The Condor IO I :86-95 8 The Cooper Ornithological Society 1999

PLUMAGE, SIZE, AND SEXUAL DIMORPHISM IN THE QUEEN CHARLOTTE ISLANDS HAIRY ’

EDWARD H. MILLER Biology Department, Memorial University of Newfoundland, St. Johns,’ NF, AIB 3X9, , e-mail: [email protected]

ERIC L. WALTERS Department of Biological Sciences, Florida State University, Tallahassee, FL 32306-2043

HENRI OUELLET Canadian Museum of Nature, P. 0. Box 3443, Station D, Ottawa, ON, KIP 6P4, Canada

Abstract. Plumage, size, and sexualdimorphism of the endemic Hairy Woodpeckersub- speciesPicoides villosus picoideus are described and compared with a less isolated insular population.Museum specimensof P. v. picoideus from the Queen Charlotte Islands, British Columbia were compared with P. v. harrisi specimensfrom Vancouver Island, British Co- lumbia. Specimensfrom the two localities are similar in size, althoughthe culmen of P. v. picoideus is 56% shorter.In plumage, P. v. picoideus is distinct in having markingson the belly and flanks, dorsal barring, and barring on the outer rectrices.The breast and belly are much darker and the wing coverts have fewer white markings in P. v. picoideus than in P. v. harrisi. Adults and juveniles, and adult males and females, differ only in minor features of plumage. Adult plumage of P. v. picoideus and insular populations of P. villosus in Newfoundlandand the Grand Bahamasresembles juvenal plumagein mainlandlocations, andincludes ancestral traits in this woodpeckerlineage. Sexual dimorphismin size is similar in P. v. picoideus, P. v. harrisi, and continentalpopulations, suggesting no ecologicalrelease of the sexeson the Queen Charlotte Islandsor VancouverIsland. Relative sexualdimorohism in bill size varies little over the species’ range, hence appears to be influenced litile by ecological factors. Key words: , island biology, Pacijc Northwest, villosus, Queen Charlotte Islands, sexual dimorphism, Vancouver Island.

