PeLM: Modeling of Pesticide-Losses Through Runoff and Sediment Transport

Abstract

A GIS-aided pesticide loss model (PeLM) was developed to simulate pesticide losses through surface runoff and sediment transport in watershed systems. The PeLM could tackle the movement of eroded soil along with surface runoff as well as the pesticide losses in adsorbed and dissolved phases. The contributions of different soil types in the sediment were also examined. The model was applied to the Kintore Creek Watershed of southern Ontario, Canada. The simulation results were verified through observed data, indicating a correlation level of 0.89–0.98. The results also showed that clay particles usually held the largest share of contributions to pesticide losses through soil erosion. This study is significant in the efforts for modeling nonpoint source pollution in watershed systems. It provides useful information and support for the related decisions of watershed management.     

Dynamic modeling of organophosphate pesticide load in surface water in the northern San Joaquin Valley watershed of California

The hydrology, sediment, and pesticide transport components of the Soil and Water Assessment Tool (SWAT) were evaluated on the northern San Joaquin Valley watershed of California. The Nash-Sutcliffe coefficients for monthly stream flow and sediment load ranged from 0.49 to 0.99 over the watershed during the study period of 1992-2005. The calibrated SWAT model was applied to simulate fate and transport processes of two organophosphate pesticides of diazinon and chlorpyrifos at watershed scale. The model generated satisfactory predictions of dissolved pesticide loads relative to the monitoring data. The model also showed great success in capturing spatial patterns of dissolved diazinon and chlorpyrifos loads according to the soil properties and landscape morphology over the large agricultural watershed. This study indicated that curve number was the major factor influencing the hydrology while pesticide fate and transport were mainly affected by surface runoff and pesticide application and in the study area.     

Management-oriented sensitivity analysis for pesticide transport in watershed-scale water quality modeling using SWAT

The Soil and Water Assessment Tool (SWAT) was calibrated for hydrology conditions in an agricultural watershed of Orestimba Creek, California, and applied to simulate fate and transport of two organophosphate pesticides chlorpyrifos and diazinon. The model showed capability in evaluating pesticide fate and transport processes in agricultural fields and instream network. Management-oriented sensitivity analysis was conducted by applied stochastic SWAT simulations for pesticide distribution. Results of sensitivity analysis identified the governing processes in pesticide outputs as surface runoff, soil erosion, and sedimentation in the study area. By incorporating sensitive parameters in pesticide transport simulation, effects of structural best management practices (BMPs) in improving surface water quality were demonstrated by SWAT modeling. This study also recommends conservation practices designed to reduce field yield and in-stream transport capacity of sediment, such as filter strip, grassed waterway, crop residue management, and tailwater pond to be implemented in the Orestimba Creek watershed.     

Foliar washoff of pesticides (FWOP) model: Development and evaluation

Abstract

The Foliar Washoff of Pesticides (FWOP) Model was developed to provide an empirical simulation of pesticide washoff from plant leaf surfaces as influenced by rainfall amount. To evaluate the technique, simulations by the FWOP Model were compared to those by the foliar washoff algorithm of the Chemical, Runoff and Erosion from Agricultural Management Systems (CREAMS) Model. The two algorithms were linked individually to the Pesticide Runoff Simulator (PRS) for the comparison. Five years of test data from a Mississippi watershed were used to evaluate six insecticides (carbaryl, profenofos, methyl parathion, permethrin, phorate, and toxaphene).

Initially, the FWOP model was used to evaluate the relative impact of chemical distribution (foliage versus soil) on the subsequent foliar washoff and soil surface contributions to runoff losses. Results indicated that runoff losses were low If all of the insecticide was applied to the foliage whereas high losses occurred if applied only to the soil. When an assumed application was distributed between the plant and soil (i.e., 90% to foliage and 10% to soil), predicted runoff losses compared well with observed field data (<3% of the application rate).

Except for toxaphene, the FWOP model generally predicted less washoff and subsequent runoff losses than the CREAMS approach. Simulated toxaphene washoff losses were in good agreement with observed field data. Statistical comparisons of the two modeling approaches using the Kolmogorov‐Smirnov test showed differences in the two cumulative frequency distributions for washoff but smaller differences for runoff. Average 5‐year runoff losses, however, were greater using the CREAMS approach—by factors of 2, 3, and 3 for profenofos, methyl parathion and phorate, respectively.

