Hamer 



Chapter 17 



Inland Habitat Associations in Western Washington 



<1.1 detections/morning. The highest occupied stand in 

 Washington was located at 1,105 m in elevation, in the 

 North Cascades Physiographic Province, near the upper 

 headwaters of Crevice Creek. The highest occupied stand in 

 the Olympic Peninsula Physiographic Province was located 

 1,025 m in elevation near Spot Lakes on the Hood Canal 

 Ranger District, Olympic National Forest. The South Cascades 

 physiographic province had an occupied stand 1,051 m in 

 elevation, located 13 km south of Alder Lake in Lewis 

 County, near the East Fork Little Creek drainage. More than 

 98 percent of all detections in Washington were recorded 

 below 1,067 m in elevation. 



Forest Type and Physiographic Province 



Forest types surveyed in Washington included stands 

 dominated by western hemlock, Douglas-fir, Sitka spruce, 

 silver fir, and mountain hemlock. These stands commonly 

 had a large component of western red cedar. 



The mean detection rates for 229 old-growth stands 

 were compared between the five physiographic provinces in 

 Washington. The North Cascades Province had a mean 

 detection rate of 23.5 detections/survey morning (n = 117 

 sites, s.d. = 7.9) and 40 percent of the old-growth stands 

 surveyed were verified as occupied. The Olympic Peninsula 

 had a similar rate of 18.3 detections/survey morning (n = 67, 

 s.d. = 8.2) and a 37 percent occupancy rate of old-growth 

 stands. The South Cascades had a detection rate of 12.5 

 detections/survey morning (n = 30, s.d. = 15.7), and 10 

 percent of the stands surveyed were occupied. The Puget 

 Trough had the lowest detection rate of any province, with 



1.3 detections/survey morning (n = 7, s.d. = 0, and no 

 occupied stands, but the number of stands sampled was 

 small. The Southwest Coast had the highest detection rate 

 (90.1 detections/survey morning; n = 8, s.d. = 7.4) and the 

 highest occupancy rate (50 percent), but the number of old- 

 growth stands surveyed was also small. 



Stand Characteristics 



Statistical Model 



The results of the logistic regression model results from 

 the 1994 study gave a total accuracy rate of 74.2 percent for 

 a predicted probability of occupancy for each stand analyzed. 

 The classification accuracy of occupied stands was 67.2 

 percent. The classification accuracy of unoccupied stands 

 was 79.2 percent Of the 32 stands with a predicted probability 

 of occupancy >0.75, 74 percent were occupied. Of the 54 

 stands with a predicted probability of occupancy of <0.25, 

 93 percent were unoccupied. A total of 65 stands (43 percent) 

 had probability values >0.25 and <0.75. 



Eight forest variables were included in the model by the 

 stepwise logistic regression procedure. These variables best 

 predicted occupancy of a stand by murrelets (table 3). The 

 stepwise selection procedure was completed in 10 steps. 



The probability of occupancy of an old-growth stand 

 increased with increasing percent topographic slope, total 

 number of potential nest platforms/ha, stem density of 

 dominant trees, mean d.b.h. of western hemlock, and the 

 moss coverage (percent moss) on the limbs of dominant 

 trees (table 3). The probability of occupancy of a stand 

 decreased with increasing stand elevation, canopy closure, 



Table 3 Stepwise Logistic Regression Model of Marbled Murrelet habitat in western Washington. The eight variables are listed 

 in order of their probability values 



USDA Forest Service Gen. Tech. Rep. PSW-152. 1995. 



169 



