Pesticides Research, 86 – Fate of Pyrethroids in Farmland Ponds

Fate of Pyrethroids in Farmland Ponds

2 Summary

Part I

Pyrethroids constitute a group of insecticides, which have been used widespread in Danish agriculture. They are toxic to aquatic organisms. The aim of the present study has been to investigate the fate of these compounds in ponds in agricultural areas. A pond is a complex ecosystem. It can be considered to consist of a series of compartments.

It was intended to investigate all compartments that are important either to the mass balance because they are place of residence for aquatic fauna. The most important compartments of pond ecosystems were a priori selected to be surface microlayer, water column and sediment. Concentration of dissolved pesticide in the water phase was studied by aid of semipermeable membrane devices (SPMDs).

The fate of 4 pyrethroids, fenpropathrin, permethrin, esfenvalerate and deltamethrin was studied during a 2-year period. The results have been used to validate a distribution model and the model analysis has been used for interpretation of the pesticide measurements. The model analysis is published as part II of this report.

The experimental ponds are artificial and have as far as possible been established to resemble natural ponds. The artificial ponds can unlike natural ponds be experimentally exposed to pesticides. The pesticides have been sprayed onto the pond surface. Samples of surface microlayer, water column and sediment have been collected and analysed from 1 hour after spraying to 8-16 days after spraying with increasing intervals.

In surface microlayer is the concentration of pesticides initially very high because that is where the pesticides enter the ponds. The concentration drops quickly because of mixing into the rest of the water compartment. The concentration of pesticides is however, 8-10 times higher in the surface microlayer compared to the water phase throughout the observation period.

In the water phase the pesticides become evenly distributed within the first day after spraying. The concentration drops quickly, mainly because the pesticides are adsorbed to the sediment particles.

Permethrin, esfenvalerate and deltamethrin rearrange into other isomers in the surface microlayer and the water column.

Concentrations of pesticides in the SPMDs reflect the concentration profile of dissolved pesticides in the water column. The dissolved pesticide is bioavailable.

Model analysis has demonstrated that pyrethroids are adsorbed in the upper few mm. of the sediment.

Part II

The focus in this report is on the mechanisms governing pyrethroid exposure in small farmland ponds. Experimental results in form of time series of pesticide concentration values are interpreted in a model analysis using mathematical models. The experiments are reported in detail in part I. Known doses of the active ingredients deltamethrin, permethrin, fenvalerate and fenpropathrin are spread at the surface of artificial ponds. In both the year 1995 and year 1996 spraying experiments have been undertaken in two ponds. All four pyrethroids were out sprayed in each experiment. However, only the fenpropathrin measurements were valid for 1995, in relation to mathematical model interpretation, due to analytical problems. Different compartments in the ponds were measured and mainly results from the water column and surface micro layer are used in this analysis. The experimental results from the artificial ponds are supplied by laboratory experiments using small glass containers (Morgenroth, 1992a and b).

The purpose of the mathematical model analysis is to identify the most realistic model which can explain the experimental results. Furthermore, the models can help to generalise the experimental results to conditions different from the test conditions.

A paradigm for selecting the model structure is suggested, where two statements are used: (1) the model shall be able to describe the experiment, and (2) the model structure shall be as simple as possible. The first statement is obviously necessary and the last statement is needed in order to avoid suggestions of models of unnecessary high complexity. Such over-complex models may describe the experiment but the suggested model structure can easily be misleading. This analysis is based on a kind of catalogue of possible mechanisms taking place in the pond. Different realistic combinations from this catalogue forms a series of alternative models having different levels of complexity. The task is now to select the least complex model, which can describe the experiment.

An idealised pond system is defined consisting of air, water column and sediment sub systems, where substance is transmitted between the different sub systems. The water column is assumed completely mixed having no concentration gradient from top to bottom, which is shown to be valid a few hours after spraying (part I). The exchange of substance to the air is assumed to be a first order release to the air from the water column. The sediment is assumed to be homogenous, in which vertical one dimensional diffusion can take place from the water column and through a laminar boundary layer at the sediment surface. The substances adsorbed to the sediment solids are assumed to be in local equilibrium in relation to the substance dissolved in the pore water. Degradation is assumed to be a first order degradation in both the water column and the sediment. A specific model is suggested in relation to the surface micro layer in which diffusion, adsorption and first order volatilisation are included.

It has not been possible to suggest a model for the surface micro layer. The data are too limited and knowledge about the 'real' layer thickness is missing. However, some negative conclusions can be drawn because a diffusion type model including the diffusion through the micro layer, linear adsorption and 1. order volatilisation from the surface seems to release the substances from the micro layer too quickly. The measurement of surface micro layer concentrations seems very uncertain so the missing coincidence between model and experiment can be a result of either uncertainty in model assumptions or the uncertainty in the experimental result.

For the sediment water column system the model including sediment pore water diffusion, linear adsorption to sediment solids and degradation in the sediment seem most promising. Both experiments in the artificial ponds and in the laboratory support this type of model. However, the degradation mechanism was not an important description for fenpropathrin which also seems to have higher sediment adsorption than the other substances.

For deltamethrin, permethrin and fenvalerate there was a distinct difference between pond 3 and pond 4, which were replicates. The most probable explanation is the difference in biomasses due to a snail invasion in pond 3, which increased the adsorption mechanisms in pond 3, mainly due to the increased turbidity in water column caused by snail activity.

Laboratory experiments for fenvalerate including sediment measurement show a transport into the sediment as predicted by the pond investigation for the other pyrethroids. Furthermore, sediment degradation is observed in the laboratory experiments from sediment contamination measurements. The sediment pore diffusion coefficient in the lab scale experiment is much smaller (60 fold) than the diffusion coefficient in the pond sediment. This indicates that the diffusion into the sediment in the pond is increased due to sediment heterogeneity.

A generalised curve is constructed for fenpropathrin based on the suggested sediment/water column model. The model predicts the concentration to be proportional to the dosage and nearly independent of pond depth. This makes is possible to perform a generalised concentration time series for the water column which is independent of both dosage and depth. The tendency of independence between depth and concentration levels seems to be a result of the quick and concentration dependent disappearance from the water column. Therefore, all hydrophobic substances will tend to have this type of independence when the initial concentration is formed as a surface area specific (atmospheric) deposition.

The experimental result given in this analysis was not sufficient to make a clear distinction between the diffusion coefficient and the adsorption in the sediment. Only the product between the diffusion coefficient and the retention factor comes out of model calibrations. Therefore, additional adsorption tests using the sediment material will be beneficial in further experiments.

The sediment water column model can be used in experimental design. Such an optimisation is outside of the scope of the present work, but it should be indented in planning of future experiments. The knowledge needed is the experimental error, the cost for sampling, analysing and spraying and the constrains in form of the number of ponds and the overall time schedule. Different experimental strategies can be tested to determinate the uncertainty of the process related parameters (degradation, adsorption, diffusion). The cost and the associated uncertainty can be estimated and the optimal combination identified.