INTRODUCTION spatial scales (Dehnel 1948). Differentiation be- The Hairy Woodpecker (Picoides villosus) is a tween eastern and western forms may reflect permanent resident over a broad range, from tree postglacial colonization from and adaptation to line in Alaska and northern Canada south to the two great forested Pleistocene refugia in Costa Rica and . Over this range, the southwestern and southeastern North America species inhabits diverse forested habitats from (Voous 1947, Moore and Price 1993), although sea level to 4,000 m (Short 1982, Jackson and genetical evidence supporting this evolutionary Ouellet 1999). Extensive geographicvariation in scenario is wanting (Klicka and Zink 1997). plumage and size occurs (Ouellet 1977, Short The distinctive subspeciesP. v. picoideus is 1982), hence many subspecieshave been rec- endemic to the Queen Charlotte Islands, British ognized (AOU 1957, Winkler et al. 1995, Pyle Columbia (Osgood 1901, Short 1982). It is one 1997). One broad pattern of variation is an in- of several species of on the Queen Char- crease in body size from south to north. Another lotte Islands that have differentiated relative to is of drab underparts occurring in birds from mainland forms (Aldrich 1968, Phillips 1991). southeasternAlaska, “west of the western foot- Ecological factors responsible for avian differ- hills of the Rocky Mountains to western Texas, entiation on the archipelago have not been in- northern Baja California, and western Panama” vestigated, but feeding competition is one likely (Ouellet 1977), and white underparts east of cause. In some insular populations of wood- there (Coues 1877, Jenkins 1906). Strong pat- peckers, ecological release from feeding com- terns of differentiation also appear on smaller petition has promoted sexual divergence in for- aging behavior and size of the feeding apparatus I Received19 January 1998. Accepted 10 August (Selander 1966, 1972). 1998. The main purpose of this study was to inves- QUEEN CHARLOTTE ISLANDS HAIRY WOODPECKER 87 tigate the possibility that ecological release led March (l-l-6-5) April (4-7-6-5), May (15-17- to enhanced sexual differences in size of P. v. 10-12), June (13-26-3-6) July (9-7-2-g), August picoideus on the Queen Charlotte Islands. Sec- (4-7-4-2), September (l-4-1-4), October (0-2-4- ondary aims were to describe body size and sex- 3), November (l-O-3-2), and December (1-1-3- ual differences in plumage, which are commonly 6). affected by insularity. The following variables were measured with In this paper we describe size and plumage of digital calipers or ruler except as noted: length P. v. picoideus, with particular reference to sex- of flattened wing, to 1 mm; chord of exposed ual differences. We compare our findings with culmen, to 0.1 mm; chord from bill tip to ante- those from parallel analyses on the coastal sub- rior edge of nares, to 0.1 mm; tarsal length, to species P. v. harrisi from Vancouver Island, 0.1 mm; tail length, to 1 mm; chord of claw on British Columbia. Vancouver Island was select- digit IV, from tip of claw to dorsal junction of ed for comparison becauseit is roughly the same claw with toe, to 0.1 mm (measured with 8X size as the Queen Charlotte Islands and because dissecting microscope and ocular micrometer); P. v. harrisi there is similar in body size to P. and several variables on primary VIII, to 1 mm v. picoideus. In addition, more speciesof wood- (Fig. 1F). Some seasonal variation in measure- pecker inhabit Vancouver Island than the Queen ments occursbut showsno obvious trends (Jack- Charlotte Islands (six vs. three; Godfrey 1986), son 1970), so specimens collected at different and Vancouver Island has few or no barriers to seasonswere combined. The two measurements gene flow with the mainland because of direct on bill length were highly correlated, so most proximity and the presence of intervening is- analyses were conducted only on chord of ex- lands. In contrast, the Queen Charlotte Islands posed culmen. are about 50 km south of Dal1 Island, Alaska, Plumage attributes were scored against refer- 50 km west of Percher Island (adjacent to the ence series for each variable (Fig. lA-E). Sep- British Columbia mainland), and 225 km north arate reference series were established for the of Vancouver Island. two geographic samples for purposes of char- acterizing sexual dimorphism in each locality. METHODS Reference series included juveniles, although This study is based upon an examination of 246 these were excluded from most analyses (see study skins from 18 museums (128 from the Results). High repeatability was established by Queen Charlotte Islands [QCI], 118 from Van- scoring specimens against the series before set- couver Island [VI]): Academy of Natural Sci- ting the ranges for final scoring. The final num- ences; American Museum of Natural History; ber of score categoriesranged from three to five British Museum (Natural History; London, Unit- for different variables, and differed between ed Kingdom); Carnegie Museum; Cowan Mu- geographic samples. Plumage variables were: seum (Vancouver, British Columbia, Canada); darkness of breast and belly (1 = palest, 3 = Field Museum; ForschungsinstituttSenckenberg darkest for QCI, 5 = darkest for VI); darkness (Frankfurt, Germany); Merseyside County Mu- of white plumage in median dorsal stripe (1 = seums (Liverpool, United Kingdom); Museum palest, 4 = darkest); darkness of supercilium (1 of Vertebrate Zoology; National Museum of = palest, 3 = darkest); darkness of moustachial Natural History; National Museum of Natural stripe (1 = palest, 3 = darkest); extent of belly Sciences (Ottawa, Ontario, Canada); Natural and flank markings (1 = least, 4 = most; QCI History Museum (Basel, Switzerland); Peabody only); amount of black markings on white areas Museum; Royal British Columbia Museum of the rectrices (1 = none [hypothetical only; (Victoria, British Columbia, Canada); Royal On- none observed], 5 = greatest; QCI only); area tario Museum (Toronto, Ontario, Canada); Uni- of median dorsal stripe (1 = smallest, 4 = larg- versity Museum (Oxford, United Kingdom); est); amount of black markings in median dorsal University of Illinois at Urbana-Champaign; stripe (1 = least, 5 = greatest); and amount of University of Michigan; and University of Puget white markings in wing coverts (1 = none, 5 = Sound. The seasonal distribution of specimens greatest). As for measurements,seasonal chang- of known sex and with known collecting dates es in plumage are minor and unsystematic(Jack- was (QCI females-QCI males-VI females-VI son 1970), although Ouellet (1977) noted in- males): January (O-1-6-5), February (O-1-3-4), creased whiteness from fall to spring in eastern 88 EDWARD H. MILLER ET AL.

