Data collected from skin specimens of the 23 Recent species of Alcidae, skeletal material for Recent and fossil alcids, and published data on body mass and wing area were used to describe the morphometric characteristics of flightlessness in the Great Auk (Pinguinus impennis) and the fossil mancalline auks. A regression equation confirmed a body-mass estimate (5 kg) for P. impennis (Bédard 1969). The size and relatively small wings produced wing-loadings of roughly 22 g cm -2, comparable to those of medium-sized penguins. Multivariate analysis of external measurements underscored the uniquely large size, relatively short wings, and moderately deep bill of Pinguinus compared to other Recent alcids. Analysis of skeletal measurements revealed that the genera of flightless Alcidae (Pinguinus, Mancalla, Praemancalla, and Alcodes) were characterized by relatively short distal wing elements and dorsoventral flattening of all major wing elements, in combination with relatively large core and pelvic dimensions. These differences were most pronounced in Mancalla, moderately developed in Praemancalla, and smallest in Pinguinus. Estimated body mass (1-4 kg) for selected fossil mancallines exceeded the largest flighted alcids (Uria) but was less than for Pinguinus. Pinguinus was a comparatively large piscivore sharing many morphological features with the Razorbill (Alca torda) and murres (Uria spp.). Its flightlessness evidently was a consequence of extreme specialization for pursuit diving, convergent with that of the Spheniscidae. Loss of flight imposed significant requirements on breeding sites and foraging habitats of the Great Auk and presumably the mancallines, and rendered Pinguinus exceptionally vulnerable to human exploitation. Received 30 December 1987, accepted 23 May 1988.

Museum of Natural History, University of Kansas, Lawrence, Kansas 66045 USA THE Great Auk (Pinguinus irnpennis), a flight- less alcid of the North Atlantic, is remembered most for its extinction in the 19th century. This is understandable, for the demise of the Great Auk remains one of the most dramatic extir- pations in historical times, one related to hu- man exploitation for food (Grieve 1885, Nettle- ship and Evans 1985) and possibly to long-term climatic trends (Bengtson 1984). Superstition also took a toll: the last Great Auk taken in Great Britain reportedly was killed as a tempest-con- juring witch (Ley 1935, Greenway 1967). The extermination of the species was complete by 1844 (Greenway 1967), despite the prediction of this eventuality in the late 18th century (Net- tleship and Evans 1985). Flightlessness of the Great Auk, traditionally less compelling to ornithologists than the de- mise of the species, is of comparable importance to evolutionary biology. Although the flightless condition of Pinguinus was noted frequently by early naturalists, particularly with respect to its vulnerability to hunters on land, the anatomical correlates of flightlessness in the Great Auk have received relatively little attention. Existing an- atomical studies consist of basic descriptive os- teology (Owen 1879, Wiman and Hessland 1942) and limited mensural comparisons (Lucas 1890, Blanc 1927, Storer 1945, Miller and Howard 1949, Verheyen 1958). Even compilations of standard external measurements for the Great Auk are not available (cf. Coues 1868, Forbush 1912). The situation is related to the rarity of skin spec- imens and the variable distribution of the abun- dant skeletal material of Pinguinus in museums. The fossil flightless mancalline alcids of coastal California are morphologically similar but less well known. Hundreds of elements of the group have been recovered, and 8 species in 3 genera are recognized from the late Mio- cene through the early Pleistocene: Praeman- calla lagunensis, P. wetrnorei, Mancalla californiensis, M. diegensis, M. rnillerL M. cedrosensis, M. ernlongL and Alcodes ulnulus (Lucas 1901; Miller 1933; Howard 1947, 1966, 1968, 1970, 1971, 1976, 1978, 1982, 1983; Miller and Howard 1949; Olson 1981; Howard and Barnes 1987). The Miocene genus Praemancalla is considered to be a possible ancestor of the largely Pliocene genus Mancalla (Howard 1966, 1976.) Neither genus is thought to be closely related to Pinguinus. The Pacific genera generally are placed in a separate family or subfamily (Howard 1966, 1983) and may be related most closely to the puffins (Fraterculini; R. M. Chandler unpubl. data). Pinguinus is judged to be closely related to the Razorbill (Alca torda) The Auk 105: 681-698. October 1988 and murres (Uria spp.) on behavioral and mor- phological grounds (Storer 1945, Strauch 1985). Alcids are wing-propelled diving birds. Strokes of the partly folded wings provide vir- tually all of the propulsion and much of the maneuverability for submarine locomotion (Townsend 1909, Kelso 1922, Storer 1945). Structural convergences between the Great Auk and the similarly flightless, wing-propelled penguins (Spheniscidae) have received consid- erable attention (e.g. Owen 1879, Wiglesworth 1900, Storer 1945). Aerial flight and submarine propulsion clearly impose different selective pressures on alar structure and the morphology of Pinguinus reflects the evolutionary substitu- tion of aerial flight for extreme specialization for diving (Wiglesworth 1900; Bent 1919; Storer 1960, 1971; Bdard 1969; Bengtson 1984). Man- calla was probably at least as specialized osteo- logically for diving as Pinguinus (Wiman and Hessland 1942, Miller and Howard 1949, How- ard 1970). I compared flightless and flighted alcids, em- phasizing multivariate morphometric analyses. Data from study skins and skeletons of all 22 extant alcid species, the Great Auk, and the fos- sil mancalline auks were included. My objec- tives were: to describe quantitatively the mor- phological characteristics associated with loss of flight in the Alcidae; to estimate the body mass of adequately represented mancalline auks; and, using available ecological and biogeo- graphical information, to consider selected as- pects of the evolution of flightlessness in alcids. METHODS Specimens and related data.--I collected data from 1 ! mounted skin specimens of the Great Auk. In addi- tion, colleagues provided comparable data from an additional !4 mounted Pinguinus skins (see Acknowl- edgments). At least 20 study skins (usually !0 of each sex) of each of the 22 extant species of Alcidae were sampled for comparisons. I measured total length (ex- tended specimens only, from bill to tail, feet exclud- ed), culmen length (exposed, on midline), bill height (at gonys), wing length (chord of unflattened wing), tarsus length (cranial surface), digit-III (middle-toe) length (excluding nail), and tail length (roedial arc) (Baldwin et al. 193!). Data on body mass of extant alcids were collected from specimen labels and published compilations (Johnson !935, !944; Belopol'skii !96!; Bdard !967, !969; Kuroda !967; Dement'ev and Gladkov !968; Sea- ly !976; Threlfall and Mahoney !980; Vermeer and Cullen !982; Murray et al. !983; Dunning !984). Wing areas were traced (Raikow !973), from fresh birds or thawed, fresh-frozen specimens, and the resultant areas were measured with a compensating polar pla- nimeter. Additional wing areas were taken from Mag- nan (!9!2, 1922), Poole (!938), Kuroda (!967), Spring (!97!), Stempniewicz (!982), and Pennycuick (!987); several tracings were provided by R. M. Chandler. Wing-loading was calculated as the ratio of body mass divided by total wing area (g.cm-2; Clark !97!). Associated skeletal material of the Great Auk is not available. One possible exception is an apparently complete, largely articulated skeleton of an immature bird at the British Museum. Mounted skeletons of Pinguinus are composites of unassociated skeletal ele- ments. Extensive series of disassociated skeletal ele- ments are held at the U.S. National Museum of Nat- ural History, American Museum of Natural History, and Museum of Comparative Zoology (Harvard Uni- versity). Anatomical nomenclature follows Baumel (!979). I sought to measure at least 40 unworn specimens of each major skeletal element of P. impennis. The more fragile scapula, furcula, and distal phalanges were not available in such numbers. Comparable data were col- lected from all available material of Mancalla, Prae- mancalla, and Alcodes. Limited mensural data also were collected from a new species of Mancalla (here re- ferred to as Mancalla lg. sp.; Chandler unpubl. data) and Australca sp. (Brodkorb !955, Olson !977). Pin- guinus alfrednewtoni, a poorly represented Pliocene form similar to P. impennis in its measurements (Olson !977), was excluded from study. I also sampled !0 complete skeletons (5 of each sex) of each of the 22 extant species of alcid, although complete samples were not available for several species (e.g. Cepphus carbo, Syn- thliboramphus wumizusume, Aethia pygmaea). Forty-six skeletal measurements were employed, most of which were described previously (Livezey and Humphrey !984, !986) and !! of which were illustrated by Spring (!97!; measurements 2, 6, 7, 12, 13, !4, !6, 38, 39, 40, 4!). All skeletal measurements were made with dial calipers to 0.! mm. Sexual di- morphism was considered to be small in extant species of alcid (Storer !952), and in this study sexual differ- ences were found to be negligible. This similarity of the sexes, and the lack of information on sex for spec- imens of Pinguinus and the fossil species, prompted the pooling of the sexes in the morphometric com- parisons. Statistical analyses.