Integrating chemical fate and population-level effect models for pesticides at landscape scale: New options for risk assessment

Abstract

Any attempt to introduce more ecological realism into ecological risk assessment of chemicals faces the major challenge of integrating different aspects of the chemicals and species of concern, for example, spatial scales of emissions, chemical exposure patterns in space and time, and population dynamics and dispersal in heterogeneous landscapes. Although these aspects are not considered in current risk assessment schemes, risk assessors and managers are expressing increasing interest in learning more about both the exposure to and the effects of chemicals at landscape level. In this study, we combined the CASCADE-TOXSWA fate model, which predicts the fate of pesticides in an interconnected system of water bodies with variable hydrological characteristics, with the MASTEP mechanistic effect model, which simulates population dynamics and effects of pesticides on aquatic species at the scale of individual water bodies. To this end, we extrapolated MASTEP to the scale of realistic landscapes and linked it to dynamic exposure patterns. We explored the effects of an insecticide on the water louse Asellus aquaticus for a typical Dutch landscape covering an area of about 10km2 containing 137 water bodies (drainage ditches) with a total length of about 65km and different degrees of connectivity. Pesticide treatments used in potato crop were assumed to result in a spray-drift input of 5% (non-mitigated) and 1% (mitigated) of the amount of pesticide applied into parts of the water body network. These treatments resulted in highly variable exposure patterns both in space and time. The effects of the pesticide on the species were investigated by comparing two scenarios with low and high individual-level sensitivity. We found that downstream transport of the pesticide led to exposure of water bodies that did not receive direct spray-drift input, even though this particular pesticide was assumed to dissipate rapidly from water. The observed differences in population-level effects and recovery patterns ranged from no observable effects in the low spray-drift and low sensitivity scenario to severe reduction of abundances in the high spray-drift and high sensitivity scenario. These results illustrate the sensitivity of our modelling approach, but also show the need for precise calculations of pesticide inputs and model parameterisation. Our study demonstrates the potential of coupled fate-and-effect to explore realistic scenarios at the scale of heterogeneous landscapes. Such scenarios could include the application of multiple pesticides to one or more crop types. Spatial realism of the landscape represented in the model ensures realistic consideration of population growth and dispersal as the two main recovery mechanisms. Future options for the landscape-scale fate-and-effect simulation approach include exploring the effects of mitigation measures on the risk estimates at landscape scale and hence represent a step towards risk management.