unlogged forest function as mitigating influences. 

 Other alternatives, however, merit consideration. 



A factor that may have contributed to overlooking 

 biologically important differences in abundance was 

 the small sample consisting of 10 replicates within 

 each stand condition. This reduced the power, or 

 the ability to detect differences in means, for our 

 statistical tests. The relative scarcity of many 

 species contributed to this problem. Nineteen 

 species were detected less than eight times during 

 censusing; these were not included in the species- 

 level analysis. Nonetheless, among the species 

 included, zero counts at census points were com- 

 monplace, and this property of census data, in 

 combination with the small sample size, served to 

 inflate variance and render different sample means 

 statistically indistinguishable. 



An example of the conservative nature of our 

 statistical tests is the killdeer, which was detected 

 only at two points in clearcut habitat during both 

 years. As a ground-nesting and ground-foraging 

 species, probably more killdeer will use cleared than 

 uncut habitat, but large variance (mean number 

 counted per point in clearcuts = 0.15, standard 

 error = 0.08) caused a failure to reject the null 

 hypothesis of equal means. The power for this 

 analysis of variance model was 44 percent, which 

 was rather low. We grouped species into guilds 

 (table 3) to partially circumvent the problem of 

 uncommon species' counts and, as shown in table 3, 

 ground foragers and ground nesters showed a highly 

 significant difference in abundance between habi- 

 tats. They were especially common in clearcut and 

 partially cut areas. 



Even if population sizes do not vary among stand 

 conditions, it is conceivable that the abundance of 

 a species may not be a valid indicator of habitat 

 suitability (Van Home 1983). A given habitat may 

 act as an ecological sink where individuals experi- 

 ence low reproductive success or high mortality in 

 comparison to more favorable habitats (Wiens and 

 Rotenberry 1981). For example, birds may be 

 abundant in cleared areas but raise fewer offspring 

 per nest than conspecifics in contiguous forest. The 

 resupply of individuals to inadequate habitat from 

 optimal areas could be behaviorally mediated. 



Factors that ultimately determine habitat selec- 

 tion include food and cover availability, but Hilddn 

 (1965) suggested that vegetation structure is the 

 proximate cue birds use to select territories. Birds 

 arriving during spring in northern latitudes must 

 select nesting territories based on structural fea- 

 tures of the habitat rather than a direct assessment 

 of food availability. If we manipulate habitat to 

 provide appropriate cues for these species, we could 

 see birds within the habitat even though it is not 

 suitable. Thus some human-altered environments 



may frinction as ecological sinks (Gates and Gysel 

 1978). 



Consider woodpeckers, for example. We detected 

 no difference in abundance between stand condi- 

 tions for red-naped sapsuckers, hairy woodpeckers, 

 and northern flickers (table 2). A proximate cue 

 that may influence a woodpecker settling into a 

 nesting territory could be the previous year's nest 

 tree. Lawrence (1967), studying yellow-bellied 

 sapsuckers {Sphyrapicus varius) in Ontario, sug- 

 gested that pair bonds were reestablished through 

 site rather than mate recognition, and females in 

 particular cued in on the former nest tree. Thus, 

 by reserving snags and deciduous timber during 

 the harvest and site preparation treatments on the 

 THSA, adult woodpeckers returning in the spring 

 may have been motivated to establish territories 

 within inadequate habitat. A detailed study that 

 evaluates habitat quality through measures of 

 reproductive success among red-naped sapsuckers 

 is in progress (Tobalske and others 1990). 



Wiens and others (1986) suggested that interpre- 

 tation of a species' response to habitat alteration 

 may depend on the spatial scale of the manipula- 

 tion. The largest clearcut on the THSA was 35 

 acres (14 ha) (fig. 2), but commercial clearcuts may 

 cover areas much larger (Dickson and others 1983). 

 Extrapolating the results of a small-scale study to 

 that of a large-scale system is ignoring biological 

 reality. Moreover, landscape-level changes on a 

 very large scale, such as a portion of a State or 

 country, may significantly affect bird species yet 

 be impossible to address without regional data. 



MANAGEMENT 

 RECOMMENDATIONS 



The variation in abundance within the community 

 of breeding bird species relative to logged or un- 

 logged stand conditions indicates that management 

 for any particular species or guild will probably fail 

 to provide for the integrity of the entire system. In 

 short, diversity of habitat will likely promote di- 

 verse bird assemblages. During timber harvest, 

 tree-dependent bird species are losing access to an 

 important resource, and it is generally this portion 

 of the bird community for which managers seek to 

 mitigate adverse effects. 



The most striking feature of the clearcuts on 

 Terrace Hill is the large number of standing larch 

 snags and live paper birch trees (fig. 4). It is fea- 

 sible that the snags and living trees, along with the 

 small size of the cutting units in relation to inter- 

 spersed unlogged forest, mitigated the effect of the 

 timber harvest for tree-dependent bird species, 

 ranging from cavity nesters, such as red-naped 

 sapsucker and red-breasted nuthatch, to foliage 



8 



