Kristina McNyset1

/	Mentors: Bruce Jones2, Tony Olsen1, Henry Walker3, James Wickham4,

Brian Hill5, Henry Lee1, Lester Yuan10
1NHEERL, Corvailis, OR; 2NERL, Las Vegas, NV; 3NHEERL, Narragansett, Rl; 4NERL,
Research Triangle Park, NC; 5NHEERL, Duluth, MN; 6NHEERL, Newport, OR; 7NCEA,

Washington, DC

Predictive ecological niche modeling in aquatic systems

Research Goal: Develop spatially explicit models of biotic conditions

in aquatic systems

Projects

•	Determine best methods for predictive modeling of species distributions in
stream systems

•	Integrate point-sampled and landscape-level data in predictive ecological
niche modeling analyses in freshwater systems

•	Develop ecological data sets for use in spatially explicit modeling in
estuarine systems

Anticipated Outcomes

A set of modeling tools, datasets, and models that will increase our

understanding of biotic conditions in aquatic systems

Predicted Distribution of the nuisance diatom species
Didvmosphenia geminata (Lvngbve) M. Schmidt

This is the 10 best-model subset from a GARP (Genetic Algorithm for Rule-set Prediction) model run of 200
models. The color gradient from pinkto dark red indicates increasing model agreement between best-subset
models. The green circles are the training data used the build the models taken from the USGS NAWQA
project, the yellow squares are the testing data taken from the EPA WEMAP project. Overall omission was
110% at a threshold of 5 models.

-------