513
Modeling Brook Trout Presence and
Absence from Landscape Variables Using
Four Different Analytical Methods
Paul J. Steen*
School of Natural Resources and Environment, University of Michigan
170 Dana Building, Ann Arbor, Michigan 48109, USA
Dora R. Passino-Reader
U.S. Geological Survey, Great Lakes Science Center
1451 Green Road, Ann Arbor, Michigan 48105, USA
Michael J. Wiley
School of Natural Resources and Environment, University of Michigan
170 Dana Building, Ann Arbor, Michigan 48109, USA
Abstract.—As a part of the Great Lakes Regional Aquatic Gap Analysis Project, we evaluated
methodologies for modeling associations between fish species and habitat characteristics at a
landscape scale. To do this, we created brook trout Salvelinus fontinalis presence and absence
models based on four different techniques: multiple linear regression, logistic regression, neu-
ral networks, and classification trees. The models were tested in two ways: by application to an
independent validation database and cross-validation using the training data, and by visual
comparison of statewide distribution maps with historically recorded occurrences from the
Michigan Fish Atlas. Although differences in the accuracy of our models were slight, the logis-
tic regression model predicted with the least error, followed by multiple regression, then clas-
sification trees, then the neural networks. These models will provide natural resource managers
a way to identify habitats requiring protection for the conservation of fish species.
American Fisheries Society Symposium 48:513–531, 2006
© 2006 by the American Fisheries Society
*Corresponding author: psteen@umich.edu
INTRODUCTION
Knowledge of the habitats required to maintain
the growth, survival, and reproduction of fresh-
water fish species and populations is necessary
for conservation planning and decision making.
In practical application, however, habitat require-
ments are often incompletely known. Therefore,
biologists commonly use data on a fish’s habitat
selection, based on field observations of species
occurrence or densities (Rosenfeld 2003). Given
data on habitat characteristics and observed fish
distributions, correlative habitat associations can
be used to predict the occurrence or densities of
fish in locations where samples have not been col-
lected. These predictions are useful for identify-
ing habitat units important to target species but
vulnerable to alteration and degradation by hu-
mans, and lacking protective status. Such habi-
tats represent “gaps” in conservation strategy.
The goal of the U.S. Geological Survey, Gap
Analysis Program (GAP) is to keep common spe-
cies common by identifying those species not ad-
equately represented in existing conservation areas
(Scott et al. 1993). In the past decade, gap analy-
ses have been performed in terrestrial systems
across the United States and in the mid-1990s
an aquatic gap pilot began in Missouri. In 2001,
GAP funded the first regional aquatic gap analy-
sis in the eight Great Lakes states: Minnesota,