Mercury cycling in aquatic ecosystems and trophic state-related variables — Implications from structural equation modeling

Abstract

Structural equation modeling (SEM) provides a framework that can more properly handle complex variable interactions inherent in mercury cycling and its bioaccumulation compared to more traditional regression-based methods. SEM was applied to regional data sets for three different types of aquatic ecosystems within Florida, USA – lakes, streams, and the Everglades – to evaluate the underlying nature (i.e., indirect and direct) of the relationships between fish mercury concentrations and trophic state related variables such as nutrients, dissolved organic carbon (DOC), sulfate, and alkalinity. The modeling results indicated some differences in key variable relationships — for example, the effect of nutrients on fish mercury in lakes and streams was uniformly negative through direct and indirect pathways consistent with biodilution or eutrophication-associated effects on food web structure. Somewhat surprisingly, however, was that total phosphorus did not serve as a meaningful variable in the Everglades model, apparently because its effects were masked or secondary to the effects of DOC. What is perhaps a more important result were two key similarities across the three systems. First, the modeling clearly indicates that the dominant influence on fish tissue mercury concentrations in all three systems is related to variations in the methylmercury signal. Second, the modeling demonstrated that the effect of DOC on fish mercury concentrations was exerted through multiple and antagonistic pathways, including facilitated transport of total mercury and methylmercury, enhanced rates of methylation, and limitations imposed on bioavailability. Indeed, while the individual DOC pathways in the models were all highly significant (generally p<0.001), the net effect of DOC in each model was greatly reduced or insignificant. These results can help explain contradictory results obtained previously by other researchers in other systems, and illustrate the importance of SEM as a modeling tool when studying systems with complex interactions such as the aquatic mercury cycle.