HARVARD UNIVERSITY

Library of the Museum of Comparative Zoology

r

TKelMIsoti Bulletin

PUBLISHED BY THE WILSON ORNITHOLOGICAL SOCIETY

VOL. 109, NO. 1 MARCH 1997 PAGES 1-194

(ISSN 0043-5643)

The Wilson Ornithological Society Founded December 3, 1888

Named after ALEXANDER WILSON, the first American Ornithologist.

President— Keith L. Bildstein, Hawk Mountain Sanctuary, RR 2, Box 191, Kempton, Pennsylvania 19529-9449.

First Vice-President — Edward H. Burtt, Jr., Department of Biology, Ohio Wesleyan University, Delaware, Ohio 43015.

Second Vice-President — John C. Kricher, Biology Department, Wheaton College, Norton, Mas- sachusetts 02766.

Editor — Charles R. Blem, Department of Biology, Virginia Commonwealth University, P.O. Box 842012, Richmond, Virginia 23284-2012.

Secretary — John A. Smallwood, Department of Biology, Montclair State University, Upper Mont- clair, New Jersey 07043.

Treasurer — Doris J. Watt, Department of Biology, Saint Mary’s College, Notre Dame, Indiana 46556.

Elected Council Members — Carol A. Corbat and William E. Davis (terms expire 1997), and Margaret C. Brittingham and Herbert T. Hendrickson (terms expire 1998), Peter C. Frederick and Danny J. Ingold (terms expire 1999).

Membership dues per calendar year are: Active, $21.00; Student, $15.00; Family, $25.00; Sus- taining, $30.00; Life memberships $500 (payable in four installments).

The Wilson Bulletin is sent to all members not in arrears for dues.

The Josselyn Van Tyne Memorial Library

The Josselyn Van Tyne Memorial Library of the Wilson Ornithological Society, housed in the University of Michigan Museum of Zoology, was established in concurrence with the University of Michigan in 1930. Until 1947 the Library was maintained entirely by gifts and bequests of books, reprints, and ornithological magazines from members and friends of the Society. Two mem- bers have generously established a fund for the purchase of new books; members and friends are invited to maintain the fund by regular contribution, thus making available to all Society members the more important new books on ornithology and related subjects. The fund will be administered by the Library Committee, which will be happy to receive suggestions on the choice of new books to be added to the Library. William A. Lunk, University Museums, University of Michigan, is Chairman of the Committee. The Library currently receives 195 periodicals as gifts and in ex- change for The Wilson Bulletin. With the usual exception of rare books, any item in the Library may be borrowed by members of the Society and will be sent prepaid (by the University of Michigan) to any address in the United States, its possessions, or Canada. Return postage is paid by the borrower. Inquiries and requests by borrowers, as well as gifts of books, pamphlets, reprints, and magazines, should be addressed to: The Josselyn Van Tyne Memorial Library, University of Michigan Museum of Zoology, Ann Arbor, Michigan 48109. Contributions to the New Book Fund should be sent to the Treasurer (small sums in stamps are acceptable).

The Wii.sdn Bulletin (ISSN (K)43-.5643)

THE WILSON BULLETIN (ISSN (KITt-.'jfrW) is piiblislicd qiiaiK-rly in Marrii, June, September, ami Deeember by tbe Wilson Ornilbologieal Soeiely. 810 East lOlb Street. Uwrenee. KS W,(Et4-8897. The siibseription priee. both in the United States and elsewhere, is [)er year. Penmbeals postage paid at Liwrenee. KS. POSTM ASTER; Send address elianges to THE WILSON

BULLETIN. P.O. Box 1897, Uwrenee. KS WKm-8ii97. '

Bac k issues or single copies are available for $I2.(X) each. Most back issues of the Hullelin are available and may be ordered from the* rrf*asijre*r. Spee'lal price's will he <|uote*(l for quantity orders.

All articles and eommunirations for publications. Ixmks and publications for reviews should be addres.sed to tbe Editor Ex- changes should be addressed to The Josselyn Van Tyne Memorial Library. Museum of Zmilogy. Ann Arbor. Michigan ABItw’ Siiliseriptions. changes of addre.ss ami claims for undelivered copies should 1m- sent to tbe OSNA PO Box 1897 Uwrenee KS «)<m-8897. Phone: (91.1) 8-t;i-l22l; EAX: (91.i) 84:M274.

THE WILSON BULLETIN

A QUARTERLY MAGAZINE OF ORNITHOLOGY Published by the Wilson Ornithological Society

VoL. 109, No. 1 March 1997 Pages 1-194

Wilson Bull: 109C1), 1997, pp. 1-27

POPULATION CHANGES IN BOREAL FOREST BIRDS IN SASKATCHEWAN AND MANITOBA

David A. Kirk,' Antony W. Diamond,^ Alan R. Smith,^ George E. Holland,^ and Paul Chytyk^

Abstract. — We counted breeding birds at four plots in central Saskatchewan and four in western Manitoba in 1990-1992 to examine changes in species composition and abundance since the plots were originally surveyed in 1972—1973. In Saskatchewan, more species of Neotropical migrants decreased (16) than increased (9; P > 0.05 < 0.1). Combined densities of Neotropical migrants declined (14—44%) on all of the Saskatchewan plots; Tennessee Warblers {Vermivora peregrina). Red-eyed Vireos (Vireo oUvaceus), and Ovenbirds (Seiurus aurocapillus) declined on the most plots and by the greatest magnitude and Black-throated Green Warblers (Dendroica virens) and Rose-breasted Grosbeaks {Pheucticus ludovicianus) also decreased. Six of seven Neotropical migrants showed the same direction of change as in a province-wide Breeding Bird Survey. Successional changes did not account for de- creased densities of these five species, but they may partly explain increases in some other species. Fluctuations in food supply (e.g., spruce budworm [Choristoneura fumiferana]) could not explain changes, because some species that should have responded numerically to budworm outbreaks that occurred in the 1990s had increased whereas others had de- creased at the same site. The surrounding forest remained continuous over the 17-18 years, so changes in forest area cannot account for the declines. In Manitoba, more Neotropical migrants increa.sed (19) than decreased (11) according to combined densities from four plots. Combined densities of Neotropical migrants also increased at three of the four plots (33- 123%). Ten of 15 Neotropical migrant species showed different ‘trends’ than a province- wide analysis of BBS data. All changes in Manitoba could be attributed to vegetation succession on the plots and forest fragmentation in surrounding landscapes. Trends in Sas- katchewan may be representative of general declines within continuously forested boreal landscapes, whereas those in Manitoba may reflect reduced opportunities for breeding in

> Aqidla Applied Ecologists. C.P. 47, Wakefield, Quebec .lOX 3G0 and % National Wildlife Research Centre, Canadian Wildlife Service, 100 Gamelin Blvd.. Hull. Quebec. Canada KIA 0H3.

^Atlantic Cooperative Wildlife Ecology Research Network, RO. Box 45111, Univ. ot New Brunswick. Fredericton. New Brunswick, Canada E3B 6E1.

3 Environment Canada, Canadian Wildlife Service, Prairie and Northern Region. Prairie and Northern Wildlife Research Centre, 1 15 Perimeter Road, Saskatoon, Saskatchewan, Canada S7N 0X4.

^ 129 Burlington Way, Winnipeg, Manitoba, Canada R3Y ICl.

’Apt. No. 7, 245 Ontario Street, Victoria, BC, Canada V8V INI.

1

2

THE WILSON BULLETIN • Vol. 109. No. I. March 1997

continuous forest as the landscape is increasingly fragmented by agriculture. Received 3 Nov. 1995, accepted 22 May 1996.

Widespread declines have been reported among forest birds that breed in North America and spend the boreal winter in the tropics (Neotropical- Nearctic migrants’, hereafter ‘Neotropical migrants’). Much of the evi- dence for these declines comes from long-term breeding bird censuses and Breeding Bird Surveys (BBS) in fragmented and isolated forests of the eastern and central United States (Johnston and Hagan 1992; for re- views see Askins et al. 1990; Askins 1993; Peterjohn and Sauer 1994). Deforestation and land-use change in the landscape surrounding the sites, and vegetation successional change on the plots themselves, are the likely causes of most local declines (Askins et al. 1990). That songbird popu- lations have not decreased in some continuously forested areas (e.g., Wil- cove 1988) supports the hypothesis that the small size and isolation of many eastern forests are crucial factors in the decline of some Neotropical migrants. However, songbird declines in other continuous forests (e.g.. Holmes and Sherry 1988) and other factors provide increasing evidence to suggest that populations of some Neotropical migrants are also affected by events in the tropical nonbreeding areas because of competition for limited, and ever-diminishing, habitat (Marra et al. 1993; Rappole and McDonald 1994; Petit et al. 1995; Sherry and Holmes 1995; 1996).

Possible causes of declines in Neotropical migrants proposed for east- ern North America include factors in the breeding areas, during migration, and in the nonbreeding areas (e.g.. Holmes et al. 1986; Robbins et al. 1989; Askins et al. 1990; Hussell et al. 1992; Rappole and McDonald 1994; Sherry and Holmes 1995; 1996). In the boreal mixed-wood forest region of western Canada, which has a high diversity of breeding species and is the centre of distribution for some Neotropical migrants (e.g. Con- necticut Warbler — scientific names are in Appendices I and II; Diamond 1991), it remains uncertain whether comparable declines have taken place. Moreover, the relative importance of the factors that affect songbird pop- ulations in western Canada may differ from those in the eastern or other parts of the continent. Boreal forest has been less subject to intensive human settlement than areas farther south where permanent fragmentation has resulted. Consequently, increases in populations of ground predators and the brood-parasitic Brown-headed Cowbird that accompany forest fragmentation elsewhere and which have been implicated in the decline of songbird populations (Robinson et al. 1995a) may be less important. Furthermore, the landscape changes that occur predominantly involve for- estry operations; cutover areas do not change in their land-use and, in-

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

3

deed, may superficially resemble natural disturbances such as forest fire even' though their effects on the forest landscape may be quite different (Hunter 1993; Telfer 1993). Some boreal species that appear ‘area-sen- sitive’ near the edges of their range in the United States may be less so within the boreal forest biome because this is a naturally heterogeneous habitat (Welsh 1987a; e.g., Merriam and Villard 1991). Therefore, long- term studies in continuous forest may be helpful to isolate the effects of factors other than forest size, cowbird parasitism, and possibly predation on bird populations.

We attempt to assess whether or not Neotropical migrants have declined in part of the boreal forest of western Canada by comparing densities in the early 1990s on eight Breeding Bird Census (BBC) plots in two lo- cations counted by A. J. Erskine in the early 1970s, and comparing these changes at each site with those indicated by Breeding Bird Survey (BBS) data over the entire province (although largely south of continuous boreal forest in Saskatchewan and, to a lesser extent, Manitoba). Neither data set is ideal; the BBS routes have been run erratically (Erskine 1993) and, like the BBS routes elsewhere, they sample forest-interior species poorly because they are roadside counts. Also the BBC plots were counted in only two years about 20 years apart. We recognize these inherent flaws in design, but we believe it is important to present the data because there are no other sources of information than can address the question. Similai comparisons (of plots counted far apart in time), by Ambuel and Temple (1982) in southern Wisconsin and Wilcove (1988) in the Great Smoky Mountains, have set important precedents for making maximum use of opportunities to repeat early censuses despite problems in their inherent lack of replication.

METHODS

Studv areas.— counted birds on two study sites each containing four plots ranging from 15 to 30 ha in area. All plots, while within the Mixedwood zone of the Boreal forest, differed to some degree in vegetation and so there are some problems with treating them as replicates. Those in Saskatchewan were in a eontinuously forested area near Dore Lake (Erskine 1973a-d). Michel Point (17.5 ha, 54°41'N, 107°15'W) is a mature mixed stand of white birch (Benda papyrifera), trembling aspen (Poptdus iremidoides), and white spruce (Picea glai(ca) with a few balsam poplar (Populus halsamifera). Except for the north end, the understory is generally sparse. Mirasty Lake (17.6 ha, 54°28'N, 1()7°14'W) is an even- aged stand dominated by trembling aspen with small numbers of white spruce, balsam fir (Abies balsamea), and black spruce (Picea mariana). A dense shrub layer of balsam fir occurs in the northern 150 m, and at the southern end speckled alder (Alnus rngcmi). balsam poplar, and black spruce saplings are more abundant. Appleby Bay (17.5 ha. 54°29 N, 107°16'W) is an uneven-aged .stand dominated by balsam fir with white birch, white spruce, trembling aspen, and some balsam poplar. Young baksam fir forms a dense shrub layer, and there are many mature blowdowns. A younger stand, dominated by white birch, trembling

4

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

aspen, and white spruce occurs to the south. Dore Lake Airstrip (23.4 ha, 54 37 N, 107°23'W) is dominated by mature black spruce with little understory. Two small creeks within this site are bordered by willow (Salix scouleriana) and speckled alder.

The sites in western Manitoba, near Mafeking, were in a landscape increasingly influenced by agriculture but within the fringes of a large expanse of forest centred on the Porcupine Hills (Erskine 1972a-d). In 1972, Novra (17.5 ha, 52°31'N, 101°5'W) was a mature stand, undisturbed by human activity and dominated by balsam poplar and trembling aspen. The dense understory was composed mainly of mountain maple {Acer spicatuiii), beaked hazel (Coryhis cornuta), and red osier {Cornus stolonifera). Speckled alder and willow {Salix spp.) occurred along three small brooks on the plot. By 1992, this stand had opened up considerably due to numerous blowdowns, creating more dense shrub in some locations and isolated large trees. The 1992 plot overlapped at least 95% with the 1972 plot; however, exact duplication of lines was not possible. Mafeking (17.5 ha, 52°47'N, 101°3'W) was a lowland black spruce stand; speckled alder and tamarack (Larix laricina) occurred in boggy gaps, and the main shrub was Labrador tea (Ledum groenlandicum). Around 1987, 30-40% of the 1972 plot was clear-cut, so in 1992 the plot was extended to the north in similar habitat to replace the cutover area. The northern 50% of the 1972 plot had changed little by 1992, except that trees were 1-2 m taller. Thus, this plot was 50% different and generally similar to that first surveyed in 1972, but lacking wet swales. Bellsite (29.25 ha, 52°35 N, 101°5'W) was dominated by jack pine (Pinus banksiana). Shrubs included Saskatoon ser- viceberry (Amelanchier alnifolia) and chokecherry {Primus virginiano). Because the 1972 plot had been clear-cut and replanted with jack pine around 1987-1989 (30 cm height in 1992), an entirely new plot was established about 1.5 km farther north in roughly similar habitat but including small areas of planted pines. The Steeprock Bay plot (15.75 ha, 52°47'N, 100°55'W) was in a balsam fir and white birch stand with a few white spruces. The lower canopy was composed of white spruce and trembling aspen. According to Erskine (1972a-d), most large spruces were logged many years before. Where gaps occurred in the canopy, mountain maple, speckled alder (in wet areas), willow {Salix discolor), and beaked hazel grew. Changes since 1972 included clear-cutting of 10% of the 1972 plot and a further 10% cut-over for birch firewood. Thus, the 1992 plot was 10% smaller than the one censused in 1972. Also, windfalls of mature balsam fir had opened up the canopy.

Census techniques. — Territory mapping methods were used on all sites (Inti. Bird Census Comm. 1969; Bibby et al. 1992), and new 50-m grids for the 1990s censuses were placed as close as possible to the same lines used in the 1970s. In 1972 and 1973 all surveys were conducted by A. J. Erskine; PC censused the .Saskatchewan plots in 1990 or 1991 and GEH censused the Manitoba plots in 1992. In Saskatchewan, plot coverage was from 27 May to 8 July in 1973 (8-10 visits per plot) and from 10 June to 28 June in 1990/1991 (9 visits per plot). Most counts in 1973 were between 04:05 and 12:46 h; in 1990 and 1991, most visits were between 04:13 and 1 1:40 h, with one evening visit (18:25-20:15 h). Less time was spent censusing each plot in 1973 because Erskine surveyed two plots each morning and all four plots (plus another wet-marshy bog census plot) in one year, whereas PC visited only one plot each morning and surveyed only two plots each year. The total counts from 1973 therefore are more likely underestimates compared w'ith 1990-91. The weather in 1973 was generally cool with frequent rains (Erskine I973a-d); in 1991 weather was similar, but in 1990 there was less cloud cover. In Manitoba, coverage was from 26 May to 7 July in 1972, and from 29 May to 24 June in 1992 (9 visits per plot for both periods). Counts were made between 04:55 and 13:22 h in 1972 and 04:30 to 12:45 h in 1992. Because GEH recorded more edge species and visitors, total counts for 1992 were higher than 1972 (A. J. Erskine pers. commun.). Erskine (1972a-d) recorded late May 1972 as being unusually

Kirk et al. ■ BOREAL FOREST BIRD POPULATIONS

5

warm. Frosts occurred on several days in mid-June 1972 (15, 19, 20), but censuses were generally on warm days.

Vegetation sampling. — Vegetation was quantified using the 0.04-ha (0.1 -acre) circle .sam- pling method of James and Shugart (1970). Five ( 1990/199 1 ) or 20-28 circles ( 1972/1973) were placed at 50-m intersections of the bird census plot (these were located systematically in 1972/1973, randomly in 1990/1991). Diameters of all trees > 3 cm diameter at breast height (dbh) were measured in each circle, and canopy height was estimated visually. Shrub density was calculated as in James and Shugart (1970). Both canopy and ground cover were recorded on the second of the two transects used to estimate shrub density. Slight differences in class diameters for shrubs and trees used by PC in 1990-1991 caused us to combine some categories to provide data equivalent (but not fully comparable) to those collected by Erskine in 1973. The same protocol was used for measuring vegetation at the Manitoba sites in 1972, but different methods used in 1992 allowed us to calculate only the relative abundance of different tree species.

Interpretation of territories. — Because the interpretation of field data on breeding terri- tories can differ between people (Oelke 1981; Bibby et al. 1992), original field sheets from all plots in both surveys were re-interpreted by DAK. This procedure avoided the bias of using interpretations by different individuals; we could not control for differences between the three field observers (A. J. Erskine and PC, and A. J. Erskine and GEH). Although we did not use the published data analyses of Erskine (1972 a-d, 1973 a^), we compared the number of territories for each species and found that, with few exceptions, they were quite similar. For this analysis, it was the relative differences in densities between periods that were important.

We based territory boundaries on simultaneous registrations (contemporary contacts) by two singing males, because this gave the best indication that two different individuals held adjacent territories (Inti. Bird Census Comm. 1969). Territories were also determined by locating distinct clusters of song contacts recorded on different occasions (Inti. Bird Census Comm. 1969). We considered both the number of censuses and the spread of dates over which the censuses were made to recognize a cluster of registrations as delineating a terri- tory. We considered 10 days between the first and last survey to be adequate following Marchant (1983).

In general, we used Inti. Bird Census Comm. (1969) criteria for the minimum number of registrations necessary for denoting breeding pairs; we considered two registrations (irre spective of the number of censuses) adequate for Black-capped Chickadee, Boreal Chick- adee, Brown Creeper, Red-breasted Nuthatch, and woodpeckers, as these (mostly resident) species probably started breeding prior to the first census. Likewise, two registrations were considered sufficient for evidence of breeding by Cedar Waxwings because their breeding season extended later than the census period. The minimum period criterion between first and last censuses was also waived for this species. Finally, we counted all clusters within the plot and estimated the proportional area occupied at the edges of the plot to calculate total number of territories. Where .species did not hold any full territories (partial clusters), we used a trace value (0.1 ) to represent possible breeders and visitors. However, we did not include these records in the main breeding totals for guild analyses for each plot because of uncertainties about their status. However, we included probable breeders (those with a + status in Erskine 1972a-d, 1973a-d). We excluded the four waterbirds, diurnal raptors, and large corvids from analyses because these species range over large areas and are not cen- sused accurately by spot-mapping on plots of the sizes u.sed. Flyovers of White-winged Crossbills, Pine Siskins, and Evening Grosbeaks were included, in both the 1970s and 1990s and territories of social groups (some of these species may nest semi-colonially) delineated to roughly estimate numbers of breeding pairs.

6

THE WILSON BULLETIN • Vol. 109. No. I, March 1997

Analysis of life histors’ strategies.— We classified species according to migratory habit and examined changes in combined densities of Neotropical migrants, short-distance mi- grants, residents, and irruptive species to test the possibility that the pattern of change between the two census periods might differ by life history strategy. Migratory strategy was assigned based on the literature and Canadian Wildlife Service databases (D. A. Welsh and J. Pedlar, unpubl. data, C. Downes, pers. commun.). Neotropical migrants were species mainly spending the boreal winter in Central and South America (south of 30°N latitudeX short-distance migrants wintered mainly within Canada or the United States (north of 20 N latitude), and resident species were present throughout the year. We also distinguished irruptive species (i.e.. Evening Grosbeak, White-winged Crossbill, Pine Siskin, and Purple Pinch). However, in stati.stical analyses we combined irruptive species with residents because of small sample sizes.

Breeding Bird Surrey (BBS) analyses.— to determine whether the patterns we observed at local census plots were part of a regional population change, we examined results from the BBS from 1973-1991 for Saskatchewan and 1972-1992 for Manitoba (B. T. Collins, pers. comm.). Standard route-regression techniques were used for BBS analyses with mod- ifications described in Downes and Collins (1996).

Statistical analyses. — We calculated the combined densities of each species for Saskatch- ewan and Manitoba plots separately for each census period (1970s or 1990s). We believe this is justified because there was no replication of each habitat type. Then, using these totals, we tested whether Neotropical migrants, short-distance migrants, or resident (or ir- ruptive) species declined overall using a Wilcoxon’s Matched Pairs test (Zar 1984); here we u.sed the 0.1 values because of small sample sizes (we omitted species that were recorded only as visitors at the edge of plots). We used chi-square one-sample tests (after Wilcove 1988) to compare the combined numbers of Neotropical migrants, short-distance migrants, and resident/irruptive species between years. Yates’ correction was used because there was only one degree of freedom in all cases (Zar 1984). Chi-square tests were performed on actual number of territories and not the corrected density estimates (presented in Appendices 1 and II for comparison among plots). Except where stated, all statistical tests are two-tailed. It may be inappropriate to apply statistical tests to our data because our samples were not randomly chosen from wider populations and our sample sizes are small. We emphasize that our study design was constrained by the establishment of the plots in 1972 and 1973 and that our purpose is to explain possible trends; we recognize the impossibility of proving definite trends from eight plots counted 20 years apart.