Results from this study will be useful for upgrading current exposure assessment models to more accurately address foliar washoff losses of pesticides as well as for assessing the impact of foliar‐applied chemicals on environmental quality.     

Persistence and runoff losses of 3 herbicides and chlorpyrifos from a corn field in the Lake Balaton watershed of Hungary

Corn is intensively cultivated in western Hungary in the basin of Lake Balaton, one of the most important water resources in eastern Europe. Pesticide runoff was measured in 1996 and 1997 from a typical corn field near Zalaegerszeg, Hungary, which drains into the Zala River, an important water source of Lake Balaton. Three herbicides, namely atrazine, acetochlor, and propizochlor, and the insecticide chlorpyrifos were applied to bare soil in a field with 5% slope and soil and runoff water pesticide concentrations were monitored. In 1997, a rainfall-runoff simulation experiment was conducted on a small sub-plot in order to measure pesticide runoff under reasonable worst-case conditions. Under natural rainfall almost all losses occurred in a large runoff event in 1996 one month after application in which 3% of atrazine and 1% of acetochlor was transported off the field. Propizochlor and chlorpyrifos losses in the same event were much lower: 0.2% and <0.01%, respectively, because of these chemicals' shorter persistence times in near-surface soil. The rainfall simulation produced only trace amounts of losses even though 4.1 cm was applied in 2 hours; the soil was extremely dry and only 0.2 cm runoff occurred containing less than 0.01% of all chemicals applied. The results suggest that intensive use of corn herbicides, which have been found to result in widespread contamination of water resources elsewhere, may be expected to have the same impact in the Balaton watershed depending on the amounts and intensities used in the basin.     

Extension of coupled multispecies metal transport and speciation (TRANSPEC) model to soil

Atmospheric deposition of metals emitted from mining operations has raised metal concentrations in the surrounding soils. This repository may be remobilized and act as a source of metals to nearby surface aquatic systems. It is important to understand metal dynamics and the impact of various chemistry and fate parameters on metal movement in the soil environment in order to evaluate risk associated with metals in terrestrial ecosystems and accurately establish critical discharge limits that are protective of aquatic biota. Here we extend our previously developed coupled multispecies metal fate-TRANsport and SPECiation/complexation (TRANSPEC) model, which was applicable to surface aquatic systems. The extended TRANSPEC, termed TRANSPEC-II, estimates the partition coefficient, K(d), between the soil-solid and -soluble phases using site-specific data and a semi-empirical regression model obtained from literature. A geochemical model calculates metal and species fractions in the dissolved and colloidal phases of the soil solution. The multispecies fugacity/aquivalence based fate-transport model then estimates inter-media transport rates such as leaching from soil, soil runoff, and water-sediment exchanges of each metal species. The model is illustratively applied to Ni in the Kelly Lake watershed (Sudbury, Ontario, Canada), where several mining operations are located. The model results suggest that the current atmospheric fallout supplies only 4% of Ni removed from soil through soil runoff and leaching. Soil runoff contributes about 20% of Ni entering into Kelly Lake with the rest coming from other sources. Leaching to groundwater, apart from runoff, is also a major loss process for Ni in the soil. A sensitivity analysis indicates that raising soil pH to above 6 may substantially reduce metal runoff and improve water quality of nearby water bodies that are impacted by runoff.     

Pesticide runoff model (PeRM): a case study for the Kintore Creek Watershed,Ontario, Canada

An integrated model, the Pesticide Runoff Model (PeRM), has been developed to predict pesticide losses due to runoff by considering the emission, degradation, adsorption and desorption of pesticides, as well as their movement in dissolved and adsorbed phases. The developed modeling system has been used to calculate the losses of atrazine from agricultural lands in the Kintore Creek Watershed, Ontario, Canada between 1988 and 1992. The modeling outputs have been verified against actual monitoring data, which were obtained from a water quality monitoring project carried out in the same watershed over the same period of time.     