--Linear measurements and log- transformed (base e) wing-loadings and skeletal ratios were compared using analysis of variance (ANOVA) and analysis of covariance (ANCOVA). Spearman cor- relation coefficients (r) were used to measure bivariate associations. Allometry of body dimensions, i.e. the relative rates of size change among variables, was quantified using bivariate allometric equations (Gould !966). I based these estimates on linear regressions on log-trans- formed data, using "geometric mean" estimates to accommodate error in both variables (Livezey and Humphrey 1986). Stepwise linear regressions of log-transformed data were used to estimate the body mass of P. impennis based on species means of external measurements of skins. Body mass of fossil Mancalla and Praemancalla was estimated from stepwise regressions of log- transformed body mass on significantly correlated (P < 0.05) principal components of available skeletal measurements. Components were derived from co- variance matrices based on log-transformed data. For external measurements that involved associated measurements for samples of skins of Great Auks, canonical analyses of log-transformed data for Recent species of alcid were used. Canonical analysis (CA) is a multivariate technique that is robust to moderate departures from the assumptions of multivariate nor- mality and homogeneity of covariance matrices of groups. CA provides multivariate axes that maximally discriminate predefined groups (Pimentel 1979). CAs were also used for interspecific comparisons of as- sociated measurements of humeri, ulnae, and sterna. Variables included in each CA were backstep-selected from the complete suites of measurements using F-statistics (P < 0.05). I used principal component analyses (PCAs) of mean measurements for the species of flightless auks (sexes pooled) to assess multivariate skeletal variation among taxa. Components were extracted by singular value decompositions of covariance matrices based on log- transformed data. Pinguinus and the extant Alcidae were compared using 46 skeletal measurements, and a reduced data set of 23 measurements was used to include fossil species in the comparisons. Because of poor representation of the similarly sized Mancalla californiensis, M. cedrosensis, and M. diegensis, the three species were pooled to derive a single mean vector representing "medium-sized" Mancalla. My primary objectives for the PCAs were to determine the multi- variate axis or axes of changes associated with flight- lesshess, and to determine the relative positions of flightless species on the major axes of variation for alcids generally. The flightless species, because of their large size and exceptional shape, acted as influential outliers in the definition of axes in normal PCAs. This seriously confounded "size" with the shape correlates of flightlessness on PC-I and adversely affected the definition of subsequent axes. Therefore, ! excluded the flightless species for the definition of the first principal component. The flightless species were pro- jected onto this PC-I a posteriori, and the residuals for all species were subjected to another PCA to extract the subsequent axes. I refer to these modified com- ponents as PC-I*, PC-II*, and PC-III*. Because the flightless species were not used in the derivation of PC-I* but were considered in subsequent compo- nents, the latter axes were correlated with PC-I* (but were mutually orthogonal). Despite this sacrifice of orthogonality, this approach was preferable to a stan- dard PCA. It produced a virtually isometric "size" axis for PC-I*, defined a single important "shape" PC- II* for alcids generally (using complete skeletons), and isolated the morphometric changes associated with flightlessness on separate axes (PC-III* and PC-II*, for complete and reduced data sets, respectively). To- tal variances incorporated on the standard principal components and corresponding modified compo- nents were virtually identical. The standard PCs ex- ceeded the modified axes by less than 0.05% of the total variance in the subspaces discussed. I use the term "size" in its traditionally broad sense, i.e. in reference to spatial extent or dimension. Var- ious measures of "size," none perfect for all purposes, emerged from the morphometric comparisons; and rationales for considering them as representative of "size" are given. Correlations between mean body mass (using log-transformed data) and these emer- gent "size" measures are probably the most useful direct measure of overall size (Clark 1979), and are given to facilitate interpretation. I estimated diver- gences between multivariate "size" axes and hypo- thetical axes of isometric size using direction cosines between vectors (Pimentel 1979). For CAs of skins and separate skeletal elements, specimens lacking a minority of measurements were subjected to missing-data estimation. Missing data were estimated from stepwise regressions on avail- able measurements for specimens grouped by genus. These estimates comprised 0.9% of the skin and 1.5% of the skeletal data sets. Statistical procedures used were part of the Biomed- ical Computing Programs (Dixon 1985) and per- formed on an IBM computer at the University of Kan- sas. RESULTS EXTERNAL CHARACTERS Univariate comparisons.--External measure- ments demonstrate the substantially larger size of the Great Auk compared with other Recent alcids (Table 1), including the large extant pi- scivorous genera Cepphus, Uria, and Alca. Alca is considered to be the closest extant relative of Pinguinus (Storer 1945, Strauch 1985). Pinguinus shared with Alca its deep, laterally compressed bill and, despite its much larger body size, had wing lengths comparable to those of Cepphus (Table 1). Despite the thousands of Great Auks slaugh- tered during the eighteenth and nineteenth centuries, the only putative datum for body mass was the report by Feilden (1872) of a Pinguinus killed in 1808 which weighed nine Danish T^II, Ii 1. External measurements (mm, g) of three flighted alcids and the Great Auk (œ + SD [n]). Total Total Culmen Bill Wing Tail Tarsus Middle-toe Species length mass' length height length length length length Cepphus 309 + 18 427 30.6 + 1.4 7.3 + 0.6 159 + 5 46 + 3 32.7 + 1.4 36.4 + 1.6 grylle (20) (155) (20) (20) (19) (20) (20) (20) Uria aalge 423 + 25 1,030 44.9 + 1.7 13.3 + 0.8 200 + 6 44 + 3 38.7 + 2.1 43.8 + 2.5 (20) (613) (20) (20) (20) (20) (20) (20) Alca torda 403 + 29 722 34.7 + 1.7 21.9 + 1.7 197 + 10 79 + 7 33.7 + 1.6 41.9 + 2.3 (20) (217) (20) (20) (20) (20) (20) (20) Pinguinus 786 + 58 [5,000] 84.7 + 4.3 39.3 + 2.4 160 + 9 73 + 13 59.6 + 11.2 70.2 + 5.1 impennis (24) (25) (25) (23) (17) (22) (24) Means based on published data and specimen labels; standard deviations not available. Mass of Pinguinus estimated (see text). pounds (4.5 kg). Bdard (1969) estimated the body mass of Pinguinus to be "c. 5000 [g]" but provided no details concerning this estimate. I attempted an independent approximation of the mass of Pinguinus from a regression equation that related body mass to six non-alar external measurements for the 22 extant species of alcids, using log-transformed means. I excluded wing length because its relationship to body size is obviously atypical in Pinguinus. Despite the small number of data (n necessarily being 22), all six remaining variables entered significantly (P < 0.10). The resulting regression model was: M = -7.909 + 2.344(TOTLEN) + 0.991(LDIGIT3) - 0.325(LTARSUS) - 0.508(LTAIL) + 0.172(HTBILL) - 0.121(LCULMEN). The variables are listed in order of entry into the model; adjusted R: for the model was 98.8%. Substituting log-transformed mean measure- ments for the Great Auk into this equation (Ta- ble 1) and taking the antilog provided an esti- mated body mass of 4,999 g, almost identical to the value of B&dard (1969). Relative wing size.