RESULTS

Avifaiinal changes: Saskatchewan. — Of 25 Neotropical migrant species counted at the four sites, 16 decreased and nine increased (Appendix I). Based on the null hypothesis that overall densities of Neotropical migrants should increase with succession in western boreal forests (see Kirk et al. 1996), as in eastern North America (Monkkonen and Helle 1989), this decrease approached statistical significance (T = 105.5, N = 25, one- tailed probability, 0.1 > P > 0.05). However, there were marked decreas- es in five Neotropical migrant species; Red-eyed Vireos declined by 48% at Michel Point, where they were most abundant, and at two other sites (Mirasty Lake 65%, Appleby Bay 91%; Appendix I). At another site (Dore Lake) they occurred at low abundance in 1973 and disappeared

Kirk et cl. • BOREAL FOREST BIRD POPULATIONS

7

Table 1

Changes in Habitat Variables in Saskatchewan and Manitoba Census Plots

Site	Year	Ground cover (%)	Shrub density^ (slems/ha)	Canopy cover (%)
Saskatchewan
Michel Point	1973	82	1750	80
	1990	78	5625	74
Mirasty Lake	1973	63	2875 (5525)	71
1991	76	6700	81
Appleby Bay	1973	54	2050 (4610)	72
1991	57	6850	80
Dore Lake	1973	80	1738 (3350)	53
	1990	90	6525	53
Manitoba
Novra	1972 1992	<50	23,000	75
Mafeking	1972 1992	98	2623	55
Bellsite	1972 1992	95	1255	42
Steeprock Bay	1972 1992	95	4143	64

Canopy height

18.9

23.5 10.2 16.7 16.2

19.6

12.9

17.7

17.1

10.8 8.1

12.6

Range

15.0-21.6

20.5-26.4

6-12

14.4- 18.9 9-24

17.4- 21.6

10.8-27.3

9.0- 21.0

9.0- 13.5

6.0- 12.0 9.0-16.5

“Note that in 1972/1973 Erskine measured trees in ein anu .-/.a r .

separately. PC measured shrubs <7.5 cm; thus Erskine’s combined trees <7.5 cm and shrub classes should be equivalent to PC s estimates. However, other differences interfered with comparisons.

completely by 1990. Tennessee Warblers declined at all sites; by 98% at Michel Point, 69% at Mirasty Lake, 27% at Appleby Bay, and 39% at Dore Lake. Black-throated Green Warblers also declined at Michel Point (76%) and at Appleby Bay (67%). Ovenbirds declined (46%) at Michel Point and (29%) at Mirasty Lake. Rose-breasted Grosbeaks declined 92% at Michel Point and 83% at Appleby Bay (Appendix I).

In contrast to Neotropical migrants, there were no consistent patterns of declines across sites in 13 short-distance migrant species. Seven species declined, six increased, and in one there was no change (T — 34.5, N — 13, P > 0.1; Appendix I). However, there was a difference (7 = 3, N =

1 1, P < 0.02) in the magnitude of the increases (9) and decreases (2) in irruptive and year-round resident species. This may be because of the widely fluctuating nature of local populations ot nomadic species, de- pending on food supply.

Migfcitory status. — There were striking differences between the two count periods when we compared the overall number ot individuals by migratory strategies (Fig. 1 ). The combined densities of Neotropical mi-

8

THE WILSON BULLETIN • VoL 109, No. 1, March 1997

grants declined on all four plots (Fig. 1). Combined densities of Neo- tropical migrants declined 44% at Michel Point (x‘ = 11.36, df 1, P < 0.001), 40% at Mirasty Lake (x" = 3.52, df 1, P > 0.05) and 25% at Appleby Bay (x' = 1.78, df 1, P > 0. 1). The non-significant decrease (x~ = 0.17, df 1, P > 0.5) in numbers of Neotropical migrants at Dore Lake was much smaller (14%; Fig. I). However, short-distance migrant species increased on all plots, although the densities were much lower; 184% at Michel Point (x' = 1.76, df 1, P > 0.1), 135% at Mirasty Lake (x^ =

0. 17, df 1, P > 0.5), and 37% at Dore Lake (x^ = L19, df 1, P > 0.1). There was little change in the combined densities of this group (10% increase) at Appleby Bay (x" = 0, df 1, P > 0.9). Similarly, combined densities of resident and irruptive species increased at three sites, although the difference was significant only for Dore Lake (176%, x^ = 5.98, df

1, P < 0.025; Fig. 1 ). There was a slight decrease at Appleby Bay (5%).

Vegetation changes. — The vegetation at Michel Point had become

slightly more uneven-aged over the period between censuses, with an increase in the proportion of larger trees (Fig. 2). Opening of the lower canopy may have occurred through windfall. There was an increase in relative density of white spruce, balsam fir (> 3 cm dbh), and aspen, but balsam poplar densities declined in 1991 (Fig. 3), possibly due to blow- downs. However, these differences could be attributed to sampling vari- ation. Mirasty Lake was a successional aspen stand following a fire 30 years prior to the first census. This area had an increase in relative den- sities of white spruce and balsam fir (> 3 cm dbh; Fig. 3) and in canopy cover. Balsam fir dominated at Appleby Bay in both periods but there was a small increase in the proportion of aspen and a larger increase in white birch (Fig. 3). The increase in the proportion of shade-intolerant hardwood trees (Fig. 3) was accompanied by an increase in canopy cover (Table 1 ). The site that changed least in vegetation characteristics was Dore Lake Airstrip (Figs. 2, 3). However, there was a slight increase in relative density of black spruce, and more deciduous tree species were recorded in 1991 than 1973. The apparent increase in shrub densities at all sites (Table 1) may be unreliable because of differing sample sizes and methods, the relative values within years may, however, be useful.

Avifaiinal changes: Manitoba. — There were no consistent patterns of declines or incieases in Neotropical migrants over the 20-year period at the four sites in the Mafeking region (Appendix II). More species in- creased (19) than decreased (11), and the difference between years was not significantly different {T = 139, N = 30, P > 0.1). Of the species that decreased, the Solitary Vireo was most notable (by 89% at Steeprock Bay, 86% at Bellsite and 18% at Mafeking). Other species such as Ten- nes.see Warbler disappeared at Bellsite and Mafeking but stayed the same

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

9

■ neotropical □short-distance □ resident/irruptive

1972 1992 1972 1992 1972 1992 1972 1992

NO MK BS SB

Fig. 1 . Combined densities of Neotropical migrants, short-distance migrants, and resi- dents/irruptive species. MP = Michel Point, ML = Mirasty Lake, AB — Appleby Bay, DL = Dore Lake; NO = Novra, MK = Mafeking, BS = Bellsite, SB = Steeprock Bay.

10

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

100

MICHEL POINT

80

60

TREE SIZE CLASS

Fig. 2. Changes in proportions of tree size classes at Michel Point, Mirasty Lake, Ap- pleby Bay, and Dore Lake Airstrip (size classes: A = 7.5-15 cm, B = 17.5—22.5, C = 32.5-37.5, D = 40.0-52.5, E = 55.0-67.5, F = 70.0-82.5). Erskine (1973) also considered trees of 3. 5-7. 5 cm diameter but these are excluded from analysis here.

(at Novra) or increased (at Steeprock Bay). Both of these species occurred at relatively low densities. The magnitude of change in the species that increased was far greater: at Novra, Least Flycatcher increased (200%) as did Swainson’s Thrushes (248%), Red-eyed Vireos ( 100%) and Canada Warblers (406%). At Steeprock Bay, increases were found for Magnolia Warbler (32%) and Ovenbird (340%). Of the short-distance migrants, eight increased and nine decreased {T = 54.5, N = 17, P > 0.1; Appendix 1). There was a signihcant increase in abundance of residents and in uptive species {T — 16.5, N = 15, P < 0.02), with 12 increasing and only three decreasing.

Migratory statu.s. — Combined densities of Neotropical migrants in- creased signihcantly at three plots in Manitoba; 33% at Novra (x^ = 4.36, df 1, P < 0.05), 67% at Steeprock Bay (y- = 6.16, df 1, P < 0.025), and 125% at Bellsite (x^ = 6.02, df 1, P < 0.025; Fig. 1). The only site where this group declined (41%) was at Mafeking (x^ = 1.59, df 1, P > 0.1).

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

TREE SPECIES

Fig. 3. Changes in proportions of tree species at Michel Point, Mirasty Lake, Applehy Bay, and Dore Lake Airstrip (tree species; Pj = jack pine, Sb = black spruce, Sw = white spruce, Fb = balsam fir, Wi = scouler willow, Po = balsam poplar. At = trembling aspen, Bw = white birch. As = speckled alder)

Short-distance migrants increased non-significantly at Steeprock Bay (106%, = 2.63, df 1, P > 0.1); low densities at Novra prohibited

statistical testing. There were non-significant decreases in this group at Mafeking and Bellsite = 0, df 1, P > 0.9 and = 0.74, df 1, P > 0.5, respectively). Resident and irruptive species increased at all sites, but none of the increases were significant and actual densities were very low (Fig. 1).

Vegetation changes. — The plot that differed most in its vegetation structure over the 20-year period was Novra. This balsam poplar site was mature at the time of the first census. By 1992, this stand had become open because of aging and windthrow; hence the decline in relative den- sities of poplar, aspen, and birch (Fig. 4). Consequently, there were more open areas and probably increased shrub density at this site, as well as isolation of mature trees. At Steeprock Bay, increased shiub densities were a result of a more open canopy cover due to logging and windthrow.

12

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

Table 2

Breeding Bird Survey Results for Saskatchewan and Manitoba (B. T Collins, pers.

COMMON.)

Saskatchewan 1973—1991 Manitoba 1972—1992

Mean abundance Mean abundance

Species % change N 73 90/91 % change N 72 92

Long-distance migrants Least Elycatcher Great Crested Flycatcher Veery

Swainson’s Thrush Red-eyed Vireo Tennessee Warbler Nashville Warbler Yellow Warbler Chestnut-sided Warbler American Redstart Ovenbird

Connecticut Warbler Mourning Warbler Common Yellowthroat Rose-breasted Grosbeak Clay-colored Sparrow Short-distance migrants Ruby-crowned Kinglet Hermit Thrush Cedar Waxwing Yellow-rumped Warbler Chipping Sparrow Vesper Sparrow Song Sparrow White-throated Sparrow Dark-eyed Junco Irruptive species Pine Siskin Purple Finch Year-round residents Ruffed Grouse Downy Woodpecker Hairy Woodpecker Pileated Woodpecker Gray Jay

Black-capped Chickadee White-breasted Nuthatch

	42	2.6	4.6
-5.8	1 1	2.7	0.9
0.4	33	3.6	3.9
-3.0	12	1.2	0.7
1.5**	43	4.1	5.4
-7.6	13	3.6	0.9
-0.1	18	0.9	0.9
6.5**	12	0.6	1.9
-10.6	37	4.6	0.6
2.3	13	2.1	3.1
-1.5	10	0.8	0.6
1.0	13	0.5	0.6
-1.6	17	0.5	0.4
—	I	—	—
—	—	—	—
-0.5	25	0.8	0.7

-3.2	34	8.8	4.6
1.6	26	0.8	1.0
1.9	24	0.6	0.9
-2.2	15	3.8	2.4
-0.7	32	6.8	5.9
1.1	20	1.1	1.4
-7.0	14	8.8	2.1
2.5	15	1.5	2.4
-7.4	18	0.6	1.2
2.2	18	3.1	4.7
-7.6	15	2.5	0.5
-2.2	19	3.6	2.3
-2.4	33	7.6	4.6
-4.9	29	2.1	0.9
-2.9	33	21.2	1 1.7
0.9	13	1.5	1.7
1.1	14	2.1	2.6
-6.5	30	5.9	1.5
5.1	17	0.9	2.5
-0.9	34	4.3	3.6
0.9	30	5.9	7.0
-1.8	34	13.9	9.7
-1.6	19	7.4	5.4

6.2 14 4.7 0.2

—	—	—	—
1.5	26	0.4	0.5
-0.9	29	0.7	0.6
4.1	16	0.3	0.7
-0.9	1 1	2.3	2.0
-14.8	15	0.2	0.7

•Significance level.s are * 0.05 < P < 0.1; «* /> < 0.05). Species with abundance values of I99()s in Appendices I and II are excluded.

0.2/10 ha for 1970s -

Kirk er al. • BOREAL FOREST BIRD POPULATIONS

13

It is not suiprising that few vegetation changes occurred at the Mafeking site because it was dominated by black spruce and succession in such stands is slow. At Bellsite, the jack pine plot was more mixed than that surveyed in 1972; the deciduous component was greater (Fig. 4) and pines were of larger size.

Breeding Bird Survey results. — We compared our data with analyses from the Breeding Bird Survey for the periods 1973-91 and 1972-92 for each province, respectively, to see if similar trends were reported at the scale of entire provinces (B. T. Collins, pers. commun.). The same direc- tion of regression coefficients was found in the BBS as in the changes at our study sites in Saskatchewan for six of the seven Neotropical migrants reported on 10 or more routes (compare Appendices I and II vs Table 2). For example, there was a significant increase in Least Flycatchers and Yellow Warblers according to the BBS. Most strikingly, two of the species that showed the greatest declines in our study (Tennessee Warbler and Ovenbird) had negative regression coefficients according to the BBS for Saskatchewan (Table 2). However, Red-eyed Vireos had a positive re- gression coefficient (Table 2). In contrast to the results for Saskatchewan, the direction of BBS regression coefficients for Neotropical migrants in Manitoba were often opposite to the results we found (10 out of 15 cases; Table 2). For example, while Least Flycatchers decreased significantly in Manitoba over the long-term, they increased at one of our sites and began breeding at two others. However, unless the slope in route-regression analyses are significant, the direction of BBS regression coefficients may not be meaningful, so we urge caution in the interpretation of these com- parisons.

DISCUSSION

We acknowledge the limitations of our data, which compare only two points in time two decades apart; we recognize that populations may have fluctuated considerably in the intervening years (e.g.. Holmes and Sherry 1988; Blake et al. 1994). We do not know how representative our plots may be of boreal mixed-wood forests in the region in general, and we acknowledge that changes apparent on small plots may be stabilized at larger geographic scales (the metapopulation level; Villard et al. 1992). Nevertheless, our data are the first long-term plot-based ‘trends to be reported from an important area of the boreal forest, and they illustrate some problems of more general interest.

Although Neotropical migrants as a group did not decline significantly in Saskatchewan, five species declined on all plots where they were cen- sused over the 17-year period, and combined densities of Neotropical migrants decreased on all sites. In contrast. Neotropical migrant numbers

RELATIVE DENSITY

14

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

1 oo

80 -

60

40 -

20

NOVRA

■ 1 972 □ l 992

ilL

Sb Sw Fb PJ WI Po At Bw Ag Mo Mr

1 OO

80

60

40

20

STEEPROCK BAY

-I L-

k

JZL

Sb Sw Fb PJ \M Po At Bw Ag Mo Mr

TREE SPECIES

Kirk et ai • BOREAL FOREST BIRD POPULATIONS

15

tended to increase or remain stable in Manitoba. The Saskatchewan results recall those from fragmented forests in the eastern U.S., rather than those reported by Wilcove (1988) in the extensive forests of the Great Smoky Mountains; yet our Saskatchewan plots were in contiguous, unfragmented forest. The changes in bird abundance in Saskatchewan also run counter to predictions that Neotropical migrant numbers should increase as the forest matures (Monkonen and Helle 1989) as found at other sites in western Canada (Kirk et al. 1996).

Regional differences in population trends of Neotropical migrants have been described elsewhere (James et al. 1992; Sauer and Droege 1992). We believe that the differences we found between the provinces are partly attributable to forest succession and the more fragmented nature of the landscape of the Manitoba plots. These were already somewhat frag- mented by human settlement in 1972, and much forest has since been cleared for agriculture, mainly on the Swan Lake plain but also by the highway close to Mafeking. Logging has also become more widespread since a pulp mill was opened at La Pas in the mid 1980s, and a new mill is about to open in the Mafeking area (P. Rakowski, pers. commun.). The Manitoba plots, although still in forest, are now adjacent to an increas- ingly agricultural landscape; the Saskatchewan plots, in contrast, remain in a predominantly unfragmented forested landscape. The Manitoba study area is close to continuous forest in the Porcupine Hills which has high populations of Neotropical migrants (A. R. Smith pers. obs.). Thus, this census area may function as a population sink (sensu Pulliam and Dan- ielson 1991; Villard et al. 1992). High apparent densities of Neotropical migrants have been found in fragmented boreal forest elsewhere in Sas- katchewan (K.A. Hobson, pers. commun.) and in isolated, fragmented forests in the United States (Robinson et al. 1995a; Brawn and Robinson 1996).

In a comparison of bird populations trends between 1969-1986 at local (Hubbard Brook Experimental Station, New Hampshire) and regional scales (BBS statewide trends). Holmes and Sherry (1988) found that 10 of 19 bird species followed the same direction of trends. More species showed significant declines than increases at both the local (8 vs 1 ) and regional scale (5 vs 1 ). However, only three Neotropical migrant species decreased significantly at both scales (Least Flycatcher, Swainson’s

Fig. 4. Relative densities of different tree species at (a) Novra, (b) Bellsite and (c) Steeprock Bay (Sb = black spruce, Sw = white spruce, Fb = Bal.sam Hr, Pj = jack pine, Wi = willow species, Po = balsam poplar. At = trembling aspen, Bw = white birch, Ag = speckled alder. Mo = mountain maple. Mm = Manitoba maple).

16

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

Thrush, and Wood Thrush). Our analysis of BBS results over the same periods as encompassed by our study supports our finding that Tennessee Warblers and Ovenbirds may have declined in Saskatchewan boreal for- est. Banding data since the 1950s also suggest that Tennessee Warblers may be declining in Saskatchewan (Smith and Diamond 1993), although trends in this species are difficult to determine because it is a budworm specialist. These results suggest that the declines we observed in Sas- katchewan may not be limited to our plots but may be part of larger-scale phenomena. The BBS data are not conclusive either, because the regres- sion coefficients were not statistically significant and may reflect cyclic population fluctuations (Hussell et al. 1992); densities of Tennessee War- blers fluctuated dramatically (Welsh 1987b) and were ‘puzzlingly high’ on the Saskatchewan plots in 1973 (A. J. Erskine, pers. commun.). Ov- enbird populations may also fluctuate greatly from year to year in frag- mented habitats (Gibbs and Faaborg 1990).

Because the Saskatchewan sites were not logged or the surrounding forest fragmented in the 17 years between counts, what might have caused the declines in the five species of Neotropical migrants? In a 16-year study of bird populations on a 10 ha plot at the Hubbard Brook Experi- mental Station, New Hampshire, Holmes et al. (1986) identified five fac- tors that affected songbird population densities; (1 ) food supply (especial- ly defoliating Lepidoptera); (2) poor spring weather conditions; (3) forest successional changes; (4) competitive interactions and (5) nonbreeding mortality. We will address each of these in turn.

At Hubbard Brook, Lepidopteran outbreaks accounted for high densi- ties of Scarlet Tanager {Piranga olivacea). Least Flycatcher, Red-eyed and Philadelphia vireos and most warbler species during the first three years of study; long-term declines were attributed to lows in Lepidopteran populations which fluctuate on a 6-10 year irregular basis (Holmes et al. 1986). Evidence for changes in food supply causing bird density changes in our study area was weak; unfortunately, we had little information on the timing of Lepidopteran outbreaks. Although spruce budworm (Cho- ristoneiira fumiferana) outbreaks struck Mirasty Lake and Appleby Bay in 1990 and 1991, among budworm specialists only Bay-breasted War- blers showed numerical increases, while Cape May Warblers were new breeders at Appleby bay in 1991. Contrary to expectations, Tennessee Warblers declined dramatically at both sites; declines in other warbler species that should also respond numerically to budworm irruptions (Hus- sell et al. 1992) argue against the food supply hypothesis for the Sas- katchewan sites. So far as we know the only other Lepidopteran outbreak was of tent caterpillars {Malacosoma disstria) which defoliated aspens near Michel Point in 1973. Subsequent movement by Red-eyed Vireos

Kirk et cl. • BOREAL FOREST BIRD POPULATIONS

17

seeking alternative foraging habitat in the birch on that plot could explain the high densities in 1973.

Cold, wet weather in late May (1974) was linked to declines in Scarlet Tanagers and several other species at Hubbard Brook (Holmes et al. 1986). While rain occurred on most survey days in 1973 in this study (A. J. Erskine, pers. commun.), the weather was dry in 1990 or 1991. Had weather influenced counts, bird numbers should have increased be- tween censuses, not decreased.

Changes in habitat suitability possibly accounted for the decline of Least Flycatchers, Philadelphia Vireos, and Wood Thrushes at Hubbard Brook (Holmes et al. 1986). Our 1990s vegetation surveys were not strict- ly comparable with those of 1973; they do not show the requisite changes in forest structure that could have caused declines in the five species in Saskatchewan, but the differences in bird numbers recorded in Manitoba were consistent with succession.

Interspecific aggression by Least Flycatchers affected habitat use by American Redstarts at Hubbard Brook. We have no data on interspecific competition, but the bird communities were similar in the two study areas, so this factor is unlikely to have been important.

Finally, evidence for winter limitation on breeding numbers at Hubbard Brook was found for Dark-eyed Junco and Hermit Thrush, both of which declined after cold winters in the United States; little evidence was found to suggest that Neotropical migrants were affected by events during the nonbreeding season. While the effect of tropical habitat change is contro- versial (Hutto 1988, Robbins et al. 1989, Askins et al. 1990, Hussell et al. 1992), it is now recognized that nonbreeding survival rates for many species depend on the availability of suitable forest habitat in the Neo- tropics, which is declining rapidly.

Except for the Red-eyed Vireo, which spends the boreal winter in South America, there is considerable overlap in the nonbreeding distribution of Tennessee Warbler, Black- throated Green Warbler, Ovenbird, and Rose- breasted Grosbeak (Rappole et al. 1995), all species that declined at the Saskatchewan plots. The main difference is that Tennessee Warblers do not ‘winter’ in the Caribbean, while Black-throated Green Warblers and Ovenbirds do. Also the range of Rose-breasted Grosbeaks extends faithei into South America than those of the warblers, occurring in much of Venezuela and Colombia, as well as Ecuador and Peru (Rappole et al. 1995). Most nonbreeding habitat for Red-eyed Vireos is in the Amazon Basin (Brazil), while for Tennessee Warblers, Black-throated Green War- blers, and Ovenbirds it is in western or southern Mexico. Most nonbreed- ing habitat for Rose-breasted Grosbeaks is in Colombia and Panama and, to a lesser extent, Mexico (Diamond 1986). Thus, the possibility exists

18

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

that where the nonbreeding areas of some Saskatchewan subpopulations of these species overlap, they are all being affected by loss of preferred habitat. Both Black-throated Green Warbler and Ovenbird were included by Petit et al. (1995) in a list of 45 species vulnerable to destruction of tropical broadleaved forests.

Recent studies suggest that the species that declined at the Saskatche- wan plots use a range of forest habitats in the Neotropics (Hutto 1992; Greenberg 1992; Wunderle and Waide 1993; see review by Petit et al.