F QUEEN CHARLO’ITE.JSLANDS HAIRY WOODPECKER 89

P. villosus. Seasonal differences were not found polynomial regression, and other analyses were in this study. done with Statview 4.5 (Abacus Concepts). In First-year birds were distinguished using cri- regressions, residuals were examined for nor- teria of Osgood (1901), Jackson (1970), George mality, homogeneity of variance, and indepen- (1972), and Short (1982). Ouellet (1977) sug- dence. gested that darkness of underparts is unrelated to age, and Jackson (1970) commented on the RESULTS similarity of first-winter and adult plumage in DIFFERENCES BETWEEN the species. However, Bent (1939) noted that GEOGRAPHIC SAMPLES “the white spots are not quite so. pure white” Hairy Woodpeckerswere of similar size on QCI (presumably referring to coverts) in first-year P. and VI, although some differences were found: v. villosus. That observation, and slight differ- QCI females had longer wings and males had ences noted in the present study, led us to ex- longer tarsi than did VI specimens (Table 1). clude first-year specimens from most analyses. Length of exposed culmen differed strongly be- tween the two areas, however, averaging about DATA ANALYSIS 3 mm shorter on QCI for both sexes (Table 1). Simple linear regression was used where some Markings on primary VIII showed few differ- predictive effect or cause of the independent ences between localities. In QCI adults, the variable seemed reasonable, such as sexual di- proximal edge of the marking on the feather’s morphism in size relative to body size. Reduced posterior margin was closer to the feather tip major axis regression was used when both var- than in VI specimens (variable P3; Table 1). In iables were measured with error and there was females, this marking was smaller in QCI than no presumed effect of one variable on another, in VI specimens (variable P4; Table 1). tie- such as male and female sizes within popula- way ANOVAs confirmed those trends and dis- tions (Ricker 1973). Polynomial regression was closed a significant interaction term for variable used to describe the relationship of bill size to P4: sexual differences within the two geographic body size acrossgeographic samples (data from samples were in opposite directions (Tables 1 Ouellet [1977] and this study). Following pro- and 2). ceduresin Sokal and Rohlf (198 l), quadratic ex- pressionswere determined to be a better fit than AGE VARIATION IN PLUMAGE higher-order polynomials or simple linear re- First-year and adult specimens differed slightly gressionsfor those data. in charactersof primary VIII (Table l), and in Sexually dimorphic structures often exhibit body plumage: QCI adults had paler underparts positive allometry with body size (Andersson (females) and a darker supercilium (males); VI 1994). To investigate this possibility, simple lin- adults of both sexeshad less black in the median ear regression of the chord of exposed culmen dorsal stripe (Table 3). on length of flattened wing was carried out for each sex in the QCI and VI samples. SEXUAL AND INDIVIDUAL Reduced major axis regression was fitted with VARIATION IN PLUMAGE Kaleidagraph 3.0 (Abelbeck Software, 2457 Per- Minor sexual differences in plumage were not- kiomen Avenue, Reading, PA 19606). Analysis ed. On primary VIII, the proximal edge of the of variance (one- and two-way ANOVAs and marking on the feather’s posterior margin was MANOVAs) was carried out with SuperAnova farther from the feather tip in VI females than 1.11 (Abacus Concepts, 1918 Bonita Avenue, in males (variable P3); this marking was larger Berkeley, CA 94704). Simple linear regression, in QCI males than females (variable P4), and in

t FIGURE 1. Examplesof referenceseries of Hairy Woodpeckersfor scoringplumage characteristics (A-E), and variables on primary VIII (F). A-amount of white markingson wing coverts (VancouverIsland; scores l- 5). B-area of median dorsal stripe (Vancouver Island; scores l-4). C-amount of black markings in median dorsal stripe (Queen Charlotte Islands; scores l-5). D-extent of belly and flank markings (Queen Charlotte Islands; scores l-4). E-amount of black markings on white areas of the rectrices (Queen Charlotte Islands; scores2-5). F-right primary VIII in dorsal view, showing five variables measuredon it. 90 EDWARD H. MILLER LZ AL.

TABLE 1. Descriptivestatistics for continuousvariables (adult specimensonly). ’

