Chemodiversity of riverine dissolved organic matter: Effects of local environments and watershed characteristics

Abstract

Riverine dissolved organic matter (DOM) is crucial to global carbon cycling and aquatic ecosystems. However, the geographical patterns and environmental drivers of DOM chemodiversity remain elusive especially in the waters and sediments of continental rivers. Here, we systematically analyzed DOM molecular diversity and composition in surface waters and sediments across 97 broadly distributed rivers using data from the Worldwide Hydrobiogeochemistry Observation Network for Dynamic River Systems (WHONDRS) consortium. We further examined the associations of molecular richness and composition with geographical, climatic, physicochemical variables, as well as the watershed characteristics. We found that molecular richness significantly decreased toward higher latitudes, but only in sediments (r = -0.24, p < 0.001). The environmental variables like precipitation and non-purgeable organic carbon showed strong associations with DOM molecular richness and composition. Interestingly, we identified that less-documented factors like watershed characteristics were also related to DOM molecular richness and composition. For instance, DOM molecular richness was positively correlated with the soil sand fraction for waters, while with the percentage of forest for sediments. Importantly, the effects of watershed characteristics on DOM molecular richness and composition were generally stronger in waters than sediments. This phenomenon was further supported by the fact that 11 out of 13 watershed characteristics (e.g., the percentages of impervious area and cropland) showed more positive than negative correlations with molecular abundance especially in waters. As the percentage of forest increased, there was a continuous accumulation of the compounds with higher molecular weight, aromaticity, and degree of unsaturation. In contrast, human activities accumulated the compounds with lower molecular weight and oxygenation, and higher bioavailability. Our findings imply that it may be possible to use a small set of broadly available data types to predict DOM molecular richness and composition across diverse river systems. Elucidation of mechanisms underlying these relationships will provide further enhancements to such predictions, especially when extrapolating to unsampled systems.