Predicting insect population dynamics by linking phenology models and monitoring data

Sustainable pest management in crop systems requires that producers target control tactics to key periods of pest activity. To accomplish this, producers often sample pests to estimate abundance and use degree-day models to predict timing of discrete pest phenology events. However, there are few examples of management tools that link phenology models with sampling data to make predictions of pest abundance. Here, we propose a method to predict pest captures by linking trap count data collected in-season with models that predict the cumulative emergence of insects from heat accumulation. Specifically, we used a 20-year dataset of codling moth Cydia pomonella (L.) pheromone trap captures to build and validate a model that produces a prediction band of cumulative captures until the end of the overwintering generation, assuming constant sampling and no migration or controls. Uncertainty was calculated as a function of the predicted mean, sample size, prediction length, and model variance. Model validation revealed that > 75 % of the tested moth capture trajectories fell within prediction bands when they were produced at or after 350 degree-days. The model provides a tool for codling moth management that integrates sampling data with an established phenology model to produce sound within-season population predictions. Producers can use such tools to make decisions on pesticide applications that are both timed to the proper pest life stage and informed by population dynamics predictions.