Analysis of insect population data with structured population models

• Perry de Valpine

Abstract
Agricultural pest control relies on understanding complex communities of herbivores and their natural enemies. Field experiments of agricultural communities have produced voluminous multi-species, spatiotemporal population data. However, conventional data analysis uses model frameworks that lack biological structure and thus are only indirectly related to the hypothesized processes of birth, death, predation and movement. This project will use population models to analyze extensive existing data from three agricultural insect communities and one protist microcosm experiment. The questions for each system are 1) Is there evidence for particular movement behaviors? 2) Is there evidence of particular species interactions? The study systems are 1) pea aphids and natural enemies in Wisconsin alfalfa, 2) cotton aphids, other herbivores, and natural enemies in California of cotton, 3) whiteflies and natural enemies in Arizona cotton, and 4) protist predator and prey in laboratory microcosms. For each system I will develop structured population models, fit the models to data under the null and alternative hypotheses, make statistical hypothesis tests, biologically interpret the conclusions, and test the power and biases of the entire procedure. This project will evaluate important ecological hypotheses that have previously been inpenetrable to statistical analysis and develop and test new analysis tools for future research.