1995). In the Caribbean, Yucatan Peninsula, and western Mexico, Black- throated Green Warblers and Ovenbirds occur at highest densities in forest (Hutto 1992; Greenberg 1992; Wunderle and Waide 1993) and are con- sidered forest generalists (Lynch 1989; Greenberg et al. 1995). Red-eyed Vireos are most abundant in early to mid-successional or edge habitats and are relatively rare in primary forests in the Amazon lowlands and adjacent Andes (S. K. Robinson, pers. commun.; Robinson et al. 1995b). Tennessee Warblers prefer patches of remnant forest and open woodland (e.g. Costa Rica; Powell et al. 1992), and Rose-breasted Grosbeaks use field-forest vegetation in western Mexico and the Yucatan (Lynch 1989; Hutto 1992). However, evaluating habitat for Neotropical migrants based on the distinction between forest and non-forest habitats may be a false dichotomy. While none of the above species uses completely open agri- cultural field habitats, they do use forest patches, gallery strips, or hedge- rows within human-altered landscapes. As has been pointed out by others (Lynch 1989; Greenberg 1992; Petit et al. 1995), decreased habitat suit- ability may not just involve loss of mature forest; in much of the Neo- tropics second growth and forest habitat patches are also disappearing rapidly. Also, habitat u.se of Neotropical migrants is usually assessed from point counts or mist-net surveys in small areas, without measures of sur- vival or differing social systems. That more individuals of Neotropical migrants are counted in early successional or patchy habitats, therefore, does not contradict the hypothesis that populations could be partly limited during the boreal winter (see Rappole and McDonald 1994) For example Tennessee Warblers occur at highest densities in coffee plantations in parts of Mexico (Greenberg et al. 1995; in press). The shift from shade coffee to sun coffee in recent years in parts of the Neotropics may have become a limiting factor for this species because of loss of flowering tree species important for foraging (R. Greenberg pers. commun ) ^ ‘ ^

Howev^, since the same species that declined in Saskatchewan did not do so m Manitoba, and considerable mixing of populations is likelv in the Neotropics (Wilcove and Terborgh 1984), it seems hard to a^Lte the trends we found to effects in the nonbreeding areas It is possibt however, (ha, the Manitoba densities were raised by successional and

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

19

fragmentation influences sufficiently to mask or reverse an undei lying downward trend reflected in the contiguous forests of Saskatchewan.

ACKNOWLEDGMENTS

Funding for this project was provided by the Canadian Wildlife Service of Environment Canada. We are particularly grateful to A. J. Erskine for careful reviews of early drafts and detailed information on the methodology used during bird and vegetation surveys. B. T. Collins performed the trend analyses on BBS data tor which we are grateful, and E. Woods- worth provided statistical advice. Earlier drafts of this manuscript were improved greatly by comments from C. E. Braun, R. Greenberg, K. A. Hobson, and D. S. Wilcove. We also thank two anonymous referees and C. R. Blem for helpful comments during the formal review procedure.

LITERATURE CITED

Ambuel, B. and S. a. Temple. 1982. Songbird populations in southern Wisconsin forests: 1954 and 1979. J. Field Ornithol. 53:149-158.

Askins, R. a. 1993. Population trends in grassland, shrubland, and forest birds in eastern North America. Current Ornithol. 11:1-34.

, J. E Lynch, and R. Greenberg. 1990. Population declines in migratory birds in

eastern North America. Current Ornithol. 7:1-57.

Bibby, C. j., N. D. Burgess, and D. A. Hill. 1992. Bird census techniques. British Trust for Ornithology and Royal Society for the Protection of Birds. Academic Press, San Diego, California.

Blake, J. G., J. M. Hanowski, G. J. Niemi, and P. T. Collins. 1994. Annual variation in bird populations of mixed conifer-northern hardwood forests. Condor 96:381-399.

Brawn, J. D. and S. K. Robinson. 1996. Source-sink population dynamics may complicate the interpretation of long-term census data. Ecology 77:3-12.

Diamond, A. W. 1986. An evaluation of the vulnerability of Canadian migratory birds to changes in Neotropical forest habitats. A report to the Latin American Program of the Can. Wildl. Serv., Conservation and Protection, Environment Canada, 3rd floor. Place Vincent Massey, 351 St. Joseph Blvd., Hull, Canada KIA 0H3.

. 1991. Assessment of the risks from tropical deforestation to Canadian songbirds.

Trans. North Amer. Wildl. and Nat. Resour. Conf. 56:177—194.

Downes, C. and B. T. Collins. 1996. The Canadian Breeding Bird Survey, 1966-94. Ca- nadian Wildlife Service Progress Note No. 210. 36 pp.

Erskine, A. J. 1972a. Mature poplar stand. Amer. Birds 26:989-990.

. 1972b. Black spruce stand. Amer. Birds 26:990.

. 1972c. Jack pine stand. Amer. Birds 26:990—991.

. 1972d. Balsam fir stand. Amer. Birds 26:991-992.

. 1973a. Mature birch and poplar forest. Amer. Birds 27:1007.

. 1973b. Young aspen forest. Amer. Birds 27:1007-1008.

. 1973c. Balsam fir forest. Amer. Birds 27:1009-1010.

. 1973d. Black spruce stand with alder swales. Amer. Birds 27:1010.

. 1993. Breeding bird surveys and the breeding bird census. Pp. 227-233 in Birds

in the boreal forest. (D. H. Kuhnke, ed.). For. Canada Northwest Region, North For. Cent., Edmonton, Alberta.

Gibbs, J. P, and J. Faaborg. 1990. Estimating the viability of Ovenbird and Kentucky warbler populations in forest fragments. Conserv. Biol. 4:193-1.

Greenberg, R. 1992. Forest migrants in non-forest habitats in the Yucatan peninsula. Pp.

20

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

in Ecology and conservation of Neotropical migrant landbirds. (J. M. Hagan III and D. W. Johnston, eds.)- Smithsonian Instit. Press, Washington, D.C.

, J. Salgado, I. Warkentin, and P. Bichier. 1995. Managed forest patches and the

conservation of migratory birds in Mexico. Pp. 178-189 in Conservation of Neotropical migratory birds in Mexico. (M. Wilson and S. Sader, eds.). Maine Agric. Expt. Stn. Misc. Publ. 727.

, P. Bichier and J. Sterling (in press). Bird populations in rustic and planted shade coffee plantations of eastern Chiapas. Biotropica.

Holmes, R. T. and T. W. Sherry. 1988. Assessing population trends of New Hampshire forest birds: local vs regional patterns. Auk 105:756-768.

' . and L W. Sturges. 1986. Bird community dynamics in a temperate

deciduous forest: long-term trends at Hubbard Brook. Ecol. Monogr. 56:201-220.

Hunter, M. L. 1993. Natural fire regimes as spatial models for managing Boreal forests. Biol. Conserv. 65:115-120.

Hussell, D. j. T, M. H. Mather, and P. H. Sinclair. 1992. Trends in numbers of tropical and temperate-wintering migrant landbirds in migration at Long Point, Ontario, 1961- 1988. Pp. 101-114 in Ecology and conservation of Neotropical migrant landbirds. (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian Instit. Press, Washington, D.C.

Hutto, R. L. 1988. Is tropical deforestation responsible for the reported declines in Neo- tropical migrant populations? Amer. Birds 42:375-379.

1992. Habitat distributions of migratory landbird species in western Mexico. Pp, 221-239 in Ecology and conservation of Neotropical migrant landbirds. (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian Instit. Press, Washington, D.C.

International Bird Census Committee. 1969. Recommendations for an international stan- dard for a mapping method in bird census work. Bird Study 24:727-736.

James, E C. and H. H. Shugart, Jr. 1970. A quantitative method of habitat description Aud. Eield Notes 24:727-736.

, D. A. Wiedenfield and C. E. Mcculloch. 1992. Trends in breeding populations of warblers: declines in the .southern highlands and increases in the lowlands. Pp. 43- 56 in Ecology and conservation of Neotropical migrant landbirds. (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian In.stit. Press, Washington, D.C.

Johnston, D. W. and J. M. Hagan III. 1992. An analysis of long-term breeding bird censuses from eastern deciduous forests. Pp. 75-84 in Ecology and conservation of

Neotropical migrant landbirds. (J. M. Hagan III and D. W. Johnston, eds.) Smithsonian Instit. Press, Washington, D.C.

Kirk, D. A., A. W. Diamond, K. A. Hobson, and A. R. Smith 1996. Breeding bird com- munitie,s of the western and northern Canadian boreal forest: relationship to forest type Can. J. Zool. 74:1749-1770.

Lynch J- F. 1989. Distribution of overwintering Nearctic migrants in the Yucatan Penin- sula, 1. Oeneral patterns of occurrence. Condor 91:515-544

<^7"”" Oi'-d Census Ins, ructions. British Trust for Ornithology, CBC Technical Review Group.

Marra P. R. T. W. Sherrv. ano R. t. Holmes. 199.1. Territorittl exclusion by a long- dts ance tntgran, warbler Jamaica: a removal experiment with Ameriean Reds Js (Setophaga riiticilla). Auk 1 10:565-572.

^"""vrrhhU Heterogeneity in ecological space and time: a vital

variable m population persistence. Advances in Ecology I 199-208

MOnkONNEN. M. and P. Helle. 1989. Migratory habits of birds breeding in different stages mciMtlM-lir""’'’”'”'’ "" 'h' Nearctic. Ann. Zool. Fen-

Kirk et cil. • BOREAL FOREST BIRD POPULATIONS

21

Oelke, H. 1981. Limitations of the mapping method, in Estimating numbers of terrestrial birds. (C. J. Ralph and M. J. Scott, eds.). Studies Avian Biol. 6:1 14-1 18.

Peterjohn, B. G. and J. R. Sauer. 1994. Population trends of woodland birds from the North American Breeding Bird Survey. Wildl. Soc. Bull. 22:155-164.

Petit, D. R., J. E Lynch, R. L. Hutto, J. G. Blake, and R. B. Waide. 1995. Habitat use and conservation in the Neotropics. Pp. 145-197 in Ecology and management of Neo- tropical migratory birds: a synthesis and review of critical issues. (T. E. Martin and D. M. Finch, eds.). Oxford Univ. Press, New York, New York.

Powell G. V. N., J. H. Rappole, and S. A. Sader. 1992. Neotropical migrant landbird use of lowland Atlantic habitats in Costa Rica: a test of remote sensing for identification of habitat. Pp. 287-298 in Ecology and conservation of Neotropical migrant landbirds (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian Instit. Press, Washington, D.C.

Pulliam, H. R. and B. J. Danielson. 1991. Sources, sinks and habitat selection: A land- scape perspective on population dynamics. Amer. Nat. 137 (Supplement): S50-S66.

Rappole, J. H. and M. V. McDonald. 1994. Cause and effect in population declines of migratory birds. Auk 111:652-660.

, E. S. Morton, T. E. Lovejoy, III, and J. L. Ruos. 1995. Nearctic avian migrants

in the Neotropics: maps, appendices, and bibliography. Smithsonian Instit. Press, Wash- ington, D. C.

Robbins, C. S., J. R. Sauer, R. S. Greenberg, and S. Droege. 1989. Population declines in North American songbirds that migrate to the Neotropics. Proc. Natl. Acad. Sci. 86: 7658-7662.

Robinson, S. K., E R. Thompson III, T. M. Donovan, D. R. Whitehead, and J. Faaborg. 1995a. Regional forest fragmentation and the nesting success of migratory birds. Sci- ence 267:1987-1990.

, J. W. Fitzpatrick, and J. Terborgh. 1995b. Distribution and habitat use of Neo- tropical migrant landbirds in the Amazon basin and Andes. Bird Conservation Inter- national 5:305-323.

Sauer, J. R. and S. Droege. 1992. Geographic patterns in population trends of Neotropical migrants in North America. Pp. 26-42 in Ecology and conservation of Neotropical migrant landbirds (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian Instit. Press, Washington, D.C.

Sherry, T. W. and R. T. Holmes. 1995. Summer versus winter limitation of populations: what are the issues and what is the evidence? Pp. 85-120 in Ecology and management of Neotropical migratory birds: a synthesis and review of critical issues. (T. E. Martin and D. M. Finch, eds.). Oxford Univ. Press, New York, New York.

. AND . 1996. Winter habitat quality, population limitation, and conservation

of Neotropical-Nearctic migrant birds. Ecology 77:36—48.

Smith, A. R. and A. W. Diamond. 1993. Monitoring songbirds in the prairie provinces. Proc. 3rd Endangered Species and Prairie Conservation Workshop. Prov. Mus. Alberta, Nat. Hist., Occ. Pap. No. 190:305-311.

Telfer, E. S. 1993. Wildfire and the historical habitats of the Boreal forest avifauna. Pp. 27-38 in Birds in the boreal forest. Proc. workshop. Prince Albert, Saskatchewan. (D. H. Kuhnke, ed.). For. Canada Northwest Region, North For. Cent., Edmonton, Alberta.

ViLLARD, M.-A., K. Freemark, AND G. Merriam. 1992. Metapopulation theory and Neo- tropical migrant birds in temperate forests: an empirical investigation. Pp. 474-482 in Ecology and conservation of Neotropical migrant landbirds (J. M. Hagan III and D. W. Johnston, eds.). Smithsonian Instit. Press, Washington, D.C.

Welsh, D. A. 1987a. The influence of forest harvesting on mixed coniferous-deciduous boreal bird communities in Ontario, Canada. Acta CEcologica CEcol. Gener. 8:247-252.

22

THE WILSON BULLETIN • Vol. J09, No. 1, March 1997

. 1987b. Tennessee Warbler. Pp. 364-365. in Atlas of the breeding birds of Ontario. (M. D. Cadman, P. E J. Eagles and E M. Helleiner, eds.). Eederation of Ontario Natu- ralists and Long Point Bird Observatory. Univ. of Waterloo Press, Waterloo, Ontario. WiLCOVE, D. S. 1988. Changes in the avifauna of the Great Smoky Mountains; 1947-1983. Wilson Bull. 100:256-271.

AND J. Terborgh. 1984. Patterns of population decline in birds. Amer. Birds 38:10-13. WuNDERLE, J. M., Jr. and R. B. Waide. 1993. Distribution of overwintering Nearctic migrants in the Bahamas and Greater Antilles. Condor 95:904-933.

Zar, j. H. 1984. Biostatistical analysis. Prentice Hall, Englewood Cliffs, New Jersey.

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

23

D

00

Z

UJ

U

z

<

u

z

u

H

<

00

<

00

Llf

<

'u:

o

Q

H

<

X

O

X

aa

D

2

Q

X

H

<

S

H

oo

LU

<u

O

Q

ca

02

Sj

CL

<

<u

a

-J

O

o.

13

x:

o

ii

On

I

O

ON

On

ON

o

ON

r-

On

ON

m

r-

On

ON

r^i

r-

ON

o

ON

r^i

ON

o

'x:

o

c/5

■o

C

o

E

c

U

^ I o — I

ooo — n^rir^f^poofs-^^n— ; — fNoo

r^fSiOa'O — — ■r'iriooooo'^oo — o»oo

— — sO O'

o r- ON —

o — — o —

00 r' — — sd o

ON ON

d d

— O'

d d

— ON

d d

— p d in

d

Tj-

d

\n —

— d

or-- — so

rt — d> ^

— rn

— NO

— — rr-,

d — <N

vq

— d d

— — rt

o — —

— O'

d (N

m o <r)

d d oi

— O' —

d d d

— ; — : q — d 00

nO

d

NO

d

O' nO nD d

q

q rj r-;

GO d —

'cf q

d o6

— O'! —

— d d

nO —

d d

— vC

d d

o — NO d

I II

Q c

r i I i

• C 5 ^ ' •C -i: ‘2 S

H - s: ^

u Uj Uj ^

>N w

3

5 ^ o ^

3=^

. S ^

X 5

O >

w o U.

H o H

3 3-

^ >N C

LL X

U ^ c u V5 •:= T3 « 5

< X c/2

^ O ^ ^

5 I Cl

^ ^ o o > y

^.l>

Cl’V

2 V

•j: — *0 o ^ u c/2 X 0^

X Z u.

^ U

?3 *a

I

.2 S'!

oS-S

S d m

- “ . -S tr ! 5 u s

i I ■§ 1

H

j _ -o ^

^ E -L ^

= - S' 22

: X X X

5 5 5 ®c >: 2 ?

-= I- ? 5

Ci. ? i

nil

c ^

b ^ D w

i i € 3

u ^

oc 'a i P

■c 1 -o

“ S = I

c > ° 3

< O S u

	1 §. s
5	?,:= 1
6C
3	1 i'S

£|

6

C

^ 'y;

r- 3

O -C .

X O ,

I r-; q r>^

I — d ri

— no

d d d o d d

Tf — —

d oi d

n — d d

_ so q

— d (N

— ON

— NO

d d

2 2

X 5

5- §

S: >2

^ c ^

•5

c o

t t

CL a

c/2 c/2

CC

•o

1) a

51 3

1 Z

V -O

• ii •= q' T2 = ti; ^ «

(U “

-C -3 — .-

q q ■§ c 5

o :>. -7

"* £

*5

el

r

3 S o ^

i.i

CD ^

Appendix I

Continued

24

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

(U

u

O

a

CQ

JO

a.

Cl

<

ON

I

o

ON

On

r-

On

ON

On

r-

ON

o

ON

On

1)

E

C

o

''E

c

a;

“O

c

o

E

o

U

OONCN — lO — 'sOm

— O O O ~ o'

r-; — C — <N O (N — O 00 O <N

O' r- d Tf

— o d d

Tt —

d d

m

vS d d

<N r- —

— — d

r-J O C O

d d r-i — d

? 3

Vj

2 ■§ *2 ^ ®C

5 2

s oc

S- :2 ^ 2

~ w b -C) Oij ^ -

CJj

^ - T3

’ N .

- ,3 <5 5 ■

■o

E i

^ JC

2 H

i I

rs ^

Q-

3 U

a: X

C -O

t

3 3 ' O. C ■

\q — fo

o d 00 d

— vO <^ —

d — d d

^ q — d d d

ON ON —

d d d

— r>- — d — ■ d

— q d d ri

2 ^

^ .2 ^

Ci.

^ y. 2

r^ — — iqcN — sOsD

d d d d d — cN d

(N <N

d d

— • — • n >0

d O — — (N

— q q q q d d d d

ON

d

ON —

d uS

ON U' f<j

d — nc

NO

d

NO NO —

d d d

n

nO nO ON

d d r-:

O 2

^ >1

:5 ^

2 .5-

C o

D

’ ^2 i5

• C 3 c

2 9j

Li -5 *0

c/D ^

CL "V

E 2

^ i

c/5 > I

,E -,2 O

^ c/5

■2 O

li

£ UJ

;2 « 3

0^ ^ u.

Jc ■g y y

|i*°^

■ u ^ >» W « -O

iPal

o ^ C •S CJ -o t -o ^

Q. -C

M

-p 5 J § 2 O CQ CQ ^2

Appendix II

Estimated Numbers of Pairs/ 10 ha at Mafeking, Manitoba Census Sites

Kirk et ai • BOREAL FOREST BIRD POPULATIONS

25

e2

E

ca

c3

ca

o

p

5.

1)

B

00

>

o

Z

(N

ON

(N

r-

ON

(N

ON

On

CN

r-*

ON

(N

ON

On

(N

r-

ON

CN

On

(N

r-

ON

(N

On

ON

(N

r-*

On

a>

£

a

c

o

£

c

TD

C

CJ

C

O

u

— 0(N — O'tfN — — ;fNa;cor^fNr^ — — QO — ^TfOir, rj

oooor^oO'Ood~o<NO — or*'<c4oNOfnooof^'rrrS — — oocNfsr^c'o

CN ri — —

cN (^00 sOi/Nr^ — ^0(7' oom os — ornr^r^ — vooo — ir,

oooofsi — o-^\brndr4odrnO(Nd'.do<^ — o — — d — dir] — wSoo

(N Ov

d d

Os

d

sO — —

d d d

so

d

os

d

— r-

d —

d

so

d

o o o o

r- O' ro

— d d

CN — rn

r4 d d

— Tt in

d d d

P-) —

d d

— — (Nin<N — sor-isq sosofn cncnso voinmTf — — cni — oosocn

dd doiddddd ddooddd dr'i'd'^dddddd— •

in O' ni —

— <n CO m m — sO O

dsD d— ~dd —

q — c

— d —

— q — q d sd d in d

r«.o — — — sCsO — sO — sd — dddddd

r- 'C r~( — O' O —

ni — Tf d d ri GO —

so

d

qqq— qcNqq

dodsd— • ddod —

o Tf m r' — r- —

d d sd — d — —

c 2 .2

5. Ci.

a

•2 — o 3

c

Ck

2

.2

N3 “=: 2

«. s: i ,5 o

£ 2

.2? X

^ w Uj O

: 0 ^

Ci. Oc ^

l_ ^

w ?

£ -o rr P

2

—

" ^

u.

"u -c -n

o o

cS O

-C — >% -

^ W y? ^

— *0 (U

D — U u-

>* < -J o

5 -5

— P -yj • CO

^ - I

i o ?

CQ > CO

>>^ ^ oi» a >s c t;

IS- g 5 .- S li ^ ii s o

■g K -t -O € T T3 -5 ^ o w r3 > (u

^ I ^ :g ^ ^ «

Jj & o A S'!

?; K,i 1 1 2 V

I Hi

U S 0 CQ

T c c ^

— ■'S C ^

j:; 1) U o a. H O Z

,2^0 •2 "T 2 ^ -e ^ 2 ^ -=

HI

■fc P -o

^ o r3

■i ^ ^

r3 ^ — CU 03 CQ

2- ^ 2 d

-o -z

<U ^

a: ^ ,

^ t3 '

5 :§ £

• I'H'

r O o ;

U i :

-e U 5

'hS#:

3

^ S. '5.

S J

: w "3

' ^

1

£ I

g g c

C > O

< o z

« F £

0^0 U 1 U

:3 C

U OC

o

? « ?

O Q. C

t CO fc

d "O Q. CO CO Stl^ y- = O C

a V o

a. >, y £ — .E O U -J

26

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

(N

O

ON

(N 00 (N ^ 'O

cooodd — <N

ON Tf — rs ON

iririooddduS

fO —

d o d

d d d r-i

v*i —O'r'

d o d ~ d

CN

O'

ON

o d d o d

O' 00

d —

— o O)

Tf — ddddo —

d o o o

r*', — \0 d d d d o

OD

c

£

(N

ON

(N

r-

ON

\0 Tf — ro —

d d — rn d (N d

o

d d — Tf

d

— f<i

d d

d

(N

ON

rn

d

Tf — r- — ri O' r*-. —

rsd ddddd d

m

d d d

f^.

d

00

a;

CQ

(N

O'

— r- Tf m

d — d ri d —

X

5

z

m

0-

a

LU

ID

z

H

Z

o

< ^

>N

QD

u

O

u

a

o

S

Jo

(N

ON

On

(N

ON

(N

ON

^ CN (N fO sO »ri

d d d d — d ri

CN »r>

— — O NO

d d — d

ON nD —

— d d

O' —

— c

ri

d

Tt — —

— d d

NO — c

d — ri

nC

d

>

O

Z

(N

r-

O'

d

so

d

1)

E

TS

C

o

c

o

•o

c

o

e

E

o

u

I II

^ A i

^ ^

l|5

' “ In

c» U > T)

O u u

^ = 'ti: = . E

E S JJ 3 ° c ° u o S >• Z

'ts 5

2 2 *5' 5

ss:

X. w

^ a

2 ^ ^ ^ ■5 ^ ^ I

“3 .

