Interest in neotropical
migratory songbirds (NMSB) in forest environments of the western
United States has increased because of widespread declines of
several species. Currently, forest managers frequently use expert-based
models that predict NMSB occurrence in forest stands based on
forest type and successional stage. Newer probabilistic models
have been developed that predict NMSB occurrence based on structural
characteristics of stands that are commonly collected during
forest inventories or that can be collected using rapid inventory
techniques. However, these models have not been validated, nor
is it certain that they can be applied outside the areas in
which they were developed. An additional question concerning
model validation is related to the intensity of the survey effort
required to reliably detect a bird species in a forest stand
(Smith and Connors 1986, pages 45-50 in Wildlife 2000).
Fifteen logistic regression habitat-relationship models for
forest birds have been developed by Weyerhaeuser biologists.
The modeled species are: orange-crowned warbler, Bewick's
wren, black-throated gray warbler, black-headed grosbeak,
brown creeper, evening grosbeak, gray jay, hairy woodpecker,
hermit warbler, red-breasted nuthatch, rufous hummingbird,
song sparrow, Steller's jay, warbling vireo, and western tanager.
These habitat-association models were developed to provide
a basis for habitat assessment and diagnosis.
The
purpose of this study is to test the efficacy of the models
to predict the presence or absence of species given particular
habitat characteristics. Three approaches were used to evaluate
the models: 1) evaluating the accuracy of prediction, 2) whether
there was a significant positive linear relationship between
the predicted probability of occurrence and the frequency
of occurrence at the stand level, and 3) whether the variable
coefficients resulting form our data fell within the confidence
intervals of the coefficients in the Weyerhaeuser models.
For additional information about this research study see the
2001 CFER
Annual Report. (2.7 MB)