Hamer 



Chapter 17 



Inland Habitat Associations in Western Washington 



Table 5 Summary of seven characteristics measured for six species of conifers available as nest trees by the 

 murreletin Washington state. Only trees >81 cm d.b.h. were measured. The mean, range, and sample size are shown. 

 See text for moss and lichen cover categories 



records indicate that fledglings may travel some distance 

 before becoming grounded. 



Detection and stand occupancy rates increased with 

 more older forest available on the landscape. For all provinces, 

 the low detection and occupancy rates near the coast were 

 probably due to the presence of large amounts of unsuitable 

 or marginal habitat in the Puget Trough and near coastal 

 lowland areas of the Olympic Peninsula. In a study 

 encompassing the entire South Fork of the Stillaguamish 

 River basin in northern Washington, significantly higher 

 numbers of murrelets were observed in old-growth and 

 mature forests than either rock/talus, clear-cut/meadow, or 

 small saw/pole cover types (Hamer and Cummins 1990). 

 Murrelet detection rates increased rapidly when the 

 percentage of old-growth and mature forest cover types 

 found within a 2,000-m-radius circle around each survey 

 station made up more than 30 percent of the landscape. 

 Mean detection rates for sites located in these areas ranged 

 between 1 and 20 detections/morning (x = 5.7; s.d. = 5.8). 

 All sites with <30 percent old-growth and mature forest 

 cover had <1.5 detections/morning ( x = 0.2; s.d. = 0.4). An 

 analysis of the landscape features associated with occupied 

 and unoccupied stands in Washington found that the amount 



of old-growth and large sawtimber available best predicted 

 murrelet occupancy at the stand level (Raphael and others, 

 this volume). Sites with a higher proportion of these mature 

 forest classes were more likely to have evidence of nesting 

 or occupancy than unoccupied sites. 



Stand Characteristics 



Statistical Model 



Overall the model correctly predicted occupancy on 

 about 74 percent of the sites. However, this success rate 

 may be biased because the same sites that were used to 

 build the model were used to test it. Because the model 

 treats occupancy as a categorical variable, individual sites 

 that scored near 0.5 were difficult to judge. In these cases it 

 was more convenient to think of occupancy as a continuous 

 variable where the higher probability scores indicated more 

 suitable habitat and a higher probability of being occupied 

 by murrelets. Errors in the classifications of stands could 

 be due to several factors: (1) some stands determined to be 

 unoccupied from field surveys may have actually been 

 occupied; (2) it is possible in some instances that birds may 

 be occupying stands of marginal habitat and; (3) the 

 vegetation sampling for some stands may have been 



172 



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