--The extremely small rel- ative wing lengths of Pinguinus was conspicuous in a bivariate plot of wing lengths on body mass for 23 species of alcids, using the estimated body mass of 5,000 g for Pinguinus (Fig. 1). The allo- metric coefficient for the flighted species, = 0.322 (SE [] = 0.001), was significantly less (P < 0.001) than the coefficient for isometry be- tween a linear variate with mass (b = 0.333). The relatively shorter wings of Pinguinus re- tained the 10 functional primary remiges typ- ical of the Alcidae (6 specimens examined), al- though the remiges were substantially shorter than in extant species. Wing areas of 10 flighted species of the AI- cidae are available, and the resultant estimates of wing-loading closely mirrored the negative allometry of wing length with body size. This relationship produced progressively greater wing-loadings as body mass increased. Wing- loadings were, from smallest to largest mean body mass (n = sample sizes for wing areas): Aethia pusilla, 0.71 (n = 2); Alle alle, 0.94 (113); Synthliboramphus antiquus, 1.02 (2); Aethia crista- tella, 1.32 (1); Cyclorrhynchus psittacula, 1.11 (1); Cepphus columba, 1.23 (1); Fratercula arctica, 1.34 (21); Alca torda, 1.63 (4); Uria lomvia, !.69 (2); and U. aalge, 2.06 (20). The allometric coefficient for wing area on body mass for flighted alcids was = 0.632 (SE [] = 0.003; regression significant, P < 0.001; R 2 = 0.99), significantly less (P < 0.00!) than that for isometry of wing area with body mass (b = 0.667). The estimated wing area for Pinguinus was 230 cm 2, based on the doubled area of a tracing of a partially folded wing that was "corrected" graphically to approximate an extended wing. Together with the 5-kg estimate for body mass, this indicates that the wing-load- ing of Pinguinus was roughly 22 g-cm -2. This estimate was much higher than one from an allometric extrapolation of a "flighted" alcid to a body mass of 5 kg, which yielded a wing- loading of 3.26 g.cm -2. This projection, and the empirical estimate of 22 g-cm 2 for Pinguinus, exceed the threshold of flightlessness of 2.5 g.cm 2 hypothesized by Meunier (!951). Multivariate patterns.--A CA of the external measurements from 481 skin specimens defined three important axes of interspecific variation. All seven variables entered the model signifi- cantly (P < 0.001), and provided highly signif- Fig. 1. Log of the mean body mass and wing length of the 23 Recent species of alcid. Regression line (Type- II) was fitted for 22 flighted species, and mass of Pin- guinus #npennis was estimated. The abscissa was bro- ken to accommodate Pinguinus. TABLE 2. Standardized coefficients and associated statistics for canonical variates of seven external measurements of 23 alcid species (n = 481). Canonical variate Character I II III Total length -0.26 0.24 -0.31 Culmen length -0.64 0.48 -0.34 Bill height -0.25 -0.99 0.15 Wing length - 0.18 0.39 0.97 Tail length -0.11 -0.17 0.07 Tarsus length -0.03 0.24 -0.18 Middle-toe length -0.27 -0.24 -0.12 Eigenvalue 152.6 26.5 13.2 Variance (%) 73.3 12.7 6.4 Canonical R 1.00 0.98 0.96 icant discrimination of the 23 Recent species of alcid (Wilks' lambda < 10-7; df = 7, 22, 458; P 0.001) All pairwise interspecific differences were significant (P < 0.001). The first canonical variate (CV-I) had coeffi- cients of like sign for all measurements and reflects in large part "general size," although the relatively great contribution of culmen length and small contributions of tail and tarsus lengths resulted in an approximate 35 ø diver- gence from isometry. Mean scores on CV-I were strongly correlated with mean body mass (r = -0.91 for flighted species, r = -0.95 including estimated mass for Pinguinus). CV-I incorporat- ed almost three-fourths of the total interspecific variance (Table 2), and scores on the axis dif- fered significantly among species (ANOVA of scores; F = 3177.6; df = 22, 458; P < 0.0001). Small species (e.g. Aethia, Alle, Brachyramphus) were scored highly on this axis (reflecting the negative coefficients), whereas species with large overall size (Uria, Fratercula, and especially Pin- guinus) had low scores (Fig. 2). The second most important axis (CV-II) was in large part a measure of "relative bill height," with a lesser, correlated contribution from lengths of the tail and middle toe (Table 2). Interspecific differences in scores on CV-II were highly significant (F = 551.6; df = 22, 458; P < 0.0001). Species with relatively deep bills and, to a lesser extent, relatively long tails and mid- dle toes (e.g. Fratercula, Cyclorrhynchus, Aethia) had low scores on CV-II, and groups with op- posite proportionalities (e.g. Uria, Cepphus) had high scores (Fig. 2). The Great Auk was inter- mediate on CV-II, with a score similar to those of Alca, Ptychoramphus, and Alle (Fig. 2). The third variate (CV-III) reflected "relative wing length" (Table 2), and once again under- scored the unique body form of Pinguinus com- pared to other Recent alcids. The highly sig- nificant interspecific differences in scores on CV-III (F = 295.1; df = 22, 458; P < 0.0001) resulted largely from the extremely low mean score of Pinguinus, and reflected its relatively short wings (Fig. 2). Of the 22 flighted species, only the two endomychurine Synthliboramphus and Aethia pusilla showed any tendency toward relative shortening of the wing. SKELETAL CHARACTERS Univariate comparisons of species.--Most skel- etal measurements also reflected the large size of Pinguinus (Tables 3, 4). Except for distal wing elements, Pinguinus exceeded all other Alcidae, Mancalla and Praemancalla, in its skeletal di- mensions. Measurements of the trunk and leg demonstrated the difference most clearly, e.g. tibiotarsi of Pinguinus averaged 4 cm (45%) lon- ger than those of Uria aalge (Table 4). Although the humeri of Pinguinus were the longest in the Alcidae, they averaged only 22% longer than those of Uria. That is, the relative length of the humerus of Pinguinus was less than in Uria, Alca, and Cepphus, but was approached by those of Mancalla and Praemancalla (Tables 3, 4). The ten- dency toward alar shortening in the flightless auks is more pronounced in the mid-wing ele- ments, especially lengths of the radius, ulna, and carpometacarpus, which were absolutely ,o/ / / Fig. 2. Plot of mean scores of 23 species of alcid on the fit 3 canonical variates based on seven external measurements. Flighted species are numbered as follows: (1) Alle alle, (2) Alca torda, (3) Uria aalge, (4) U. lornvia, (5) Cepphus carbo, (6) C. columba, (7) C. grylle, (8) Brachyramphus brevirostris, (9) B. marmoratus, (10) Synthlboramphus antiquus, (11) S. craveri, (12) S. hypoleucus, (13) S. wumizusume, (14) Ptychoramphus aleuticus, (15) Cyclorrhynchus psittacula, (16) Aethia cristatella, (17) A. pusilia, (18) A. pygmaea, (19) Cerorhinca monocerata, (20) Fratercula arctica, (21) F. corniculata, and (22) F. cirrhata. shorter in Pinguinus than in the much smaller Uria (Table 3). Reductions in the lengths of the mid-wing elements were even greater in Prae- mancalla and Mancalla. Greater shaft widths also characterized the wing elements of the three flightless genera (Table 3), as demonstrated by their compara- tively large maximal widths (MWMs). An AN- COVA of "relative flatness" (ratios of maximal and least shaft widths) confirmed these shape differences in humeri (F = 84.8; df = 6, 113; P < 0.001) and ulnae (F = 24.1; df = 6, 120; P < TABLE 3. Summary statistics (œ _+ SD [n]) for selected measurements of major skeletal wing elements of three flighted and 6 flightless alcids. MWM = maximal width at midpoint. Humerus Ulna Carpometa- Species Length MWM Length MWM carpus length Cepphus grylle 59.6 +__ 1.6 4.4 _+ 0.2 51.4 _+ 1.7 4.2 _+ 0.3 34.0 _+ 1.3 (1) (11) (11) (11) (11) Uria aalge 87.3 _+ 2.4 7.9 _+ 0.5 65.9 _+ 1.6 6.3 _+ 0.3 43.3 -+ 2.8 (12) (12) (12) (12) (12) Alca torda 75.5 _+ 3.0 6.9 _+ 0.3 60.1 _+ 2.6 5.8 -+ 0.3 40.2 _+ 1.8 (10) (10) (10) (10) (10) Pinguinus impennis 106.1 _+ 7.3 12.3 _+ 0.7 57.4 _+ 1.8 8.6 _+ 0.4 42.7 _+ 1.4 (69) (69) (59) (59) (59) Mancalla diegensis 72.9 _+ 6.9 9.7 _+ 1.0 29.3 _+ 1.7 6.0 _+ 0.6 35.7 -+ 2.4 (12) (15) (15) (15) (7) M. cedrosensis 71.7 _+ 2.3 9.6 _+ 1.3 30.0 _+ 1.0 6.2 _+ 0.3 35.8 (2) (2) (9) (9) (1) M. milleri 63.0 _+ 2.8 8.3 _+ 0.8 26.4 _+ 1.3 5.5 -+ 0.4 31.6 _+ 1.7 (18) (24) (21) (21) (12) M. emlongi 87.1 _+ 4.4 11.3 _+ 0.1 -- -- 40.6 -+ 1.1 (2) (2) (2) Praemancalla spp. 78.0 _+ 5.3 9.8 _+ 1.1 40.6 _+ 5.2 7.3 _+ 0.0 36.3 (4) (5) (2) (2) (1) TABLE 4. Summary statistics (œ + SD [n]) for lengths of two trunk and three leg elements of three flighted and six flightless alcids. Sternal carina Coracold Femur Tibiotarsus Tarsometa- Species length length length length tarsus length Cepphus grylle 81.2 + 3.0 30.4 + 0.9 36.2 _+ 1.0 64.5 + 1.9 31.3 _+ 1.2 (11) (11) (11) (11) (11) Uria aalge 122.5 + 5.8 40.1 + 1.1 47.8 + 1.5 91.1 + 2.9 38.0 + 1.3 (12) (12) (12) (12) (12) Alca torda 108.7 + 2.8 36.3 + 1.2 41.2 + 1.9 76.1 _+ 1.4 33.2 + 1.0 (10) (10) (10) (10) (10) Pinguinus impennis 189.7 + 11.3 60.3 + 2.8 73.0 + 2.4 132.8 + 4.6 52.2 + 1.8 (28) (62) (62) (62) (60) Mancalla diegensis -- 47.8 + 4.6 52.3 + 5.5 89.3 + 8.7 41.4 + 2.7 (6) (3) (7) (5) M. cedrosensis -- -- 54.8 87.0 41.2 (1) (1) (1) M. milleri -- 42.8 + 5.1 48.3 + 2.9 78.4 + 4.5 35.2 + 2.3 (6) (8) (S) (10) M. emlongi -- 58.6 65.7 + 2.3 -- 47.3 (1) (3) (1) Praemancalla spp. -- -- 68.8 108.0 47.6 (1) (1) (1) 0.001). This "flattening" of the wing bones in flightless auks also was confirmed by an AN- COVA of maximal shaft widths (MWMs) for the species tabulated (excluding the inadequate samples of M. emlongi and Praemancalla), while correcting for interspecific differences in lengths of elements, for both the humerus (F = 194.6; df = 6, 113; P < 0.001) and ulna (F = 334.2; df = 6, 120; P < 0.001). Canonical analyses of single elements.--Separate CAs of humeri, ulnae, and sterna of flightless and selected flighted alcids permitted a multi- variate assessment of morphological differences among a maximal number of taxa (Fig. 3). The CA of 125 humeri of 10 species of alcid incor- porated all four measurements significantly (P < 0.001), and effectively discriminated the six adequately sampled species (Wilks' lambda = 0.0009; df = 4, 5, 113; P < 0.001) and five ad- ditionally plotted taxa (Australca sp., Mancalla cedrosensis, M. emlongi, M. lg. sp., Praemancalla spp.). The first axis (CV-I) reflected the lengths, head widths, and maximal shaft widths of hu- meri (Table 5), and interspecific differences in scores were significant (F = 518.6; df = 8, 115; P < 0.0001). Species with large measurements had low scores (P. impennis), smaller species in- curred higher scores (Cepphus grylle), and the remaining taxa (e.g. Alca, Uria, and Mancalla) were intermediate (Fig. 3A). CV-II contrasted lengths and maximal shaft widths of humeri, i.e. measured "relative shaft width" (Table 5). Species differed significantly in scores on CV- II (F = 135.6; df = 8, 115; P < 0.0001). Those with comparatively narrow humeri (Uria, Alca, and Cepphus) scored highly; P. impennis was in- TABLE 5. Standardized coefficients and summary sta- tistics for canonical variates of humeri, ulnae, and sterna of selected Recent and fossil alcids. Element Variable CV-I CV-II Humerus Length -0.55 1.01 (n = 125) Head width -0.18 -0.12 LWM 0.06 0.11 MWM -0.50 - 1.08 Eigenvalue 36.7 9.1 Variance (%) 77.4 19.2 Canonical R 0.99 0.95 Ulna Length 1.19 0.15 (n = 13i) LWM 0.02 -0.15 MWM -0.55 -0.98 Eigenvalue 125.8 13.7 Variance (%) 90.0 9.8 Canonical R 1.00 0.97 Sternum Carina length -0.44 1.14 (n = 43) Basin length -0.54 -0.79 Least width -0.34 -0.39 Caudal width 0.15 -0.52 Eigenvalue 86.5 3.4 Variance (%) 94.9 4.8 Canonical R 0.99 0.89 2 (C) Sterna __-,c. .',o Fiõ. 3. ?lots of first 2 canonical 3/4ariates for fliõht- less end selected fliõhted alcids of: (A) 4 measure- merits of humeri, (B) 3 measurements of ulnae, and (C) 4 sternal measurements. Polyõons connect ex- treme individuals in each taxon. termedlate; and the mancallines had the broad- est humeral shafts and lowest scores (Figs. 3A, 4). The undescribed, large Mancalla ("M. lg. sp.") closely resembled Pinguinus in its humeral di- mensions. A third variate (CV-III, not figured) contributed only 3% of total intergroup vari- ance, but provided significant interspecific dif- ferences in scores (F = 30.0; df = 8, 115; P < 0.0001). CV-III primarily separated Alca, Uria, and M. milleri from Cepphus, Pinguinus, and the larger mancallines by relative least shaft width and relative head width. A CA of ulnae of selected alcids significantly separated the seven species analyzed (Wilks' lambda = 0.0004; df = 3, 6, 121; P < 0.001); an additional three taxa were plotted on the axes (Australca, Alcodes, Praemancalla) which were represented by small samples (Fig. 3B). The three variables entered the model significantly (Table 5; P < 0.001) and contributed to two axes in- corporating interspecific differences in scores; (ANOVA of scores; F = 1669.8 and 213.2, re- spectively; df = 8, 122; P < 0.0001). CV-I con- trasted lengths and maximal shaft widths of the ulna (Table 5). Mancalla had short, wide ulnae and hence low scores, Praemancalla and Pingui- nus were intermediate, and the three flighted genera and the fossil Australca had high scores which reflected their comparatively long, slen- der ulnae (Figs. 3B, 4). The vertical axis (CV-II) implied residual robustness of the ulnar shafts, which was greatest in the stout bones of Pin- guinus and progressively less so in Mancalla and the flighted genera (Fig. 3B). Sterna of the Great Auk, similar in overall conformation to that of Uria (Pinguinus illus- trated in Eyton 1875, Wiman and Hessland 1942; Uria in Kuroda 1954), were contrasted with those of three flighted species of alcid using a CA of five measurements, four of which entered the model significantly (P < 0.01; Table 5). No ster- num of the fossil species was preserved ade- quately for analysis. The resultant CA signifi- cantly differentiated the four species (Wilks' lambda = 0.003; df = 4, 3, 39; P < 0.001). CV-I for sterna essentially ordinated the taxa by gen- eral sternal size, exclusive of caudal width (Table 5). Taxa differed significantly on this axis (F = 864.1; df = 3, 39; P < 0.0001), and the low scores for Pinguinus on this axis reflect their large sterna (Fig. 3C). CV-II contrasted carina length with the remaining sternal dimensions (Table 5). Scores differed among taxa (F = 43.7; df = 3, 39; P < 0.0001), and indicated that Alca and Uria slightly exceeded Pinguinus and Cepphus in "relative carina length" (Fig. 3C). PCA of complete skeletons.--A modified PCA of mean vectors of 46 skeletal measurements Fig. 4. Illustrations of major wing elements of (A) Alca torda (Univ. S. Florida 4358), (B) Pinguinus impennis (U.S. Nat. Mus. 1285, Los Angeles County Mus. 90055, 90057), and (C) Mancalla diegensis (San Diego Nat. Hist. Mus. 24868, 21044, 28603, 25002); dorsal views of left elements in probable positions used in submarine propulsion with diagrams of mean intra-alar skeletal proportions (H = humerus, U = ulna, C = carpometa- carpus, M = proximal phalanx of digiti majoris, not illustrated). Note the comparatively great extension of distal elements (positions inferred from qualitative osteology) and dorsoventral flattening of alar elements in flightless Pinguinus and Mancalla; Mancalla is further derived in the proximal position of the processus supracondylaris dorsalis or "ectepicondylar process" (ep), shortening of the frewing, and curvature of the humerus. ep M C I U I H I 10 I 20 31 39 %1 B ep M I C I U I H %1 I 10 19 25 46 c ep iM i C I U I H 12 22 19 4 %1 I I cm TABLE 6. Correlation coefficients of 46 original skeletal variables with first (modified) principal component (PC-I*, see text), and the first and second principal components of the residual variance from PC-I* for 23 Recent alcids (PC-U*, PC-III*). Signs to right of coefficients for PC-III* indicate variables so correlated with residuals from PC-I* (Irl >- 0.20). PC- PC- PC- PC- PC- PC- Variable I* II* III* Variable I* II* III* Bill length 0.95 0.30 -0.48 Tibiotarsus length 0.98 -0.03 -0.52 - Cranium length 0.99 0.18 -0.48 LWM 0.97 -0.08 -0.54 - Height 0.95 0.07 -0.59 - Tarsometatarsus length 0.89 -0.11 -0.42 Width 0.96 0.0! -0.50 APW 0.96 -0.07 -0.54 - Humerus length 0.99 0.!5 -0.39 + LMW 0.94 -0.!! -0.6! - Head width 0.99 0.10 -0.39 + Digit III, Ph. ! length 0.96 -0.!3 -0.5! - LWM 0.98 0.08 -0.37 + Ph. 2 length 0.98 -0.08 -0.50 MWM 0.96 0.35 -0.51 - Ph. 3 length 0.95 -0.15 -0.55 - Radius length 0.92 -0.02 -0.15 + Scapula length 0.98 0.24 -0.44 LWM 0.98 0.!1 -0.33 + Blade width 0.9! 