Monday, June 4, 2007 - 4:00 PM
214

Modeling the spread of invasive African tilapia in northeastern Mesoamerica

Peter Esselman and J. David Allan. School of Natural Resources and Environment, University of Michigan, Ann Arbor, MI 48109

Advances in computing technology have facilitated the accessibility of a family of modeling techniques collectively referred to as "machine learning" approaches.  When combined with geographic information systems, these approaches have been shown to yield highly accurate predictions about ecological patterns, which can help with hypothesis generation.  Machine learning approaches were used to predict the spatial distribution patterns of non-native African tilapias (Oreochromis niloticus) in northeastern Central America using neural networks (ANNs) and maximum entropy approaches trained with field data and geospatial data about landscape environmental context.  The utility of the different approaches for visualization of probable distributions, and for identification of the strongest correlates to resultant spatial patterns is explored, and hypotheses relevant to understanding the tilapia invasion process are discussed.

See more of Special Session - Cascades, Chains, and Webs: Alternative Approaches to Understanding Complex Systems
See more of Special Sessions

See more of The NABS 55th Annual Meeting