Pesticide residues from agricultural applications may accumulate in wetlands, presenting a risk to wetland biota. In Alberta, at the northern limits of the fertile prairie pothole region, sales of herbicides nearly doubled between 2003 and 2013. Simultaneously, concern regarding contamination of prairie wetlands with neonic insecticides has spiked. Yet no provincial monitoring program exists to track risks to wetlands receiving agricultural run-off and data on the extent of pesticide contamination is scarce. In 2014 and 2015, we undertook an unprecedented sampling effort to determine: 1) which pesticide residues were most common in Albertan wetland water and soil in relation to pesticide sales data; 2) how the incidence of pesticide contamination related to surrounding agricultural land use (based on characterization of a 500 m radius buffer around each wetland); and 3) whether pesticide detections differ between a “normal” climate year (2014) and a year with limited snow-melt (2015). Snowmelt run-off is the main source of water to these geographically isolated wetlands, and thus expected to be an important vector for pesticides from agricultural lands into wetlands. We sampled water and soil in 96 wetlands across the northern prairie pothole region, spanning a gradient in the extent of agriculture from 0% to 100%. Samples were analysed for >100 pesticides using GC-MS/MS, including popular herbicides as well as de-registered products. 2014 water samples were also analysed for the herbicide glyphosate and its primary breakdown product Aminomethyl Phosphonic Acid by HPLC-MS/MS and for clothianidin, imidacloprid and thiamethoxam by GC-MS.
Results/Conclusions
In agreement with recent sales data, the most frequently detected pesticides were glyphosate, MCPA and 2,4-D. Sediment samples, however, revealed several “legacy” pesticides, most commonly breakdown products of DDT, t- and cis-Chlordane, and Dieldrin. The extent of surrounding agricultural activity was a significant predictor of pesticide contamination in water [Chi-square = 25. 42, p < 0.00001, water sample detection = -2.350 + 0.036*(% Cropland in 500 m buffer)]. Repeating this analysis for 2014 and 2015 separately yielded a similar but weaker logit function in 2015 (Chi-square 2014: 16.675, p < 0.00001; Chi-square 2015: 7.924, p = 0.005). This supports the theory that the major conveyance of pesticides into the wetlands is by surface runoff from surrounding land. In dry years, there is less surface run off, and thus less connection between surrounding land use and the probability of detecting one or more pesticides in the wetland’s water.