ConserveOnline - A Review of Modeling Approaches to Predict Fish Habitat and Distribution, by William L. Fisher

You are here: Home → workspaces → Alaska Salmon Habitat Prediction Workshop → Session II: Existing Approaches → A Review of Modeling Approaches to Predict Fish Habitat and Distribution, by William L. Fisher

A Review of Modeling Approaches to Predict Fish Habitat and Distribution, by William L. Fisher

By Web Admin on 6/5/2007 | Keyword(s): Session ii: existing approaches

Predictive fish habitat distribution models have increased in recent years, partly because of the increased use of geographic information systems (GIS) and statistical analysis techniques. In modeling fish distribution, issues of scale, model type, and available habitat information influence our ability to develop both precise and realistic models. A first step in distribution modeling is the development of a conceptual model. I reviewed 15 recent studies that predict fish distribution at local (site) or regional (landscape) scales. Multiple logistic regression was the most common type of statistical analysis used to predict the probability of occurrence of species or their habitats, although discriminate analysis and artificial neural network models were also used. At the local scale, microhabitat variables such as depth, velocity, substrate, and cover were used to predict either occurrence, density or biomass. At the regional scale, probability of occurrence was predicted from macrohabitat variables such as basin size, elevation, water temperature, precipitation, and land cover/land. Because of its enormous size, diversity of habitat types, limited data, and remoteness, precisely predicting the distribution of salmon species in Alaska will be a large challenge.

ConserveOnline is a part of the Conservation Commons. Some content may be subject to Creative Commons licenses.
© ConserveOnline.