PREDICTING THE LIKELIHOOD OF EURASIAN WATERMILFOIL PRESENCE IN LAKES, A MACROPHYTE MONITORING TOOL

Abstract

In regions with abundant and diverse freshwater resources, it is difficult and costly to survey all lakes at the level required to detect invasive plants. Effective allocation of monitoring resources requires tools that identify waterbodies where exotic species are most likely to invade. We developed and tested models that predict conditions in which Eurasian watermilfoil, Myriophyllum spicatum, is most likely to survive and successfully colonize. We used logistic regression to model the likelihood of M. spicatum presence or absence using a suite of biological, chemical, and physical lake characteristics which are easily obtainable from public databases. We evaluated model fit by the Aikake criterion and model performance by the percentage of misclassification errors as well as the costs associated with acquiring data for variables modeled. Several models fit our data well, misclassifying only 1.3–11.0% of the lakes where M. spicatum was observed, and used relatively inexpensive landscape variables (percent forest cover in a drainage basin, presence and type of public boat launch, and bedrock type) that typically exist as information layers in geographic information systems (GISs) or recreational atlases. We found that the most important factors affecting the presence or absence of M. spicatum were those that influence water quality factors known to impact M. spicatum growth, rather than factors associated with human activity and dispersal potential. In particular, the amount of forest cover in the lake watershed was consistently important and could control the level of dissolved inorganic carbon in lakes, one of the factors known to affect M. spicatum growth rates. Factors such as the number of game fish species and number and types of boat ramps or proximity to roads were generally less important lake characteristics. Our models can be useful tools for developing management strategies to prevent or slow the spread of M. spicatum and aquatic invaders, such as the zebra mussel, that can attach to it and thus be dispersed. Our models also exemplify a general approach for slowing or stopping the spread of other invading species.