Kuletz and others 



Chapter 15 



Inland Habitat Suitability in Southcentral Alaska 



activity levels and behaviors in non-forested and forested habi- 

 tats may not reflect actual differences in murrelet abundance. 

 For example, murrelets may be more vulnerable to predation 

 in open areas and therefore less active around ground nests. 



Effects of Survey Methods 



Levels of murrelet activity did not vary among survey 

 methods. However, significantly more occupied behaviors 

 were observed when surveys were done from upland sites 

 rather than from the shoreline or a boat. Occupied behaviors 

 may be hard to detect during surveys conducted from a boat 

 because the observer is often 50-100 m from forest habitat. 

 However, occupied behaviors were equally low in frequency 

 when surveys were done from the shoreline. Thus, our results 

 may reflect real differences in habitat use. Although murrelets 

 sometimes nest within a few hundred meters of the shore 

 (Cody, unpubl. data; Kuletz, unpubl. data; Marks, unpubl. 

 data; Naslund and others, in press), they may use areas along 

 the shoreline less frequently than those further inland (Hamer, 

 this volume). The effect of survey method was confounded 

 with effect of survey area, because boat and shore-based 

 surveys predominated at Prince William Sound and Kenai 

 Fjords National Park, whereas upland surveys predominated 

 at Naked Island and Afognak Island. The latter had very 

 high activity levels, large trees and high epiphyte cover 

 (Naslund and others, in press), and the high occupied status 

 rate could have been due to truly higher nesting densities. 



Sources of Unexplained Variation 



Our best multiple regression model explained 52 percent 

 of the variation in murrelet activity. There were many potential 

 sources of unexplained variation. Because sites were surveyed 

 only once, day-to-day variation within the same area could 

 have contributed to incorrect estimation of general activity 

 level of a given site. We did not account for observer 

 variability, which can introduce additional bias to murrelet 

 surveys (Kuletz and others 1994c; Ralph, pers. comm.). 

 Because each area was generally surveyed by different 

 observers, area effects could be due partially to observer 

 variability. In addition, differences in sampling design may 

 have contributed to area effects or other variation. For 

 example, all forest was treated equally in our analyses, yet 

 forest characteristics (e.g., age structure, volume, tree species) 

 are quite variable. The Naked Island group was the only area 

 for which specific forest types were stratified and sampled. 



Prevailing winds, local topography and vegetation patterns 

 varied throughout the study area. Therefore, the geographic 

 range of study sites likely contributed to the variation in 

 murrelet activity we observed. In addition, murrelet nesting 

 distribution may vary with availability of suitable habitat. For 

 example, murrelets may be more dispersed in Prince William 

 Sound if prime nesting habitat is abundant and widespread, 

 whereas nesting density may be higher in good habitat on the 

 Kenai Peninsula if suitable habitat is sparse. Thus, the lower 

 activity levels in Prince William Sound, relative to the Kenai 

 Peninsula, may reflect differences in habitat availability, rather 

 than habitat suitability, between the two areas. 



An important factor not considered in our models was 

 the adjacent marine environment and the availability of 

 foraging habitat. These factors must ultimately determine 

 the use of suitable nesting habitat. Thus, the apparent 

 increase in murrelet activity from Prince William Sound to 

 Afognak Island may also reflect large-scale differences in 

 prey availability. 



Conclusions 



These models primarily serve as descriptive tools until 

 they can be tested with independent data. However, we were 

 able to explain 52 percent of the total variation in Marbled 

 Murrelet activity levels based on temporal, topographic, and 

 habitat characteristics. Further, our results suggest an 83 

 percent success rate of classifying murrelet nesting habitat 

 in the areas examined on the basis of occupied behavior. The 

 features indicative of murrelet nesting habitat include low 

 elevation locations near the heads of bays, with extensive 

 forest cover of large old-growth trees. In some areas, such as 

 the Kenai Fjords, location relative to bay heads may be less 

 important. The best predictors of nesting habitat in forested 

 areas are high epiphyte cover and large numbers of potential 

 nesting platforms on trees. 



Our results were derived from surveys designed to 

 estimate murrelet use of forested habitat. Potential variation 

 in murrelet behavior associated with habitat type (i.e., forest 

 or non-forest) has not been adequately examined and could 

 influence accurate interpretation of survey results. Therefore, 

 caution should be exercised when extrapolating observed 

 trends on a broad scale across the landscape. 



Acknowledgments 



The contribution of several studies was integral to this 

 paper. We thank the USDA Forest Service (Chugach National 

 Forest), who was a cooperative partner in the studies in 

 Prince William Sound and Naked Island areas, and the U.S. 

 Fish and Wildlife Service, Division of Realty who conducted 

 murrelet surveys on Afognak Island. This research was funded 

 through the U.S. Fish and Wildlife Service, Division of 

 Migratory Bird Management as part of the Exxon Valdez oil 

 spill restoration program and the Division of Realty. For 

 assistance on projects we thank G. Esslinger, A. Belleman, I. 

 Manley, S. Anderson, D. Huntwork, K. Rausch, K. Fortier, J. 

 Maniscalco, E. Tischenor, L. Fuller, B. Grey, J. Fadely, B. 

 Fadely, D. Zwiefelhofer, G. Johnson, V. Vanek, M. Nixon, 

 T. Nelson, G. Landua, D. Goley and D. Kaleta. For the GIS 

 support, we also thank T. Gerlach of the Division of Realty 

 and T. Jennings and C. Wilder, all of them with U.S. Fish and 

 Wildlife Service. From the USDA Forest Service we thank 

 C. Hubbard, R. DeVelice, Z. Cornett, and B. Williams. S. 

 Klosiewski provided guidance on study design and data 

 analysis. The comments of R. Barrett, P. Connors, Chris 

 Iverson, Michael McAllister, C. John Ralph, and an anonymous 

 reviewer greatly improved this manuscript. 



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



149 



