Do Historical Landscape Patterns Help Explain Persistent Groundwater Nitrate Concentrations in a Cross-Border Aquifer?

The effects of agricultural production on nitrate contamination of groundwater is a pressing global concern and agricultural best management practices (BMPs) are often implemented as a means to help alleviate this problem. In Western North America, the Abbotsford-Sumas Aquifer spans the US-Canada border and provides drinking water for over 100,000 people. Intensive agriculture combined with high precipitation and well-drained soils make this aquifer susceptible to nitrate contamination. Long-term studies indicate elevated nitrate concentrations may be influenced by overlying land use; thus, in recent decades, various agricultural BMPs and stewardship programs have been implemented in the region to help reduce potential nitrate sources to the aquifer. Despite these improved nutrient management practices, nitrate concentrations have remained relatively high.

To explore how time lags associated with surface inputs might explain this pattern, we ask two questions: (1) How have agricultural landscape patterns changed historically? (2) Do historical agricultural patterns explain continued elevated nitrate concentrations? Responses of nitrate concentrations in deep wells (with screens > 15 m below the water table) were contrasted with shallow wells (screens < 15 m) in the US and Canada. A seamless cross-border mosaic of land cover was created by harmonizing a variety of US and Canadian land use and land cover data. Surrounding each well, terrestrial zones of influence (aligned with the directional flow of groundwater) were delineated then historical and contemporary landscape patterns were characterized within these zones. To link landscape patterns with nitrate, multiple regression was used to compare the strength of relationships between historical land uses and mean nitrate concentrations from both deep and shallow wells. Because previous research showed that contemporary land cover was significantly correlated with shallow well nitrate concentrations, we hypothesize that historical land use may show a stronger correlation due to time-lags associated with historical surface inputs.