Dual water quality responses after more than 30 years of agricultural management practices in the rural headwaters of the Choptank River basin in the Chesapeake Bay watershed

Abstract

Eutrophication is the water quality response to over-enrichment by nitrogen (N) and phosphorus (P) in fresh, estuarine, and coastal waters globally. Agricultural best management practices (BMPs) are the primary tool for controlling eutrophication in rural areas, particularly in the Chesapeake Bay watershed, where BMPs are vital to achieving TMDL goals. However, despite the application of BMPs, local water quality in the headwaters of the Choptank River, a major tributary of the Chesapeake Bay on the Delmarva Peninsula, has not improved. Thus, further investigation of agricultural BMP impacts on water quality in the Greensboro watershed is needed. My overarching research question is, “Why have N and P concentrations increased at the USGS Greensboro gauge if agricultural Best Management Practices (BMPs) have been implemented?” I applied statistical approaches to three linked, testable hypotheses to systematically evaluate agricultural BMPs and their impacts on nutrient (N and P) export from the Greensboro watershed. My first hypothesis was that agricultural BMPs have increased significantly in the Greensboro watershed. To test this hypothesis, I obtained publicly available modeling data via the Chesapeake Assessment Scenario Tool (CAST) and estimated the subsequent edge-of-stream N and P export. My findings indicated that the number of BMPs in the agricultural sector increased significantly between 1985 and 2021, supporting the hypothesis. Overall, modeled agricultural N and P export significantly decreased between 2010 and 2021 (p < 0.001). However, the modeled edge-of-stream agricultural nutrient export resulted in no significant change in N export and an increase of 3% in agricultural P export resulting from BMP implementation levels in 2021 compared to 2010. This study demonstrated the use of CAST to acquire reported BMP implementation levels and increased nutrient inputs into the Greensboro watershed between 1985 and 2021. The watershed nutrient inputs mirror the upward trends in N and P export captured by the USGS long-term monitoring station at Greensboro. With this improved access to BMP implementation and nutrient data, decision-makers can consider adaptive management measures to decrease nutrient export downstream. My second hypothesis was that agricultural BMPs have an adequate basis for estimating their capacity to reduce N export. To test this hypothesis, I conducted a meta-analysis on 689 cover crop N efficiencies reported in 18 empirical and modeling studies. The cover crop N efficiency was calculated as the ratio of an N interception by cover crop biomass or a reduction in soil or groundwater N divided by an N input, e.g., previous spring fertilizer or a previous soil or groundwater N concentration or flux. These variable approaches resulted in wide ranges in mean cover crop N efficiency (10-80%) due to empirical and modeling experimental approaches, varying methods, and parameters used to calculate efficiency. The modeling approach generally resulted in N efficiency values significantly higher than the empirical approach, as did the parallel control-treatment experiments compared to the sequential before-and-after implementation method. Because of these variables, there appears to be no standard methodology to report the effects of cover crops or standardized metadata describing the variables used in the N efficiency calculations. I suggest a standard methodology and metadata that should accompany future reports of cover crop N efficiencies to improve the modeled effects of BMPs on nutrient export. My third hypothesis was that three methods of estimating N and P concentrations and yields are in agreement and show a relationship to BMP implementation in the Greensboro watershed. To test this hypothesis, I compiled annual nutrient (N and P) datasets based on (1) USGS field measurements of concentrations and discharge, (2) USGS flow-normalized weighted regression based on time, discharge, and season (WRTDS) of concentrations and yields, and (3) CAST-modeled nutrient yields. Statistical analyses revealed time, discharge, agricultural BMPs, and animal waste management practice trends of the three methods. Results indicated that the USGS field measurements and WRTDS flow-normalization methods consistently showed an increase in N and P concentrations and yields. In contrast, all CAST-modeled regressions showed significantly decreasing nutrient concentrations and yields (p ≤ 0.05), which did not support the hypothesis that all three methods are in agreement. Despite CAST-modeled results decreasing with increasing BMPs, which supports the hypothesis that N and P concentrations and yields show a relationship with BMP implementation, USGS methods resulted in increasing nutrient concentrations and trends. These results indicated significant underestimates of modeled N and P export by CAST. I recommend using adjusted BMP efficiencies during cultural and structural BMP lifespans to improve model outputs. I also suggest two approaches to reflect the role of annual poultry manure applications: (1) model nutrient transport via artificial drainage ditches that interfere with natural nutrient flow pathways and exacerbate N and P transport, and (2) model the accumulation of soil-P and saturated soil-P, resulting in increases in dissolved P and particulate P in downstream surface waters. Agronomic recommendations include developing efficient manure recycling approaches within the local agricultural systems via nutrient management practices and concurrent research and development to support alternative uses of animal waste, including composting, bioenergy generation, granulating/pelletizing, and establishing a marketplace to support the sale of these products and to offset the costs of transporting manure from areas of manure surplus to manure deficit areas. This dissertation revealed that modeling studies overestimate cover crop N efficiencies in the United States Coastal Plain province and that CAST modeling is not in agreement with the USGS field measurements. CAST-modeled nutrient concentrations and yields decrease over time, indicating improvements in water quality. In contrast, USGS methods consistently show that nutrient concentrations and yields increase, indicating that BMPs are insufficient, inadequate, overwhelmed by nutrient inputs, or efficiencies are overestimated. Indeed, nutrient-reducing BMPs have increased between 1985 and 2021. With over 35 years of BMP implementation, measurable water quality response is expected. However, BMPs that relocate and apply higher amounts of manure annually have also increased with nutrient-reducing BMPs. Rising manure application rates combined with higher fertilizer application rates due to economic pressures on farmers to increase crop yields appeared to have overwhelmed implemented BMPs. Continued manure applications onto croplands in the Greensboro watershed suggest nutrient export will continue to rise; thus, reaching water quality goals is unlikely.