Background/Question/Methods   Insecticides that do not cause direct mortality in wildlife species can still cause indirect effects by reducing prey availability for insectivores. Reduced resources for songbirds can result in a lower reproductive rate or poor nestling condition at fledging. While these effects may appear negligible at the individual level, few studies have examined consequences at the population level. We use a long-term dataset from a forest ecosystem where Bacillus thuringiensis kurstaki (Btk) was applied experimentally to control gypsy moth (Lymantria dispar L.). This study tests the hypothesis that small indirect effects of insecticide on reproductive rates can result in qualitative differences in intrinsic growth rate. We incorporated longer temporal scales in population models as well as different spatial representations of treatment scenarios to examine possible compensatory effects.

Results/Conclusions   Altered nestling feeding behavior by adult worm-eating warblers (Helmitheros vermivorus) leading to lower nestling mass at fledgling was documented in Btk treated plots through video surveillance of feeding trips and weighing of nestlings. Warbler productivity on Btk plots was also lower than untreated plots, resulting in an intrinsic growth rate < 1; however interannual variability within the treatment group exceeded that of between-group differences. Simulations including reduced juvenile survival expected as a result of lower nestling mass following treatment further reduced intrinsic growth rate. In addition to treatment-specific comparisons, we explored different spatial representations of treated and untreated areas. First, we developed a two-patch matrix model from field data to predict population-level changes, incorporating dispersal between treated and untreated patches. At the multi-site level, overall population growth was most influenced by survival within the untreated area, which could counteract reductions in reproductive success and juvenile survival on treated plots if control groups are increasing. We also simulated population dynamics within a single patch with different proportions of treatment areas, which could inform management strategies in similar systems. Differences in intrinsic growth rates at local, long-term, and multi-site scales confirm that changes in vital rates should be examined with the spatial connectivity representative of habitat availability and for the time period appropriate for risk assessment of pesticide effects on a wildlife population.