Investigating Factors Affecting Unit-to-Unit Variability in Non-Systemic Pesticide Residues by Stochastic Simulation Modelling

ABSTRACT

The estimation of acute dietary intake plays a key role in the safety assessment of pesticide residues and needs knowledge of the unit-to-unit variability in residues. There is limited knowledge of contributions of factors to often observed large unit-to-unit variability in residues. A stochastic simulation study was conducted to investigate the effects of sample size and heterogeneity in factors driving residue dissipation of non-systemic pesticides on the unit-to-unit variability among individual apples. The heterogeneity in driving factors was expressed in terms of variability in three dissipation parameters. The variability factor (VF), calculated as the ratio of residues in individual fruit at the 97.5th percentile and the mean residue, was used to represent the unit-to-unit variability. As the rate of dissipation increased, the variability in the corresponding parameter led to larger increases in the variability VF of residues over time. Thus, under field conditions, the relative in pesticide residues is expected to increase with time although the absolute level of residues decreases. The coefficient of variation (CV) was used to describe residue variability in samples of small sizes (5, 10, 20, 40, and 80). When sample size was fewer than 40 fruit, the sample CV increased steeply with decreasing sample size. Measuring residues as concentrations, instead of per fruit, also led to an average 4–8% increase in the CV of residues because of variability in fruit expansion.

Key Words:

• stochastic models
• residue variability
• residue dissipation
• stochastic simulation
• pesticides

ACKNOWLEDGMENTS

This research was funded jointly by the British Biotechnology and Biological Sciences Research Council and Syngenta.

Notes

^aInitial deposit on individual fruit was assumed to follow a log-normal distribution.

^b k ₁ is the rate of daily loss in residues due to non-rain causes, k ₃ is the loss due to the first rain after an application, k ₄ is the maximum loss due to the rain other than the first after an application, and Y _max is the threshold of rainfall (mm) such that further increase over this threshold will not lead to increased wash-off of residues. The variability in Y _max is not considered in this study.

^cOnly the residue on a single fruit was considered with its residue set to the input average residue.

^dSamples of an appropriate size were randomly drawn from the simulated 1000 residue values on day zero (initial deposit) and day 30. Furthermore, to simulate the effects of variable fruit growth, the weight of each fruit was sampled from a normal distribution (N (75,15) and N (110, 22) on day zero and 30, respectively) to estimate the residue concentration.