= "3 02

O ^

2

U > V

c ^ ^

> *J t*

> - T3

o .5 -■'

2 i-

S t .is £

-5? ts ? § ^ Q

: "5 ' V

■ a S 1

= S 5

i7

: :£ c

! c§ I I

i a ^ .1 ;

C 4? 3

^ ^ V

1 i

o J3 o

c "O =

a350a'I<o>i

, ^ s

I J i J

n a. S

0 a " ^ c/5 ;

1 T -o i

> 1> I C O S -

^^2- ^ aj= : «- 00 Y ;

^ 01, H , ■•'■ =

■4! ® >

> c/5 >

=■5

s ^

s

o

U

o

O

§■ ■= C I

^

S a “ 2 -g

0^3

n3 O .g 1) ^ o

c .E o

■“ ^

5 . 00

u ^ .5

> c/5 ^ Q CQ <

. Q. w

"2 2 Q CQ ■

Q Q: cj •?•

^ 't

W

•o o o

a 3 c/5

■ -o

O

1 ■§'’S

^ o ^

< O cj

>'=^

3 >, -1^

^ .h y o « J2 Q I CQ

Kirk et al. • BOREAL FOREST BIRD POPULATIONS

27

X

Q

Z

LU

a.

CL

<

Q

UJ

z

H

Z

O

u

(N

ON

vO — ^

d d fs d

r4

d

e2

(N

ON

mcsco — o o d — d d

csi)

c

5

c3

CN

ON

(N

r-

ON

\0 '■O 'C

d odd

— sO

— d

ON

d

n

O

1/1

13

CQ

ON

O'

<N

r-

O'

n — (N

d d d

<N — Tf

d d H

cd

QD

(j

O

u

a

v

(D

c35

<N

O'

<N

r-

O'

'O 04 (N O —

d d d d d

in

c3

u

>

O

Z

CN

ON

rsi

O'

04 04 On

d d d 00

o

Cl

•o

c

2

75

<u

£

o

u

o

75

(A

C3

75

13

'£

J3

r3

XI

O

a

d

04

o^

O'

O

04)

C

Wilson Bull., 109(1), 1997, pp. 28-41

NESTS OF NORTHERN SPOTTED OWLS ON THE OLYMPIC PENINSULA, WASHINGTON

Eric D. Forsman' and Alan R. Giese'

Abstract. — We located 155 nests in 82 territories occupied by Northern Spotted Owls (Stri.x occidentalis caurina) on the Olympic Peninsula, Washington. All nests were in trees. Of 116 nests that were measured, 105 were in cavities and 11 were in external platforms on tree limbs. Cavity nests were typically in large holes in the side of the trunk or in the broken top of the trunk. Aspect of cavity entrances was non-random, with the majority of cavities facing east-north-east. Location of nest trees did not differ from expected values tor slope aspect or position on slope. Proportions of nest sites in different percent slope categories differed from availability, with more nests than expected in the higher percent slope categories. Nests usually were in stands with high overall canopy closure (> 70%), but canopy closure in the immediate vicinity of the nest varied from 35-90%. Most nests (87%) were in multilayered stands dominated by large trees. Nests in younger stands were typically in stands where remnant old trees were present. Owls changed nests between successive nesting events in 80% of all cases. Changes in pair members on a teiTitory did not influence the frequency with which pairs switched to a new nest tree in the next nesting year. Based on observed rates of attrition, the expected life span of nests was 120 years. Received 23 Jan. 1996. accepted 23 Aug. 1996.

Spotted Owls {Strix occidentalis) use a variety of structures for nests, including cavities, platforms constructed by other birds or mammals, plat- forms that are the result of natural accumulations of debris, and ledges on cliffs or cave walls (Bent 1938, Forsman et al. 1984, LaHaye 1988, Buchanan et al. 1993). In most regions, nesting is limited primarily to trees (Forsman et al. 1984, LaHaye 1988, Buchanan et al. 1993, Folliard 1993). Nesting on ledges on cliffs or cave walls is largely restricted to rocky canyons in the southwestern United States and Mexico (Dickey 1914, Ligon 1926, Bent 1938, Ganey 1988). However, two nests on cliffs have been observed in western Oregon (M. Brown pers. comm., J. Niles pers. comm). Although there have been no studies in which use versus availability of different nest types have been tested, it appears that the types of nests used are influenced by availability and possibly regional differences in climatic conditions.

With the exception of reports by Buchanan (1991 ) and Buchanan et al. (1993, 1995), relatively little information is available on characteristics of nests used by Northern Spotted Owls (5. rx caurina) in Washington. During a long-term study of demographic characteristics of Northern Spotted Owls on the Olympic Peninsula in western Washington (1987-

' USDA Forest Service. Pacific Northwest Research Laboratory. 3200 SW Jefferson Wav Corvallis Oregon 97331, and Dept, of Ftshenes and Wildlife, Oregon State Univ., Corvallis, Oregon 97331. * ’

28

Forsman and Giese • SPOTTED OWL NESTS

29

0

Eig. 1. Location of 116 nest trees used by Northern Spotted Owls on the Olympic Peninsula study area in northwestern Washington. In some cases a single dot may represent more than one nest because nests were very close together. We subdivided the area into western and eastern subprovinces based on differences in precipitation and vegetation.

1994), we located 155 nest trees, some of which were used in multiple years. Herein, we describe characteristics of a sample of those nests, in- cluding rates of nest attrition and frequency of reuse of nests.

STUDY AREA AND METHODS

The study area included the entire Olympic Peninsula (Fig. 1). Most Held work occurred on lands administered by the USDA Forest Service, but we also monitored some pairs of owls within the Olympic National Park and on lands administered by the Washington De- partment of Natural Resources. The Olympic Peninsula is a mountainous region character- ized by a wet, maritime climate and dense coniferous forests. Mean annual precipitation is highly variable, ranging from 1 15 cm at Quilcene on the east side of the peninsula (USDA Forest Service records for 1985-1995) to 365 cm on the west side of the peninsula (U.S. National Park Service records for 1987-1992).

Forests on the peninsula are generally dominated by mixtures of western hemlock, (Tsiiga heterophyla), western redcedar (Thuja plicata), sitka spruce (Picea sitchensis), Douglas-fir (Pseiidotsuga menziesii) and PaciHc silver fir (Abies amabilis). Western hemlock, western redcedar, and Sitka spruce predominate in more mesic areas, and Douglas-fir predominates

30

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

on drier sites. Pacific silver fir occurs at a broad range of elevations but becomes increasingly dominant at higher elevations.

History of nest use was determined by monitoring the same owl territories each year. Pairs were checked several times each year to determine their nesting status and to locate nest trees. Owls were marked with unique color bands, which made it possible to determine movements between alternate nests and to determine when pair members were replaced by new individuals. Nests were located by observing owls as they carried prey into cavities or platforms (Eorsman 1983).

The 155 nests located included 132 on the Olympic National Eorest, 13 in the Olympic National Park, four on lands administered by the Washington Department of Natural Re- sources, five on private lands, and one on the Quinault Indian Reservation. We measured 1 16 nest trees in 70 territories. The number of nest trees measured per territory was one (33 territories), two (28 territories), or three (9 territories). We climbed 86 nest trees in order to measure dimensions of nest cavities or platforms and measured 30 from the ground, with no nest dimensions taken. Selection of nests for measurement was not random. We simply measured as many nests as we could, incidental to the other objectives of our study. To as.sess whether types of nests used by spotted owls differed between the wet coastal region of the peninsula and the comparatively drier east slope of the Olympic Mountains, we stratified the study area for some analyses (Fig. 1 ).

We used a metric tape or clinometer to measure vertical distances. Tree diameter at breast height (DBH) was measured 1.4 m above ground, except that trees with swollen bases were measured immediately above the swelling. Tree diameter at nest height (DNH) was measured at the level of the nest. If the tree had secondary tops, the number of tops was recorded. Secondary tops were limbs that grew upward and formed a new top after the original top of the tree broke off or died. Trees were classified as alive if they had any live limbs, regardless of whether the nest was in a live portion of the tree or not.

Nest types were classified as side cavities, top cavities or external platforms. A side cavity was a hole in the side of a tree trunk. A top cavity was a hole or depression in the broken top of a tree bole that was accessed through the top of the bole. External platform nests were accumulations of sticks or other debris located on limbs outside a tree bole.

Nest entrance width and height were defined as the horizontal and vertical dimensions of the entrance in centimeters. These parameters were measured only for nests with obvious horizontal and vertical entrance limits. Such measures were not applicable to external nest platforms or to cavity nests with large, irregular shaped entrances. Cavity depth was the vertical distance from the cavity entrance to the cavity floor. Platform depth applied only to external platform ne.st.s and was the vertical distance from the bottom of the nest structure to the top edge of the structure. Mean diameter of each cavity or external platform was estimated by taking two measurements at right angles to each other across the cavity or platform and dividing the sum by two.

Percent cover above the nest was visually estimated a.s overhead cover from the perspec- tive ot an owl in the nest, including contributions from cavity structure and overhanging vegetation. All estimates of percent cover above the nest were made by the same observer, and were thus not subject to among-observer variation. Entrance aspect was an azimuth from the center of the tree through the center of the cavity opening or across the center of the nest platform (for external nests).

Each nest stand was assigned to one or more structural categories depending on the size distribution of the trees. Categorizations of ne.st stand structure were based on visual in- spection rather than on exhaustive measurements. Canopy closure within a 30 m radius of the nest tree was estimated visually. Elevations were recorded using altimeters or USGS maps. Other variables recorded were slope azimuth, slope gradient (in percent), and position

Forsman and Giese • SPOTTED OWL NESTS

31

of nests relative to topographic position on slope (lower third, middle third, upper third). We used X‘ tests to determine if slope aspect, slope percent, or topographic position of nests differed from what was available (Marcum and Loftsgaarden 1980). Expected values for categories of slope aspect, slope percent, and topographic position were determined from a random sample of 200 30-m^ grid cells on a digital elevation map of the Olympic National Forest. Aspects were grouped into eight 45° classes, and slopes were grouped into seven 15% classes. The Rayleigh test (Batschelet 1981) was used to test whether azimuths of nest entrances were randomly distributed.

The annual survival rate of nest trees (ct) was estimated by calculating the average pro- portion of known nest trees that survived each year. The estimated mean life expectancy of nest trees from the time they were first located was calculated using the formula -l/lntj) (Brownie et al. 1978:204).

RESULTS

We found 155 nests in 82 different owl territories, including 24 terri- tories where pairs nested only once, 19 where pairs nested in two different years, 23 where pairs nested in three years, 1 1 where pairs nested in four years, and four territories where pairs nested in five years, and one ter- ritory where a pair nested in six years. Two pairs nested twice in the same year after initial nesting attempts failed (Forsman et al. 1995). The total number of different nest trees per territory ranged from 1-4. Thirty-three territories had one nest, 29 had two nests, 16 had three nests, and four had four nests. All nests were in trees.

Nest characteristics. — We measured 1 16 nests in 70 different territories, including 65 nests in 40 territories in the western subprovince of the peninsula and 51 nests in 30 territories in the eastern subprovince (Fig. 1). Of the nests that were measured, 62 (53.4%) were in side cavities, 43 (37.1%) were in top cavities, and 11 (9.5%) were in external platforms. Frequency of use of different nest types differed between the eastern and western subprovinces of the peninsula (x^ = 19.4, 2 df, P < 0.001), with nests in side cavities predominating in the western subprovince (71%), and nests in top cavities predominating in the eastern subprovince (51%). Percent cover directly above the nest averaged slightly higher in the west- ern subprovince than in the eastern subprovince (Table 1) (t = —2.31, 92 df, P = 0.015).

Nest trees that were measured included 40 western hemlock, 36 Doug- las fir, 33 western red cedar, two Pacific silver fir, two grand fir {Abies grandis), one Sitka spruce, one western white pine {Finns monticola), and one unknown species. Ninety nests (78%) were in live trees and 26 (22%) were in dead trees. Frequency of occurrence in different tree species did not differ between nests in live trees, and nests in dead trees (x‘ = 1.72, 3 df, P = 0.633).

Of 90 nests in live trees, 67 (74%) were located above the lowest live limb (i.e., in the crown of the tree), and 23 (26%) were located below

32

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

Table 1 Measurements of Nests and Nest Trees Used by Northern Spotted Owls on the Olympic Peninsula, Washington, Subdivided by Physiographic Subregion
Variable		E. subregion		W. subregion		Entire sampi	e
N	.r	SE	N	X	SE	N	x	SE
Nest height (m)	43	23.2	1.72	63	23.5	1.12	106	23.2	0.96
Tree height (m)	44	33.1	2.01	62	46.0	1.29	106	40.6	1.28
Bole height (mp	26	19.6	1.59	54	20.1	0.96	80	19.9	0.82
DBH (cm)^	46	107.2	6.31	64	157.7	8.20	1 10	136.6	5.93
DNH (cm)*’	30	69.3	4.52	57	105.6	6.00	87	93.1	4.61
No. secondary top.s’’	42	1.0	0.24	62	1.4	0.20	104	1.2	0.15
Avg. nest diameter (cm)	31	43.9	2.40	55	46.3	1.16	86	45.4	1.14
Percent cover above nest	37	82.2	3.00	57	90.0	1.89	94	86.9	1.68
Cavity depth (cm)**	24	28.0	8.73	52	10.8	2.16	76	16.2	3.23
Platform depth (cm)"	7	21.0	2.05	2	23.5	3.50	9	21.6	1.71
Entrance width (cm)'	1 1	24.0	3.16	40	25.2	1.74	51	24.9	1.51
Entrance height (cm)*^	7	95.7	51.89	41	88.5	10.29	48	89.5	1 1.28

“ Height to first live limb.

^ DBH and DHN indicate diameter of tree at breast height and nest height, respectively.

' Number of live tree tops developing subsequent to the development of the original tree top.

Distance from cavity entrance to floor of cavity.

' Distance from top to bottom of external platform type nests.

'Entrance dimensions apply to nests in cavities only.

the crown. When both live and dead trees were included, 67 of 1 16 nests (58%) were in the crown of a live tree. Mean DBH of nest trees was 136.6 cm (SE = 5.93, range = 30-379 cm). On average, total height, DBH and DNH of nest trees were larger in the western subprovince than in the eastern subprovince (/ht = -5.39, P < 0.001 = -4.88, P <

0.001, /p)NH — —4.84, P < 0.001) (Table 1).

When only nests in live trees were included, the majority (83%) of nests in top cavities were in trees with secondary tops. Only 53% of nests in side cavities and 50% of nests in external platforms occurred in trees with secondary tops. When both live and dead trees were included, the percent of nests in trees with secondary tops was 65% for top cavity nests, 44% for side cavity nests, and 50% for external platforms (y^ = 3.93, 2 df, E = 0.140).

Mean entrance dimensions of cavities were 24.9 cm (width) and 89.6 cm (height) (Table 1). However, entrance dimensions were recorded only for cavities that had well-defined entrances. As a result, our mean entrance dimensions did not represent many cavities that were accessed through large, irregular shaped holes that were formed when tops or large limbs broke off of trees. Nests of the latter type were typically entered through a long crack of variable width in the side of the tree or through the jagged

Forsman and Giese • SPOTTED OWL NESTS

33

Table 2 Measurements of Spotted Owl Nests in Cavities and External Platforms on the Olympic Peninsula, Washington
Variable		Cavity nests	Platform nests"	Comparison	of means'^
N	X	SE	N	X	SE	/	p
Nest height (m)	95	23.3	1.02	1 1	24.1	3.00	-0.27	0.791
Tree height (m)	95	40.7	1.36	11	39.8	3.99	0.21	0.837
Bole height (mT	70	19.9	0.87	10	20.2	2.66	-0.13	0.896
DBH (cm)'*	99	141.8	6.15	11	88.7	15.74	2.77	0.007
DNH (cm)'*	79	97.1	4.74	8	53.4	10.73	2.86	0.005
Avg. nest diamter (cm)	76	45.3	1.15	10	48.0	4.59	-0.76	0.449
Percent cover above nest	83	87.0	1.79	1 1	86.2	5.02	0.12	0.906
No. secondary tops	94	1.2	0.15	10	1.7	0.75	-0.98	0.332

' Platforms on tree limbs.

" 2-tailed r-test. Data on percent cover were log-transformed for /-tests.

' Height to first live limb.

DBH and DHN indicate diameter of tree at breast height and nest height, respectively.

top of a tree with a broken bole. Entrance dimensions of cavities did not differ between the eastern and western subprovinces = -0.32, 49

df, P = 0.748; tht = 0.14, 48 df, P = 0.896). The distribution of nest cavity entrance azimuths was non-random, with the majority of nests facing east-north-east (mean angle = 72°, SE = 7.5, r = 0.2021, P = 0.038, N = 94).

Height of the cavity entrance above the cavity floor averaged 16.2 cm (SE = 3.23, median = 4.5 cm) and ranged from 0-66 cm with a single exceptional value of 203 cm (Table 1). In the latter case, the owls accessed the bottom of the cavity by hopping down a ladder-like series of old branch cores that projected into the hollow interior of the tree.

On average, cavity nests occurred in larger trees than external platform nests (Table 2). Variables that did not differ between cavity nests and external platform nests were nest height, total tree height, bole height, percent cover directly above the nest, mean number of secondary tops, and average diameter of nests (Table 2).

The nest substrate in cavity nests typically consisted of a decomposing mixture of wood, bark, conifer needles, twigs, and insect tillings. Small amounts of moss or lichens were sometimes mixed with the debris. In all cases, it appeared that substrate accumulations were the result of decom- position of the interior of the tree and/or debris falling into the cavity. Nests typically contained remains of prey and pellets forming a layer up to 7 cm deep. Eggshell fragments were rarely found in nests except in conjunction with nest failures.

34

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

Of 1 1 nests classified as external platforms, five were old stick nests that appeared to have been built by Common Ravens (Corvus corax) or Northern Goshawks {Accipiter gentilis), three were on debris platforms on large clumps of deformed limbs caused by dwarf mistletoe (Arceu- thobium spp.) infections, and three were collections of debris that accu- mulated where a tree split into multiple tops. The nest substrate in external platform nests was typically a decomposing mixture of tree bark, conifer needles, twigs, lichens, and moss.

Site characteristics. — Of 1 16 nests measured, 82 (71%) were in forests characterized by multi-layered canopies where the dominant overstory trees were > 100 cm DBH. Twenty-two nests (19%) were in multi-layered forests dominated by 50—99 cm DBH trees that included scattered indi- viduals or patches of large (> 100 cm DBH) old trees. Three nests (2%) were in relatively even-aged forests of 50-99 cm DBH trees, and nine nests (8%) were in forests that included a mosaic of small trees (DBH = 13-49 cm) and larger trees (DBH > 50 cm).

Estimated mean canopy closure in the vicinity of the nest tree ranged from 30—95% (x — 70%, SE = 1.44, N = 104) and did not differ sig- nificantly {t — 1.50, 102 df, P — 0.136) between the eastern subprovince (x = 72.6, SE = 2.43, N = 41) and western subprovince (x = 68.2, SE — 1.75, N — 63). Of 13 nest sites with estimated canopy closure 50%, all were in or adjacent to small natural openings in stands that otherwise had high (> 70%) canopy closure.

In the western subprovince, 63 of 65 (97%) nest trees were in forests in which the majority of trees in the overstory and understory were west- ern hemlock, and two (3%) were in forests dominated by Douglas-fir. Western redcedar was present in variable numbers in nearly all stands. Sitka spruce and Pacific silver fir were common associates on lowland and upland sites, respectively.

In the eastern subprovince, 37 nests (73%) were in forests in which the majority ol trees in the overstory were Douglas fir, and 14 (27%) were in forests dominated by western hemlock. Common associates m Douglas-fir stands were western hemlock, western redcedar, and Pacific silver fir, with the silver fir component generally increasing with elevation Stands dominated by western hemlock typically included variable amounts of western redcedar, Douglas-fir, and Pacific silver fir Grand fir (Abies grcmdis) was a relatively uncommon overstory component in a few low-elevation nest sites in the eastern subregion

In both subprovinces, western white pine, grand fir, red alder (Alnus rubra) and bigleaf maple (Acer macrophylum) were present in many nest

stands, usually m small amounts. Red alder was typically limited to areas along streams or swampy areas.

Forsmcm and Giese • SPOTTED OWL NESTS

35

Table 3 Proportions of Spotted Owl Nest Locations and Random Locations in Different Physiographic Categories on the Olympic Peninsula, Washington, 1987-1994
Variable	Nests-*	Random	95% CP
Slope aspect
North	0.108	0.140
Northeast	0.171	0.105
East	0.090	0.120
Southeast	0.126	0.155
South	0.1 17	0.125
Southwest	0.108	0.120
West	0.144	0.140
Northwest	0.135	0.095
Slope gradient (%)
0-14	0.129	0.140	-0.106, 0.084
15-29	0.086	0.165	-0.168, 0.010
30-44	0.129	0.205	-0.177, 0.025
45-59	0.155	0.210	-0.161, 0.512
60-74	0.216	0.180	-0.076, 0.148
75-89	0.164	0.065	0.006, 0.192
90 +	0.121	0.035	0.007, 0.165
Position on slope
Lower third	0.500	0.450
Middle third	0.345	0.400
Upper third	0.155	0.150

“ Sample sizes for nest variables were 1 1 1 (aspect) and 1 16 (% slope, slope position). Sample size for random samples was 200.

^ Ninety five percent confidence intervals are presented only for variables that had significant X‘ ^or use versus availability. Intervals indicate whether a category was used more than expected (interval is positive), less than expected (interval is negative) or in proportion to availability (interval overlaps 0).

Location of nest trees did not differ from expected values for slope aspect (x" = 5.17, 7 df, f = 0.64) or position on slope (x“ = 0.99, 2 df, P = 0.61) (Table 3). Proportions of nest sites in different percent slope categories differed from availability, with more nests than expected in the steeper percent slope categories (x^ = 22.61, 6 df, P = 0.001) (Table 3).

