Distinct variations in fluorescent DOM components along a trophic gradient in the lower Fox River-Green Bay as characterized using one-sample PARAFAC approach

Abstract

Variations in molecular weight distributions of dissolved organic matter (DOM) and PARAFAC-derived fluorescent components were investigated along a transect in the seasonally hypereutrophic lower Fox River-Green Bay using the one-sample PARAFAC approach coupling flow field-flow fractionation for size-separation with fluorescence excitation-emission matrix (EEM) and PARAFAC analysis. Concentrations of dissolved organic carbon and nitrogen, chromophoric-DOM, specific UV absorbance at 254 nm, and humification index all decreased monotonically from river to open bay, showing a strong river-dominated DOM source and a dynamic change in DOM quality along the river-lake transect. The relative abundance of colloidal DOM (>1 kDa) derived from ultrafiltration exhibited minimal variation, averaging 71 ± 4 % of the bulk DOM, across the entire estuarine transect although the colloidal concentration decreased in general. Using the one-sample EEM-PARAFAC approach, the identified major fluorescent components were distinct between stations along the river-estuary-open bay continuum, with four components in river/upper-estuary but three components in open bay waters. Among the four common fluorescent components (C475, C410, C320 and C290), the most abundant and refractory humic-like component, C475, behaved conservatively and its relative abundance (%ΣFmax) remained fairly constant (50 ± 4 %) along the transect, while the semi-labile humic-like component, C410, consistently decreased from river to estuary and eventually vanished in open Green Bay. In contrast, the two autochthonous protein-like components (C320 and C290) increased from river to open bay along the trophic gradient. The new results presented here provide an improved understanding of the diverse and fluctuating characteristics in DOM composition, lability, and estuarine mixing behavior across the river-lake interface and demonstrate the efficacy of the one-sample PARAFAC approach.