Variations in soil properties and herbicide sorption coefficients with depth in relation to PRZM (pesticide root zone model) calculations

There are few experimental data available on how herbicide sorption coefficients change across small increments within soil profiles. Soil profiles were obtained from three landform elements in a strongly-eroded agricultural field and segmented into 2-cm intervals to 0.6 m depth in the knoll (eroded-upper slope), to 1.0 m depth in the toeslope (deposition zone) and to 1.6 m depth in the trough (eroded water-way). Soil samples were analyzed for soil organic carbon content (SOC) (n = 154), soil pH (n = 155), soil carbonate content (n = 126), CEC (n = 126), soil texture (n = 32), bulk density (n = 160), 2,4-D [2,4-(dichlorophenoxy) acetic acid] or glyphosate [N-phosphonomethylglycine] sorption by soil (Kd) (n = 90), and 2,4-D or glyphosate sorption per unit soil organic carbon (Koc) (n = 90). Considering all soil profiles, 2,4-D Kd values ranged from 0.12 to 2.61 L kg- 1 and were most strongly influenced by variations in SOC. In contrast, glyphosate Kd values ranged from 19 to 547 L kg- 1 and were predominantly controlled by variations in soil pH and clay content. Two hundred and fifty-two PRZM (pesticide root zone model) version 3.12.2 simulations were also performed. PRZM predicted that glyphosate would be immobile in soils even under an extreme rainfall scenario of 384 mm at one day after herbicide application. In contrast, for 2,4-D, PRZM predicted that up to 6% of the applied herbicide would move to a 15 cm depth under an actual rainfall scenario. PRZM output was particularly sensitive to input values of Kd, relative to input values of soil properties. The greatest change to PRZM outputs occurred when Kd values of toeslope profiles, ranging from 0.16 to 1.77 L kg- 1, were replaced by those measured in knoll profiles, ranging from 0.12 to 0.50 L kg- 1, when the amount of 2,4-D leached to a 15 cm depth increased by 29,081% (from 0.09 to 26.17 g ha- 1) under an actual rainfall scenario. We conclude that, when pesticide fate models such as PRZM are being used in policy analyses at larger-scales, data on Kd values in different landform elements and at the soil horizon level could be important for strengthening pesticide leaching predictions.