In screening and registration programs, it is still common practice to estimate pesticide mobility by conducting short-term leaching tests on homogeneous soils in laboratory experiments. It is even more common to simply determine physical and chemical properties of the pesticides (e.g., adsorption constants, water solubilities and degradation rates) and predict leachability based on such information. However, the environmental conditions in such tests may be quite different from natural soils and field conditions. Therefore, pesticide mobility predicted by such laboratory experiments may not reflect the behaviour in field conditions.

Models of nutrient and fertiliser transport based on the Convection-Dispersion equation have proven very satisfactory for predictions of the overall flow of water and the transport of nutrients and tracers in both soil columns and in field experiments. However, predictions of pesticide mobility with these models, except for a very few cases, fail to represent the observed values. The difference in predictive accuracy between pesticides and fertilisers result partly from the quantity of applied chemicals that will cause health standards to be exceeded. For example, for nutrients and fertilisers, since a very large percentage must flow through the vadose zone to the water table before toxic levels are reached, it is unlikely that the capacity of macropores in soils is large enough to transport such quantities in a sufficient amount or time to cause significant contamination. However, with pesticides, only a comparatively small amount needs to find its way into a macropore (or preferred channel) to cause significant contamination of groundwater. Not with-standing the considerable difficulties of modelling non-ideal chemical transport in homogeneous soils, attempts to model and monitor the transport of small quantities of pesticides in heterogeneous soils is almost impossible, particularly on a field-scale. Current soil-physical models are clearly not adequate for describing transport of pesticides in natural soils.

Steenhuis, et al. (1994) developed a simple conceptual model for predicting the concentration and time-of-arrival of pesticides in preferentially moving water. The conceptual framework of the model is shown in Fig. 1 and is based on the observation that rainfall infiltrating the soil surface is distributed in a thin layer before being transported vertically (Fig. 1 top). This layer, called the distribution zone permits moisture to move laterally and mixing of surface-applied chemicals before being transported downward. During the time that the distribution layer becomes saturated, solutes can be absorbed into the soil matrix. After, saturation, the pesticides may be desorbed and released into preferential flow paths or dispersed into the surrounding soil matrix. Water and pesticides passing in preferential pathways, are transported to the groundwater with little modification or reaction with the surrounding matrix, and hence the concentration of solutes in this water is similar to the concentration of surface-applied chemicals.

Figure 1. Conceptual framework of pesticide transport model. In the mixing stage (top), water and solutes are absorbed in the distribution layer. After the distribution layer is saturated (bottom), water and solutes may be distributed to preferential flow paths (either macropores or fingers).

Using the above approach, we were able to predict the concentration of effluent in several field experiments (Shalit and Steenhuis, 1996). Fig. 2 illustrates the predicted and actual concentration of Chloride; 2, 4-D and Atrazine for a field plot under conventional tillage at Cornell University Experiment Farm. It is interesting to note that all graphs experienced similar slopes indicating that the soil volumes contributing to this discharge were similar regardless of the solute and that preferential flow may be an more important mechanism in the delivery of pesticides to the groundwater in conventionally tilled soils.

Figure 2. The natural logarithm of predicted and actual solute concentration in the drainage outflow from a conventionally tilled farm plot near Cornell University. The top plot is for Chloride, the middle for 2, 4-D and the bottom for Atrazine. The thin straight lines indicate sections in which a linear regression was performed and the thick black line on the top of each graph indicates the duration of irrigation.

Even though several studies have found significant amounts of pesticides in groundwater much earlier than models predict, other studies have observed that concentrations of pesticide in macropores decrease rapidly due to biodegradation. Also large populations of micro-organisms have been found below the root zone (Webster, et al., 1985), suggesting that biodegradation at depth can occur. Since the type of chemical kinetics and the rate of biodegradation depend on the substrate concentration, initial microbial population, availability of electron acceptors and inorganic nutrients, all of which vary spatially in a soil, biodegradation cannot be considered uniform with depth and the role of macropores may be important. Models that assume uniform or even no biodegradation below the root zone in heterogeneous soils are unlikely to accurately predict the transport or fate of contaminants in the unsaturated zone. For more accurate modelling, the interaction of preferential flow and non-uniform biodegradation needs to be addressed.

Pivetz et al. (1996) and Pivetz & Steenhuis (1995) applied p-nitrophenol (PNP) and 2, 4-D (2, 4-dichlorophenoxyacetic acid) to the surface of soil columns containing preferential flow paths to study the influence of macropores on biological biodegradation. Both chemicals are rapidly biodegradable, only slightly absorbed, and easily analysed. Separate break-through curves were obtained from small areas sampling matrix flow and macropore (or channel) flow. The velocity, retardation, hydrodynamic dispersion, and biodegradation of these chemicals were determined by the difference in concentration obtained by fitting theoretical solutions of break-through curves for "sterile" soils using CXTFIT (Parker and Van Genuchten, 1984) and comparing them to actual break-through curves determined experimentally. They confirmed the validity of this assumption by comparing the break-through curves obtained experimentally from a germicide-treated soil with those obtained by fitting theoretical curves by using CXTFIT. Their results are briefly discussed below:

Biodegradation of p-nitrophenol (PNP)

Biodegradation of 2, 4-D ( 2, 4-dichlorophenoxyacetic acid)

References:

Parker, J. C., & van Genuchten, M. T. (1984). Flux-averaged and volume-averaged concentration in continuum approaches to solute transport. Water Resources Research, 20(7), 866 - 872.

Pivetz, B. E., & Steenhuis, T. S. 1995. Soil and matrix macropore biodegradation of 2, 4-D. Journal of Environmental Quality 24:564-570.

Pivetz, B. E., Kelsey, J. W., Steenhuis, T. S., & Alexander, M. 1996. A procedure to calculate biodegradation during preferential flow through heterogeneous soil columns. Soil Science Society of American Journal. 60:381-388.

Shalit, G., & Steenhuis, T. (1996). A simple mixing layer model predicting solute flow to drainage lines under preferential flow. Journal of Hydrology, 183:139-149.

Steenhuis, T. S., Boll, J., Shalit, G., Selker, J. S., & Merwin, I. A. (1994). A simple equation for predicting preferential flow solute concentrations. Journal of Environmental Quality, 23(5), 1058 - 1064.

Webster, J. J., Hampton, G. J., Wilson, J. T., Ghiorse, W. C., & Leach, F. R. (1985). Determination of microbial cell numbers in subsurface samples. Groundwater, 23(1), 17 - 25.