Phosphorus Loss Risk Assessment for Alabama Soils: A Step Towards Protecting Alabama Water Resources

Abstract

Agricultural phosphorus (P) loss is a key driver of eutrophication, threatening water quality worldwide. Effective management strategies and robust assessment tools are essential to mitigate its impact on water quality. This dissertation presented a comprehensive study on P losses across various scales, from plot to field, within agricultural systems, and evaluated P loss risk assessment tools, including the soil test P method (STP), phosphorus saturation ratio (PSR), and the phosphorus index (P-index) for Alabama soils. The first study evaluated the relationship between P concentration in runoff and STP concentrations determined by different methods (water-soluble P, Mehlich-1 (M1), and Mehlich-3 (M3)) using artificial rainfall simulation (RS) and also quantified the differences in P loads in runoff from soils having distinct STP levels (low, high and extremely high). Mehlich-3 demonstrated a stronger correlation than M1 with both dissolved reactive P (DRP) and total P (TP) in runoff, indicating its greater reliability for assessing P loss risk in Coastal Plain soils. Among the four study sites, the cumulative DRP and TP loadings from site 2 that tested high in STP ranged from 118.4 and 232.3 g ha-1, respectively for both 1st and 2nd RS and were significantly higher than the other three sites (range: 3.48 – 37.5 g ha-1 for DRP and 6.03 – 83.1 g ha-1 for TP). No statistical differences were found between the 1st and 2nd RS for both DRP and TP loads, indicating that rainfall events of similar intensity occurring 24 h apart produced the same amount of P loss in runoff. The second study investigated the temporal and spatial distributions of various forms of P loss in runoff, including DRP, total particulate P (TPP), dissolved organic P (DOP), and TP, and identified critical periods of P loss from Edge-of-Field (EOF) monitoring experiments. The EOF runoff studies were conducted at three farms located in North, South, and Central Alabama and consisted of two adjacent watersheds. The event mean concentration (EMC) and loads were calculated for individual rainfall-runoff events. The EMC in runoff ranged between <0.01 to 8.27 mg L⁻¹ for DRP, 0.03 to 11.49 mg L-1 for TPP, <0.01 to 1.98 mg L-1 for DOP and 0.18 – 15.23 mg L-1 for TP. Across all watersheds, the loads for DRP ranged from <0.001 to 0.84 kg ha-1, TPP ranged from <0.001 to 1.50 kg ha-1, DOP ranged from <0.001 to 0.39 kg ha-1 and TP ranged from <0.001 and 1.99 kg ha-1. For overall events, the DRP constituted <0.1 to 92%, TPP constituted 2 to 96% and DOP constituted 2 to 51% of TP loads. The months from November to March were found critical periods for P losses. The dominant form of P in total loads was influenced by management practices such as manure application and highlighted the need for adjustment of farm operations such as manure management, cover crop planting date and maintaining enough ground cover during the winter months to minimize P loss. The third study evaluated the relationship between water-soluble P (WSP), often used as a surrogate for water quality, and soil test matrices (M1, M3, and P saturation ratio using Melich-3 (PSRM3)) for non-calcareous Alabama soils (Appalachian Plateau, Coastal Plain, Limestone Valley, and Piedmont Plateau). The relationship showed that WSP was strongly correlated (r > 0.68 to 0.91) with PSRM3 than STP methods for all four soil types and at all three sampling depths (0 – 5, 5 – 15, and 15 – 30 cm). This indicated that PSRM3 is a more reliable indicator of environmental P loss than STP methods and accounts for P sorption characteristics (primary drivers of P sorption and desorption are Fe and Al content of the soil) of soils. Further, this study aimed to establish threshold PSRM3 for four soil regions and to develop and validate predictive models for PSRM3 using M3-P data. The threshold PSRM3 determined using a segmented regression model was found to be 0.07 for Appalachian Plateau, 0.05 for Coastal Plain, 0.12 for limestone Valley and 0.05 for Piedmont Plateau. Despite apparent numerical differences in threshold PSRM3, the overlapping 95% confidence interval indicated no statistical difference among soil types. The spatiotemporal analysis showed a distinct region-specific PSRM3 response to P management. The prediction models were significant and strongly correlated for all soils and depths (0 – 5, and 5 – 15 cm). The evaluation of predicted vs observed data demonstrated a strong agreement between predicted PSRM3 and observed PSRM3 with r2 values of >0.94 for Limestone Valley soils and >0.62 for Coastal Plain soils. The fourth study evaluated the Alabama P-index using a sensitivity analysis approach to identify the most sensitive parameters of P-index and prioritize parameters for future research. Stochastic sensitivity analysis was performed using the Monte Carlo simulations, with and without weighted factors for all soil regions of Alabama. Across all major soil regions, P application distance to water was found the most sensitive factor, explaining greater than 27% of the variance in Alabama P-index score. The other sensitive parameters with weighted factors followed the order: soil erosion rate > hydrologic soil group > P application rate > P application method > vegetative buffer width > critical habitat waters. In contrast, Alabama P-index output was found to be insensitive to soil test P, field slope, grazing animals, and underground outlet system. The transport factors within the Alabama P-index altogether accounted for 73% of the explained variance, and 22% was explained by the source factors. Comparison of sensitivity analysis results with and without the weighted factors showed larger, albeit inconsistent, variability, pointing to the need for further validation of weightage factors used in the Alabama P-index. The fifth study evaluated the Alabama P-index using EOF monitoring data and explored the applicability of multiplicative (Tennessee) and component-based (Georgia) P-indices. In addition, this study tested the performance of a modified version of Alabama P-index. Results showed that the relationships between annual dissolved reactive P (DRP), total particulate P (TPP), and total P (TP) loads with the Alabama P-index scores were weakly correlated. The poor alignment of P-index scores with actual P loadings indicated directional inaccuracies for P loss risk assessment. Similar discrepancies were observed with the multiplicative P-index, while the component P-index provided better predictions of DRP losses. However, further research is needed to confirm the applicability of the component P-index for Alabama farms. Further, a modified version of the Alabama P-index was proposed, where soil test P was replaced with the P saturation ratio and the underground outlet system with the timing of P application, along with slight adjustments of weightage factors. The modified P-index demonstrated significant and satisfactory correlations with DRP, TPP, and TP loads and was directionally correct and can be a reliable interim tool for P loss risk assessment.