Elevation at nest sites ranged from 104-975 m in the western sub- province and 1 14-1 189 m in the eastern subprovince. When nests were grouped into 150 m elevation bands, the distribution of nests differed in the eastern and western subprovinces (x^ = 35.6, 7 df, P <0.001, Fig. 2). The proportion of nests above 600 m elevation was 59% in the eastern subprovince, compared to only 10% in the western subprovince. In both regions, upper elevations at which nests were located generally corre- sponded with the transition to stands that were largely dominated by Pa-

36

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

25

CDwest Sub Region 0 East Sub Region

CO

4—'

CO

o

CD

XI

E

u

Elevation (m)

Fig. 2. Distribution of 108 Spotted Owl nests by elevational zone on the Olympic Pen- insula, Washington, 1987—1994.

cific silver fir. Due to (differences in temperature ancd precipitation this transition occurrecd at about 900 m in the western subprovince and 1200 m in the eastern subprovince (Henderson et al. 1989).

Site history. — Of 1 16 nest trees measured, three became nonfunctional during the study. In two cases, the tree fell down. In one case, the tree remained standing but the cavity collapsed. The average annual survival rate of nest trees from 1987-1993 was 0.992 (SE = 0.003). Estimated mean life expectancy of nest trees from the time they were first located was 124 years.

Although their historical nests were usually still intact, owl pairs changed nests in 88 of 1 18 (75%) sequential nesting attempts. This cal- culation was based on the entire sample of 155 nests, and all years of

Forsmcm and Giese • SPOTTED OWL NESTS

37

data for all pairs that nested in > 2 yrs. Frequency of switching to dif- ferent nests in different years did not differ between pairs that underwent a change in a pair member versus pairs that did not change members (x^ = 0.190, 1 df, f* = 0.663). Use of a different nest following replacement of a pair member was 80% following a male replacement (N = 10) and 60% following a female replacement (N = 10) (x^ = 0.952, 1 df, F* = 0.329).

The frequency with which pairs changed nests in sequential nesting years did not differ between 17 cases in which pairs failed at nesting (0.765) and 83 cases in which pairs nested successfully (0.699) (x^ = 0.298, 1 df, P = 0.585). This comparison did not include two pairs that failed and renested in the same year. In the latter cases, both pairs moved to a new nest for their second nesting attempt.

At 40 territories where we observed at least two nest change events, we found that at least 40% of the time pairs changed back to a nest that had been used in previous years. The median distance between alternate nests was 0.52 km (range = 0.03-3.36 km, N = 92). At 46 territories where we confirmed two or more nests used in different years, the area of the smallest circle that encompassed all of the known nests ranged from 0.001-8.87 km^ (median = 0.33 km^).

DISCUSSION

Species of trees used for nesting on the peninsula were about equally divided among western hemlock (35%), Douglas-fir (31%) and western red cedar (28%). In contrast, some studies in other regions have reported the majority of nests in Douglas-fir (Forsman et al. 1984, LaHaye 1988, Buchanan et al. 1993). In managed stands in California, 35% of nests were in redwoods {Sequoia sempervirens), 27% were in Douglas-fir, 17% were in grand fir, and 13% were in hardwoods (Folliard 1993:31). The considerable variation in species composition of nest trees in different regions suggests that selection is based primarily on the presence of a suitable cavity or platform rather than tree species.

The proportion of nests in dead trees on the Olympic Peninsula (23%) was higher than has been reported from study areas in Oregon (4%), the Washington Cascades (12%), and managed stands in northern California (10%) (Forsman et al. 1984:31, Buchanan et al. 1993:5, Folliard 1993: 30). We do not know if this reflected a difference in the relative abun- dance of suitable nests in live vs. dead trees, or was due to other factors.

The proportion of nests in external platforms on the Olympic Peninsula (10%) was considerably lower than has been reported for other regions. For example, external platform nests comprised 19% of nests in samples from western Oregon (Forsman et al. 1984:32), and 50-80% of nests in

38

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

samples from the east slope of the Cascades in Oregon and Washington (Forsman et al. 1984:32, Buchanan et al. 1993:5). It is unclear whether the high proportion of nests in cavities on the Olympic Peninsula is sim- ply a function of availability. An alternative hypothesis is that Spotted Owls on the Olympic Peninsula actually select for cavity type nests be- cause they provide more protection from the frequent heavy rains that occur during the nesting season.

Availability of different nest types does appear to influence nest selec- tion by Spotted Owls (LaHaye 1988, Folliard 1993). Folliard (1993:51) noted that “Generally, platform nests tended to be used more often in even-aged stands with few or no residual trees remaining.” On the east slope of the Washington Cascades, where many stands had been thinned to remove large old trees, 80% of nests examined by Buchanan et al. (1993) were in platforms. These observations suggest that, where large trees with cavities have been removed or are otherwise lacking, and where heavy precipitation is not a common occurrence during the nesting season, platform nests provide a viable alternative for Spotted Owls, if adequate numbers of platforms are present.

Without some sort of overhanging cover, nests in the top of broken off trees are likely to be more exposed to the elements than nests in side cavities. Therefore, we hypothesized that the proportion of nest trees with secondary tops should be higher for nests in top cavities than for nests m side cavities. Although there appeared to be a trend in this direction, the differences were not significant. Thus, while Spotted Owls do appear to select nests with good overhead cover on the Olympic Peninsula, the

presence of secondary tops does not appear to be a reliable indicator of this.

Mean nest tree height, nest height, and nest tree dbh on the Olympic Peninsula were about the same as, or slightly greater than, was reported for Oregon and northern California (Forsman et al. 1984, LaHaye 1988 Folliard 1993), but were considerably greater than values reported for the east slope of the Cascades in Washington (Buchanan 1991, Buchanan et al. 1993). The considerable variation in these parameters among regions suggests that size of the tree or height of the nest are relatively less important in nest selection than the presence of a suitable cavity or plat- form. ^

Forsman et al. (1984) noted that cavity nests used by Spotted Owls in Oregon tended to be m the upper two-thirds of the canopy while platforms tended to be m the lower third. Buchanan et al. (1993) noted that cavity nests on the east slope of the Cascades were located at all levels in the forest canopy, whereas most platform nests were in the lower third of the canopy. In our sample, nests were found at all levels in the forest canopy

Forsman and Giese • SPOTTED OWL NESTS

39

but the majority of both cavity nests (88%) and platform nests (82%) were in the lower two-thirds of the canopy. We do not know if this reflected selection by the owls or simply reflected the availability of nests. However, the preponderance of platform nests in the lower two-thirds of the canopy is likely explained by availability; large limbs capable of sup- porting large platform nests tend to be located in the lower part of tree boles, and debris platforms tend to form in the lower canopy as a result of materials falling from above.

Selection for cavities that opened to the east-northeast could possibly be explained by the fact that storms on the Olympic Peninsula typically approach from the west-southwest. However, since we did not have data on the actual availability of cavities with respect to entrance aspect, we could not discount the possibility that selection was based on availability.

Mean width of cavity entrances on the Olympic Peninsula was slightly less than in Oregon (30 cm; Forsman 1983:32) and was slightly above the upper end of the range reported by (Folliard 1993:28) for managed stands in northern California (15-23 cm). The smallest entrance we mea- sured was 16 X 18 cm, which is similar to the smallest entrance found in California (15 X 18 cm; Folliard 1993:28). Mean diameter of nests on the Olympic Peninsula (45 cm) is similar to values reported for Oregon (50 cm; Forsman et al. 1984:32) and California 52 cm; Folliard 1993: 28). Folliard (1993) noted that width of the nest platform did not differ according to nest type, a result supported by our observations. Although we recorded one cavity that was 203 cm deep, most cavities used by Spotted Owls are less than 100 cm deep (this study, Forsman et al. 1984). Infrequent use of deeper cavities may be a function of availability or may be because owls find it difficult to climb in and out of such cavities.

The fact that nest sites in our study did not differ from expected values for slope aspect or position on slope suggests that these variables were relatively unimportant in nest site selection on the Olympic Peninsula. Although we found higher than expected proportions of nests on steeper slopes, we are unsure whether this represented selection by the owls. It is possible that the distribution of suitable owl nesting habitat was skewed towards the steeper slope categories as a result of historical patterns of forest management in which harvest on steep, unstable slopes was avoid- ed. In northwestern California, Blakesley et al. (1992) also found no dif- ferences between observed and expected values for slope aspect at Spot- ted Owl nest sites. In contrast to our findings, Blakesley et al. (1992) found no differences between observed and expected values for percent slope at nests and found that owls nested on the upper third of slopes less than expected and nested on the lower third of slopes more than expected. Although they did not have data on the availability of different

40

THE WILSON BULLETIN • Vol. 109, No. 1. March 1997

slope categories in their study areas, some investigators have compared the distribution of slope azimuths at nests with a uniform distribution (Forsman et al. 1984) or random distribution (LaHaye 1988, Buchanan 1991, Folliard 1993). In these cases, some investigators have found sig- nificant differences (LaHaye 1988), while others have not (Forsman et al. 1984, Buchanan 1991, Folliard 1993).

The mean and range of canopy closure at nest sites in Oregon (L = 69%, SE = 2.65, range = 35-91) (Forsman et al. (1984:30), were nearly identical to measurements from the Olympic Peninsula (x = 70%, SE = 1.44, range = 30-95%). Although canopy closure at nests in Oregon and on the Olympic Peninsula was highly variable, Eolliard (1993) and Bu- chanan et al. (1993) stressed the consistency of high canopy cover at Spotted Owl nest sites in their study areas. We are unsure to what extent canopy closure estimates can be compared among study areas, because estimates may have been influenced by differences in methodology or observers.

In contrast to the low rate of attrition of nests on the Olympic Penin- sula, Eorsman et al. (1984) noted a relatively high rate of attrition of nest trees in Oregon. Of eight nests that became unusable in the Oregon study, four fell down, three were cut down during logging operations, and one cavity collapsed. Although he did not mention overall attrition rates of nests, Folliard (1993) noted that platform nests in northern California were especially ephemeral in nature.

The tendency of Spotted Owls on the Olympic Peninsula to use dif- ferent nests in different nesting years contrasts with reports from other regions indicating frequent reuse of the same nests in different years (e g Forsman et al. 1984, Ganey 1988). This behavior did not appear to be influenced by turnover of pair members or by success or failure of nests in prior years. We do not know of any obvious factors that should have caused owls on the peninsula to use alternate nests more frequently than owls in other regions.

Habitat selection by spotted owls is likely influenced by a variety of factors, including prey availability, availability of suitable nests and roosts, and presence of escape cover. Use of nest stands by Spotted Owls on the Olympic Peninsula appears to be almost entirely restricted to stands of large trees or younger stands in which there are residual old trees. Retention of small clusters of live trees in harvest units may provide future nesting habitat in stands that would otherwise be uninhabitable bv Spotted Owls.

ACKNOWLEDGMENTS

This study would not have been possible without a dedicated group of field assistants who helped locate nests, including Meg Amos, Duane Aubuchon, Sue Grayson, Martha

Forsman and Giese • SPOTTED OWL NESTS

41

Jensen, Debaran Kelso, Timm Kaminski, Jeff Lewis, Rich Lowell, Dave Manson, Kevin Maurice, Matt Nixon, Ivy Otto, Doreen Schmidt, Stan Sovern, Jim Swingle, Margy Taylor, and Joe Zisa. Funding was provided by the USDA Eorest Service Pacific Northwest Re- search Laboratory, USDA Eorest Service Pacific Northwest Regional Office, and USDI National Biological Service Forest and Rangeland Ecosystem Science Center, Corvallis, Oregon.

LITERATURE CITED

Batschelet, E. 1981. Circular statistics in biology. Academic Press, London, England.

Bent, A. C. 1938. Life histories of North American birds of prey. Part 2. U.S. National Museum Bull. 170.

Blakesley, J. a., a. B. Eranklin, and R. J. Gutierrez. 1992. Spotted owl roost and nest site selection in northwestern California. J. Wildl. Manage. 56:388-392.

Brownie, C. D. R. Anderson, K. P. Burnham, and D. S. Robson. 1978. Statistical infer- ence from band recovery data — a handbook. USDI Fish and Wildl. Serv., Resource Publ. No. 131.

Buchanan, J. B. 1991. Spotted owl nest site characteristics in mixed conifer forests of the eastern Cascade Mountains, Washington. M. Sc. thesis, Univ. of Washington, Seattle.

, L. L. Irwin, and E. L. McCutchen. 1993. Characteristics of Spotted Owl nest

trees in the Wenatchee National Forest. J. Raptor Res. 27:1-7.

, , AND . 1995. Within-stand nest site selection by Spotted Owls in

the eastern Washington Cascades. J. Wildl. Manage. 59:301-310.

Dickey, D. R. 1914. The nesting of the Spotted Owl. Condor 16:193-202.

Folliard, L. 1993. Nest site characteristics of Northern Spotted Owls in managed forests of northwest California. M.S. thesis, Univ. of Idaho, Moscow.

Forsman, E. D. 1983. Methods and materials for locating and studying Spotted Owls. USDA For. Serv. Gen. Tech. Rept. PNW-162.

, A. Giese, D. Manson, S. Sovern, and D. R. Herter. 1995. Renesting by Spotted

Owls. Condor 97:1078—1080.

, E. C. Meslow, and H. M. Wight. 1984. Distribution and biology of the Spotted

Owl in Oregon. Wildl. Monogr. 87.

Ganey, j. L. 1988. Distribution and habitat ecology of Mexican Spotted Owls in Arizona M.S. thesis. Northern Arizona University, Flagstaff.

Gutierrez, R. J. 1985. An overview of recent research on the Spotted Owl. Pp. 39-49 in Ecology and management of the Spotted Owl in the Pacific Northwest (Gutierrez, R. J., and A. B. Carey, eds.). USDA Eor. Serv. Gen. Tech. Rept. PNW-185.

Henderson, J. A., D. H. Peter, R. D. Lesher, and D. C. Shaw. 1986. Forested plant associations of the Olympic National Forest. USDA For. Serv. Ecol. Tech. Paper R6-ECOL-TP 001-88.

LaHaye, W. 1988. Nest site selection and nesting habitat of the Northern Spotted Owl in northwest California. M.Sc. thesis, Humboldt State University, Areata.

Ligon, j. S. 1926. Habits of the Spotted Owl (Syrnium occidentale). Auk 43:421-429.

Marcum, C. L. and D. O. Loftsgaarden. 1980. A nonmapping technique for .studying habitat preferences. J. Wildl. Manage. 44:963—968.

Wilson Bull.. 109(1), 1997, pp. 42-51

NEST-SITE SELECTION AND REPRODUCTIVE SUCCESS OF CALIFORNIA SPOTTED OWLS

W. S. LaHaye,'-* R. J. Gutierrez' and D. R. Call' ^

Abstract. — We evaluated quality of nesting habitat and nest-site selection of an insular population of California Spotted Owls {Stri.x occidentalis occidentalis). We assessed habitat structure for successful and unsuccessful nests from 103 independent territories at three spatial scales, and habitat selection by comparing nest stand structure with identical variables from random points. Fledging success was unrelated to nest type, nest tree, nest stand characteristics, or habitat type. However, nest productivity was greatest in lower elevation oak/big-cone fir habitat (1.7 fledglings per successful nest). Nest stands were characterized by greater variation in tree size, higher canopy closure, and greater basal area of large trees compared with random points. We were able to differentiate consistently between nest and random points using discriminant function models (==79% correct classification). Our results confirm previous ob.servations that California Spotted Owls will use a variety of habitats, but these habitats are consistently characterized by greater structural complexity compared with available habitat. Received 24 May 1996, accepted. 30 Sept. 1996.

Conservation of the Spotted Owl {Strix occidentalis) is controversial because of its affinity for economically important, late serai-stage conifer forests (Gutierrez et al. 1995). Both Northern and Mexican Spotted Owls iS. o. caurina, S. o. lucida, respectively) are Federally listed threatened species because of past and projected habitat loss (U. S. Department of Interior 1990, 1993). In contrast, the California Spotted Owl (S. o. occi- dentalis) is not currently under consideration for Federal protection, pre- sumably because it inhabits a variety of habitat types other than late seral- stage conifer forests and there is no evidence for decline in the largest population occurring in the Sierra Nevada (Verner et al. 1992). Never- theless, at least one insular population of California Spotted Owls is de- clining rapidly (LaHaye et al. 1994).

Even though California Spotted Owls have been observed in a variety of habitat types, we do not know which of these are preferred habitats. More importantly, we do not know what contribution each habitat type represents to the overall viability of the subspecies. For instance, terri- torial displacement may force individuals to use less preferred habitats (Van Horne 1983). Individuals in suboptimal habitats may represent sink populations (Pulliam 1988), and while sink populations may help to sta- bilize a regional population (or metapopulation), they would not be viable by themselves (Pulliam and Daniekson 1991).

' Dept, of Wildlife, Humboldt State Univ., Areata, California 95321.

Present address— Dept, of Zoology, Washington State Univ., Pullman, Washington 99164 4^56 ' Present addres.s— PO. Box 525, Big Bear City, California 92314.

42

LaHaye et al. • SPOTTED OWL NEST-SITE SELECTION

43

In this study we evaluated habitat selection and habitat quality of an insular population of California Spotted Owls. We incorporated measures of fitness (reproductive success and productivity) to evaluate the relative quality of habitat characteristics and different habitat types that the owls were using. We also evaluated habitat selection by comparing owl nest stands to random points throughout the San Bernardino Mountains.

STUDY AREA AND METHODS

The California Spotted Owl occurs as insular populations in southern California (LaHaye et al. 1994) with the largest of these island populations in the San Bernardino Mountains. This mountain range, which is part of the Transverse Range Geologic Province (Norris and Webb 1990), includes a large elevation gradient (800 m to 3500 m) as well as the majority of habitats used by the subspecies throughout its geographic range. Mean annual precipi- tation ranges from less than 20 cm to more than 100 cm and is strongly influenced by elevation, topography, and rain shadow effects (Minnich 1988). The vegetation is diverse ranging from Mojave Desert scrub (Vasek and Barbour 1977) at lower elevations to alpine (Major and Taylor 1977) on San Gorgonio Mountain. Most Spotted Owls occupy mixed conifer forests between 1000 m and 2500 m elevation.

Owl survey methods. — We located Spotted Owls and assessed their reproductive activity following methods of Franklin et al. (1996). Nests were located by following male owls to nest trees or by observing females leaving or entering nests. To minimize disturbance we did not measure nesting habitat until after juveniles fledged. Nests were classified as plat- form, cavity or broken-top (LaHaye 1988). Broken-top nests were typically found near the breakpoint of a broken trunk. Cavity nests were usually formed by a large branch tearing free of the main stem. Both of the above nest types required advanced heart rot for proper development. Platform nests were either abandoned stick nests constructed by other animals or natural accumulations of debris in the branches.

Vegetation measurement. — We measured vegetation characteristics using a variable cir- cular plot (Mueller-Dombois and Ellenberg 1974) at nest and random locations. Random points were .selected from universal transverse mercator coordinates throughout forest habitat in the San Bernardino Mountains. At nest and random points we estimated basal area of trees using a 20-factor basal area prism (Dil worth 1981). We measured the height and diameter at breast height (dbh) of each tree tallied with the prism. Diameter estimates were then grouped for further analysis with the conifer dbh classes sapling (0.1-25.0 cm), pole (25.1-50.0 cm), medium (50.1-75.0 cm) and large (>75.0 cm). Hardwood dbh classes were similar except we based them on 15 cm intervals instead of 25 cm intervals. We estimated percent canopy closure using a concave, spherical densiometer (Lemmon 1957). Other hab- itat characteristics were measured using standard techniques (see LaHaye 1988). Nests were also classified by their location in mixed conifer forests above 1800 m (Pinus jeffreyi, P. ponderosa, P. lambertiana, Abies concolor), oak/big-cone fir forest below 1500 m (Quercus chrysolepis, Pseudotsuga macrocarpa) and mid-elevation conifer/hardwood habitat.

Statistical analysis. — We explored differences between successful and unsuccessful nests across three spatial scales. If one or more fledglings were produced at a nest, then it was classified as successful. At the smallest scale, we tested for independence between nest success and both nest type and nest tree characteristics (dbh, nest height, tree height). We then expanded the analysis to nest stand variables and finally we considered patterns of nest success between broad habitat categories. We also evaluated differences in nest productivity

44

THE WILSON BULLETIN • Vol. 109, No. I. March 1997

(average number of fledglings/successful nest) at the broadest scale. In order to avoid pseu- dorephcation, only one nest site per owl territory was used in our analyses. When data appeared normally distributed we used parametric statistics for our comparisons, otherwise nonparametric tests were used. Lor multiple pairwise comparisons we adjusted our signifi- cance level using a 95% Bonferroni interval to avoid excessive Type I error (Neter et al. 1990) and used Tukey’s studentized range test for significant ANOVAs. We compared slope aspects (compass bearings) using circular statistics (Batschelet 1981).

In order to examine differences in forest structure, we calculated the standard deviation for all tree dbh measurements at each sample point. This standard deviation was then treated as a random variable and used to examine differences in variability of tree size between nest and random points. We excluded points where less than two trees were present ( 1 nest and 33 random sites). This is a conservative test of tree structure differences because Spotted Owl sites usually contain more trees than random sites.

We tested Spotted Owl nest-site selection by comparing 29 vegetation variables from nest points with corresponding values from random points. We also assessed our ability to clas- sify Spotted Owl nesting habitat using a series of discriminant function models (OEMs) (Capen et al. 1986, Call et al. 1992). The OEMs were restricted to five variables that were significantly different between nest and random points (P < 0.002) and minimally correlated (r < 0.6, Spearman rank test; percent slope, percent canopy closure, hardwood basal area, conifer basal area). Using these five variables we constructed a series of 25 nonparametric OEMs based on a randomly .selected subsample of nest and random points (approximately 6 of each type). The DEM algorithm used a k-nearest neighbor density function (k = 20, SAS 1989). The ability to discriminate between random and nest points was then evaluated by examining the classification rate for the data set used to construct each DEM (cross- validation) and the ability of each DEM to classify correctly the remaining nest and random points (independent points from approximately 25 nest and 221 random sites). Percent cor-

-^‘^"'‘^‘'ance using CoheiTs Kappa statistic (Titus et

RESULTS

Between 1987 and 1994 we ioeated 216 California Spotted Owl nests at 103 sites occupied by territorial owls. All nests were in trees and the elevation of nest sites ranged from 885 m to 2560 m. We found nests in ten different tree species (71% conifer, 29% hardwood, Table I) and the majority (59%) of these nests were platforms. For the 103 independent nests (one from each territory) we found that the average Spotted Owl nest was Ioeated 16.1 m (standard deviation, j = 6.9) above ground in a tree that was 24.4 m (.r - 9.1) tall with a dbh of 90 8 cm (s = 27 6) Platform nests were located in trees having a dbh (j-' = 75 0 cm s = 34.9) that was significantly smaller than either cavity nest trees (x = 108 3 cm, V = 29.1) or broken-top nest trees (x = 122.3 cm v = 29 0 F = 18.2, df = 102 P < 0.0001). No platform nests were found on snags ae , standing dead trees), whereas 13.5% of broken-top nest trees and 234% of cavity nest trees were snags. Mean slope aspects where nest and random points occurred {(f) = 348°, r = 0.45, N = 103- = 341°

- 0.23, N = 296, respectively) were not significantly different (P > 0.05).

