Towards a framework for invasive aquatic plant survey design in Great Lakes coastal areas

Abstract

At least 65 aquatic plant species have been identified as part of a surveillance list of non-native species that pose a threat to biodiversity and ecosystem services in the Laurentian Great Lakes. Early detection of these potentially invasive aquatic plants (IAP) could minimize impacts of novel incursions and facilitate successful eradication. We developed, implemented, and then adaptively refined a probabilistic boat-based sampling design that aimed to maximize the likelihood of detecting novel IAP incursions in large (400+ hectares) Great Lakes coastal areas. Surveys were conducted from 2017 to 2019 at five Great Lakes locations – St Joseph River (MI), Saginaw River (MI), Milwaukee (WI), Cleveland (OH), and the Detroit River (MI). Aquatic plant communities were characterized across the five sites, with a total of 61 aquatic plant species detected. One-fifth of the species detected in our surveys were non-native to the Great Lakes basin. Sample-based species rarefaction curves, constructed from detection data from all surveys combined at each location, show that the estimated sample effort required for high confidence (> 95%) detection of all aquatic plants at a site, including potentially invasive species, varies (< 100 sample units for Detroit River; > 300 sample units for Milwaukee, roughly equivalent to 6 to 18 days sampling effort, respectively). At least 70% of the estimated species pool was detected at each site during initial 3-day surveys. Leveraging information on detection patterns from initial surveys, including depth and species richness strata, improved survey efficiency and completeness at some sites, with detection of at least 80% of the estimated species pool during subsequent surveys. Based on a forest-based classification and regression method, a combination of just five variables explained 70% or more of the variation in observed richness at all sites (depth, fetch, percent littoral, distance to boat ramps and distance to marinas). We discuss how the model outcomes can be used to inform survey design for other Great Lakes coastal areas. The survey design we describe provides a useful template that could be adaptively improved for early detection of IAP in the Great Lakes.