Estimating pesticide runoff in small streams

Surface runoff is one of the most important pathways for pesticides to enter surface waters. Mathematical models are employed to characterize its spatio-temporal variability within landscapes, but they must be simple owing to the limited availability and low resolution of data at this scale. This study aimed to validate a simplified spatially-explicit model that is developed for the regional scale to calculate the runoff potential (RP). The RP is a generic indicator of the magnitude of pesticide inputs into streams via runoff. The underlying runoff model considers key environmental factors affecting runoff (precipitation, topography, land use, and soil characteristics), but predicts losses of a generic substance instead of any one pesticide. We predicted and evaluated RP for 20 small streams. RP input data were extracted from governmental databases. Pesticide measurements from a triennial study were used for validation. Measured pesticide concentrations were standardized by the applied mass per catchment and the water solubility of the relevant compounds. The maximum standardized concentration per site and year (runoff loss, R_Loss) provided a generalized measure of observed pesticide inputs into the streams. Average RP explained 75% (p < 0.001) of the variance in R_Loss. Our results imply that the generic indicator can give an adequate estimate of runoff inputs into small streams, wherever data of similar resolution are available. Therefore, we suggest RP for a first quick and cost-effective location of potential runoff hot spots at the landscape level.     

Arsenic and mercury concentrations in major landscape components of an intensively cultivated watershed

To provide an understanding of arsenic (As) and mercury (Hg) concentrations in soil, sediment, water, and fish tissues, samples were collected from a Mississippi River alluvial floodplain located in northwest Mississippi. As concentrations increased approximately an order of magnitude from water (5.12 micrograms/l) to fish tissues (36.99 micrograms/kg) and an additional two orders of magnitude in soils, lake sediments, and wetland sediments (5728, 5614, and 6746 micrograms/kg), respectively. Average Hg concentrations in water, soils, lake sediments, and fish were 2.16 micrograms/l, 55.1, 14.5 and 125 micrograms/kg, respectively. As and Hg concentrations were within published ranges for uncontaminated soil, water, and sediments. As concentrations represented a low risk. Hg concentrations were also low but showed a greater tendency to concentrate in fish tissue. The dominant mode of entry of these materials into aquatic systems is through storm-generated runoff. Since both metals accompany sediments, agricultural conservation practices such as reduced tillage, buffer riparian strips, and bordering sediment ponds or drainage wetlands will minimize watershed input to aquatic systems.     

晋江西溪流域茶园降雨径流产污特征

在野外模拟降雨条件下,开展了晋江西溪流域茶园和裸地的径流产沙及氮磷养分流失过程对比实验,研究结果表明,在相同降雨强度下,3种下垫面径流和产沙量顺序均为:裸地＞2年茶园＞4年茶园,径流量与产沙量之间呈显著正相关.对地表径流水相而言,2年茶园、4年茶园和裸地的TN流失量分别为:461.29、129.38和107.86 mg/m2;NO3-N流失量分别为:286.42、98.58和103.00 mg/m2,均占TN流失量的60％以上;NH4-N流失量分别为:48.67、16.19和4.42 mg/m2;Tp流失量分别为:34.71、16.47和23.88 mg/m2.对径流泥沙相而言,2年茶园、4年茶园和裸地的TN流失量分别为:379.28、44.81和747.16 mg/m2,占流失总量的比重在25.72％～87.93％之间;TP流失量分别为:27.94、4.17和58.85 mg/m2,占流失总量的比重在53.42％ ～68.36％之间.茶园的N、P主要随径流流失,而裸地以泥沙迁移为主.这说明茶叶种植具有一定的水土保持效应,且种植年限较长的茶园可显著减少随径流泥沙进入水体中的N、P元素.     

Sensitivity of agricultural runoff loads to rising levels of CO2 and climate change in the San Joaquin Valley watershed of California

The Soil and Water Assessment Tool (SWAT) was used to assess the impact of climate change on sediment, nitrate, phosphorus and pesticide (diazinon and chlorpyrifos) runoff in the San Joaquin watershed in California. This study used modeling techniques that include variations of CO₂, temperature, and precipitation to quantify these responses. Precipitation had a greater impact on agricultural runoff compared to changes in either CO₂ concentration or temperature. Increase of precipitation by ±10% and ±20% generally changed agricultural runoff proportionally. Solely increasing CO₂ concentration resulted in an increase in nitrate, phosphorus, and chlorpyrifos yield by 4.2, 7.8, and 6.4%, respectively, and a decrease in sediment and diazinon yield by 6.3 and 5.3%, respectively, in comparison to the present-day reference scenario. Only increasing temperature reduced yields of all agricultural runoff components. The results suggest that agricultural runoff in the San Joaquin watershed is sensitive to precipitation, temperature, and CO₂ concentration changes.     