UiHave et al. • SPOTTED OWL NEST-SITE SELECTION

45

Table 1

Nest Tree Species and Nest Type for All Nests Used by California Spotted Owls in THE San Bernardino Moltntains, California (1987-1994)

Tree species	Number of nests (%)	Platform	Percent nest types Cavity	Broken-top
Abies concolor	75 (34.7)	34.5	13.8	51.7
Almis rhombifolia	3 (1.4)	66.7	0.0	33.3
Calocednis decurrens	19 (8.8)	89.4	5.3	5.3
Pinus coulteri	1 (0.5)	0.0	100	0.0
P. jeffreyi or P. ponderosa	29 (13.4)	58.6	34.5	6.9
P. lambertiana	19 (8.8)	57.9	36.8	5.3
Pseudotsuga macrocarpa	29 (13.4)	34.5	13.8	51.7
Quercus chrysolepis	34 (15.7)	67.6	23.5	8.8
Q. kelloggii	7 (3.2)	0.0	85.7	14.3
Total	216	58.8	24.1	17.1

Thirty-nine percent of the owl territories occurred in higher elevation mixed conifer forests, while 41% occurred in oak/big-cone fir forests. Twenty percent of the territories were in mixed conifer/hardwood habitat (Table 2).

Nest success. — At the two smaller spatial scales (i.e., nest tree and nest stand) we found no significant differences between successful nests (N = 77) and unsuccessful nests (N = 26) (F and tests, P > 0.05). Nest success also was independent of habitat type (x^ = 1.7, df = 2, P = 0.4), but productivity was not. We found more juvenile Spotted Owls fledged from nests located in the oak/big-cone fir forests when compared with the mixed conifer and conifer/hardwood forests (Wilcoxon rank test, x^ = 7.3, df = 2, P = 0.026, Table 2).

Characteristics of nesting habitat. — In general, nest sites were multi-

Table 2

Habitat Types and Fledging Success for California Spotted Owl Nests Located in THE San Bernardino Mountains, California (1987-1994)

Habitat type	Number of nests (% successful)	Average fledglings per nest (.?)“	Average fledglings per successful nest’’ (.r)
Oak/big-cone fir	42 (81)	1.39 (0.87)	1.72 (0.61)
Conifer/hardwood	21 (67)	0.98 (1.05)	1.46 (0.97)
Mixed conifer	38 (76)	0.95 (0.77)	1.31 (0.58)

* Standard deviation. ^ Nest productivity.

46

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

Table 3

Summary of Vegetation Variables that Were Significantly Different between Random Points and California Spotted Owl Nest Points in the San Bernardino

Mountains, California (1987-1994)

Habitat variable	Nest points (N	= 103)	Random points (N = 296)
Mean*	% CV"	Mean	% cv
Percent canopy closure	79.3	22.3	52.4	49.9
Percent slope	54.2	49.8	32.1	68.7
Broken-top tree basal area*-	2.9	174.3	0.5	322.9
Snag basal area	4.8	116.7	1.8	217.8
Hardwood basal area (30.1—45 cm dbh)	3.2	216.7	0.9	332.8
Hardwood basal area (>45 cm dbh)	4.9	144.7	0.8	380.4
Total conifer basal area	37.1	59.5	20.1	85.8
Conifer basal area (50.1—75 cm dbh)	9.6	100.3	4.9	130.1
Conifer basal area {>15 cm dbh)	19.1	77.4	6.7	124.2

•* Includes zero values for all variables. ^ Percent coefficient of variation. Square meters per hectare.

storied stands composed of both conifers and hardwoods (Table 3). In addition, basal areas for large conifers and hardwoods, broken-top trees and snags were significantly higher in nest stands than in random loca- tions. Nest points had a greater mean standard deviation for tree sizes than random points (x = 30.9, .? = 1 1.6, N = 102; T = 23.0, 5 = 12.8, N = 263, respectively; t = 5.6, P < 0.0001) showing that nest stands had greater variability in tree sizes.

We were able to differentiate consistently between nest and random points based on our DFMs (Table 4). Overall rates of correct classification were very similar for both cross-validation and independent classification schemes (—79%, P < 0.0003). There was greater variation in the correct classification for independent nests which was not surprising given the smaller sample size for this group (average N = 26; Table 4). Overall variation in correct classification was low (coefficient of variation < 10%).

DISCUSSION

Habitat .selection in Spotted Owls has been studied extensively (Gu- tierrez et al. 1995). The extent of inference, however, is usually limited in these studies for many reasons (Wolff 1995). For example, logistical and financial constraints of field research often restrict selection studies to small sample sizes (e.g., Solis and Gutierrez 1990, Carey et al. 1990, Call et al. 1992) and pseudoreplicated designs (e.g., Solis and Gutierrez

UiHa\e et al. • SPOTTED OWL NEST-SITE SELECTION

47

Table 4

Percent Correct Classification Rates for 25 Discriminant Function Models (DEM) Comparing California Spotted Owl Nest Points with Random Points in the San Bernardino Mountains, California (1987-1994)

Type of DFM	Mean percent correct classification (% CV)	Range	N=	Kappa
Cross-validation^
Nest	82.6 (3.6)	74.6-87.5	77
Random	78.2 (6.3)	67.9-86.1	76
Total	80.4 (3.8)	74.2-85.4	153	0.6 1*<^
Independent
Nest	83.1 (8.8)	71.9-95.7	26
Random	77.7 (3.3)	73.1-81.7	220
Total	78.2 (2.8)	74.8-82.7	246	0.35*

“ Mean sample size; actual sample size will vary slightly between each DFM.

^ Includes only those points that were used to formulate the DFM.

Proportion of points that are correctly classified over the number of correct classifications expected by chance; P < 0.0003.

Independent classifications only include points that were not used to formulate the DFM.

1990, Call et al. 1992). Habitat studies rarely account for temporal vari- ation and the scale of investigation is usually limited to one or two spatial scales (e.g., Lemkule and Raphael 1993, Hunter et al. 1995). Habitat studies rarely include a full range of available habitats (i.e., include ex- tremes). Finally, habitat quality is rarely assessed using some measure of fitness (Van Horne 1983).

We investigated habitat selection in an entire population of Spotted Owls over eight years at several spatial scales. Our extensive survey ef- forts have allowed us to sample 95% of all territories consistently in the San Bernardino Mountains each year since 1989 (LaHaye et al. 1994) which includes all habitat types used by the owls. The extent of our sampling avoided pseudoreplication and allowed us to evaluate two mea- sures of fitness (reproductive success and productivity) as indicators of habitat quality.

Nest-site selection. — Spotted Owls selected large trees in which to nest which is consistent with observations throughout their range (LaHaye 1988, Bias and Gutierrez 1992, Gutierrez et al. 1992, Seamans and Gu- tierrez 1995). Owls in our study also differentially used platform struc- tures as nest sites. The use of platform nests, however, does not appear to be related to nesting success. Thus, selection for nest type may be related to availability of the different nest types; something which we were unable to estimate (see also LaHaye 1988).

Nest habitat characteristics. — Spotted Owls in our study showed a pat-

48

THE W^ILSON BULLETIN • Vo/. 709, No. 7, March 1997

tern of habitat selection similar to other populations where the owls se- lected habitats with a structure different than what was generally available to them (e.g., Solis and Gutierrez 1990, Bias and Gutierrez 1992, Gu- tierrez et al. 1992). Nest sample points were characterized by more com- plex vegetative structure (greater variation in tree sizes, larger trees, high- er canopy closure). Our DFM models demonstrated that the multivariate distribution of habitat characteristics for nest and random points were quite dissimilar. Some overlap, however, is evident between these distri- butions as we would expect since some random points were in fact suit- able owl habitat in terms of stand structure. Of course, it was not possible to determine which of these characters, if any, was the reason for habitat selection. Nevertheless, it appeared that the Spotted Owls were behaving as habitat specialists at the scale of nest habitat selection.

Habitat quality. — Spotted Owls were equally likely to fledge juveniles in all three habitats, but breeding owls located in the lower elevation oak/ big-cone fir habitat produced more fledglings per nest. This is consistent with earlier reports that showed a negative relationship between produc- tivity and elevation (Bart and Forsman 1992). Given the potential for improved fitness in the oak/big-cone fir habitat, we would predict that owls would select this habitat preferentially. This appears to be the case- Smith (1995) estimated the ecological densities for this same population to be 0.43, 0.20 and 0. 1 1 owls/km^ for oakd3ig-cone fir, conifer/hardwood, and mixed conifer, respectively. Higher densities may reflect smaller ter- ritory sizes which could result from increased prey densities associated with higher mast production at lower elevations. Thus, owls may have more energy to invest in reproduction in the lower elevation oakA^ig-cone r habitat. Ultimately, we will need data on survivorship and reproductive success of fledglings from each of these habitat types before we can assess their true contribution to the total population of California Spotted Owls m the San Bernardino Mountains.

There is a potential for increased disturbance of Spotted Owl habitat associated with the burgeoning human population in southern California (McKelvey and Weatherspoon 1992). In particular, as demand for housing and general suburban expansion continues in San Bernardino County the

rZ ';'i'oot°u fi-- habitat may be the first to be impacted.

Smith {I99.‘i) has shown a strong negative association between habitat fragmentation and occurrence of Spotted Owls. Thus, human disturbance IS likely to fragment these important habitats and negatively affect what

appears to be the most productive segment of the San Bernardino Spotted Owl population. ^

We infer from our study results that although Spotted Owls used a variety of habitat types, they .selected forests that were different from

UiHaye ef al. • SPOTTED OWL NEST-SITE SELECTION

49

available habitat. Further there appears to be differential fledgling pro- ductivity attributable to different habitats, but not to nest structure /?er se. Therefore, we conclude that Spotted Owls are structural habitat specialists inhabiting areas of differing qualities.

ACKNOWLEDGMENTS

We thank the many dedicated field biologists who helped gather nest information during the years: S. Bailey, M. Brosnan, P. Carlson, G. Caulkins, R. Enderlein, M. Engle, R. Gamez, J. Gronski, D. Holcomb, M. Hollister, L. Johnson, A. Kirn, D. Kwasney, T. Lagarda, C. Canning, K. Maas, K. Norgaard, E. Pausch, D. Roberts, J. Saia, J. Schmid, J. Silva, J. Stephenson, S. Stroich, R. Tanner, A. Tur, T. Van Blankenstein, and A. Vejar. Richard and Donna Smith were especially committed to the project. This study was funded in part by Snow Summit Ski Corporation, Bear Mountain, LTD., San Bernardino National Forest, Pacific Southwest Forest and Range Experiment Station, U.S. Forest Service Region 5, Southern California Edison, California Department of Fish and Game, Salad King, Inc., U.S. Army Corps of Engineers, San Bernardino Audubon Society, and the Sidney S. Byers Trust.

LITERATURE CITED

Bart, J. and E. D. Forsman. 1992. Dependence of northern spotted owls Strix occidentalis caurina on old-growth forests in the western USA. Biol. Conserv. 62:95-100. Batschelet, E. 1981. Circular statistics in biology. Academic Press, San Francisco, Cali- fornia.

Bias, M. A. and R. J. Gutierrez. 1992. Habitat associations of California spotted owls in the central Sierra Nevada. J. Wildl. Manage. 56:584—595.

Call, D. R., R. J. Gutierrez, and J. Verner. 1992. Foraging habitat and home-range characteristics of California Spotted Owls in the Sierra Nevada. Condor 94:880-888. Capen, D. E., j. W. Fenwick, D. B. Inkley, and A. C. Boynton. 1986. Multivariate models of songbird habitat in New England Forests. Pp. 171-176 in Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates (J. Verner and M. L. Morrison, eds.). Univ. Wisconsin Press, Madison, Wisconsin.

Carey, A. B., J. A. Reid, and S. P. Horton. 1990. Spotted Owl home-range and habitat use in southern Oregon coast ranges. J. Wildl. Manage. 54:1 1-17.

Dilworth, T. 1981. Log scaling and timber cruising. Oregon State Univ. Press, Corvallis, Oregon.

Franklin, A. B., D. R. Anderson, E. D. Forsman, K. P. Burnham, and F. F. Wagner. 1996. Methods for collecting and analyzing demographic data on the northern spotted owl. Pp. 12-26 in Demography of the Northern Spotted Owl. (Forsman, E. D., S. DeStefano, M. G. Raphael, and R. J. Gutierrez, eds). Studies in Avian Biology 17. Gutierrez, R. J., A. B. Franklin, and W. S. LaHaye. 1995. Spotted Owl {Strix occiden- talis). In The Birds of North America, No. 179 (A. Poole and F. Gill, eds). The Academy of Natural Sciences, Philadelphia, and The American Ornithologists’ Union, Washing- ton, D.C.

, J. Verner, K. S. McKelvey, B. R. Noon, G. S. Steger, D. R. Call. W. S. Lahaye,

B. B. Bingham, and J. S. Senser. 1992. Habitat relations of the California spotted owl. Pp. 79-147 in The California spotted owl: a technical assessment of its current status (J. Verner, K. S. McKelvey, B. R. Noon, R. J. Gutierrez, G. 1. Gould, Jr., and T. W. Beck, eds.). U.S.D.A. Pacific Southwest Forest and Range Experiment Station. Gen. Tech. Rept. PSW-GTR-133.

50

THE WILSON BULLETIN • Vol. 109, No. 1. March 1997

Hunter, J. E., R. J. Gutierrez, and A. B. Eranklin. 1995. Habitat configuration around Spotted Owl sites in northwestern California. Condor 97:684-693.

LaHaye, W. S. 1988. Nest site selection and nesting habitat of the northern spotted owl

(S. o. caurina) in northwestern California. M.S. thesis, Humboldt State Univ., Areata, California.

-, R. J. Gutierrez, and H. R. Akcakaya. 1994. Spotted Owl metapopulation dy- namics in southern California. J. An. Ecol. 63:775-785.

Lehmkuhl, j. E and M. G. Raphael. 1993. Habitat pattern around Northern Spotted Owl locations on the Olympic Peninsula, Washington. J. Wildl. Manage. 57:302-315.

Lemmon, R E. 1957. A new instrument for measuring forest overstory density J Eorestrv 55:667-668.

Major, J. and D. W. Taylor. 1977. Alpine. Pp. 601-675 in Terrestrial vegetation of Cal- ifornia (M. G. Barbour and J. Major, eds.). John Wiley and Sons, New York, New York McKelvey, K. S. and C. P. Weatherspoon. 1992. Projected trends in owl habitat Pp. 261- 273 m The California spotted owl: a technical assessment of its current status (J. Verner, K. S. McKelvey, B. R. Noon, R. J. Gutierrez, G. I. Gould, Jr., and T. W. Beck eds )

U.S.D.A. Pacific Southwest Forest and Range Experiment Station. Gen. Tech.’ Rept. PSW-GXR- 1 33.

Minnich, R. a. 1988. The biogeography of fire in the San Bernardino Mountains of Cali- forma. Geography Volume 28, Univ. California Press, Los Angeles, California Mueller-Dombois, D. and j. Ellenberg. 1974. Aims and methods of vegetation ecology John Wiley and Sons, Inc., New York, New York

H, Kutner. 1990. Applied linear siatistical models. Third edition. IRWIN, Homewood, Illinois

Norris. R^M^AND R. W. Webb. 1990. Geology of California. John Wiley and Sons, Inc New York, New York.

Pulliam, H. R. 1988. Sources, sink.s, and population regulation. Amer. Nat 132-652-661 AND B. J. Danielson. 1991. Sources, sinks, and habitat selection: a landscape perspective on population dynamics. Amer Nat 137 S50-S66

SAS iNSTiTtiTE INC 1989. SAS/STAT* User's guide. Version 6. Fourth edition. Volume I Carey, North Carolina.

Seamans, M, E. and R. J. Gutierrez. 1995. Breeding habitat of the Mexican Spotted Owl in the Tularo.sa Mountains, New Mexico. Condor 97'944_953

Summer habitat ecology of Northern Spotted Owls in northwestern California. Condor 92:739-748

m '■’''‘■'/PO';"'' of an insular Spotted Owl population in relation to

Area, Xf" California. M.S. thesis, Humboldt State Univ

Areata, California.

Titus K,. J. A. Mosher, and B. K. Williams. 1984. Chance-corrected classilication for use tn discrninnant analysis: ecological applications. Amer. Midi Nat 1071-7 U.S. Department OF Interior. 1990. Endangered and ihreatened wildlife and plants- De-

i.!te7 55 ^1 "PP“‘' O-P Tie Federal Reg-

199,3. Endangered and threatened wildlife and plants: final rule to list the Mexican spotted owl as a threatened species. Federal Register 58- 14248-14271

''*77"8T3t90i J. Wildl, Manage.

VASEK. F C. AND M. G. Barbour. 1977. Mojave Desert .scrub vegetation. Pages 835-867 in Terrestrial vegetation of California (M. G. Barbour and J Maior eds f Clif Native Plant Society Special Publication Number 9.

UiHave et al. • SPOTTED OWL NEST-SITE SELECTION

51

Verner, J., R. Gutierrez, and G. I. Gould, Jr. 1992. The California spotted owl; general biology and ecological relations. Pp. 55—77 in The California spotted owl: a technical assessment of its current status (J. Verner, K. S. McKelvey, B. R. Noon, R. J. Gutierrez, G. I. Gould, Jr., and T. W. Beck, eds.). U.S.D.A. Pacific Southwest Forest and Range Experiment Station. Gen. Tech. Rept. PSW-GTR-133.

Wolff, J. O. 1995. On the limitations of species-habitat association studies. Northwest Science 69:72—76.

Wilson Bull., 109(1), 1997, pp. 52-67

AGE-RELATED TIMING OF MIGRATION: GEOGRAPHIC AND INTERSPECIFIC PATTERNS

Mark S. Woodrey' and C. Ray Chandler^

Abstract.— Ditferential timing of passage by age or sex classes at sites along migration routes IS ambiguous with respect to whether these groups differ in onset or rate of migration and provides little insight into the dynamics of differential timing. Therefore, we compared age-specihc differences in timing of autumn migration by five species of passerines at three sites m eastern North America. Two species showed consistent differential timing of migra- lon at all sites, with adults preceding immatures in Red-eyed Vireos (Vireo olivaceus) and immatures preceding adults in Magnolia Warblers (Dendroica magnolia). Age differences in timing of passage by Swainson’s Thrushes {Catharus ustulatus), American Redstarts (Setophagarimcdla), and Common Yellowthroats (Geothlypis trichas) varied with year and ocation. The degree of differential timing varied substantially among sites within species For Swainson s Thrush and Magnolia Warbler, species which breed largely to the north of

Onl^The^R^ H ^ H v"*' ^''’^ence for differential rate of migration between age classes.

niy the Red-eyed Vireo and Magnolia Warbler showed evidence of differential onset of migration. Our results point to the need for further data on the dynamics of differential timing as a prerequisite for testing hypotheses on the causes of differential migration Re- ceived 1 Dec. 1995, accepted I Sept. 1996. ^

Differential migration among individuals of the same species may in- volve differences m migratory timing (review by Gauthreaux 1982) dis- tance migrated (reviews by Gauthreaux 1982, Ketterson and Nolan 1983) or ot (e.g., Chandler and Mulvihill 1990a, Nolan and Ketterson 1990) Because ornithologists have a long tradition of quantifying the timing of passage y migrants at sites along migration routes, patterns of differential timing are particularly well-known. Differential timing may occur be- tween sexes (e.g., Annan 1962, Johnson 1965, Bildstein et al. 1984, Fran- cis and Cooke 1986, Chandler and Mulvihill 1990a), between rec’ogniz-

UusscU et al. 1967; Hussell 1980, 991 Bildstein et al. 1984; Chandler and Mulvihill 1990a), or among

populations of different geographic origin (e.g., Mueller et al. 1981, Hog stedt and Persson 1982, Hedenstrom and Petersson 1984)

Mosi studies of differential timing in passerines have documented dif- ferences m the tim.ng of passage at a specific site. Unfortunately, timing of passage at a stngle s.te along the migration route is ambiguLs with respect to whether groups (ages, sexes, or populations) differ in onset of

WObrN *'•'= of origin (Chandler and Mulvihill

Ob). Nevertheless, data on the processes that underlie patterns of dif-

' Missi.s.sippi Mu.seum of Natural Science. 1 1 I North Jefferson Rt tn,w xa •

^ Dept, of Biology. Georgia Southern Univ.. State.sboro. Srorgif.WbO 3^202.

52

Woodrev and Chandler • DIFFERENTIAL TIMING OF MIGRATION

53

ferential timing are needed to identify the causes of differential timing — causes that have received remarkably little attention despite the recent revitalization of interest in avian migration.

The purpose of this paper is to explore patterns of age-specihc differ- ential timing of migration in several species of passerine birds and to discuss the processes that may account for these patterns. Specifically, we quantified geographic (three sites in eastern North America), interspecific (five species), and temporal (two years) patterns of age-specific differ- ential timing of fall migration. Our objectives were (1) to compare mi- gratory timing of adult and immature passerines among several species of passerine birds, (2) to determine whether these patterns are consistent across space and time, and (3) to assess whether these data can be used to provide information on the onset and rate of migration in species that exhibit differential timing of migration. We show that comparisons of migratory timing over a broad geographic area may provide insights into the mechanics of migratory timing and discuss the importance of this information to an understanding of the causes of differential migration.

STUDY AREAS AND METHODS

Study sites. — We assembled data on the timing of autumn migration from three sites in eastern North America. The most northern site was Long Point Bird Observatory (LPBO; 42°33'N, 80°10'W) on the north shore of Lake Erie, Ontario, Canada. The second site was Powdermill Nature Reserve, field station of the Carnegie Museum of Natural History, in western Pennsylvania (PNR; 40°10'N, 79°16'W). The third and southernmost site was Bon Secour National Wildlife Refuge, located on the western tip of Fort Morgan peninsula in coastal Alabama (FTMN; 30°10'N, 88°00'W). These locations were selected for two reasons. First, these sites (separated by 12° of latitude) lie along a migration route for birds moving out of northeastern North America to the Gulf of Mexico where they may take either a circum- or trans-gulf route into the Neotropics. The timing of passage by birds at these sites should be broadly representative of a variety of Neotropical migrants. Second, daily banding operations at each site provide detailed records of migratory timing. A description of band- ing operations at LPBO can be found in Hussell (1991) and at PNR in Leberman and Wood (1983). At FTMN, approximately 25 12-m mist nets were opened daily throughout the autumn migration in forest and scrub habitat (Woodrey 1995).

Data collection. — We selected five species for analysis: Swainson’s Thrush (Catharus ustulatus). Red-eyed Vireo (Vireo olivaceus). Magnolia Warbler (Dendroica magnolia), American Redstart (Setophaga ruticilla), and Common Yellowtbroat (Geothlypis trichas). We chose these five species because identifiable age classes were captured in relatively large numbers at all three banding sites.