0.29 -0.61 - MWM 0.99 0.21 -0.52 - Coracold length 0.99 0.02 -0.50 Ulna length 0.92 -0.02 -0.!4 + Basal width 0.95 0.36 -0.41 LWM 0.99 0.!3 -0.37 + Sternal carina length 0.93 0.44 -0.44 MWM 0.99 0.!8 -0.46 + Basin length 0.98 0.20 -0.51 - Carpometacarpus Least width 0.98 0.2! -0.39 + length 0.95 0.06 -0.2! + Caudal width 0.95 0.05 -0.38 + APW 0.98 0.13 -0.35 + Carina depth 0.97 0.!4 -0.53 - DVW 0.98 0.!6 -0.33 + Furcula height 0.97 -0.06 -0.55 - Digit II, Ph. ! length 0.97 0.07 -0.29 + LWM 0.96 0.29 -0.46 Ph. ! MWM 0.97 -0.0! -0.33 + MWM 0.93 0.40 -0.47 Ph. 2length 0.95 -0.0! -0.26 + Synsacrum length 0.98 0.2! -0.50 - Femur length 0.96 -0.15 -0.54 - Interacetabular width 0.89 0.07 -0.40 Head width 0.97 -0.11 -0.57 - Eigenvalue a 3.90 0.!2 0.06 LWM 0.98 -0.04 -0.55 - Percentage of variance 9!.9 2.8 !.9 MWM 0.99 0.!2 -0.53 - a Eigenvalues and percentages of variance include contributions from all 23 species; variance shared between PC-I* and PC-II* attributed to PC-I*. (see methods) concisely summarized the multi- variate differences among Recent species (Table 6, Fig. 5). The first three components together incorporated 96.7% of the total interspecific skeletal variation. Loadings of variables on the first component (PC-I*) were all positive and of high magnitude (r -> 0.89, 38 r, -> 0.95), and identified PC-I* as a measure of "general skel- etal size" (Table 6). Scores on this component were highly correlated with mean body mass (r = 0.99 for extant species, r = 0.98 including estimate for Pinguinus), and the axis deviated from strict isometric size by only 8 ø . Small species (e.g. Aethia, Alle, Synthliboramphus) had low scores in PC-I*, whereas the largest taxa (Uria, Pin- guinus) had the highest scores (Fig. 5). The second axis (PC-II*) for complete skele- tons was a shape variable, characterized by loadings of varying magnitude and sign (Table 6). Bill length, maximal humeral shaft width, sternal carina length, and widths of the cora- coid and furcula had relatively large loadings, whereas dimensions of leg elements, especially lengths, had negative correlations with PC-II*. PC-II* contrasted bill length and the size of the pectoral girdle with the pelvic limb. Genera characterized by long bills and robust pectoral girdles relative to their leg measurements (e.g. Brachyramphus, subgenus Endomychura of Synthl- iboramphus) had high scores on this axis, and genera with opposite proportions (e.g. Aethia, Cyclorrhynchus) had low scores (Fig. 5). Pinguinus was moderately high in this dimension, as were Uria and Alca. The third modified component (PC-III*) sum- marized the unique morphometric character- istics of Pinguinus impennis relative to other Re- cent alcids, notably the skeletal correlates of flightlessness (Fig. 5). This axis contrasted least shaft widths and lengths (to a lesser extent) of wing elements, and widths of the sternum with cranial height, maximal shaft widths of the hu- merus and radius, scapular blade width, basin length and carina depth of the sternum, most dimensions of the pelvic limb, and synsacrum length (Table 6). The large negative score for Pinguinus revealed that, compared with flighted alcids, the Great Auk had flattened humeri and Fig. 5. Trivariate plot of mean scores of Recent species of alcid on first 3 (modified) principai components of 46 skeletal measurements; PC-III*, the axis which separates Pinguinus from flighted alcids, is explained in text. Flighted species are numbered as in Fig. 2. radii, was (slightly) shortened in the mid-wing and manus, and had comparatively large leg elements (except tarsometatarsus length), broad scapulae, long and narrow pelvises, and sterna that were long and narrow with relatively deep carina. Although none approached the extreme position of Pinguinus on PC-III*, the Dovekie (A. alle) and puffins (Cerorhinca, Fratercula) had the lowest scores on this axis of the flighted Alcidae (Fig. 5). PCA of reduced skeletal vectors.--A PCA of mean vectors for the reduced data sets for "me- dium-sized" Mancalla, M. milleri, Praemancalla spp., and 23 Recent species identified two major axes of skeletal variation. Together, these ac- counted for 96.1% of the interspecific dispersion in the reduced morphometric space (Table 7). As in the PCA of complete skeletal vectors, PC- I* of the reduced data set largely reflected "gen- eral skeletal size" (Table 7). Mean body mass was strongly correlated with scores on this com- ponent (r = 0.99 for flighted species, r = 0.98 including the estimate for Pinguinus), and the axis deviated from strict isometric size by only 5 ø. PC-I* indicated M. milleri was comparable in "general skeletal size" to Alca; the medium-sized Mancalla spp. were similar to Uria; and Prae- mancalla and Pinguinus were progressively larg- er (on this axis) than any flighted alcid (Fig. 6). The second component (PC-II*) contrasted lengths and least shaft widths of wing elements (especially humeri and radii) with maximal widths of wing elements, lengths and widths of leg elements, and synsacrum length (Table 7). This axis corresponded closely with PC-III* for complete skeletons (Table 6). The 4 flightless species or species-groups had extremely low scores on PC-II*, whereas the flighted alcids varied little in this dimension. The low scores of the flightless species reflect their relatively short, flattened wing elements, and their rela- tively long leg elements and long synsacra. Scores on PC-II* indicate that, of the flightless genera, Mancalla was most extreme in these pro- portions, Praemancalla was next most distinc- tive, and Pinguinus was the least modified in this dimension (Fig. 6). Estimated body masses of mancalline auks.- Regressions of total body mass on principal components of mean skeletal measurements for Recent alcids (including the estimated body mass of 5 kg for Pinguinus impennis) provided esti- mates of body masses for selected fossil man- calline species. For each taxon, principal com- TABLE 7. Correlation coefficients of 23 skeletal vari- ables with first (modified) principal component (PC- I*, see text) and the first principal component of the residual variance from PC-I* for Recent and fossil alcids (PC-II*). Signs to right of coefficients of PC-II* indicate variables so correlated with re- siduals from PC-I* (Irl > 0.20). Variable PC-I* PC-II* Humerus length 0.99 - 0.42 + Head width 0.99 -0.41 + LWM 0.98 -0.39 + MWM 0.94 -0.55 - Radius length 0.94 -0.14 + LWM 0.98 - 0.35 + MWM 0.98 - 0.55 - Ulna length 0.93 -0.14 + LWM 0.99 - 0.40 + MWM 0.99 - 0.48 Carpometacarpus length 0.96 -0.23 + APW 0.97 -0.35 DVW 0.98 -0.35 + Femur length 0.97 -0.51 - Head width 0.98 -0.56 - LWM 0.99 -0.57 - MWM 0.99 -0.57 - Tibiotarsus length 0.99 -0.54 - LWM 0.98 -0.56 - Tarsometatarsus length 0.91 -0.50 - APW 0.97 -0.56 - LMW 0.94 - 0.54 - Synsacrum length 0.97 -0.55 - Eigenvalue" 2.23 0.22 Percentage of variance 87.6 8.5 "Eigenvalues and percentages of variances include contributions from all 26 species. Variance shared between PC-I* and PC-II* attributed to PC-I*; r (PC-I*, PC-II*) = 0.48. ponents of available skeletal measurements which were significantly correlated (P < 0.05) with body masses were used as estimator vari- ables. However, unlike the rough "size" com- parisons permitted by PC-I* for the reduced skeletal data set (Fig. 6), these regression esti- mates also incorporated "shape" variables that distinguished the larger flightless alcids from flighted confamilials (Table 8). My estimates in- dicate that the mancalline auks were more mas- sive than the flighted alcids and less than P. impennis, varying between ! and 4 kg in total mass. An estimate was not attempted for the very large, as yet undescribed species of Man- calla (Fig. 3A). Furthermore, the estimates con- firmed that M. milleri was the smallest of the group, followed by the "medium-sized" Man- calla and Praemancalla, and the largest was the relatively poorly represented M. emlongi (Table 8). small. "general skeletal size"-- -20 25 Fig. 6. Plot of mean scores of Recent and selected fossil alcids on first 2 (modified) principal compo- nents of 23 skeletal measurements. Flighted species are numbered as in Fig. 2; "medium" Mancalla rep- resents mean vector for pooled data of similarly sized M. californiensis, M. cedrosensis, and M. diegensis. DISCUSSION Locomotor compromise and convergence.