Impacts of Best Management Practices on Nitrogen Discharge From a Virginia Coastal Plain Watershed

Abstract

Long-term watershed and field nitrogen (N) balances were used in this study to quantify the surface (baseflow) and ground water lag times and effects of BMPs on N discharge from a Virginia Coastal Plain watershed. Ten-year water quantity/quality data (1986-1996) collected at the Nomini Creek (NC) watershed were used. Field (Field-N) and watershed (Watershed-N) scale N models were developed for computing the N balances. BMPs evaluated in this study included no-till corn and split N application. The role of atmospheric N (atm-N) deposition (dry+wet) in masking the effects of BMPs on watershed N loading was also investigated. Nitrogen retention and discharge from the forest areas in the NC watershed were simulated using the 5-year water and N input and output data from forested subwatersheds. Field and watershed N balances (WNBAL) were used to evaluate the effects of BMPs on measured surface and ground water N in the NC watershed. A 6-month laboratory study was conducted to develop N mineralization (Nmin) models for agricultural, forest, and fallow soils in the NC watershed. Mineralization potential (N0) and rate constants (k) for surface and subsurface soils from agricultural, forest, and fallow soils were estimated by fitting the laboratory measured data to a first-order model, using the nonlinear regression procedure. A large variability (300%, 163 - 471 kg/ha) in N0 of agricultural surface soils was observed. On average, forest soils had much higher potentially mineralizable N than agricultural soils. The first-order model was incorporated into the Field-N model to predict daily Nmin using the measured N0 and k and daily values of soil water and temperature. Atmospheric deposition was a major source of N in the NC watershed, accounting for 23% of the total N input. Variation in atm-N deposition during the 10-year period was from 10 to 42 kg/ha (average = 25 kg/ha); much larger than the variation in fertilizer N (37 to 51 kg/ha). Atm-N deposition was found to be a controlling factor affecting surface water DIN (dissolved inorganic N) and TDN (total dissolved N) loading in the NC watershed; an indication that atm-N deposition is a masking factor in the BMP impact evaluation. Large uncertainty in atm-N deposition existed due to uncertainty involved in quantifying dry N deposition. Forested areas of the NC watershed retained 77% of the atm-N deposition. Forest area N discharge was simulated using the 77% retention and annual atmospheric deposition. Comparison of Field-N predicted N balance and leaching (steady-state and transient conditions) with observed ground water NO3 concentration revealed that the ground water lag time ranged from 2 to 8 months. Unusually rapid transport of solute in the watershed was facilitated by the network of discontinuous clay lenses. Based on the lag time, the pre-BMP (1986-1990) and post-BMP (1991-1995) periods were defined. Results from Field-N indicated that implementation of split fertilizer N on corn reduced the post-BMP ground water NO3 concentration by 10-12% at two of the four ground water monitoring sites. The split N application reduced the frequency of detection of high NO3 (> 9 mg/l) concentration by 44% during the post-BMP period. Considerably large uncertainty existed in evaluating the effects of BMPs on ground water NO3 due to N contributions from neighboring agricultural and forest areas. Effects of no-till corn could not be evaluated since this BMP was already implemented at the sites prior to the beginning of the study. Results of statistical trend analysis of the ground water N supported the modeling results. Watershed-N model was able to accurately predict the effects of land use activities on watershed N balances (WNBAL) and baseflow and ground water N. A one-to-one relationship between the WNBAL and observed N loading and concentration time series was observed. Comparison of WNBAL and measured baseflow N revealed that the baseflow lag time or residence time was between 4-11 months. Multivariate regression models were developed to predict baseflow N using Watershed-N results. The multivariate model predicted the N loading and concentration exceptionally well (R2 > 90%). Corn N input and output and acreage was an important predictor of ground water N and baseflow N loading and concentration. Post-BMP WNBAL was considerably less than the WNBAL for the pre-BMP period. However, these reductions were mainly due to the 43% reductions in atm-N deposition and 31% increase in the plant uptake during the post-BMP period. Reductions in WNBAL caused by BMPs were only 5%. Reductions in N loading caused by BMPs were 10%. Statistical trend analysis of monitoring and modeling results indicated significant post-BMP reductions in WNBAL and DIN and TDN loading. However, poor to moderate evidence was available to suggest that BMPs caused a significant reductions in WNBAL and N loading. Marginal effects of BMPs could mainly be attributed to insufficient BMP implementation. Watershed-N was used to evaluate N reduction scenarios and to design BMPs. Irrigating corn was one of the best BMPs, as it could reduce N loading from NC watershed by 50%. Quantification of lag time and long-term watershed N balances from this study provide crucial information for understanding N cycling and factors controlling N discharges which is essential for designing programs for controlling N discharges from Mid-Atlantic Coastal Plain watersheds.