Agricultural land use changes stream dissolved organic matter via altering soil inputs to streams.

Agricultural land use leads to significant changes in both the quality (e.g., sources and compositions) and quantity of dissolved organic matter (DOM) exported from terrestrial to aquatic ecosystems. However, the effect of agricultural activities often interacts with those of hydroclimatic drivers, making it difficult to delineate agriculture-induced changes and identify associated mechanisms. Using partial least square path modeling (PLS-PM), we examined the relative importance of agricultural land use, stream order, precipitation, and temperature in mediating allochthonous versus autochthonous sources and pathways that influenced stream DOM quality and quantity. We analyzed stream water DOM from 15 small streams draining watersheds across a gradient of agricultural land use in Southeast USA for about one year. For DOM quantity, agricultural land use increased the export of DOC and various DOM pools (terrestrial humic, microbial humic, and protein-like DOM) from land to streams, and for DOM quality, agricultural streams showed greater proportions of microbial humic compounds than forested streams. The PLS-PM model for DOM quantity accounted for 75.5% of total variance and identified that agricultural land use increased stream water DOM quantity primarily through increasing allochthonous inputs, which can be attributed to shallower flow paths in agricultural watersheds that enabled the export of organic materials from the upper, organic-rich soil horizon. PLS-PM models for DOM quality only explained ~13% of the total variance, highlighting the complex dynamics between environmental drivers and stream water DOM. Relative to commonly used multivariate statistic modeling (e.g., redundancy analysis (RDA)), PLS-PM models offer the advantages of identifying the primary pathway by which agricultural lands alter freshwater DOM and quantifying the relative importance of interactive effects of agriculture and hydroclimatic drivers. Therefore, structural equation modeling is a powerful tool that should be more widely adopted to distinguish among multiple drivers and mechanisms regulating freshwater biogeochemistry.