--The Alcidae are morphologically committed to a largely aquatic existence. The compromise in wing shape necessary for wing-propelled div- ing is substantial (Storer 1960, 1971; Pennycuick 1975, 1987), and in flighted alcids is associated with their comparatively heavy wing-loadings (Greenewalt 1962; this study) and high wing- beat frequencies (Meinertzhagen 1955). Wing- loadings of tourres (Uria) are moderately high, but do not approach closely the threshold of flightlessness of 2.5 g.cm (Meunier 1951). Us- ing available wing areas, allometric extrapola- tions of wing-loadings of flighted alcids to greater body masses yielded estimates of 2.10 TABLE 8. Estimated body masses for fossil alcids, based on regressions of body masses of 23 Recent species on principal components of (p) skeletal measure- ments available for each fossil taxon. R 2 nents Species Mas1/2 p (%) entered Mancalla (inter- mediate) 2,400 38 99.1 I, III, IV M. milleri 1,650 29 99.2 I, III, IV M. emlongi 3,800 24 98.9 I, IV, V Praemancalla spp. 3,050 23 98.8 I, III, IV Included estimated mass for Pinguinus impennis (5,000 g, see text). Estimates rounded to nearest 50 g. Pooled for similarly sized M. californiensis, M. diegensis, and M. ced- rosensts. at 1,500 g, 2.33 at 2,000 g, and 2.53 at 2,500 g. Allometric enlargement probably would result in impaired flight, and might compromise the wing-body proportions optimal for submarine propulsion (Bdard 1969). The mean area of two half-folded wings (the position used for diving; Spring 1971) of Pinguinus was 77 cm 2. Doubled and divided into the estimated 5-kg body mass, this yields a "flipper-loading" of 32 g.cm 2, a value comparable to those of the spheniscid genera Megadyptes, Pygoscelis, and some Sphenis- cus (Stonehouse 1967). This reduction in wing area, and other morphological characteristics of the flightless alcids, are aptly termed "special- izations." They are of obvious adaptive signif- icance for one function (diving), limiting with respect to another (aerial flight), and uniquely associated with flightlessness, unquestionably a derived condition in carinate birds. Parallels have been drawn between the div- ing petrels (Pelecanoididae) and flighted alcids (Verheyen 1958, Kuroda 1967, Harrison 1977), and between the flightless alcids and the pen- guins (Storer 1960, 1971; Pennycuick 1975; Fed- uccia 1980; Sparks and Soper 1987) and the fossil Plotopteridae (Olson and Hasegawa 1979, Ol- son 1980). Despite these notable convergences in pectoral anatomy, obvious differences in lo- comotor adaptations and diagnostic osteologi- cal synapomorphies demonstrate their homo- plasious nature (e.g. Harrison 1977), the predictions by Olson (1980) concerning the fal- libility of a cladistic analysis of such groups notwithstanding. Particularly manifest are the numerous and extreme morphological novel- ties of the penguins, which include signifi- cantly reduced mobility of the wing articula- tions and radical modifications of the remiges (Coues 1872, Owen 1879, Sparks and Soper 1987, Raikow et al. 1988). Alcine vs. mancalline fiightlessness.--Members of different suprageneric groups within the A1- cidae, Pinguinus (Tribe Alcini) and the subfamily Mancallinae share several morphological cor- relates of flightlessness. Those include large size (Tables 1, 2, 8), relatively short wings (Figs. 1, 2), and dorsoventrally flattened skeletal wing elements (Figs. 3-6). These characteristics were among the first to be recognized, largely be- cause of similarities to the anatomy of penguins (Newton 1861, Owen 1879, Lucas 1901). Raikow et al. (1988) concluded that in flighted, wing- propelled diving birds the functional demands of aerial flight preclude the skeletal and myo- logical modifications observed in flightless al- cids and penguins. There also is osteological evidence that the remiges of Mancalla were at least as reduced as those of Pinguinus (Miller and Howard 1949). Other unusual characters of Pin- guinus--sequential molt of the primary remiges (Storer 1960), comparatively extensive fusion and short transverse processes of vertebrae (Storer 1945), and a large number of caudal ver- tebrae (Owen 1879)--may also have been con- vergent with the mancallines, but currently available material does not permit such infer- ences. Contrary to the reference to "... degen- eration of the wing and keel ..." in Pinguinus by Greenway (1967), the flightless auks had ro- bust wing elements and deep sternal carinae. Pinguinus was unique among the Alcidae for its large size, exceeding the described mancal- lines in skeletal dimensions (Tables 2, 3; Fig. 6) and in estimated body mass (Table 8). The man- callines were more specialized than Pinguinus in morphometric shape characters (Figs. 3, 4, 6), and had wing elements that more closely ap- proached those of the penguins in relative length, flatness, curvature, and articulative ri- gidity (Lucas 1901, Miller 1933, Wiman and Hessland 1942, Miller and Howard 1949). The exceptional modification of the wing for sub- marine propulsion in Mancalla is indicated by the degree of shaft compression of major ele- ments, well developed sulcus musculus scapu- lotrioipitis, the proximal position of the hu- meral processus supracondylaris dorsalis, and by the obsolete processus pisiformis, distally extended processus extensorius, and dorsoven- tral compression of the trochlea carpalls of the carpometacarpus (Fig. 4; Howard 1966). There is also significant qualitative variation in skel- etons within Mancalla (Howard 1970). The geo- logically older Praemancalla, as confirmed by its morphometric intermediacy (Figs. 3, 6), was not as specialized as Mancalla osteologically (How- ard 1966, 1976, 1978; Howard and Barnes 1987). The poorly known mancalline genus Alcodes, described by Howard (1968: 19) as "progressing towards flightlessness," appears on the basis of its ulnar proportions to have been completely flightless (Fig. 3B). Ontogenetic considerations.--The likely impor- tance of heterochrony, specifically neoteny, in the evolution of avian flightlessness has been suggested (Lowe 1928, Olson 1973, James and Olson 1983) largely on the basis of morpholog- ical similarities between the wing and pectoral girdle of adults of flightless species and juve- niles of flighted relatives. However, the ob- vious locomotor specializations of Pinguinus and the Spheniscidae led Olson (1973: 31) to rec- ommend that these groups "... not be included in discussions of flightlessness." Olson (1977: 690) also stated: "The great modifications seen in the wing of Pinguinis [sic] are not the result of neoteny, as seen in many other flightless birds .... Instead, these modifications repre- sent highly derived specializations for wing- propelled diving." Implicit in this conclusion are the assump- tions that "neotenic" flightlessness is necessarily "degenerate," pervasive in its impact on the pectoral girdle, and not associated with loco- motor specialization. Although the compres- sion and curvature of wing elements cannot be attributed to neoteny because neither charac- terizes developing alcids, the relatively short distal wing elements of flightless auks resemble the intra-alar proportions of embryonic alcids and other nonpasserines (B6ker 1927, Marpies 1930). Hence, they are by definition paedo- morphic (Gould 1977). Definitive support for heterochrony in the extinct auks is probably unattainable, but Livezey and Humphrey (1986) presented evidence of a heterochronic basis for flightlessness in the pectorally "nondegener- ate" steamer-ducks (Anatidae: Tachyeres). Ecological implications.--The Great Auk is probably related most closely to Alca and Uria (Strauch 1985), and all three genera are consid- ered to be comparatively specialized for pisciv- ory (Storer 1945, Bdard 1969, Hudson et al. 1969). Although Pinguinus is morphologically specialized (Figs. 2, 3, 5), the genus and its close relatives are characterized by only moderate proportions on several major axes of skeletal variation in the Alcidae (Figs. 2, 5). Pinguinus had only moderate scores on PC-II* for com- plete skeletons (Fig. 5), an axis with profound locomotory and ecological implications and which reflects, in part, the patterns in pelvic proportions (Storer 1945). Hudson et al. (1969) concluded that the close relatives of Pinguinus (Alca and Uria) were myologically "special- ized," whereas the puffins were "primitive." The myological details of Pinguinus will in all probability never be known; although it seems likely that the genus was at least as specialized in its pectoral musculature as its close relatives (Miller and Howard 1949). Probably the most conspicuous characteristic shared by Pinguinus, Alca, and Uria is compar- atively large size (Figs. 1, 2, 5). Large body mass is an advantage for diving birds, especially ma- rine pursuit-divers, because it reduces buoy- ancy and makes available a greater range of water depths for foraging (Sparks and Soper 1987). A direct relationship between body mass and maximal diving depth was documented in alcids (Piatt and Nettleship 1985), and the div- ing ability of Uria appears to be comparable to that of medium-sized penguins (Burger and