One of the important feedback pathways in the carbon cycle is the vegetation - atmosphere interactions as part of the terrestrial ecosystem. Understanding this interaction is important not only because of its magnitude but also because of the uncertainty and the variability associated with the carbon source / sink values. In this study, we hypothesize that the potential of the vegetated land surface to be a source or sink for carbon will depend on the basal vegetation characteristics (such as photosynthesis pathways), and environmental feedbacks from hydrological and radiative effects. To explore this hypothesis, we present the first results based on CO2 and heat flux observations taken from an AmeriFlux site in Oak Ridge, TN under pre-selected hydrological and radiative conditions. The surface flux time series are initially categorized under two groups corresponding to: 'drought' and 'no drought' periods (as indicated by precipitation anomaly and drought indices). These groups are further categorized based on aerosol loading over the study site (as based on aerosol air mass trajectories and observed regional diffuse to direct radiation ratios [DDR] over Mt. Mitchell site). The categories include 'polluted', 'continental' and 'maritime' corresponding to high, medium, and low aerosol loading values. With higher aerosol loading the DDR increases. Changes in DDR cause differential responses in the ability of the vegetated land surface to assimilate carbon, and respire; thus having a variable response for its potential ability for source / sink of terrestrial carbon. The drought effect is also interactive in this cycle, with higher water deficit causing changes in both the basal characteristics associated with the maximum photosynthetic assimilation capacity as well as the short terms response in the plants ability to assimilate CO2 due to environmental changes. We will present results from these observations and discuss the potential cause - effect pathways affecting the carbon source / sink variability particularly in the southeastern US. We will also outline a follow-up strategy to isolate the feedback pathways using coupled modeling studies using a factor separation approach.