Qwm Charlotte Islaltds~ Vancouver Islattd~

Variable FUlUk Male FellISle Male Body mass (g) 72.8 + 3.1 80.1 2 2.9 - - (5) (4) Length, flattenedwing 127.1 + 0.4 ’ 128.7 + 0.4 125.4 + 0.4 ’ 128.1 + 0.4 (mm) (34) (52) (45) (54) Chord, exposedculmen 28.1 + 0.2d 31.5 2 0.2 ’ 30.9 ? 0.2d 34.5 + 0.2e (mm) (33) (52) (43) (54) Chord, bill tip to anterior 24.6 + 0.2 27.7 + 0.2 27.3 + 0.2 30.4 + 0.2 edge of nares(mm) (32) (50) (43) (54) Tarsal length (mm) 21.7 + 0.1 22.7 2 0.1 ’ 21.6 ? 0.1 22.1 + 0.1 ’ (34) (53) (44) (54) Tail length (mm) 77.1 ? 0.8 76.3 + 0.7 77.4 ” 0.6 77.9 ? 0.6 (31) (46) (41) (47) Chord, claw IV (mm) 6.48 2 0.05 6.83 2 0.03 6.57 ? 0.04 6.88 + 0.04 (34) (53) (45) (54) Pl (mm) 35.4 + 1.2 34.7 2 1.4 32.6 ? 1.3” 33.3 + 1.3 (34) (53) (45) (54) P2 (mm) 2.47 + 0.15 2.70 + 0.20 2.58 + 0.13 2.65 ? 0.15 (34) (53) (45) (54) P3 (mm) 45.5 + 0.9 ’ 43.9 + 0.W 50.4 2 0.8J’ ”’ 41.2 + 0.W (34) (51) (45) (52) P4 (mm) 3.24 + 0.16~0.~ 3.75 2 O.l@ 3.96 2 0.18” 3.67 + 0.18 (34) (51) (45) (52) P5 (nun) 3.35 + 0.21 3.18 + 0.15 3.91 & 0.21q 3.27 + 0.20s (34) (51) (45) (52)

aCell entries are Mean + SE 01). For explanation of PI-P% see Figure 1. b Witbin localities. sexes differed significantly (P < 0.05) on all body measurements c Females between localities: P < 0.01 (by one-way ANOVA. here and below). de Females and males behveeo localities: P < 0.001. fMales between localities: P < 0.01. 8 Male adults vs. juveniles Ljuveniles = 29.3 + 2. I (ZO)]: P < 0.05. ~.Female adults vs. juveniles Ljuveniles = 22.1 ? 4.8 (7)]: P < 0.01. ‘J Females and males between localities: females. P < 0.001: males, P < 0.01. k Male adults vs. juveniles ljuveniles = 40.4 * 1.4 (u))]: P < 0.05. ’ ’ Femaleadults vs. juveniles Ljuveniles = 42.9 + 2.2 (7)]: P < 0.01. 111Females vs. males: P < 0.01. ” Females between lccalities: P < 0.01. oFemde adults vs. juveniles Ljuveoiles = 4.36 ? 0.44 (1411: P < 0.01. P.9Females vs. males: P < 0.05.

VI females than males (variable P5). QCI males SEXUAL VARIATION IN MEASUREMENTS had darker underparts than females, and VI Data on measurements and sexual dimorphism males had a darker dorsal median stripe than fe- are summarized in Tables 1, 3, and 4. Males males (Table 3). were much larger than females in length of cul- men (12%), moderately larger in chord of claw TABLE 2. Summaryof resultsfrom two-way ANOVAs and in tarsal length @-!I%), and slightly larger on continuousvariables, for variablesthat exhibitedstatis- in length of flattened wing (l-24); tail length tically significantdifferences. did not differ between the sexes. As noted above, QCI males had longer tarsi than VI Lncalit~ Sex Interaction males, which accountsfor the greater sexual di- -MaleQCI vs. Sex x Variable VI female locality morphism in tarsal length in the QCI sample (= 4.4%, vs. 2.6% on Vancouver Island). As ex- Length, flattenedwing *** *** Chord, exposedculmen *** *** pected, MANOVAs disclosed highly significant Tarsal length *** *** sexual differences in size in each geographic Chord, claw IV *** sample (P < 0.001). *** *** P3 All regressionsof length of exposed culmen P4 * P5 * on length of flattened wing were statistically nonsignificant (P > 0.05), indicating indepen- ueen Charlotte Islands; VI = Vancouver Island :F: Yits,*** P c 0.001. dence of the two traits within localities. QUEEN CHARLOTTE ISLANDS HAIRY WOODPECKER 91

TABLE 3. Data summary for plumage scores.8

Queen Charlotte Islands Vancouver Island

Character scored Female Male Female Male Darkness,breast and belly 1.42b,c 2.41 2.93 (l-3) (1-5) (l-3 Darkness,dorsal stripe 2.19 2.42 2.00* 2.22* (l-4) (l-4) (l-4) (l-4) Darkness,supercilium 1.88 1.91’ 1.80 1.81 (l-3) (l-3) (l-3) (l-3) Darkness, moustachialstripe 2.50 2.38 2.24 2.30 (2-3) (l-3) (l-3) (l-3) Extent, belly and flank markings 1.75 1.41 - 1’6;) (l-4) - Amount, black rectrix markings 2.19 - (i-5) (l-5) - Area, dorsal stripe 2.50 2.10 2.73 2.52 (l-4) (l-4) (l-4) (l-4) Amount, black markings in dorsal stripe 3.33 3.14 0.89’ 0.888 (l-5) (l-5) (l-3) (l-2) Amount, white markings in wing coverts 2.20 2.30 2.68 2.81 (l-5) (l-5) (l-5) (l-5)