The potential contribution of urban runoff to surface sediments of the Passaic River: sources and chemical characteristics

Urban runoff has been reported as the second most frequent cause of surface water pollution in the United States. Due to the incidence of runoff in urban areas, it was of interest to estimate the impact runoff may have to recent sediment quality within the lower reaches of the Passaic River. Study objectives included i) review of recent urban runoff studies to determine the occurrence and pattern of distribution of chemicals in runoff; ii)comparison of the "fingerprints" from urban runoff studies to the contaminant distributions in surface sediments from the River; and iii) estimation of mass loadings to the surface sediments using surrogate data. The analyses showed that metals and PAH distributions in the sediments were similar to those observed in runoff from diverse locations, suggesting that urban runoff composition within the Passaic watershed is similar to other urban areas. Mass loading calculations demonstrated that urban runoff is a significant source of the metals observed in the sediments, and that PAH and DDT sediment loadings could, in some cases, be accounted for by urban runoff. Observed sediment loads for PCBs, however, were significantly higher than were estimated from urban runoff.     

Contribution of macroporosity to infiltration into a continuous corn no-tilled watershed: Implications for contaminant movement

Rainfall and runoff were measured for many years on small watersheds on 10–15% slopes in east-central Ohio. Surface runoff from watersheds used for corn (Zea mays L.) production was high with conventional tillage and very low with no-tillage. A 50-year storm produced 15 times more runoff from a plowed watershed than from a mulch-covered no-till watershed. Reduced runoff from the no-till surface resulted in increased percolation and enhanced the potential for transport of agricultural chemicals to the groundwater. The mulched surface of the no-till watershed also created a favorable environment for the deep burrowing earthworm, Lumbricus terrestris L., whose burrows can transmit water rapidly downward through the soil profile, thus contributing to the high infiltration rates.Open biopores and smaller structural pores were counted and measured to characterize the major flow paths of water movement in the no-till soil. Photos of horizontal surfaces at 2.5-, 7.5-, 15-, and 30 - cm depths and vertical faces of impregnated samples from the 1- and 5-cm depths were evaluated by image analysis. Number of pores was inversely proportional to pore diameter, however pores in the 0.05–1.0-mm diameter range accounted for less porosity than did those in the 1.0–5.0-mm range. The large pores were nearly vertical earthworm burrows and were continuously open from near the surface to the bedrock. Surface applications of lime increased subsoil pH in the no-till watershed but had little effect below the plow sole in the tilled watershed, suggesting that rapid movement of water in large pores can enhance chemical migration into the subsoil.     

Tillage management to mitigate herbicide loss in runoff under simulated rainfall conditions

Conservation tillage mitigates soil loss in cropland because plant residues help protect the soil, but effects on pesticide movement in surface runoff are not as straightforward. Effects of soil disturbance on surface runoff loss of chlorimuron and alachlor were evaluated utilizing runoff trays. Soil in the trays was either disturbed (tilled) and kept bare or was not tilled, and existing decomposed plant residue was left on the surface. Rainfall (25mm, 20min) was simulated 1d after alachlor (2.8kg ha(-1)) or chlorimuron (54g ha(-1)) application, and runoff was collected. Runoff fractions were analyzed for herbicide and sediment. Total alachlor loss from bare plots was greater than that in no-tillage plots (4.5% vs. 2.3%, respectively). More than one-third of total alachlor lost from bare plots occurred in the first l of runoff, while no-tillage plots had less runoff volume with a more even distribution of alachlor concentration in the runoff during the rainfall simulation and subsequent runoff period. In contrast, more chlorimuron was lost from no-tillage plots than bare plots (12% vs. 1.5%) even though total runoff volume was lower in the no-tillage plots (10.6mm vs. 13.6mm). This was attributed to dense coverage with partially decomposed plant residue in no-tillage plots (1652kg ha(-1)) that intercepted chlorimuron. It was likely that chlorimuron, a polar compound, was more easily washed off surface plant residues and transported in runoff.     

