Assessing the impact of pesticides on lumbricid populations: An individual-based modelling approach

1. In an ecological approach to risk assessment, population models may play an important role. Our population dynamical models take into account the complex life history of earthworms, as well as a particular ecological interaction (predation). 2. Sublethal impact of pesticides is quantified at the individual level in terms of impaired vital rates, i.e. growth, maturation and reproduction. A simple model for the energetic relationships underlying these vital rates quantifies toxic stress mechanistically through proportionality constants relating size to energetic costs of maintenance, growth, food intake and production of offspring. 3. Risk is defined at the population level. Risk resulting from chronic exposure relates to reductions in (equilibrium) density, changes in population size- and age-structure, and probability of surpassing an extinction threshold. Risk posed by a single or intermittent application of a highly degradable pesticide is defined in terms of extinction probabilities and recovery times, and related to pesticide decay and initial applied dose. 4. The models are individual-based and complementary. A deterministic partial differential equation model is used to derive equilibrium properties of the system analytically and to investigate the general dynamic behaviour. An individual-by-individual model shows how this behaviour is influenced by demographic and environmental stochasticity. 5. Results obtained for two species, Lumbricus rubellus and L. terrestris, indicate that both are sensitive to pesticides affecting the energy available for individual growth, as opposed to the amount of energy available for reproduction. Retarded growth impedes individuals in reaching adulthood. This juvenile delay regulation translates from individual performance to population demography. 6. Life-history characteristics appear to make L. terrestris more sensitive to toxic stress than L. rubellus, resulting in longer population recovery times. 7. The insights obtained from the models and the way results depend on model assumptions, are discussed, and compared to the little available observational and experimental evidence. Extensions enabling a full ecological risk assessment for pesticide use are identified. Establishing an explicit relationship between ambient concentration and individual performance seems mandatory prior to use of the models as predictive tools.