a Cell entries are median (range). b Females vs. males: P < 0.01 (by Chi-square test here and below, except as noted). c Female adults vs. juveniles (juvenile median = 2.21): P = 0.02 (Fishers’ Exact Test, with score categories >I combined). d Females vs. males: P -c 0.05. e Male adults vs. juveniles (juvenile median = 1.67): P = 0.01 (Fishers’ Exact Test. with score categories 21 combined). Female‘ adults vs. juveniles (juvenile mediao = 1.69): P = 0.03 (Fishers’ Exact Tesl, with score categories 21 combined). 8 Male adults vs. juveniles (juvenile median = 1.75): P = 0.02 (Fishers’ Exact Test. with score categories >I combined).

DISCUSSION to have more spotting on primaries, secondaries, In general features of plumage, Hairy Wood- and coverts than males (Jackson 1970). We de- peckers are darker on the Queen Charlotte Is- tected no sexual difference in this character,per- lands than elsewhere in their range, closely fol- haps because populations in western North lowed by those from Vancouver Island and the America tend to have less spotting than eastern adjacent mainland (from southeasternAlaska to populations (Coues 1877, Voous 1947). northern California; Swarth 1922, Ouellet 1977). P. v. picoideus is unique among western We found that males of P. v. picoideus had dark- North American P. villosus in having dark mark- er underparts than females. Ouellet (1977) de- ings on the belly and flanks, and on the tail (such tected no sexual differences in darkness (or markings occur occasionally in adults from Van- whiteness) of underpartsover the species’ range. couver Island, coastal British Columbia, and Elsewhere in the species’ range, females tend elsewhere, however; Ridgway 1914, Voous 1947, Ouellet 1977). Similar markings to those of P. v. picoideus occur regularly in two other TABLE 4. Sexualdimorphism in size (adult specimens only).a insular populations: Newfoundland (P. v. terrue- novae) and Grand Bahamas (P. v. piger) (Jen- Queen Charlotte V~COUVer kins 1906, Oberholser 1911). Dorsal barring, Variable Islands Island markings on the underparts,and markings on the - Body mass (g) 10.0 outer rectrices are thought to reflect ancestral Length, flattened wing statesin Picoides (Goodwin 1968, (mm) 1.21 2.26 Chord, exposed culmen Short 1971). Their independent evolution in sev- (mm) 12.2 12.0 eral insular populations presumably reflects ne- Chord, bill tip to anterior oteny, as the charactersin question occur in ju- edge of nares (mm) 12.6 11.4 Tarsal length (mm) 4.74 2.68 venal plumage elsewhere in the species’ range Tail length (mm) -1.08 0.74 (Jackson 1971, Ouellet 1977). The sexes do not Chord, claw IV (mm) 5.37 4.63 differ in these attributes on the Queen Charlotte aValues shown are lOO[(male measurement/female measurement) - I]. Islands or Vancouver Island. 92 EDWARD H. MILLER r?rAL.

SSD in variable 6- length P. v. hanisi 4- VaNower Is.

CentralB.C. -2 I I 1 I I I Tail Wing Tarsus Claw Culmen FIGURE 2. Sexual size dimorphism(SSD) in different variables.SSD in length of culmen is greaterthan in othercharacters, but doesnot differ betweenP. villosuspicoideus and hrrisi. Linesjoin data pointsto emphasize trends. SSD was computed as lOO[(male measurement/femalemeasurement) - 11. Points for Red-breasted (Sphyrapicusruber) and Red-naped(S. nuchlis) Sapsuckersalso are shown (see text).

In summary, sexual differences in plumage of Sexual dimorphism in woodpeckers is some- P. v. picoideus and hurrisi are small, and the times greatest in simple communities (e.g., is- sexes overlap extensively. This pattern is typical lands), where ecological release in feeding ecol- of the species throughout its range and has ap- ogy is possible, or where high population den- parently not been evolutionarily aItere4Jfor life sities (with attendant increased intraspecific on the Queen Charlotte Islands or elsewhere. competition) often occur (Selander 1966, 1972, P. v. picoideus is similar in size to other pop- Stamps et al. 1997). Examples are the Cuban ulations in the Pacific Northwest, including Green Woodpecker (Xiphidiopicus percussus), coastal and more southernpopulations in British Hispaniolan Woodpecker (Melanerpes striatus), Columbia and more northern coastal populations and Puerto Rican Woodpecker (M. portoricen- in southeasternAlaska (Jackson 1970, Ouellet sis). Increasedsexual dimorphism in woodpeck- 1977). Thus, no evolutionary differentiation in ers on islands is manifested in the feeding ap- the species’ size appearsto have taken place on paratus, with males evolving larger tongues and the Queen Charlotte Islands, except in the small- bills, females smaller tongues and biis, or both er bill. (Selander 1966, 1972). The Hairy Woodpecker should show similar patterns becausemales and