Runoff and leaching of atrazine and alachlor on a sandy soil as affected by application in sprinkler irrigation

Rainfall simulation was used with small packed boxes of soil to compare runoff of herbicides applied by conventional spray and injection into sprinkler-irrigation (chemigation), under severe rainfall conditions. It was hypothesized that the larger water volumes used in chemigation would leach some of the chemicals out of the soil surface rainfall interaction zone, and thus reduce the amounts of herbicides available for runoff. A 47-mm rain falling in a 2-hour event 24 hours after application of alachlor (2-chloro-N-(2,6-diethylphenyl)-N-(methoxymethyl)-acetamide) and atrazine (6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-triazine-2, 4-diamine) was simulated. The design of the boxes allowed a measurement of pesticide concentrations in splash water throughout the rainfall event. Initial atrazine concentrations exceeding its' solubility were observed. When the herbicides were applied in 64,000 L/ha of water (simulating chemigation in 6.4 mm irrigation water) to the surface of a Tifton loamy sand, subsequent herbicide losses in runoff water were decreased by 90% for atrazine and 91% for alachlor, as compared to losses from applications in typical carrier water volumes of 187 L/ha. However, this difference was not due to an herbicide leaching effect but to a 96% decrease in the amount of runoff from the chemigated plots. Only 0.3 mm of runoff occurred from the chemigated boxes while 7.4 mm runoff occurred from the conventionally-treated boxes, even though antecedent moisture was higher in the former. Two possible explanations for this unexpected result are (a) increased aggregate stability in the more moist condition, leading to less surface sealing during subsequent rainfall, or (b) a hydrophobic effect in the drier boxes. In the majority of these pans herbicide loss was much less in runoff than in leachate water. Thus, in this soil, application of these herbicides by chemigation would decrease their potential for pollution only in situations where runoff is a greater potential threat than leaching.     

Pesticide concentrations in agricultural storm drainage inlets of a small Swiss catchment

Agricultural pesticides transported to surface waters pose a major risk for aquatic ecosystems. Modelling studies indicate that the inlets of agricultural storm drainage systems can considerably increase the connectivity of surface runoff and pesticides to surface waters. These model results have however not yet been validated with field measurements. In this study, we measured discharge and concentrations of 51 pesticides in four out of 158 storm drainage inlets of a small Swiss agricultural catchment (2.8 km²) and in the receiving stream. For this, we performed an event-triggered sampling during 19 rain events and collected plot-specific pesticide application data. Our results show that agricultural storm drainage inlets strongly influence surface runoff and pesticide transport in the study catchment. The concentrations of single pesticides in inlets amounted up to 62 µg/L. During some rain events, transport through single inlets caused more than 10% of the stream load of certain pesticides. An extrapolation to the entire catchment suggests that during selected events on average 30 to 70% of the load in the stream was transported through inlets. Pesticide applications on fields with surface runoff or spray drift potential to inlets led to increased concentrations in the corresponding inlets. Overall, this study corroborates the relevance of such inlets for pesticide transport by establishing a connectivity between fields and surface waters, and by their potential to deliver substantial pesticide loads to surface waters.

     

Variations in source apportionments of nutrient load among seasons and hydrological years in a semi-arid watershed: GWLF model results

Quantifying source apportionments of nutrient load and their variations among seasons and hydrological years can provide useful information for watershed nutrient load reduction programs. There are large seasonal and inter-annual variations in nutrient loads and their sources in semi-arid watersheds that have a monsoon climate. The Generalized Watershed Loading Function model was used to simulate monthly nutrient loads from 2004 to 2011 in the Liu River watershed, Northern China. Model results were used to investigate nutrient load contributions from different sources, temporal variations of source apportionments and the differences in the behavior of total nitrogen (TN) and total phosphorus (TP). Examination of source apportionments for different seasons showed that point sources were the main source of TN and TP in the non-flood season, whereas contributions from diffuse sources, such as rural runoff, soil erosion, and urban areas, were much higher in the flood season. Furthermore, results for three typical hydrological years showed that the contribution ratios of nutrient loads from point sources increased as streamflow decreased, while contribution ratios from rural runoff and urban area increased as streamflow increased. Further, there were significant differences between TN and TP sources on different time scales. Our findings suggest that priority actions and management measures should be changed for different time periods and hydrological conditions, and that different strategies should be used to reduce loads of nitrogen and phosphorus effectively.     

