Species distribution models effectively predict the detection of Dreissena spp. in two connecting waters of the Laurentian Great Lakes

Abstract

Among the highest profile invasive species in the Laurentian Great Lakes region are Dreissena polymorpha and D. rostriformis bugensis (collectively dreissenids). Despite their abundance and ecosystem-wide effects, little is known about dreissenid distributions in large connecting channels between lakes. The objectives of this study were to estimate and document dreissenid densities and their habitat characteristics throughout the St. Clair River, to compare dreissenid species demographics, and predict spatial distributions between two connecting waters of the Great Lakes: the St. Clair and Detroit rivers. Two types of species distribution models (SDMs), MaxEnt and classification and regression tree analysis (CART), were created using dreissenid and habitat data collected in both the Detroit and St. Clair rivers. The SDMs were then used to predict presence of dreissenids in the St. Clair River. The St. Clair River had more D. r. bugensis (mean density = 486 ± 152 individuals/m²) than D. polymorpha (mean density = 3 ± 1 individuals/m²). The SDMs created from the Detroit River data reliably predicted presence of dreissenids in the St. Clair River. Depending on the river and species, CART models identified velocity and depth to be important predictor variables, while distance to river inlet/outlet were the most influential variables in the MaxEnt models. Most research on dreissenid distribution modeling is focused on determining areas for potential spread; however, this study presents a unique perspective by modeling dreissenid presence, both D. polymorpha and D. r. bugensis separately and together, where they have been established for more than 30 years.