Simpson 1986). Avian flightlessness generally is associated with large, often absolute, increases in body size (Pennycuick 1975), and it is probable that the exceptionally large size of flightless alcids represents an adaptive body form for subma- rine foraging (Storer 1960, Bdard 1969). Al- though Olson et al. (1979) presented evidence that breeding Great Auks at Funk Island, New- foundland, may have fished primarily in water less than 18 m deep, it is agreed generally that Pinguinus typically foraged in deeper waters (Bradstreet and Brown 1985, Brown 1985). A related advantage of large body size is the abil- ity to capture and swallow larger prey, also in- ferred for Pinguinus (Bradstreet and Brown 1985). The relatively large contribution of culmen length to the size-related first canonical variate (Table 2, Fig. 2) underscores the important re- lationship between body size and the size of the feeding apparatus in alcids (Bdard 1969). Although sexual differences in skeletal mea- surements were negligible in extant species of alcid, samples of Pinguinus from Funk Island indicated bimodality in sample distributions of several measurements, especially bill length. The larger samples (n = 200) of Pinguinus mea- sured by Lucas (1890) suggest that sexual dif- ferences in length of the femur were present, although Lucas (1890: 523) concluded other- wise. The likelihood of sexual dimorphism of bill length in Pinguinus is enhanced by the im- portance of bill size in the feeding niches of alcids (Bdard 1969), wherein sexual differences in bill size may reflect intersexual niche differ- ences. Increased sexual dimorphism in flight- less species characterizes at least one other fam- ily of diving birds, the grebes (Podicipedidae; Livezey in press). Another benefit of large body size is a ther- modynamically efficient surface: volume ratio (Calder 1974, Sparks and Soper 1987). This ad- vantage would be greater in colder waters at high latitudes, and probably contributed to the extreme size of Pinguinus compared with the mancallines. Furthermore, the larger body size of Pinguinus may have compensated, in part, for its only moderate skeletal specializations, ren- dering it comparable in diving ability to the osteologically more extreme Mancalla. In spite of these advantages of large size and specialized wing morphology, the resultant flightlessness imposed significant ecological constraints on the Great Auk. Inability to fly undoubtedly limited the foraging radius of adults during nesting and its large size proba- bly increased incubation and developmental periods, thus making the species more vulner- able to climatic variations in the lengths of breeding seasons (Bengtson 1984). Even more important was the requirement for nesting sites that were free from terrestrial predators, suffi- ciently near rich food supplies, and accessible to flightless birds (Bengtson 1984, Harris and Birkhead 1985). Similar nesting habitats were inferred for the mancallines (Miller and How- ard 1949). In addition to rendering Pinguinus more vulnerable to human exploitation, these requirements, in combination with other breed- ing constraints and long-term fluctuations in climate in the North Atlantic, may have pre- disposed the Great Auk to a natural decline (Bengtson 1984). ACKNOWLEDGMENTS This research was supported by National Science Foundation grant BSR-8516623, and by collection study grants from the American Museum of Natural History and the U.S. National Museum of Natural History. I thank H. Levenson and G. Mack for their hospitality, and I appreciate the assistance and insights offered by R. W. Storer and R. M. Chandler. I am grateful to the curatorial personnel of the following institutions for permitting access to collections in their care: American Museum of Natural History, New York; U.S. National Museum of Natural History, Washing- ton, D.C.; Museum of Zoology, University of Michi- gan, Ann Arbor; Field Museum of Natural History, Chicago; Museum of Vertebrate Zoology and Mu- seum of Vertebrate Paleontology, University of Cal- ifornia, Berkeley; San Diego Natural History Mu- seum; Los Angeles County Museum of Natural History; British Museum (Natural History), Tring, U.K.; Zoo- logical Museum, University of Wisconsin, Madison; Royal Ontario Museum, Toronto; Museum of Com- parative Zoology, Harvard University, Cambridge; and Peabody Museum of Natural History, Yale Univer- sity, New Haven. Loans of specimens were arranged by: Department of Biology, University of South Flor- ida, Tampa; Museum of Natural History, University of Connecticut, Storrs; and the Department of Biol- ogy, University of California, Los Angeles. Data on specimens of the Great Auk were provided by col- leagues from the following institutions: Staatliches Museum fir Naturkunde, Stuttgart, B. R. D.; Zoolo- gisk Museum, Copenhagen, Denmark; Muses de Metz, Metz, France; Museum d'Histoire Naturelie, Autun, France; Instituto di Zoologia, Universit& di Bologna, Italy; Zoologiska Museet, Lund, Sweden; Muse d'Histoire Naturelie, Neuch&tel, Switzerland; Museum National d'Histoire Naturelie, N&ntes, France; Staatliche Museen fiir Tierkunde und Volk- erkunde, Dresden, East Germany; Naturhistoriska Rijksmuseet, Stockholm, Sweden; Zoologisches Mu- seum der Universitat, Oslo, Norway; Rijksmuseum van Natuurlijke Historie, Leiden, The Netherlands; Muse Zoologique, Strasbourg, France; Landesmu- seum Joanneum, Graz, Austria. P.S. Humphrey, R. M. Chandler, R. J. Raikow, D. S. Wood, J. Vanden Berge, A. H. Bledsoe, and R. W. Storer made com- ments on the manuscript. M. Jenkinson and R. M. Mengel provided work space and access to specimens at the University of Kansas, and K. Corbin and M. Schmalz typed the manuscript. LITERATURE CITED BALDWIN, $. P., H. C. OBERHOLSER, & L. G. WORLEY. 1931. Measurements of birds. Cleveland Mus. Nat. Hist. Sci. Publ. 2: 1-143. BAUMEL, J. J. (ED). 1979. Nomina anatomica avium. New York, Acad. Press. BDAD, J. H. 1967. Ecological segregation among plankton-feeding alcidae (Aethia and Cyclorrhyn- chus). Ph.D. dissertation, Univ. Brit. Columbia. --. 1969. Adaptive radiation in Alcidae. Ibis 111: 189-198. BELOPOL'$KII, L.O. 1961. Ecology of sea colony birds of the Barents Sea. Jerusalem, Israel Progr. Sci. Transl. BENGTSON, S.-A. 1984. Breeding ecology and extinc- tion of the Great Auk (Pinguinus impennis): anec- dotal evidence and conjectures. Auk 101: 1-12. BENT, A.C. 1919. Life histories of North American diving birds. U.S. Nat. Mus. Bull. 107. BI,ANC, G.A. 1927. Sulla presenza di Alca impennis Linn. nella formatione superiori di Grotta Ro- manelli in Terra d'Otranto. Arch. Antropol. Eth- nol. (Firenze) 58: 155-186. BKER, H. 1927. Die biologische Anatomie der Flug- arten der VOgel und ihre Phylogenie. J. Ornithol. 75: 304-371. BRADSTREET, M. S. W., & R. G. B. BROWN. 1985. Feed- ing ecology of the Atlantic Alcidae. Pp. 263-318 in The Atlantic Alcidae (D. N. Nettleship and T. R. Birkhead, Eds.). New York, Acad. Press. BRODKORn, P. 1955. The avifauna of the Bone Valley formation. Florida Geol. Surv. Rep. Invest. No. 14: 1-57. BROWN, R. G. B. 1985. The Atlantic Alcidae at sea. Pp. 383-426 in The Atlantic Alcidae (D. N. Net- tleship and T. R. Birkhead, Eds.). New York, Acad. Press. BURGER, A. E., & M. SIMPSONß 1986. Diving depths of Atlantic Puffins and Common Murres. Auk 103: 828-830. CALDER, W. A., IlL 1974. Consequences of body size for avian energetics. Pp. 86-144 in Avian ener- getics (R. A. Paynter Jr., Ed.). Publ. Nuttall Or- nithol. Club No. 15. CLARK, G. A., JR. 1979. Body weights of birds: a re- view. Condor 81: 193-202. CLARK, R.J. 1971. Wing-loading--a plea for consis- tency in usage. Auk 88: 927-928. CouPS, E. 1868. Monograph of the Alcidae. Proc. Acad. Nat. Sci. Philadelphia 20: 1-81. 1872. Material for a monograph of the Spheniscidae. Proc. Acad. Nat. Sci. Philadelphia 1872: 170-212. DEMENT'/V, G. P., & N. A. GLADKOV, EDS. 1968. Birds of the Soviet Union, vol. II. Jerusalem, Israel Prog. Sci. Transl. DIxoN, W. J., CHIEF ED. 1985. BMDP statistical soft- ware. Berkeley, Univ. California Press. DUNNING, J. B., JR. 1984. Bird weights of 686 species of North American birds. Western Bird Banding Assoc. Monogr. No. 1. EYTON, T.C. 1875. Osteologia avium: a sketch of the osteology of birds. Supp. II, pt. 3. London, Wil- liams and Norgate. FEDUCCIA, A. 1980. The age of birds. Cambridge, Harvard Univ. Press. FEILDEN, H. W. 1872. Birds of the Faeroe Islands. Zoologist 7(Ser. 2): 3277-3294. FORhUSH, E. H. 1912. Game birds, with fowl and shore birds of Massachusetts and adjacent states. Massachusetts State Bd. Agric. GOULD, S.J. 1966. Allometry and size in ontogeny and phylogeny. Cambridge Biol. Rev. 41: 587- 640. ß 1977. Ontogenyandphylogeny. Cambridge, Belknap. GREENEWALT, C.H. 1962. Dimensional relationships for flying animals. Smithsonian Misc. Collect. 