The contribution of particles washed from rooftops to contaminant loading to urban streams

Rooftops are both a source of and a pathway for contaminated runoff in urban environments. To investigate the importance of particle-associated contamination in rooftop runoff, particles washed from asphalt shingle and galvanized metal roofs at sites 12 and 102 m from a major expressway were analyzed for major and trace elements and PAHs. Concentrations and yields from rooftops were compared among locations and roofing material types and to loads monitored during runoff events in the receiving urban stream to evaluate rooftop sources and their potential contribution to stream loading. Concentrations of zinc, lead, pyrene, and chrysene on a mass per mass basis in a majority of rooftop samples exceeded established sediment quality guidelines for probable toxicity of bed sediments to benthic biota. Fallout near the expressway was greater than farther away, as indicated by larger yields of all contaminants investigated, although some concentrations were lower. Metal roofing was a source of cadmium and zinc and asphalt shingles a source of lead. The contribution of rooftop washoff to watershed loading was estimated to range from 6 percent for chromium and arsenic to 55 percent for zinc. Estimated contributions from roofing material to total watershed load were greatest for zinc and lead, contributing about 20 and 18 percent, respectively. The contribution from atmospheric deposition of particles onto rooftops to total watershed loads in stormwater was estimated to be greatest for mercury, contributing about 46 percent.     

Hydroponic uptake of atrazine and lambda-cyhalothrin in Juncus effusus and Ludwigia peploides

Phytoremediation encompasses an array of plant-associated processes known to mitigate contaminants from soil, sediment, and water. Modification of pesticides associated with agricultural runoff includes processes directly associated with aquatic macrophytes in addition to changes in soil geochemistry and associated rhizospheric degradation. Remediation attributes of two vegetative species common to agricultural drainages in the Mississippi Delta, USA, were assessed using atrazine and lambda-cyhalothrin. Concentrations used in 8-d hydroponic exposures were calculated using recommended field applications and a 5% runoff model from a 0.65-cm rainfall event on a 2.02-ha field. While greater atrazine uptake was measured in Juncus effusus, greater lambda-cyhalothrin uptake occurred in Ludwigia peploides. Maximum pesticide uptake was reached within 48h for each exposure and subsequent translocation of pesticides to upper plant biomass occurred in macrophytes exposed to atrazine. Sequestration of 98.2% of lambda-cyhalothrin in roots of L. peploides was measured after 8d. Translocation of lambda-cyhalothrin in J. effusus resulted in 25.4% of pesticide uptake partitioned to upper plant biomass. These individual macrophyte remediation studies measured species- and pesticide-specific uptake rates, indicating that seasonality of pesticide applications and macrophyte emergence might interact strongly to enhance mitigation capabilities in edge-of-field conveyance structures.     

Runoff and leaching of atrazine and alachlor on a sandy soil as affected by application in sprinkler irrigation

Abstract

Rainfall simulation was used with small packed boxes of soil to compare runoff of herbicides applied by conventional spray and injection into sprinkler‐irrigation (chemigation), under severe rainfall conditions. It was hypothesized that the larger water volumes used in chemigation would leach some of the chemicals out of the soil surface rainfall interaction zone, and thus reduce the amounts of herbicides available for runoff. A 47‐mm rain falling in a 2‐hour event 24 hours after application of alachlor (2‐chloro‐N‐(2,6‐diethylphenyl)‐N‐(methoxymethyl)‐acetamide) and atrazine (6‐chloro‐N‐ethyl‐N‐(1‐methylethyl)‐1,3,5‐triazine‐2,4‐diamine) was simulated. The design of the boxes allowed a measurement of pesticide concentrations in splash water throughout the rainfall event. Initial atrazine concentrations exceeding its’ solubility were observed. When the herbicides were applied in 64000 L/ha of water (simulating chemigation in 6.4 mm irrigation water) to the surface of a Tifton loamy sand, subsequent herbicide losses in runoff water were decreased by 90% for atrazine and 91% for alachlor, as compared to losses from applications in typical carrier water volumes of 187 L/ha. However, this difference was not due to an herbicide leaching effect but to a 96% decrease in the amount of runoff from the chemigated plots. Only 0.3 mm of runoff occurred from the chemigated boxes while 7.4 mm runoff occurred from the conventionally‐treated boxes, even though antecedent moisture was higher in the former. Two possible explanations for this unexpected result are (a) increased aggregate stability in the more moist condition, leading to less surface sealing during subsequent rainfall, or (b) a hydrophobic effect in the drier boxes. In the majority of these pans herbicide loss was much less in runoff than in leachate water. Thus, in this soil, application of these herbicides by chemigation would decrease their potential for pollution only in situations where runoff is a greater potential threat than leaching.