The methodology of ecological niche modeling has long been a topic of interest in the lab. A particular challenge has been the development of methods that successfully draw a boundary around the set of habitable conditions (the fundamental niche) using data only on species occurrences. A new study by Sean Maher, Antoine Guisan, Chris Randin and John Drake looks at a range of modeling approaches that can be fit to either presence-only or presence-absence data. Models were fit to data on 106 species in exhaustively sampled vegetation quadrats in the Swiss Alps. The combination of these data and models enabled a comparison of presence-only and presence-absence modeling approaches that could be independently validated. The upshot: When suitably fit, presence-only models can equal the performance of presence-absence models, removing the need for “pseudo-absence” sampling and similar maneuvers needed to shoehorn ecological niche modele into a classification framework.
- Maher, S.P., A. Guisan, C. Randin & J.M. Drake. 2014. Pattern recognition ecological niche models fit to presence-only and presence-absence data. Methods in Ecology & Evolution 5:761-770. [online]