Molecular-weight-dependent responses of DOM to hydrological seasonality and nutrient ratios

Dissolved organic matter (DOM) plays a central role in aquatic biogeochemical processes, yet seasonal differences in its carbon concentration, fluorescence composition, and environmental associations among operational size fractions remain insufficiently characterized. Here, water samples collected during dry and wet seasons were fractionated into three cumulative filtrates (<0.45 μm, <10 kDa, and <1 kDa). DOC measurements, UV-vis spectroscopy, excitation-emission matrix fluorescence coupled with parallel factor analysis (PARAFAC), and multivariate analyses were combined to characterize seasonal variation and component-specific environmental associations. DOC concentrations were higher in the wet season in the <0.45 μm and <10 kDa permeates but similar between seasons in the <1 kDa permeate. In the smallest filtrate, similar DOC concentrations contrasted with differences in fluorescence composition: the dry-season PARAFAC model was dominated by tryptophan-like fluorescence, whereas the wet-season model was dominated by humic-like fluorescence. PARAFAC models fitted to pooled seasonal data resolved four, three, and two components in the <0.45 μm, <10 kDa, and <1 kDa permeates, respectively, with both protein-like and humic-like fluorescence were identified in the model for each fraction. Environmental associations differed among filtrates and within broad fluorescence classes. Notably, two <10 kDa protein-like components shared an emission maximum at 340 nm but exhibited opposing associations with the physicochemical and nutrient constructs in partial least squares path modeling. Overall, operational fractionation revealed filtrate-specific seasonal DOC patterns not captured by bulk measurements alone, while component-resolved fluorescence analysis exposed environmental associations obscured by broad fluorescence classifications. These findings highlight the complementary roles of DOC concentration measurements and fluorescence characterization in interpreting aquatic DOM dynamics.