Abstracto
- Decision-support systems for pest management rely on phenology models to predict the timing of pest life cycles as well as monitoring populations through sampling. Phenology models are regularly used to predict biological events that are important for management, such as the presence of a certain life stage targeted by pesticides. However, while monitoring data are often intuitively used by producers to estimate pest pressure and make decisions, capture data are rarely linked with phenology models to generate pest forecasts. Here, we introduce a model that adjusts phenology predictions across degree-days based on the sampling efficacy of trapping networks. Specifically, we created a simulation model to account for the delay in observed trap captures with respect to predicted phenology as a function of sampling efficacy, measured as the probability of individuals being sampled per degree-day. Results from the simulation were used to derive a functional relationship between the shape of the phenology model and probability of capture per degree-day. The model was then tested using trapping data for codling moth, Cydia pomonella, from eight orchards in western USA and Canada. We found that the overall average error in predictions decreased by over 50% when forecasting curves were adjusted for site-specific sampling efficacy. These results suggest a key difference in the shape between phenology models and observed pest captures in the field can be described through cumulative binomial delays. More broadly, our study shows that pest forecasting and resulting management decisions might be refined by adjusting model predictions to site-specific sampling profiles. © 2026 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license. http://creativecommons.org/licenses/by/4.0/