Birds included in the analysis (first captures only) were captured at each site during daily banding operations in autumn 1990 and 1991. Capture dates for tbe five species were 1 Aug.-17 Oct. at LPBO, I Aug.-19 Oct. at PNR, and 31 Aug.-31 Oct. at FTMN. Birds of these five species banded over the.se dates were considered migrants; we assume that any captures of local breeders were insufficient to affect our conclusions concerning differential timing of migration.

For those species whose breeding ranges lie largely to the north of all three capture sites

54

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

(Swainson’s Thrush and Magnolia Warbler), we estimated average migration rates for adults and immatures based on the median capture dates of each age class at the three sites. This was done by dividing the distance between two sites by the number of days difference in median passage dates at the two sites (Ellegren 1990). These estimates are meant to be qualitative, since they are based on the assumption that the populations of these species sampled at LPBO and/or PNR were at least partially the same as those at LTMN. In short, we assume a simple NE-SW oriented migration route from LPBO and PNR to LTMN and into Central and South America. Although alternative migratory patterns are conceivable, our approach is to make an explicit simplifying assumption and generate testable general- izations about migration rates in the age classes of these two species. This method (com- parison of median capture dates from sites along the migration route) has produced estimates of migration speed that compare favorably with rates calculated from direct banding recov- eries (Ellegren 1990). The distance from LPBO to FTMN was calculated to be 1610 km; PNR to FTMN is 1380 km.

Stati.stics. — We used Mann-Whitney tests to compare differences in the capture dates (Julian dates) of adults and immatures at each of the three sites. The Mann-Whitney test is sensitive to differences in the location of the distributions of capture dates between age classes (Sokal and Rohlf 1995). All tests were two-tailed, and we used SYSTAT for all analyses (Wilkinson 1989).

We used box plots for the visual display of migratory timing. These plots convey sub- stantial information about the shape of a distribution by displaying the median with a vertical line, the interquartile range (IQR; a box around the median), and adjacent values (extent of a horizontal line). The upper adjacent value is the observation less than or equal to the upper quartile value plus 1.5 X IQR. The lower adjacent value is the observation greater than or equal to the lower quartile value minus 1 .5 X IQR. Observations that lie 1 .5 IQRs or 3.0 IQRs beyond the quartile values are outside values and are shown with asterisks and open circles, respectively.

Because we conducted multiple tests (five species at three sites), we sequentially Bon- ferroni-adjusted (Hochberg 1988) p-values to maintain an experimentwise error rate of 0.10 within years (Chandler 1995). In 1990, all p-values that were significant when unadjusted, remained so after Bonferroni adjustment. In 1991, only two unadjusted values became non- significant after adjustment for multiple tests. Therefore, we report unadjusted values and indicate ca.ses where adjustment for multiple tests would alter significance.

RESULTS

Swainson’s Thrush. — In 1990, Swainson’s Thrushes showed no signif- icant difference in the timing of passage of adults and immatures at LPBO {U = 8145, P = 0.46), PNR {U = 685, P = 0.09), or FTMN {U = 619, P = 0.14) (Fig. 1). The same was true for 1991, with the exception of LPBO, where median passage of adults was significantly earlier than im- matures {U = 13,755, P = 0.002; Fig. 1).

The general lack of significant age differences in passage of Swainson’s Thrushes among sites suggests that migration rates of adults and imma- tures were similar (or do not differ sufficiently to result in differential passage over the range of latitudes analyzed here). Nevertheless, the age difference in median dates of passage varied from immatures preceding adults by 4.5 days at FTMN in 1990 to no difference at LPBO in 1990

Woodrev and Chandler • DIFFERENTIAL TIMING OF MIGRATION

55

s

O

CL

o

Z

o

HY

AHY

1990

H 224

69

j I 1 L.

1991

a.

j

d

CC

LU

O

a.

HY

AHY

^108

H 1 7

H 1 1 3

H 22

51

1 4

i 1

200 220 240

—1 1_

260 280

I

300

I 1 I I I

200 220 240 260 280 300

JULIAN DATE JULIAN DATE

Fig. 1. Distribution of capture dates (first captures only) for Swainson's Thrushes at the three sites. Box plots display median (vertical line within box), interquartile range (extent of box), adjacent values (extent of horizontal line), and outside values (asterisks and open circles). Julian dates range from 18 July (200) to 26 October (300). Sample sizes are given to the right of each plot.

56

THE WILSON BULLETIN • VoL 109, No. 1. March 1997

to adults preceding immatures by 7.5 days at FTMN in 1991 (Fig. 1). Passage dates at LPBO (averaged across years) suggested an average migration rate to the Gulf coast of 1 19 km/day for adults and 108 km/day for immatures. At PNR (again averaged across years), the estimates of migration rates were 402 km/day for adults and 317 km/day for imma- tures. The large difference in estimated migration rates from LPBO and PNR was attributable to a large difference in passage dates of Swainson’s Thrushes at these two sites (7-13 days later at PNR even though these sites are separated by only 235 km).

Red-eyed Vireo. — Adult vireos migrated significantly earlier than im- matures at all three sites in 1990. Median passage dates of the age classes differed by 21 days at LPBO {U = 9\7, P < 0.001), 22 days at PNR (U - 8766, P < 0.001), and 17 days at FTMN (U = 1065, P = 0.004) (Fig. 2). The age differences in timing of migration were so large in 1990 that the median passage of adults at FTMN preceded median passage of im- matures at LPBO. The pattern was similar in 1991 but less pronounced. Adults preceded immatures by four days at LPBO {U = 1456, P = 0.04), 13.5 days at PNR (U = 2256, P = 0.001), and two days at FTMN {U = 1205, P = 0.06) (Fig. 2). After adjustment for multiple tests, only the PNR difference was significant.

Estimates of migration rates were complicated by the extensive breed- ing range of this species. Like Swainson’s Thrushes, however, there was local variation in passage dates. Even though LPBO is Just to the north of PNR, Red-eyed Vireos passed PNR 8-9 days earlier in 1990 and 1 1-20 days earlier in 1991 (Fig. 2).

Magnolia Warbler. — In 1990, adult and immature Magnolia Warblers differed significantly in timing of passage at all three sites. Immatures preceded adults by an average of 16 days at LPBO {U = 2204, P < 0.001), 10 days at PNR {U = 1263, P < 0.001), and 19 days at FTMN {U = 359, P < 0.001) (Fig. 3). The same was true in 1991, with im- matures preceding adults by five days at LPBO {U = 5259, P < 0.001), 12 days at PNR {U = 1292, P < 0.001), and eight days at FTMN (U = 186, P < 0.001) (Fig. 3).

The median passage dates at LPBO and FTMN (averaged across years) suggested average migration rates of 79 km/day for adult Magnolia War- blers and 92 km/day for immatures. Median passage dates at PNR and FTMN gave estimates of 101 km/day for adults and 103 km/day for immatures.

American Redstarts. — In 1990, there were no significant differences in the timing of passage of adults and immatures at LPBO {U = 1261, P = 0.07), PNR (t/ = 1 1 15, P = 0.40), or FTMN {U = 1287, P = 0.91) (Fig. 4). The

Woodrex and Chandler • DIFFERENTIAL TIMING OE MIGRATION

57

d HY

2

cr

LU

O

0.

AHY -

H 137

H 83

I I I I

O

u.

HY

AHY

75

— I

20

_i I I I I

69

^ 28

_j I I I I

200 220 240 260 280 300 200 220 240 260 280 300

JULIAN DATE JULIAN DATE

Fig. 2. Distribution of capture dates (first captures only) for Red-eyed Vireos at the three sites. See Figure 1 for description of box plots.

same was true in 1991, with the exception of FTMN where immature red- starts preceded adults by 10 days {U — 2284, P < 0.001) (Fig. 4).

Although the extensive breeding range of this species complicated in- ferences about migration rates, there was substantial among-site variation in passage dates. In 1990, the median passage date of adults was 13 days

58

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

I

O

0.

CD

z

O

2

tr

1X1

Q

O

D.

o

LX

AHY

^59

36

I -H 28

23

I I I I I

200 220 240 260 280 300

1 I L_

200 220 240 260

I I

280 300

JULIAN DATE

Fig. 3. Distribution of capture dates (first captures only) for Magnolia Warblers at the three sites. See Figure 1 for description of box plots.

earlier at LPBO than at PNR just to the south; the median passage of immatures was only six days earlier at LPBO that same year. In 1991, the median date of passage for the age classes differed by no more than 3.5 days between these two sites (Fig. 4).

Common Yellowthroat. — Adult and immature yellowthroats did not dif-

FORT MORGAN POWDERMILL LONG POIf^

Woodrey and Chandler • DIFFERENTIAL TIMING OF MIGRATION 59 1990 1991

HY

AHY

HY

AHY

HY

AHY -

117

1 7

_j I I I 1

H 83

30

_i I I I 1

• 87

H 30

- 1-

230

_i I I I 1

_i I I I I

140

^ 66

I I I I I I I I I I

200 220 240 260 280 300 200 220 240 260 280 300

JULIAN DATE

Fig. 4. Distribution of capture dates (first captures only) for American Redstarts at the three sites. See Figure 1 for description of box plots.

fer significantly in timing of passage at any site in 1990 (LPBO; U = 1637, P = 0.14; PNR: U = 7421, P = 0.53; FTMN: U = 1714, P = 0.95) (Fig. 5).

In 1991, median passage of immatures preceded that of adults by 14

60

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

O

CL

o

HY

AHY

1990

j I I I I

HY - “h

CC

LU

O

CL

H - 195

AHY - h

H 59

_l I I I I

p

HY

AHY

170

— I

°20

“ 19

I I I I I

200 220 240 260 280 300

1 1 I I I

200 220 240 260 280 300

JULIAN DATE

Eig. 5. Distribution of capture dates (first captures only) for Common Yellowthroats at the three sites. See Eigure I for de.scription of box plots.

days at PNR {IJ — 3909, P < 0.001), but there was no significant dif- ferential timing at the other sites (LPBO: V = 1472, P — 0.07; FTMN: V = 1298, P = 0.07) (Fig. 5). In 1991, adult yellowthroats passed LPBO 20 days earlier than at PNR just to the south (immatures had almost identical median passage dates at these sites; Fig. 5).

Woodrev and Chandler • DIFFERENTIAL TIMING OF MIGRATION

61

DISCUSSION

When analyzed on a large geographic scale, five species of passerine migrants showed different patterns of age-specific differential timing of migration. Adult Red-eyed Vireos preceded immatures at all sites, where- as the opposite was true for Magnolia Warblers. Age-specific differences in timing of passage by Swainson’s Thrushes, American Redstarts, and Common Yellowthroats varied with site and year. Our results demonstrate that the magnitude of age-specific differential timing can vary substan- tially between years (e.g.. Red-eyed Vireos) and that migratory timing can vary appreciably from site to site within a region (PNR vs LPBO). For those two species for which it was possible to estimate average mi- gration rates, the rates were similar between adults and immatures (Mag- nolia Warbler) to perhaps faster in adults (Swainson’s Thrushes based on PNR data).

The specific patterns of differential timing that we report from each site are not new (e.g., Leberman and Clench 1969, 1970). In fact, many studies document adults preceding immatures during fall migration (e.g., Hussell et al. 1967; Seel 1977; Hussell 1980, 1991; Chandler and Mul- vihill 1990a) or immatures preceding adults (e.g., Mueller and Berger 1968, Nilsson 1970, Bildstein et al. 1984). However, simply documenting patterns of differential passage at a site along the migration route offers little insight into the dynamics of differential migratory timing within species of passerine birds (Chandler and Mulvihill 1990b). Comparative analyses of data from several sites along the migration route, such as that attempted here, offer the opportunity to explore the contributions of rate and onset to differential timing in passerines (Rosenfeld and Fagerstrom 1980, Hilden and Saurola 1982, Ellegren 1990).

Onset versus rate of migration. — Our data provide evidence for differ- ential onset of migration in two of the five species that we examined. Magnolia Warblers showed large age differences in migratory timing at all sites. Because the bulk of their breeding range lies just to the north of LPBO and PNR, differences in rate of migration by age classes of Magnolia Warblers appear insufficient to account for the large differences in their timing of passage at LPBO and PNR. Apparently, immature Mag- nolia Warblers initiate migration approximately 1-2 weeks earlier than adults, depending on the year. Red-eyed Vireos also show large age dif- ferences in timing at all three sites although the magnitude of the differ- ence varied annually. It is difficult to explain this age difference in timing of passage at all sites as a result of age-specific differences in rate of migration. Instead, adult vireos appear to depart the breeding grounds substantially earlier than immatures (possibly as much as three weeks).

62

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

This early departure may be related to the fact that adult Red-eyed Vireos show a very limited prebasic molt during late summer (Mulvihill 1993) that might permit earlier onset of migration. A substantial late-summer movement of second-year birds and/or failed breeders (R.S. Mulvihill, pers. comm.) also may be a contributing factor. Whatever the cause, age differences in timing of migration by Red-eyed Vireos appear susceptible to substantial annual variation (Fig. 2).

We also have some evidence on rates of travel by age classes in two of the five species. Assuming that one or both of our northern sites sample the same populations as our Gulf site (or at least populations with the same migratory timing), our results suggest that adult and immature Mag- nolia Warblers migrate at roughly similar rates (approximately 80-100 km/day). Swainson’s Thrushes appear to migrate more rapidly than the warblers, and adults may migrate at a slightly faster rate than immatures (by about 80 km/day based on PNR data). If these age differences are real, they are insufficient to result in significant differential passage by the time Swainson’s Thrushes reach the Gulf coast. We emphasize that these estimated rates are based on two seasons of data and are meant to be qualitative, testable generalizations about migratory dynamics within species. Ellegren (1990), in a similar study of Bluethroats {Luscinia sve- cica), found that immatures had slower migration rates than adults early in autumn migration (100 km/day vs 40 km/day) but not as they neared the wintering grounds (immatures approached 100 km/day). Unfortunate- ly, Ellegren’s study is rare in reporting age-specific estimates of migration rates in passerines. Our estimated migration rates of 80-130 km/day are generally similar to those reported for other passerines (Hyytia and Vik- berg 1973, Hilden and Saurola 1982, Ellegren 1990). This suggests that our comparisons of timing among sites produced reasonable estimates of migration rates for Swainson’s Thrushes and Magnolia Warblers.

The only exceptions to this were our estimates of migration rates by adult and immature Swainson’s Thrushes from PNR to the Gulf (402 km/day and 317 km/day, respectively). These large estimates are attrib- utable to Swainson’s Thrushes passing PNR almost two weeks later than LPBO just to the north (235 km). The passage dates at LPBO give mi- gration rates for Swainson’s Thrushes most similar to those reported in other passerines (Hilden and Saurola 1982), but little is known about movements between these two sites and the Gulf coast. These banding sites may sample different populations of thrushes, or individuals passing PNR could make rapid and direct flights to the Gulf while an ecological barrier at LPBO (Lake Erie) may slow movements or direct birds farther west. That there are interesting migratory dynamics at these sites is further suggested by Red-eyed Vireos moving through PNR 8-9 days earlier than

Woodrex and Chandler • DIFFERENTIAL TIMING OF MIGRATION

63

at LPBO (opposite of the pattern in Swainson’s Thrushes) and the fact that age classes of American Redstarts and Common Yellowthroats vary substantially in their timing between the two sites.

Causes of differential timing. — A variety of causes might ultimately result in age differences in onset or rate of autumn migration (Table 1). First, differential timing might result from differential constraints oper- ating on age classes. In other words, there is a single optimal migration schedule for all individuals of a population, but constraints operate dif- ferentially on adults and immatures. Second, there may be different op- timal migration schedules for the age classes even in the absence of any differential constraints on these groups.

Constraints that might influence migration include age-related differ- ences in timing of molt, which could impose different migration schedules (Rimmer 1988, Chandler and Mulvihill 1990a, Ellegren 1990). Molt that is completed on the breeding grounds should primarily influence onset of migration, whereas molt that extends into migration (or occurs en route; Rohwer and Manning 1990, Thompson 1991) could affect rate of migra- tion. Age differences in molt would have to be pronounced to account for the large differences in onset of migration in Red-eyed Vireos or Magnolia Warblers. Nevertheless, adult Red-eyed Vireos have such a lim- ited prebasic molt during fall (Mulvihill 1993) that it might not preclude a very early onset of migration (Canned et al. 1983), and molt can delay the departure of adults from the breeding grounds by several days in some species of warblers (Yellow Warbler [Dendroica petechia]', Rimmer 1988). Molt also appears to hamper fat accumulation by Swainson’s Thrushes at stopover sites (Winker et al. 1992). Overall, the possible costs of molt to fat accumulation rates or flight dynamics need to be quantified further before the influence of molt on differential migration can be tested adequately.

Lack of experience may also place important constraints on migratory timing in immature birds. Inefficient foraging, unfamiliarity with stopover sites, or less efficient navigation could all slow onset or rate of migration. However, lack of experience as a general explanation for slower migration in immatures is not well supported (but see Ellegren 1990). Lack of ex- perience does not appear to preclude early departure from the breeding grounds by immature Magnolia Warblers and subsequent rates of migra- tion comparable to adults. Eurthermore, in redstarts and yellowthroats, immatures appear to have migratory timing equal to that of adults. Al- though inexperience is a fundamental feature of birds undertaking their first migration, specific tests of its effects on migratory timing are needed.

Social interactions might also affect migratory timing (e.g., Gauthreaux 1978, Nolan and Ketterson 1990). As was the case for experience, social

64

THE WILSON BULLETIN • Vol. 109, No. 1, March 1997

dominance needs further testing as a general explanation for age-specific differential timing. Virtually any observed pattern of age differences in timing could be consistent with the constraints imposed by social domi- nance. Young birds might leave the breeding grounds early to avoid dom- inant adults or late because reduced access to food in the presence of dominant adults slows fat accumulation. Along migration routes, imma- tures might move through stopover sites faster or slower for the same reasons. This is not to suggest that social effects on migratory timing are not important. Rather, these effects need to be tested in specific cases and not simply invoked as being consistent with observed patterns. We know of no data on the possible contribution of social interactions to the dif- ferential timing reported here.

Onset of migration in adult birds also might be slowed by the energetic demands of late summer breeding activity. This seems most likely in species raising multiple broods into late summer (immatures from early broods could initiate migration while adults are still occupied with breed- ing activity). Only the Magnolia Warbler showed differential onset of migration consistent with this constraint and this trait is not apparent in this species. We caution that early migration by failed breeders (as is possible in Red-eyed Vireos) might obscure any clear relationship be- tween adult breeding activity and delayed migration. Finally, the well- documented differences in flight morphology between age classes (e.g., Alatalo et al. 1984, Mulvihill and Chandler 1990) might affect flight speeds, flight ranges, or fattening strategies and thus result in differential rates of migration.

In the absence of differential constraints between ages, there might still be differential timing of migration. Even with equal migratory options open to them, selection might still favor age classes departing the breeding grounds, arriving at stopover sites, or arriving on the wintering grounds at different times (Table 1). Although sexual selection for early arrival on the breeding grounds by males is a clear example of such differential selection, the selection pressures for different optimal migration times between ages during the autumn are less obvious. Migration schedules that are tied to late summer exploration in immatures (Ellegren 1990), to age-specific variation in latitudinal distribution (e.g., Ketterson and Nolan 1983), or to habitat segregation on the wintering grounds (e.g.. Lynch et al. 1985) are possible sources of differential selection pressure on rates and onset of migration.

Overall, our results show that broad geographic comparisons of migra- tory timing can provide insight into the dynamics of differential migra- tion. Quantifying onset and rate of migration permits more detailed anal- yses of migration strategies and naiTows the possible causes of differential

Woodrey and Chandler • DIFFERENTIAL TIMING OF MIGRATION

65

Table 1 Potential Ultimate Causes for Differential Onset and Rate of Migration Between Age Classes During Autumn Migration
Onset of migration	Rate of migration
Differential constraints on timing	imposed by:
Molt on breeding grounds	Molt in transit
Experience	Experience
Social interactions	Social interactions
Breeding activity	Morphology
Differential selection for:
Early arrival	Early arrival
(destination, stopover)	(destination, stopover)
Early departure	Early departure
(site of origin)	(stopover)

timing (Table 1 ). However, testing of specific causes of differential timing between age classes is hampered by a lack of information on age-specific timing of molts, effects of experience on migratory performance, and the role of social dominance in migratory movements. We hope this study and those such as Ellegren (1990) will stimulate increased attention to the dynamics and causes underlying differential timing of migration.

ACKNOWLEDGMENTS

We thank Bryan Lee, Frank Moore, and The Shed for valuable assistance during this research. Jennifer Carr and Valerie Beamon assisted with data entry and manuscript prep- aration. This analysis would not have been possible without the cooperation of Jon Mc- Cracken of Long Point Bird Observatory, and Robert Leberman, Robert Mulvihill, and D. Scott Wood of The Carnegie Museum of Natural History. We also appreciate the cooperation of Jerome Carroll of Bon Secour National Wildlife Refuge and Blanton Blakenship of the Alabama Historical Commission. David Aborn, Jill Busby, Michelle Cawthorn, David Cim- prich, Jeff Clark, Sarah Mabey, Frank Moore, Robert Mulvihill, and Wang Yong provided helpful comments on earlier drafts of this paper. This work was supported by NSF grant BSR-9020530 and BSR-9 100054, Paul A Stewart Awards, the Frank M. Chapman Memorial Fund, a USFWS Southeast Nongame Wildlife Grant, and the University of Southern Mi.s- sissippi.

LITERATURE CITED

Alatalo, R. V., L. Gustafsson, and A. Lundberg. 1984. Why do young passerine birds have shorter wings than older birds? Ibis 126:410—415.

Annan, O. 1962. Sequence of migration, by sex, age, and species, of thrushes of the genus Hylocichla, through Chicago. Bird-Banding 33:130-137.

Bildstein, K. L., W. S. Clark, D. L. Evens, M. Files, L. Soucy, and E. Henckel. 1984.

66

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

Sex and age differences in fall migration of Northern Harriers. J. Field Ornithol. 55: 143-150.

Cannell, P. E, J. D. Cherry, and K. C. Parkes. 1983. Variation and migration overlap in flight feather molt of the Rose-breasted Grosbeak. Wilson Bull. 95:621-627.

Chandler, C. R. 1995. Practical considerations in the use of simultaneous inference for multiple tests. Anim. Behav. 49:524—527.

AND R. S. Mulvihill. 1990a. Wing-shape variation and differential timing of mi- gration in Dark-eyed Juncos. Condor 92:54-61.

AND . 1990b. Interpreting differential timing of capture of sex classes during

spring migration. J. Field Ornithol. 61:85-89.

Ellegren, H. 1990. Autumn migration speed in Scandinavian Bluethroats Luscinia s. sve- cica. Ringing & Migration 11:121-131.

Francis, C. M. and F. Cooke. 1986. Differential timing of spring migration in wood war- blers (Parulinae). Auk 103:548-556.

