Improving invasive species management using predictive phenology models: an example from brown marmorated stink bug...

In order to better understand the population dynamics of invasive species in their native range, we developed two predictive phenological models using the ubiquitous invasive insect pest, Halyomorpha halys (Stål) (Hemiptera: Pentatomidae), as the model organism. Our work establishes a zero-inflated negative binomial regression (ZINB) model, and a general additive mixed model (GAMM) based on 11 years of black light trap monitoring of H. halys at three locations in Japan.

The ZINB model indicated that degree days (DD) have a significant effect on the trap catch of adult H. halys, and that precipitation has no effect. A dataset generated by 1000 simulations from the ZINB suggested that higher predicted trap catches equated to a lower probability of encountering a zero-count. The GAMM produced a cubic regression smooth curve which forecasts the seasonal phenology of H. halys as following a bell-shaped trend in Japan. Critical DD points during the field season in Japan included 261 DD for first H. halys adult detection and 1091 DD for peak activity.

This study establishes the first models capable of forecasting native H. halys population dynamics based on DD. These robust models practically improve population forecasting of H. halys in the future and help fill gaps in knowledge pertaining to its native phenology, thus ultimately contributing to the progression of efficient management of this globally invasive species. © 2021 Society of Chemical Industry.