Comparing optical versus chromatographic descriptors of dissolved organic matter (DOM) for tracking the non-point sources in rural watersheds

[Display omitted] •Reasonable comparison of two different DOM descriptors for source discrimination.•EMMA was applied with various optical and chromatographic descriptors.•Optical tracers were more successful for agricultural watersheds.•HS aromaticity and C/N ratio are applicable tracers for DOM with depleted aromatics. Identification of the major dissolved organic matter (DOM) sources in watersheds is critical for an efficient management to control the non-point sources associated with DOM and for an understanding of the local carbon cycle. In this study, several DOM indicators, which are derived from fluorescence spectroscopy and size exclusion chromatography, were compared for their applicability to track DOM sources in three forested-agricultural watersheds. Several potential source end-members were collected from upper catchments, which included fallen leaves, riparian reeds, riparian weeds, paddy soils, field soil, riverside soil, cow manure, swine manure, and poultry manure, and the values of their representative DOM descriptors were estimated along with those of the stream samples during high flow periods. The relative source contributions to streams were estimated via an end member mixing analysis using the paired descriptors that were selected based on the criteria in terms of the value ranges and the significant differences in stream samples between the high and low flow periods. Biological index (BIX) and the ratio of the terrestrial humic-like to the humic-like component were found to be optimum optical tracers, while the aromaticity and the C/N ratio of humic substance fraction were an applicable pair of chromatographic tracers. The optical tracers were superior to the latter because of more consistent and less variability in the assigned sources between the different streams in the same watersheds. The optical tracer-based estimations revealed that the major DOM sources to the high flow stream samples were soil (22.3%–58.5%) and manures (12.9%–51.3%). This comparative study provides new insights into the practical feasibility and the overlooked limitations of the common DOM descriptors for source discrimination.