36 females differ in foraging behavior and bill size

34 (Kilham 1965, Selander 1965). On the Queen Charlotte Islands, in particular, ecological re- lease might be expected, as only Hairy Wood- Lengthof pecker,Northern Flicker (Co@tes auratus), and culmen ;:28 ma& (mm) Red-breastedSapsucker (Sphyrapicus ruber) oc- cur there (Short 1982, Cowan 1989). On Van- 24- .8 l _._--e o __-- couver Island those speciesplus Downy (Picoi- 22- c---- a23 pubescens), Pileated (Dryocopus pileatus), 201 ,O _I , , , , , 1 95 100 105 110 115 120 125 130 135 and Lewis’ (Melanerpes lewisi) Woodpeckers Lengthof wing(mm) reside or breed (the latter species only in the FIGURE 3. Tbe bill is relatively longerin large-bod- southern part, however; Godfrey, 1986). Con- ied (estimatedby wing length) populationsof P. vil- trary to prediction, sexual dimorphism is similar losus. Data represent geographic samples from in all characters-including the bill-in P. v. pi- throughouttbe $ecies’ r&gc durves repreSentqua- coideus and harrisi (Fig. 2). Interestingly, slight dratic regressions:for males (solid circles), Y = 104.6 - 1.60X + 0.008X2 (8 = 0.75); for females (open (- 6%) sexual size dimorphism in length of bill circles), Y = 72.8 - 1.11X + 0.006X2 (rZ = 0.72). occurs in the normally monomorphic Red- QUEEN CHARLOTTE ISLANDS HAIRY WOODPECKER 93

36-

34-

32-

30- Culmen length of males 26- (mm) 26-

24-

22-

20 ; 1 I I I 1 I I , 20 22 24 26 26 30 32 34 36 Culmen length of females (mm)

2o B

.

. l . 15- 9 . . . . l SSD in . . --- _9------, -%----- culmen length l

5 I I I I I I , 100 105 110 115 120 125 130 135 Mean length of wing (mm)

FIGURE 4. Sexual dimorphism in relative bill size of P. viflosusdoes not vary geographically.A-bivariate plot of chord of exposed culmen in males and females for geographic samples from throughoutthe species’ range; regressionline from reduced major axis regression(RMAR) is shown. B-sexual size dimorphism (SSD = lOO[(male/female) - l] ) in chord of exposed culmen plotted againstmean length of flattened wing [ = (male + female)/2], for geographic samples from throughoutthe species’ range; regressionline from simple linear regression(SLR) is shown (data from Ouellet [ I9771 and this study).

