNP varies from a weak annual carbon sink (−29 g C m−2 y-1) to a moderate source (+207 g C m−2 y-1), while the coniferous-dominated forest at KKM is a strong carbon sink (from −635 to −900 gC m−2 y-1). The component fluxes GPP and TER are much larger, and intra-annual variations are much stronger at KNP than at KKM. We found potential and actual evapotranspiration, as well as the evaporative fraction, to be higher at the broadleaf deciduous than at the coniferous forest. At both the sites, canopy-atmosphere decoupling coefficient is highest during the wet period and, as a whole, the broadleaf deciduous forest remains less tightly coupled to the atmosphere compared to its coniferous evergreen counterpart. At KNP, the biological carbon uptake primarily governs the ratio of intercellular to ambient CO2 concentrations (Ci/Ca), whereas at KKM it is the seasonality of surface conductance. The seasonal cycle of photosynthetic capacity (Vcmax25) is stronger at the deciduous forest than the coniferous forest, where photosynthetic capacity is retained at higher level during the winter. At both the sites, surface conductance and its intra-annual variability, due to environmental and physiological factors, and seasonal cycle of leaf-area, are the main factors controlling the seasonal patterns of surface-atmosphere coupling and ecosystem Ci/Ca. The Ci/Ca is found to decrease inversely with the square root of VPD at both the sites with the variation being similar in monsoon but widely divergent in winter. Our results will have direct implications in improving the parameterizations of the studied forest types in these models to improve their performances over this region.