Fishing in streams for ideas: Towards greater spatial competency in habitat models.
University of Wisconsin, Madison, WI 53706 USA 1
The application of methods developed in terrestrial ecology to habitat models for stream-dwelling organisms is complicated by difficulties in quantifying some of the important properties of streams, including their connectivity, complexity, and temporal variability. Conversely, the linearity of streams lends itself to a more explicit recognition of spatial patterns, including autocorrelation and the existence of environmental gradients; stream habitat models must directly confront issues that are often ignored in terrestrial systems. Two distinct modelling approaches demonstrate the importance of stream topology for predictive models of fish distributions. Firstly, environmental data and a statewide database of fish collection records (assembled by the DNR) for Wisconsin are used to show how logistic regression methods can be adapted to make useful, probabilistic predictions in aquatic systems. Models for several fish species are developed from a set of readily-mapped environmental variables including elevation, hydrology, surficial geomorphology, soil type, stream order and explicit measures of distance. Measures of stream topology and the distances of collection records from the mainstem of the nearest large river emerge as useful predictor variables from this exercise. Secondly, I consider the importance of stream topology using a non-linear, stochastic cellular automaton model that assumes a homogeneous habitat. The model predicts that streams with proportionally more branches are likely to be invaded (or recolonised) at faster rates than more linear streams, and should support a higher abundance of aquatic species. Concepts arising from the study of streams offer a potentially useful paradigm from which issues of connectivity, habitat fragmentation and other explicitly spatial problems in terrestrial ecology and conservation can be approached.
Keywords: modeling, streams, fish, logistic regression, aquatic ecology
This abstract is being presented at: 5:00 PM in session:
Oral Session #62: Freshwater Fish Ecology.