INFLUENCE OF HABITAT ABUNDANCE AND FRAGMENTATION ON NORTHERN SPOTTED OWLS IN WESTERN OREGON

Current management for the northern spotted owl (Strix occidentalis ca urina) is largely driven by metapopulation models or individually-base d models that assume the success of juvenile dispersal in a fragmented landscape is a primary factor determining the future existence of spo tted owls in the Pacific Northwest. We tested hypotheses about fragmen tation by comparing sites known to be occupied by spotted owls with ra ndom sites to determine if relationships existed between landscape ind ices and spotted owl presence and productivity in western Oregon. From a total of 445 known spotted owl sites within the Bureau of Land Mana gement's (BLM) checkerboard patterned lands in western Oregon, we rand omly selected (1) 50 long-term data sites to determine if landscape ch aracteristics influenced site occupancy or reproduction, (2) 50 random owl sites to evaluate possible biases in the long-term data sites, an d (3) 50 random landscape locations for comparison with the 50 random owl sites. BLM staff classified from aerial photographs the mosaic of forest successional stages within a 3.4-km-radius circle surrounding e ach of the 150 sites. From these mosaics, we calculated several indice s of landscape characteristics and forest fragmentation for 0.8-, 1.6- , 2.3-, and 3.4-km-radius circles. Results were combined with data on occupancy and reproduction to test the null hypotheses that landscape characteristics did not affect site location, site occupancy, or repro ductive success of spotted owls. Landscape indices did not differ betw een long-term owl sites and randomly-selected owl sites, indicating li ttle bias in our sample of long-term data sites. Landscape characteris tics at random owl sites differed significantly from those at random l andscape locations. Differences were greatest for 0.8-km-radius circle s sur rounding the study sites, suggesting that site selection by spot ted owls may be most strongly affected by landscape characteristics wi thin a 0.8-km-radius circle (less than or equal to 200 ha). Statistica lly significant differences also were found for radii up to 3.4 km, bu t most of those differences did not contribute significant new informa tion beyond the differences existing in the core area of the circles.R andom owl sites contained more old-growth forest, larger average size of old-growth patches, and larger maximum size of old-growth patches t han occurred in random landscape locations, for all circle radii (P < 0.01). Additionally, random owl sites contained less young-age forest within 0.8-km-radius circles than did random landscape locations. Howe ver, amount of clearcut forest did not differ between random owl sites and random landscape locations. None of the forest fragmentation indi ces except size of old-growth patches was strongly related to site sel ection, none was strongly related to frequency of occupation of sites, and only fractal dimension was moderately related to reproduction. In stead, the major influences of landscape pattern were related to amoun t of habitat. Amount of habitat dominated in resource selection probab ility functions (RSPF) for western Oregon, and these RSPF's can be use d to predict the probability that a given landscape mosaic will be a s uitable spotted owl site.