Verifying Manitoba's 1994 draft barred owl habitat suitability index model

Abstract

Biologists designated the barred owl (Strix varia) as an indicator species for older forests and in December 1994, they de eloped a draft habitat suitability index model (HSI) to describe this dependent relationship. The primary purpose of the research was to evaluate Manitoba's 1994 draft HSI model in Manitoba. The specific objectives were to describe habitat characteristics associated with barred owls in Manitoba; to verify the HSI model developed for the barred owl and; to make recommendations for modifying the HSI model. Data from Manitoba's Nocturnal Owl Survey and the Forest Resource Inventory database were used. Habitat associations were examined at 6.25 ha and 400 ha scales using logistic regression (LR). The LR model predictive capability was 80%. The 6.25 ha scaled LR model consisted of 4 variables: conifer forests, crown class 4, cutting class 4 and cutting class 5. The cutting classes were the most influential variable in the LR model. At the 400 ha scale, the LR model consisted of 3 variable sets: cutting class, species composition and unproductive forests. Cutting class was the most influential variable set. The predictive capability of the LR model was 83%. The probability of detecting a barred owl given the model and data indicates the barred owl prefers older, hardwood dominated mixed wood forests and avoids unproductive and younger forests. The LR analysis refutes the HSI model assumptions that crown class is a more influential variable than species composition. (Abstract shortened by UMI.)