δ15N of detritivores track nitrogen inputs from agricultural land into shallow Minnesota lakes

Background/Question/Methods

Eutrophication of freshwater ecosystems continues to be a major issue, and nonpoint sources of nutrients are particularly difficult to quantify and alleviate. Agricultural activities can be a major source of nutrients for lakes and could be especially influential on shallow lakes in the prairie region of North America due to high watershed area: lake surface area ratios. Prairie lakes are naturally productive and nutrient levels are dynamic through time making it difficult to estimate influences of agriculture on these lakes.  Enriched δ ¹⁵N values in lake biota can reflect the influence of agriculture on N-cycling in lakes, owing to fractionation associated with high rates of N inputs to the landscape. We assessed agricultural effects on shallow lakes in 2009 by measuring δ¹⁵N values of detritivorous fathead minnows (Pimephales promelas) collected from 33 lakes from three prairie-parkland ecoregions of Minnesota, USA.  We used an information-theoretic approach to assess ability of competing models to predict δ ¹⁵N in the fish; models were based on landscape-level effects (ecoregion), watershed features (e.g. watershed size, extent of row-crop agriculture and urbanization), and within-lake characteristics (e.g. lake volume, abundance of submerged macrophytes, and turbid versus clear-water states).  

Results/Conclusions

Results show little within-lake variation, but substantial variation between lakes, with lake mean δ¹⁵N values ranging from 5 to 19‰.  Results of model selection indicated the most parsimonious model for predicting δ¹⁵N in fish was based on just the total amount of agriculture in the watershed (summed area of hay, row crops, and pasture); the relationship was positive and agriculture explained 48% of the variation in δ¹⁵N among lakes.  Model parsimony was not improved by partitioning total agriculture into row-crop agriculture versus pasture land, and we found little support for δ¹⁵N being influenced by other watershed features, ecoregion effects, or within-lake characteristics.  Our results indicate agriculture in lake watersheds of shallow prairie-parkland lakes results in enriched δ¹⁵N in fish, suggesting δ¹⁵N may be a useful index of agricultural influence on these lakes.  An unanswered question is the degree to which the large enrichment of δ¹⁵N in fish in these systems is driven by N-cycling in watershed soils versus within the lake ecosystem.