Predicting pest phenology: A climate-based model for Drosophila suzukii detection

For integrated pest management (IPM) practices to be effective, detailed knowledge of the target pest phenology and activity is essential. Spotted-wing drosophila (SWD), Drosophila suzukii,is a damaging invasive pest whose phenology is highly temperature-dependent, and varies widely based on seasonal variation. Using thirteen years of trap data from blueberry farms in SW Michigan, we developed models for SWD first detection and phenology. Weekly abundance was modelled with a generalized additive mixed model and first detection date was predicted using a random forest model. Climate predictors in the models were spring mean temperature at season start, preceding-winter mean temperature, in-season weekly mean temperature, and in-season weekly relative humidity. We found warm springs advanced SWD onset while fruit were still developing, whereas cold springs delayed onset and shortened the active season but produced a sharper mid-season peak once temperatures became favorable. In contrast, preceding winter temperatures had limited effect on SWD phenology. In-season warm weeks similarly advanced and prolonged activity, while cold weeks delayed and compressed it. Using the 2025 data for validation, the first-detection model predicted onset within ~10 days of the observed date. These results indicate spring warmth and within-season temperature are the primary drivers of early season SWD pressure. Our modelling approach provides early warning of spring emergence and relative risk among seasons, supporting climate-resilient IPM decisions under increasingly variable winter and spring conditions.