Modeling the presence and abundance of buckthorn across the forests of Wisconsin, USA using different regression techniques

Abstract

Common buckthorn (Rhamnus cathartica) and glossy buckthorn (Frangula alnus) are two species of invasive shrubs causing extensive environmental harm across multiple states of the US and Canadian provinces. One important consideration when managing invasive plants is understanding their habitat preferences. Knowing where the species is most likely to occur can allow monitoring and management activities to focus in those areas. In Wisconsin, factors affecting the presence and abundance of buckthorn have been described for small regions but never at the State level. Therefore, the objective of this study was to develop a predictive model for occurrence and abundancy of buckthorn. Data were collected by establishing 616 sample plots across Wisconsin school forests. We constructed five types of models; logistic regression, regular Poisson and negative binomial regression, and zero-inflated Poisson (ZIP) and negative binomial regression (ZINB) using several plot and stand level model explanatory variables. The ZINB, as the best model among them, indicated that stem density, species diversity, and mean diameter at breast height of woody species (not including buckthorn) as well as distance to nearest house, housing density within 1 km, solar irradiation, percent silt, and latitude were important for estimating buckthorn presence and abundance. The models developed in this study can guide land managers on where to target mitigation actions by identifying areas invaded by buckthorn.