144: 1-46. GREENWAY, J.C.,JR. 1967. Extinct and vanishing birds of the world, 2nd ed. New York, Dover. GRIEVE, S. 1885. The Great Auk, or garefowl (Alca impennis, Linn.), its history, archaeology, and re- mains. London, Thomas C. Jack. HARRIS, M.P., & T. R. BIRKHEAD. 1985. Breeding ecology of the Atlantic Alcidae. Pp. 155-204 in The Atlantic Alcidae (D. N. Nettleship and T. R. Birkhead, Eds.). New York, Acad. Press. HARRISON, C. J.O. 1977. The limb osteology of the diving petrels and the Little Auk as evidence of the retention of characters in morphologically convergent species. Ardea 65: 43-52. HOWARD, H. 1947. California's flightless birds. Los Angeles County Mus. Quart. 6: 7-11. ß 1966. A possible ancestor of the Lucas auk (Family Mancallidae) from the Tertiary of Orange County, California. Los Angeles County Mus. Contrib. Sci. No. 101: 1-8. ß 1968. Tertiary birds from Laguna Hills, Or- ange County, California. Los Angeles County Mus. Contr. Sci. No. 142: 1-21. 1970. A review of the extinct avian genus Mancalla. Los Angeles County Mus. Contrib. Sci. No. 203: 1-12. 1971. Pliocene avian remains from Baja Cal- ifornia. Los Angeles County Mus. Contrib. Sci. No. 217: 1-17. 1976. A new species of flightless auk from the Miocene of California (Alcidae: Mancallinae). Smithsonian Contrib. Paleobiol. 27: 141-146. 1978. Late Miocene marine birds from Or- ange County, California. Los Angeles County Mus. Contrib. Sci. No. 290: 1-26. ß 1982. Fossil birds from Tertiary marine beds at Oceanside, San Diego County, California, with descriptions of two new species of the genera Uria and Cepphus (Aves: Alcidae). Los Angeles County Mus. Contrib. Sci. No. 341: 1-15. 1983. A list of the extinct fossil birds of California. Bull. S. California Acad. Sci. No. 82: 1-11. ., & L. G. BARNES. 1987. Middle Miocene ma- rine birds from the foothills of the Santa Ana Mountains, Orange County, California. Los An- geles County Mus. Contrib. Sci. No. 383: 1-9. HUDSON, G. E., K. M. HOFF, J. VANDEN BERGE, & E. C. TRIVETTE. 1969. A numerical study of the wing and leg muscles of Lari and Alcae. Ibis 111: 459- 524. JAMES, H. F., & S. L. OLSON. 1983. Flightless birds. Nat. Hist. 92: 30-40. JOHNSON, R. A. 1935. Additional Dovekie weights. Auk 52: 309. ß 1944. Weight records for some Atlantic AI- cidae. Wilson Bull. 56: 161-168. KELSO, J.E. H. 1922. Birds using their wings as a means of propulsion under water. Auk 39: 426- 428. KURODA, N. 1954. On some osteological and ana- tomical characters of Japanese Alcidae. Jap. J. Zool. 11: 311-327. 1967. Morpho-anatomical analysis of par- allel evolution between diving petrel and ancient auk, with comparative osteological data of other species. Yamashina Inst. Ornithol. Zool. Misc. Rep. 5: 111-137. LEY, W. 1935. The Great Auk. Nat. Hist. 36: 351-357. LIVEZ, B.C. In press. Flightlessness in grebes (Aves, Podicipedidae): its independent evolution in three genera. Evolution. ---, & P.S. HUMPHREY. 1984. Sexual dimorphism in continental steamer-ducks. Condor 86: 368- 377. ß & --. 1986. Flightlessness in steamer- ducks (Anatidae: Tachyeres): its morphological bases and probable evolution. Evolution 40: 540- 558. LOWE, P. R. 1928. A description of Atlantisia rogersi, the diminutive and flightless rail of Inaccessible Island (southern Atlantic), with some notes on flightless rails. Ibis 70: 99-131. LtCAS, F.A. 1890. The expedition to Funk Island, with observations upon the history and anatomy of the Great Auk. U.S. Nat. Mus. Rep. 1887-1888: 493-529. ß 1901. Aflightlessauk, Mancallacaliforniensis, from the Miocene of California. Proc. U.S. Nat. Mus. 24: 133-134. MAGNAN, M.A. 1912. Modifications organiques con- sctives chez les oiseaux l'absence de vol. Bull. Mus. Hist. Nat. Paris 18: 524-530. 1922. Les caracteristiques des oiseaux sui- vant le mode de vol. Ann. Sci. Nat. (Set. 10) 5: 125-334ß MARPLES, B.J. 1930. The proportions of birds' wings and their changes during development. Proc. Zool. Soc. London 1930: 997-1008. MEINERTZHAGEN, g. 1955. The speed and altitude of bird flight (with notes on other animals). Ibis 97: 81-117. MEUNIER, K. 1951. Korrelation und Umkonstruktion in den Gr6ssenbeziehungen zwischen Vogelflu- gel and Vogelk6rper. Biologia Generalis 19: 403- 443. MILLER, L. 1933ß The Lucas auk of Californiaß Condor 35: 34-35. ß & H. HOWARD. 1949. The flightless Pliocene bird Mancalla. Carnegie Inst. Washington Publ. 584: 201-228. MURRAY, K. G., K. WINNETT-MURRAY, Z. A. EPPLEY, G. L. HUNT, & D. B. SCHWARTZ. 1983. Breeding bi- ology of the Xanthus' Murrelet. Condor 85: 12- 21. NETTLESHIP, D. N., & P. G. H. EVANSß 1985. Distri- bution and status of the Atlantic Alcidae. Pp. 53- 154 in The Atlantic Alcidae (D. N. Nettleship and T. R. Birkhead, Eds.). New Yorkß Acad. Press. NEWTON, A. 1861. Abstract of J. Wolley's researches in Iceland respecting the gate fowl or Great Auk. Ibis 3: 374-399. OLSONß S.L. 1973. Evolution of the rails of the South Atlantic islands (Aves: Rallidae). Smithsonian Contrib. Zool. No. 152: 1-53. 1977. A great auk, Pinguinis [sic], from the Pliocene of North Carolina (Aves: Alcidae). Proc. Biol. Soc. Washington 90: 690-697. 1980. A new genus of penguin-like pele- caniform bird from the Oligocene of Washington (Pelecaniformes: Plotopteridae). Los Angeles County Mus. Contr. Sci. No. 330: 51-57. 1981. A third species of Mancalla from the late Pliocene San Diego formation of California (Aves: Alcidae). J. Vert. Paleontol. 1: 97-99. , & Y. HASEGAWA. 1979. Fossil counterparts of giant penguins from the North Pacific. Science 206: 688-689. --, C. C. Swrt, & C. MOKHmER. 1979. An attempt to determine the prey of the Great Auk. Auk 96: 790-792. OWEN, R. 1879. Memoirs on the extinct wingless birds of New Zealand with an appendix on those of Englandß Australia, Newfoundlandß Mauritius, and Rodriquez, vol. II. Londonß John van Voorst. PENNYCUICK, C.J. 1975ß Mechanics of flight. Pp. 1- 75 in Avian biology, vol. 5 (D. S. Farrier and J. R. Kingß Eds.). New Yorkß Acad. Press. 1987. Flight of auks (Alcidae) and other northern seabirds compared with southern Pro- cellariiformes: orthithodolite observations. J. Exp. Biol. 128: 335-347. PLATT, J. F., & D. N. NETTLESHIP. 1985ß Diving depths of four alcids. Auk 102: 293-297. PImENTE, R. A. 1979. Morphometrics: the multi- variate analysis of biological data. Dubuque, Ken- dall/Hunt. POOLE, E.L. 1938. Weights and wing areas in North American birds. Auk 55: 511-517. RAncow, R.J. 1973. Locomotor mechanisms in North American ducks. Wilson Bull. 85: 295-307. ß L. BICANOVSKY, & A. H. BLEDSOE. 1988. Fore- limb and joint mobility and the evolution of wing- propelled diving in birds. Auk 105: 446-451. SEALY, S. G. 1976. Biology of nesting Ancient Mur- relets. Condor 78: 294-306. SPARKS, J., & T. SOPER. 1987. Penguins. New York, Facts on File. SPRING, L. 1971. A comparison of functional and morphological adaptations in the Common Murre (Uria aalge) and Thick-billed Murre (Uria lornvia). Condor 73: 1-27. STEMPNIEWICZ, L. 1982. Body proportions in adults and fledglings of the Little Auk. Acta Zool. Cra- coviensa 26: 149-158. STONEHOUSEß B. 1967. The general biology and ther- mal balances of penguins. Adv. Ecol. Res. 4: 131- 196. STORER, R.W. 1945. Strvctural modifications in the hind limb in the Alcidae. Ibis 87: 433-456. 1952. A comparison of variation, behaviorß and evolution in the sea bird genera Uria and Cepphus. Univ. California Publ. Zool. 52: 121-222. 1960. Evolution in the diving birds. Proc. Int. Ornithol. Congr. 12: 694-707. 1971. Adaptive radiation of birds. Pp. 150- 188 in Avian biology, vol. 1 (D. S. Farner and J. R. King, Eds.). New York, Acad. Press. STRAUCH, J. G., JR. 1985. The phylogeny of the A1- cidae. Auk 102: 520-539. THREEFALL, W., & S. P. MAHONEY. 1980. The use of measurements in sexing Common Murres from Newfoundland. Wilson Bull. 92: 266-268. TOWNSEND, C.W. 1909. The use of wings and feet by diving birds. Auk 26: 234-248. VERHEYEN, R. 1958. Contribution la systematique des Alciformes. Bull. Inst. Royal Sci. Nat. Bel- gique 34: 1-15. VERMEER, K., & L. CULLEN. 1982. Growth comparison of a plankton- and a fish-eating alcid. Murrelet 63: 34-39. WIGLESWORTH, J. 1900. Inaugural address on flight- less birds. Trans. Liverpool Biol. Soc. 14: 1-33. WtMAN, C., & I. HESSLAND. 1942. On the garefowl, Alca impennis L., and the sternum of birds. Nov. Acta Reg. Soc. Sci. Uppsaliensis (Ser. 4) 13: 1-28.