Species Distribution Modeling
Species distribution models (SDMs) based on empirical field data are a popular tool for understanding species response to environment and making projections about future occurrence. SDMs have been used to investigate the impacts of climate change and invasive species and in conservation applications. However, SDM reliability is challenged by issues of data quality, such as inadequate sampling, choice of analytical model, and ability to accurately evaluate model performance. The availability of large predictor sets increases the likelihood of identifying factors underlying species distribution, however the complex and often non-linear relationship between predictors and response can make traditional parametric modeling techniques inappropriate. As a result, ecologists have turned to machine learning or algorithmic approaches, which have been successfully used in a range of SDM applications. My research focuses on developing reliable SDM techniques, using both machine learning and regression modeling methods to understand the factors influencing the distribution of freshwater fish. This work can be more broadly applied to developing reliable predictive models for a range of species and environments.
SDMs and Invasion Ecology
It is known that invasive species can cause dramatic economic and ecological harm outside their native ranges, however the ability to predict and manage invasions has proven challenging. There are numerous hypotheses to explain the establishment of species in novel ecosystems. Most of these perform very well at explaining invasion success for particular species or regions, but are inadequate for broader application. On goal of my research is the development of a broad modeling approach that can be applied to understanding and predicting invasions across ecosystems and taxa. The use of an ensemble modelling approach, with both empirical and simulated data, provides flexibility and improves model performance when dealing with varying species responses to environment.
Non-Ecological Applications
I am currently exploring modeling approaches that can be used to identify students who are at-risk of academic failure prior to beginning a course, to provide appropriate intervention and curriculum modifications.