breasted Sapsucker on the Queen Charlotte Is- at several levels of analysis. First, as noted lands, which seems to be the only place in the above, bill size of P. villosus is uncorrelated species’ range where such a sexual difference with body size in both sexes of the QCI and VI exists (Walters and Miller 1999). The Northern samples.It is usual for scaling relationships such Flicker on the Queen Charlotte Islands merits as this one to become stronger at higher taxo- study from this viewpoint, as the “purest” red- nomic levels (Gould 1966, Peters 1983, Calder shafted forms may occur there in isolation from 1984), and this is true here: acrossa broad range mainland populations (Short 1965, Moore of geographic samples, bill size increases dis- 1995). proportionately with body size in both males and The absence of a large sexual difference in females (Fig. 3). Bill size between the sexes is length of bill in P. v. picoideus initially seems highly correlated at this level of analysis (Fig. surprising, considering the isolation of the 4A), so sexual size dimorphism is uniform Queen Charlotte Islands, their simple avifauna, across populations-despite the wide range in and the subspecies’ distinctiveness. This finding body size and ecological conditions represented is consistent with conservatism in sexual size by the samples (Fig. 4B) (contrast Hogstad dimorphism in the species,however, as revealed 1993). Selander (1966) analyzed sexual size di- 94 EDWARD H. MILLER ET AL.. morphism across species of melanerpine wood- Downy Woodpeckersin eastern North America. peckers in a similar manner, and likewise found -Banding 43: 128-l 35. GODFREY, W. E. 1986. The birds of Canada, rev. ed. no evidence of a relationship to body size. Natl. Museums of Canada, Ottawa, Ontario, Can- Refugia were present in Kodiak Island, south- ada. eastern Alaska, the Queen Charlotte Islands, and GOODWIN, D. 1968. Notes on woodpeckers(Picidae). Vancouver Island during the last glacial maxi- Bull. Br. Mus. Nat. Hist. 17:1-44. GOULD,S. J. 1966. Allometry and size in ontogenyand mum about 17,000-20,000 years ago (Kavan- phylogeny. Biol. Rev. 41:587-640. augh 1992). Vegetation then was tundra, hence HFBDA, R. J. 1995. British Columbia vegetation and was unsuitable as woodpecker habitat (Hebda climate history with focus on 6 KA BP Geogra- 1995). Thus, Hairy Woodpeckers in the Pacific phie physique et quatemaire 49:55-79. Northwest presumably colonized coastal Alaska HOGSTAD,0. 1993. Why is the Three-toed (Picoides rri&~ctylus),more sexually dimorphic than other and British Columbia (including the Queen European woodpeckers? Beih. Veriiff. Natur- Charlotte Islands) from south of the ice sheet. schutz Landschaftspflege Bad.-Wihtt. 67:109- The subspecies P. v. picoideus differentiated 118. into its present distinctive form over the last - JACKSON,J. A. 1970. Character variation in the Hairy 12,000 years, when suitable forested habitat first Woodpecker (Dendrocoposvillosus). Ph.D. diss., Univ. Kansas, Lawrence, KS. appeared on the Queen Charlotte Islands (Ma- JACKSON,J. A. 1971. The evolution, , distri- thewes 1989, Hebda 1995). bution, past populationsand current statusof the Red-cockaded Woodpecker, p. 4-29. In R. L. ACKNOWLEDGMENTS Thompson [ed.], The ecology and managementof the Red-cockaded Woodpecker. U.S. Dept. Int., We are indebted to the curatorsand staff of many mu- Bur. Soort Fisheries and Wildl.. Folkston. GA. seums for permitting us to examine specimensin their JACKSON,J: A., AND H. OUELLET.19&X Down; Wood- care. We thank M. R. Browning, M. Collins, D. R. pecker (Picoides pubescens)and Hairy Wood- Paulson, R. J. Hebda, J. A. and Jackson,E C. James, pecker (P. villosus).In A. Poole and E Gill [eds.], R. W. Mathewes for their helpful comments on this The birds of North America. Academy of Natural paper. Research was supportedby Memorial Univer- Sciences, Philadelphia and American Omitholo- sity of Newfoundland, the Natural Sciences and En- gists’ Union, Washington,DC, in press. gineering ResearchCouncil of Canada (Individual Op- JENKINS,H. 0. 1906. Variation in the Hairy Wood- erating Grant to EHM, University of Victoria), and the pecker ( villosus and subspecies).Auk Royal British Columbia Museum. EHM and ELW are 23:161-171. sad to report that Henri Ouellet passedaway just be- KAVANAUGH,D. H. 1992. Carabid beetles (Insecta:Co- fore this paper was published. leoptera: Carabidae) of the Queen Charlotte Is- lands, British Columbia. Mem. California Acad. LITERATURE CITED Sci. 16. ALDRICH,J. W. 1968. Population characteristicsand KILHAM,L. 1965. Differences in feeding behavior of nomenclature of the Hermit Thrush. Proc. U.S. male and female Hairv Woodoeckem. Wilson NaU. Mus. 124(3637):1-33. Bull. 77:134-145. - - AMERICANOrort~~o~ts~s ’ UNION. 1957. Check-list KLICKA,J., ANDR. M. ZINK. 1997. The importanceof of North American birds, 5th ed. The Lord Bal- recent ice ages in speciation: a failed paradigm. timore Press,Baltimore. Science 277: 16661669. ANDERSSON, M. 1994. Sexual selection. Princeton MA-, R. 1989. Paleobotanyof the Queen Char- Univ. Press, Princeton, NJ. lotte Islands, p. 75-90. In G. G. E. Scudder and BENT, A. C. 1939. Life histories of North American N. Gessler leds.1. The outer shores.Oueen Char- woodpeckers.Bull. U.S. Natl. Mus. 174. lotte Islands Mus. Press, Skidegate,British Co- CALDER,W. A., III. 1984. Size, function, and life his- lumbia, Canada. tory. Harvard Univ. Press, Cambridge, MA. MOORE,W. S. 1995. Northern Flicker (Colupresau- COVES,E. 1877. Birds of the North-west: a hand-book rum). In A. Poole and E Gill [eds.], The birds of of American ornithology. Estes and Lauriat, Bos- North America, No. 166. Academy of Natural Sci- ton. ences, Philadelphia and American Ornithologists’ COWAN,I. McT. 1989. Birds and mammals on the Union, Washington,DC. Queen Charlotte Islands, p. 175-186. In G. G. E. MOORE,W. S., ANDJ. T PRICE.1993. Nature.of selec- Scudder and N. Gessler leds.1. The outer shores. tion in the Northern Flicker hybrid zone and its Queen Charlotte Islands-Mui: Press, Skidegate, implications for speciationtheory, p. 196-225. In British Columbia, Canada. R. G. Harrison red.], Hybrid zones and the evo- DEHNEL,I! A. 1948. Analysis of variation in some lutionary process.Oxford Univ. Press, New York. western races of the Hairy Woodpecker,Dendro- OBERHOLSER,H. C. 1911. A revision of the forms of copes villosus. M.A. thesis, Univ. California, the Hairy Woodpecker (Dryobates villosus [Lin- Berkeley, CA. naeus]). Proc. U.S. Natl. Mus. 40:595-621. GEORGE,W. G. 1972. Age determinationof Hairy and OSGOOD, W. H. 1901. Natural history of the Queen QUEEN CHARLOTIE ISLANDS HAIRY WOODPECKER 9.5

