Characterization of dissolved organic matter (DOM) for a mid-Atlantic Piedmont forested watershed using multivariate analyses

Understanding the biotic and abiotic mechanisms of DOM solubilization and immobilization in forested ecosystem have important implications to comprehend present and future carbon budgets. In this study, DOM concentrations and compositions were investigated for a mid-Atlantic Piedmont forested watershed using excitation-emission matrices (EEMs) fluorescence combined with parallel factor analysis (PARAFAC). Samples were collected across wide sources including throughfall, litter, soils , wetlands, streams, hyproheic, seeps, shallow, riparian, and deep groundwaters, over a two year period (2008-2009) during baseflow conditions. A site specific PARAFAC model was developed to evaluate DOM characteristics, resulted in six diverse DOM compositions, which include, two terrestrial humic-like (component 1 and 4), one soil derived humic-like (component 3), one non-humic like (component 2), and two protein-like (component 5 and 6) components. Component 1 and 3 showed a decreasing trend from surface (litter, throughfall, wetlands), with an exception in soils, to subsurface layers with a high variability in riparian, shallow, and deep ground waters in addition to seep locations . Component 2, an indicator of biologically labile organic matter, showed a larger variability in subsurface sources, with a highest value at seep locations, in comaprison to surface sources. The observed median values for component 4 (visible-humic-like,) was highest in soils (0.14 RU) and lowest in deep groundwaters (0.08 RU). In contrast to component 1 and 3, component 5 and 6 (indicator of protein-like moieties) were significantly higher in subsurface sources (riparian, seep, and deep groundwaters). Median values for components 5 ranged from 0.09 to 0.33 RU, whereas the same for component 6 ranged from 0.02 to 0.76 RU (litter and deep groundwaters, respectively). Other optical DOM metrics were generated including a254, SUVA254, UV253/203, HIX, FI, alongwith DOC concentrations, across all watershed compartments. HIX and a254 showed a decreasing trend from surface layers (litter) to deep groundwaters (344.7 to 2.9 m-1, and 0.9 to 0.4, respectively) and were similar to the same observed for DOC (25.9 to 1.2 mgC/L). SUVA254 and FI did not show any trend with the soil depth. Further, component 1 was found positively correlated with a254, HIX, and DOC (r = 0.7, 0.6, and 0.6, respectively) and negatively correlated with FI (r = - 0.7) at wetlands. In comaprison, component 6 was found positively correlated with FI (r = 0.8) and was negatively correlated with HIX (r = - 0.9) for shallow groundwaters. DOC concentrations were observed significantly positive for component 1 (r = 0.8), while were negative for component 6 (r = - 0.6) at shallow groundwaters. In addtion, a model comparison is done between our site specific model and a global model (Cory & McKnight, 2005). Results obtained suggest a site specific model would reveal further insights for protein-like DOM compositions and enhance our understanding for future implications of DOM variability in forested ecosystems.