Gauthreaux, S. a., Jr. 1978. The ecological signihcance of behavioral dominance. Pp. 17-54 in Perspectives in ethology. Vol 3 (PPG. Bateson and PH. Klopfer, eds.). Plenum Press, New York, New York.

. 1982. The ecology and evolution of avian migration systems. Pp. 493-516 in

Avian biology. Vol. 6 (D.S. Farner and J.R. King, eds.). Academic Press, New York.

Hedenstrom, a. and j. Petersson. 1984. The migration of Willow Warblers, Phxlloscopus trochilus, at Ottenby. Var Fagelvafld 43:217-228.

Hidden, O. and P. Saurola. 1982. Speed of autumn migration of birds ringed in Finland. Ornis Fenn. 59:140-143.

Hochberg, Y. 1988. A sharper Bonferroni procedure for multiple tests of significance. Biometrika 75:800-802.

Hogstedt, G. and C. Persson. 1982. Do Willow Warblers Phylloscopus trochilus of north- ern origin start their autumn migration at an earlier age than their southern conspecifics? Holarctic Ecol. 5:76-80.

Hussell, D. j. T. 1980. The timing of fall migration and molt in Least Flycatchers. Bird- Banding 5 1 .65—1 1 .

. 1991. Fall migrations of Alder and Willow flycatchers in southern Ontario. J. Field

Ornithol. 62:260-270.

, T. Davis, and R. D. Montgomerie. 1967. Differential fall migration of adult and

immature Least Flycatchers. Bird-Banding 38:61—66.

Hyytia, a., and P. Vikberg. 1973. Autumn migration and moult of the Spotted Flycatcher Muscicapa striata and the Pied Flycatcher Ficedula hypoleuca at the Signilskar bird station. Ornis Fenn. 50:134—143.

Johnson, N. K. 1965. Differential timing and routes of the spring migration in the Ham- mond’s Flycatcher. Condor 67:423-437.

Ketterson, E. D. and V. Nolan, Jr. 1983. The evolution of differential bird migration. Pp. 357-402 in Current ornithology. Vol. 3 (R.E Johnston, ed.). Plenum Press, New York, New York.

Leberman, R. C. and M. H. Clench. 1969. Bird-banding at Powdermill, 1968. Research Report No. 23, Carnegie Museum of Natural History, Pittsburgh, Pennsylvania.

AND . 1970. Bird-banding at Powdermill, 1969. Research Report No. 26,

Carnegie Museum of Natural History, Pittsburgh, Pennsylvania.

AND D. S. Wood. 1983. Bird-banding at Powdermill: twenty years reviewed. Re- search Report No. 42. Carnegie Museum of Natural History, Pittsburgh, Pennsylvania.

Lynch, J. F, E. S. Morton, and M. E. Van Der Voort. 1985. Habitat .segregation between the sexes of wintering Hooded Warhlers (Wilsonia citrina). Auk 102:714—721.

Woodrex and Chandler • DIFFERENTIAL TIMING OF MIGRATION

67

Mueller, H. C. and D. D. Berger. 1968. Sex ratios and measurements of migrant Gos- hawks. Auk 85:431—436.

, R. E. Berger, and G. Allez. 1981. Age, sex, and seasonal differences in size of

Cooper’s Hawks. J. Field Ornithol. 52:112-126.

Mulvihill, R. S. 1993. Using wing molt to age passerines. N. Amer. Bird Bander 18:1—10.

AND C. R. Chandler. 1990. The relationship between wing shape and differential

migration in the Dark-eyed Junco. Auk 107:490—499.

Nilsson, L. 1970. Local and seasonal variation in sex-ratios of diving ducks in south Sweden during the non-breeding season. Ornis Scand. 1:115—128.

Nolan, V., Jr. and E. D. Ketterson. 1990. Timing of autumn migration and its relation to winter distribution in Dark-eyed Juncos. Ecology 71:1267—1278.

Rimmer, C. C. 1988. Timing of the definitive prebasic molt in Yellow Warblers at James Bay, Ontario. Condor 90:141 — 156.

Rohwer, S. and J. Manning. 1990. Differences in timing and number of molts for Balti- more and Bullock’s orioles: implications to hybrid fitness and theories of delayed plum- age maturation. Condor 92:125—140.

Rosenfeldt, j. and T. Eagerstrom. 1980. Spring migration at Orskar, Central Sweden, in thirteen night-migrating passerines. Var Eagelvarld 39:217-224.

Seel, D. C. 1977. Migration of the northwestern European population of the Cuckoo Cu- culus canorus, as shown by ringing. Ibis 1 19:309-322.

SOKAL, R. R. AND E J. Rohlf. 1995. Biometry, 3rd ed. W.H. Ereeman and Company, New York, New York.

Thompson, C. W. 1991. The sequence of molts and plumages in Painted Buntings and implications for theories of delayed plumage maturation. Condor 93:209-235.

Wilkinson, L. 1989. SYSTAT: The system for statistics. Systat, Inc., Evanston, Illinois.

Winker, K., D. W. Warner, and A. R. Weisbrod. 1992. The Northern Waterthrush and Swainson’s Thrush as transients at a temperate inland stopover site. Pp. 384-402 in Ecology and conservation of Neotropical landbirds (J. M. Hagan and D. W. Johnson, eds.). Smithsonian Institute Press, Washington, D.C.

WooDREY, M. S. 1995. Stopover behavior and age-specific ecology of Neotropical passerine migrant landbirds during autumn along the northern coast of the Gulf of Mexico. Ph.D. diss. Univ. of Southern Mississippi, Hattiesburg, Mississippi.

Wilson Bull., 109(1), 1997, pp. 68-73

THE EFFECT OF AGE ON NEST CONCEALMENT AND ITS COMPLIMENTARY EFFECT ON PRODUCTION OF WOOD THRUSH

Mark S. Johnson'

Abstract. — Declines in Neotropical migratory land birds have been associated, in part, with decreases in productivity attributed to predation. Since those predators that affect off- spring production may be visually dominant (i.e., avian), vegetative nest concealment was quantified for Wood Thrushes (Hylocichla mustelina) for fifty nests in a Delaware forest fragment in 1992 and 1993. Nest concealment was tested against age of owners (i.e., ASY vs SY) and with fledgling production. Though past studies have found an age-specific dif- ference in per nest fledgling production, no age-specific differences in concealment were found. However, there was a positive relationship between nest concealment and fledgling output, accounting for 38% of the variation for 1992 and 19% for both years. Results from 1993 were not descriptive, since predation events were fewer. These results are consistent with the logic that the variability of predator communities in different habitats offers no adaptive strategies in nest concealment per se. However, birds may be choosing other, more holistic habitat qualities within territories which may offer greater nest concealment as a benefit. Received 16 March 1996, accepted 20 Sept. 1996.

Nest predation may be a major factor influencing avian reproductive success (Ricklefs 1969). Choosing the optimal nest site may be critical in habitats where predation rates are high. Additionally, parental age has been positively associated with increased reproductive performance of Wood Thrushes (Johnson 1994, Johnson and Roth, unpub. data) as well as other species (Saether 1990). Whether learned or innate, the ability to reduce predation through the concealment of nests should serve to in- crease fitness and be favored through natural selection, particularly in habitats where primary predators are choosing prey visually (e.g., corvids, icterids, etc.).

Some nesting criteria have been positively associated with the age of nest owners, including earlier nest initiation, greater yearly nesting at- tempts, and larger mean clutches (Saether 1990, Johnson 1994). However, the effect of parental age on nest site concealment has not been measured. The present study examines the effect of age on nest concealment in a migratory Neotropical, open-cup nesting species (Wood Thrush: Hylo- cichla mustelina). Past analyses have found fledgling production of Wood Thrushes to be positively associated with age (Johnson 1994, Johnson and Roth, unpub. data).

' Dept, of Entomology and Applied Ecology, Univ. of Delaware. Newark. Delaware 19717.

68

Johnson • AGE EFFECTS ON NEST CONCEALMENT

69

STUDY AREA AND METHODS

The study site was a 15-ha woodlot (UDW) on the Univ. of Delaware farm in Newark, Delaware. Surrounding UDW on two sides are large fields used for agriculture; one side is adjacent to athletic fields and the other is a four lane highway. It is a relatively mature, mesic hardwood forest fragment with a mix of Coastal Plain and Piedmont elements char- acteristic of the Middle Atlantic Region (for more detailed descriptions, see Longcore and Jones (1969), Roth and Johnson (1993), and Gorman and Roth (1989)).

The intensive banding, mapping, and nest-monitoring protocol followed annually since 1974 (Roth and Johnson 1993) was used to provide data on location, success, and ownership of nests for the period 1992 and 1993. Ages were determined from previous banding results for returning adults and from rectrix shape for immigrants (Weinberg and Roth 1994). Nests were checked daily and timing when offspring fledged was recorded.

Potential predators of Wood Thrush eggs and young at UDW include Blue Jays (Cya- nocitta cristata), American Crows {Corvus branchyrhynchos). Common Crackles {Quiscalus quiscula), gray squirrels (Sciurus carotinensis), white-footed mice (Peromyscus leucopus), and raccoons (Procyon lotor). Garter snakes {Thamnophis sirtalis), and black rat snakes (Elaphe obsoleta) are uncommon. No chipmunks (Tamias slriatus) or flying squirrels (Glau- comys volans) were found. Nests which failed were investigated for indicators of predator identity (e.g., tracks, teeth marks on bands, nature of nest destruction, etc.).

Roth and Johnson (1993) attribute most of the 32% failure rate for Wood Thrush nests at UDW from 1978-1987 to predation. During the nesting period of 1992-1993, 65% of nests experienced depredation events, although only 35% were total losses (Roth, unpub. data).

Based on daily observations during the extensive monitoring of UDW, the predominant suspected predators were Blue Jays, Common Crackles, and raccoons. In order to address avian predation effects, nests which failed due to mammalian predation (torn apart), observer error, poor construction, and/or adverse weather were eliminated in the regression analyses so as to only include nests which had the potential to be affected by avian predators.

To calculate nest concealment we quantified cover around each nest, using a “density board” of clear plexiglass bearing a 3 X 4 grid of twelve 4X4 cm squares (Fig. 1), immediately following nest abandonment. With the grid centered on the nest and 0.5 m from it, the observer, 3 m away and at eye level with the nest, estimated the proportion of each square that was obscured by vegetation (e.g., 1.0 = totally concealed, 0.3 = Vj con- cealed, etc.). Only vegetation between the density board and the plane passing through the approximate center of the nest, parallel to the board, was considered. Measurements were recorded from the four major compass headings and from directly above the nest. The latter was accomplished using the same grid configuration mounted on an adjustable mirror. Again, as we estimated cover, the grid was 0.5 m from the nest and our eye at nest level. Con- cealment of each nest was the sum of all measurements at that nest. Since the intent of this study was to address concealment from avian predators, concealment from below was not considered. Since females alone build the nest (Brackbill 1958), concealment values for SY and ASY females were tested initially with a t-test. However, to address the possibility that males may be involved in nest site selection and since territories include vegetative char- acteristics which may be involved in territory selection, male age was also tested. To test if concealment affected nest success, I regressed fledgling production on concealment.

My technique logistically restricted my sampling to nests < 4 m above ground. Owners of nests < 4 m and > 4 m in 1992 and 1993 did not differ in age (females, = 1.286, df = 1, N = 1 13, f > 0.25, males x^ = 1-745, df = I, N = \ \5, P > 0.18).

70

THE WILSON BULLETIN • Vol. 109, No. I, March 1997

Eig. I . Diagram of the relative position of nest, observer, and cover density board for estimating vegetative nest concealment.

RESULTS

Vegetational concealment of nests did not differ between SY and ASY females (jc ASY = 28.5 ± 1.6; .x SY = 27.25 ± 1.67; P > 0.95). Ob- servation of pre-nesting activities (Brackbill 1958) and of multiple use of same nesting sites for territorial males coupled with different females (M. Johnson, pers. obs.) have provided circumstantial evidence that males may influence female nest site selection. As with females, no age class differences were found (x ASY = 25.60 ± 1.30; x SY = 27.13 ± 1.27; P < 0.16). These comparisions included all nests regardless of outcome.

Though there were no relationships between age of parent and nest concealment, a beneht of greater fledgling productivity may be realized from nests with greater concealment. Linear regression of fledgling pro- duction on total concealment explained 19% of the variance (r- = 0.19, df = 49, F = 11 .066, P < 0.01 ) of those nests which could be potentially affected by avian predators (see Methods). Nests constructed in 1992, however, yielded a stronger relationship between concealment and pro- duction where nest concealment explained 38% of the variance (F = 0.38, dt = 27, F = 15.61, P < 0.0 i ). No such relationship was found for nests in 1993 (F = 0.08, df = 21, F = 1.653, P > 0.21), when overall failure rates were lower (Johnson 1994).

DISCUSSION

If nest site selection is closely tied to fltness, given its effect on fledg- ling production, then the ability to select well-hidden sites should be

Johnson • AGE EFFECTS ON NEST CONCEALMENT

71

common where visual predators predominate. Decreases in residual re- productive value have been hypothesized as a mechanism for innate age- specific dominance in reproductive effort for older birds. Moreover, if there are age-specific differences in territory acquisition (e.g., SheiTy and Holmes 1989, 1990), then the optimal nest sites that are within these selected territories should be used predominately by older birds. Given the predator composition, structural heterogeneity of the understory, and previous failure rates of Wood Thrush nests at UDW (Roth and Johnson 1993, Johnson 1994), it is reasonable to assume that older birds would conceal their nests more than yearlings. I found no evidence that this was true.

The effect of experience has been observed to enhance foraging success (Desrochers 1992) and reproduction in birds (Pyle et al. 1991). Likewise, an age-specific difference in nest concealment seems reasonable if birds learn from previous avian predation events. Qualitative observations of individuals suffering suspected avian predation losses did not consistently result in birds subsequently choosing more greatly concealed sites nor did the data reflect this logic.

Given the evidence (i.e., predator community structure, vegetational heterogeneity, Wood Thrush response to Blue Jays and Common Grackles around nest site (Johnson 1994), witnessed events, and post predation evidence), avian predation may be a strong regulating force on fledgling production of Wood Thrushes at UDW. Westmoreland and Best (1985) attribute 79% of the nest failures to avian predation, and Yahner and Delong (1992) attribute 90% of unsuccessful nests in a Pennsylvania woodlot to avian predators. Additionally, Hoover et al. (1995) attribute most of the Wood Thrush losses to the same suspected avian predators in several Pennsylvania forests. Therefore, there should be a clear overall advantage to concealing nests from visually oriented predators. In 1992, the data reflected this advantage. In 1993, there was no association. How- ever, during this period, failure rates (and therefore predation events) were lower (Johnson 1994). In this light, any clear advantage of nest conceal- ment would not have been apparent.

Nest concealment may be less important where predator communities are different or where vegetational structure selectively permits predator access. The plant species most frequently used by Wood Thrush in UDW is Viburnum dendatum (Roth, unpub. data), which cannot easily support the weight of larger mammalian predators present at UDW. In habitats where Viburnum or structurally similar understory is not present, mam- malian and reptilian predation may be an important factor. These pressures may be variable in time as well as space and be such to offset or modify the selective advantage of concealment. Further, other factors such as

72

THE WILSON BULLETIN • Vol. 109. No. 1, March 1997

microclimate, food abundance, shrub density, as a host of multiple nest patch factors, may additionally be instrumental in nest site selection (Hol- way 1991); vegetation density may be only highly correlated with these attributes.

A stronger relationship between male age and concealment might have been found had I quantified the vegetation structure of the nest patch. Previous data from UDW and from other studies have shown older birds to breed in superior habitats more often than yearlings (Sherry and Holmes 1989, 1990, Johnson 1994). Greater concealment may be a cir- cumstance of greater nest patch vegetation than nest site vegetation per se. Greater vegetation in the entire nest area may hold a greater advantage in concealment of adult movements to and from the nest as well as yield- ing a greater probability of nest concealment from sheer random nest placement within the nest patch. In addition, it must be noted that cryptic nest placement would not be measured in these analyses, as only vege- tation actually concealing the nest was measured.

My data suggest that concealment of the nest can be advantageous through lower probability of avian predation. However, the selective ad- vantage of greater nest concealment by experienced birds was not evident in this population of Wood Thrush.

ACKNOWLEDGMENTS

I thank H. Weinberg, L. Walton, M. Laut, and E. Santana for their field a.ssi.stance, J. Pesek for statistical assistance, and R. Roth for review, advice, and assistance on this re- search. Linancial support came from the Mclntire-Stennis Lorestry Research Program. Pub- lished as Miscellaneous Paper No. 1597 of the Delaware Agricultural Experiment Station and Contribution No. 694 of the Dept, of Entomology and Applied Ecology, Univ. of Del- aware.

LITERATURE CITED

Brackbill, H. 1958. Nesting behavior of the Wood Thrush. WiLson Bull. 70:70-89.

Desrochers, A. 1992. Age and foraging success in European blackbirds: variation between and within individuals. Anim. Behav. 43:885-894.

Gorman, O. T. and R. R. Roth. 1989. Consequences of a temporally and spatially variable tood supply lor an unexploited gray squirrel (Sciuru.s carolinensi.s) population. Am. Midi. Nat. 121:41-60.

Holway, D. a. 1991. Nest site selection and the importance of nest concealment in the Black-throated Blue Warbler. Condor 93:575-581.

Hoover, J. R, M. C. Brittingham, and L. J. Goodrich. 1995. Ellects of forest patch size on nesting success of Wood Thrushes. Auk I 12:146-155.

Johnson, M. S. 1994. The effect.s of age and habitat on the reproductive performance of wood thrushes (Hylocichla mu.stelina). M.S. thesis. Univ. of Delaware, Newark.

Lonocore, j. R. and R. E. Jones. 1969. Reproductive success of the Wood Thrush in a Delaware woodlot. Wilson Bull. 81:396-406.

Johnson • AGE EFFECTS ON NEST CONCEALMENT

73

Pyle, P, L. B. Spear, W. J. Sydeman, and D. G. Ainley. 1991. Effects of experience and age on the breeding performance of western gulls. Auk 108:25-33.

Ricklefs, R. E. 1969. An analysis of nesting mortality in birds. Smithson. Contrib. Zool. 9:1-48.

Roth, R. R. and R. K. Johnson. 1993. Dynamics of a Wood Thrush population breeding in a forest fragment. Auk 101:37-48.

Saether, B. E. 1990. Age specific variation in reproductive performance of birds. Pp. 251- 283 in Current Ornithology vol. 7. (R. E Johnston, Ed.) Plenum Press, New York, New York.

Sherry, T. W. and R. T. Holmes. 1989. Age-specific social dominance affects habitat use by breeding American Redstarts {Setophaga ruticilla): a removal experiment. Behav. Ecol. Sociobiol. 25:327-333.

and . 1990. Population age structure of long-distance migratory passerine

birds: variation in time and space. Acta XX Congressus Internationalis Ornithologici. 3:1542-1556.

Weinberg, H. L. and R. R. Roth. 1994. Rectrix shape as an indicator of age in the Wood Thrush. J. Field Ornithol. 65:115-121.

Westmoreland, D. and L. B. Best. 1985. The effect of disturbance on Mourning Dove nesting success. Auk 102:774—780.

Yahner, R. H. and C. a. DeLong. 1992. Avian predation and parasitism on artificial nests and eggs in two fragmented landscapes. Wilson Bull. 104:162—168.

Wilson Bull., 109(1), 1997, pp. 74-81

EFFECTS OF FEMAFE COWBIRD REMOVAF ON REPRODUCTIVE SUCCESS OF HOODED WARBLERS

Bridget J. M. Stutchbury'

Abstract. — Female Brown-headed Cowbirds (Molothrus ater) were systematically re- moved from two adjacent Hooded Warbler (Wilsonia citrina) breeding populations for 2—5 years. All female cowbirds detected using playbacks were removed. Although an average of only 17 female cowbirds were removed annually from each population, this reduced the frequency of parasitism to < 10% of nests in most years. The average percentage of nests parasitized (53%) in populations with no cowbird control (N = 3) was significantly higher than for populations where female cowbirds were being removed (9%, N = 7). Experimental reduction of parasitism, however, did not result in a significant increase in number of young fledged per nest. Predation of entire clutches and broods effectively swamped the gains achieved by controlling female cowbirds. The effectiveness of cowbird removals as a man- agement tool will likely vary with cowbird abundance and host species, but this study suggests that beneficial effects for the host species can be minimal. Received 27 Feb. 1996, accepted 20 Aug. 1996.

The alarming decline of many North American songbird populations over the past few decades (Robbins et al. 1989, Askins et al. 1990) has lead to intensified efforts to identify the causes of this decline (e.g., Hagan and Johnston 1992). There is much debate about whether the critical factor causing the songbird decline is loss of winter habitat (e.g., Robbins et al. 1989) or fragmentation of breeding habitat (e.g., Robinson et al. 1995b, Sherry and Holmes 1996). Brown-headed Cowbirds {Molothrus ater) have been implicated in the songbird decline (e.g., Brittingham and Temple 1983, Robinson et al. 1995b) because they are brood parasites of songbirds. Forest fragmentation in eastern North America, combined with an increase in cowbird numbers (Brittingham and Temple 1983), has lead to unusually high rates of brood parasitism on many forest songbirds (e.g., Robinson et al. 1995b).

There have been calls to control cowbird populations on a large scale as a management tool for slowing population declines (Robinson et al. 1995b). However, few studies have actually controlled adult cowbirds and assessed the impact on songbird breeding success. In the most famous example, cowbird control in the Kirtland’s Warbler {Dendroica kitlandii) breeding area decreased parasitism and increased the average number of young fledged per nest (Kelly and DeCapita 1982, Walkinshaw 1983), but this did not lead to an increase in population numbers (Mayfield 1993). Cowbird control programs for the Least Bell’s Vireo {Vireo belli pusillus) and Black-capped Vireo (V. atricapillus) have also reduced par-

' Dept, of Biology, York Univ., North York. Ontario, Canada M3J I P.3.

74

Stutchburx • EFFECTS OF COWBIRD REMOVAL

75

asitism and increased nesting success (reviewed in Robinson et al. 1995a). Cowbird removals are apparently linked to increasing populations of the Least Bell’s Vireo and Willow Flycatcher (Empidonax traillii extimus) (reviewed in Robinson et al. 1995a). In this study, I examined the effec- tiveness of female cowbird removals for increasing the reproductive suc- cess of Hooded Warblers (Wilsonia citrina), a species that is not threat- ened and has extensive breeding habitat remaining. The frequency of brood parasitism and nesting success of Hooded Warblers in a cowbird removal area was compared with populations without cowbird control.

METHODS

I conducted this study from 1991-1995 in Crawford County, Pennsylvania (41°N, 79°W), in conjunction with other on-going research (Stutchbury et al. 1994, Neudorf 1996, Tarof 1996, Stutchbury and Evans Ogden 1996). The main study site of 100 ha mixed hardwood forest supports about 40 breeding pairs annually. Another study area (50 ha)