Analysis of the manage drain concentration database to evaluate agricultural management effects on drainage water nutrient concentrations

Abstract

Agricultural systems are substantial contributors of nonpoint-source nitrogen (N) and phosphorus (P) pollution, and loss of dissolved forms of these nutrients is exacerbated in subsurface-drained (tile-drained) landscapes. The majority of reviews summarizing drainage nutrient losses have focused on N and P loads, but closer inspection of drainage concentrations is necessary to more directly link cropping management factors with water quality outcomes. More than 1,500 recently compiled site-years of drainage N and P concentration in the Measured Annual loads from AGricultural Environments (MANAGE) Drain Concentration database were used to analyze the impacts of crop selelction, nutrient management, and tillage type on annual drainage nutrient concentrations. The highest annual flow-weighted mean NO3-N concentrations across the database were from corn, corn and soybean (grown within the same plot in the same year), and soybean site-years (14.0, 13.5, and 12.1 mg L-1, respectively). However, crop selection was not a significant predictor for annual average dissolved reactive phosphorus (DRP) concentrations in drainage. Nitrogen application rates below 75 kg ha-1 for corn did not significantly reduce annual NO3-N concentrations compared to rates of 75 to 149 kg ha-1 or 150 to 224 kg ha-1, although the three largest application rate categories (75 to 149 kg ha-1, 150 to 224 kg ha-1, and >224 kg ha-1, respectively) resulted in significantly increasing NO3-N concentrations. The stepwise regression approach was used to reduce and select predictors to model annual NO3-N and DRP concentrations. Regression analysis of NO3-N concentrations had an overall model R2 of 0.59 (n = 254) and indicated that N application rate had the greatest effect on NO3-N concentrations in corn site-years, followed by fertilizer timing and tillage type. Regression analysis of DRP concentrations had an overall R2 of 0.94, and although the model was less robust due to the small sample size (n = 47), fertilizer timing was most closely correlated with annual DRP concentrations. The MANAGE database will continue to evolve and remain a resource for new exploratory efforts to better understand and reduce nutrient losses from agricultural systems.