Charlotte Islands, British Columbia; Natural his- SHORT,L. L. 1965. Hybridization in the flickers (Co- tory of the Cook Inlet region, Alaska. N. Am. luptes) of North America. Bull. Am. Mus. Nat. Fauna 2 1. Hist. 129. OUELLET,H. R. 1977. Biosystematicsand ecology of SHORT,L. L. 1971. Systematicsand behavior of some Picoides villosus (L.) and P. pubescens (L.), North American woodpeckers, Picoides (Aves: Picidae). Ph.D. diss., McGill Univ., Mon- (Aves). Bull. Am. Mus. Nat. Hist. 145:1-l 18. treal, Quebec, Canada. SHORT,L. L. 1982. Woodpeckersof the world. Dela- PETERS,R. H. 1983. The ecological implications of ware Mus. Nat. Hist., Greenville, DE. body size. Cambridge Univ. Press, Cambridge. SOKAL.,R. R., AND E J. ROHLF.1981. Biometry, 2nd PHILLIPS,A. R. 1991. The known birds of North and ed. W. H. Freeman, San Francisco. Middle America, Part II. Bombycillidae; Sylviidae STAMPS,J. A., J. B. Losos, ANDR. M. ANDREWS.1997. to Sturnidae; Vireonidae. Privately published, A comparative study of population density and Denver, CO. sexual dimorphism in lizards. Am. Nat. 149:64- PYLE, P 1997. Identification guide to North American 90. birds, part 1: Columbidae to Ploceidae. Slate SWARTH,H. S. 1922. Birds and mammalsof the Stikine Creek Press, Bolinas, CA. River region of northern British Columbia and southeasternAlaska. Univ. Calif. Pub. Zool. 24: hCKER, W. E. 1973. Linear regressionsin fishery re- search.J. FisheriesRes. Bd. Canada 30:409-434. 125-314. Voous, K. H., JR. 1947. On the history of distribution R~CWAY, R. 1914. The birds of Middle and North of the genus . Limosa 20:1-142. America, Part VI. Bull. U.S. Natl. Mus. 50(6). WALTERS,E. L., AND E. H. MILLER.1999. Yellow-bel- SELANDER,R. K. 1965. Sexual dimorphism in relation lied Sapsucker(Sphyrapicus varius), Red-breasted to foraging behavior in the Hairy Woodpecker. Sapsucker(5. ruber) and Red-napedSapsucker (.S. Wilson Bull. 77:416. nuchalis). In A. Poole and E Gill [eds.], The birds SELANDER, R. K. 1966. Sexual dimorphism and differ- of North America. Academy of Natural Sciences, ential niche utilization in birds. Condor 68: 113- Philadelphiaand American Ornithologists’ Union, 151. Washington,DC, in press. &LANDER, R. K. 1972. Sexual selection and dimor- WINKLER,H., D. A. CHRISTIE,AND D. NURNEY.1995. phism in birds, p. 180-230. In B. Campbell [ed.], Woodpeckers. An identification guide to the Sexual selection and the descent of man 1871- woodpeckers of the world. Houghton Mifflin, 1971. Aldine, Chicago. Boston.