sign in sign up
<<
Home , Anhinga, Booby, Cormorant, Frigatebird, Gannet, Great cormorant

PHYLOGENETIC PATTERNS OF SIZE AND SHAPE OF THE NASAL GLAND DEPRESSION IN PHALACROCORACIDAE

DOUGLAS SIEGEL-CAUSEY Museumof NaturalHistory and Department of Systematicsand , University of Kansas, Lawrence, Kansas 66045-2454 USA

ABSTRACT.--Nasalglands in Pelecaniformesare situatedwithin the orbit in closelyfitting depressions.Generally, the depressionsare bilobedand small,but in Phalacrocoracidaethey are more diversein shapeand size. (Phalacrocoracinae) have small depressions typical of the ; shags(Leucocarboninae) have large, single-lobeddepressions that extend almost the entire length of the frontal. In all PhalacrocoracidaeI examined, shape of the nasalgland depressiondid not vary betweenfreshwater and marine populations.A general linear model detectedstrongly significant effectsof speciesidentity and gender on size of the gland depression.The effectof on size was complexand was detectedonly as a higher-ordereffect. Age had no effecton size or shapeof the nasalgland depression.I believe that habitat and diet are proximateeffects. The ultimate factorthat determinessize and shape of the nasalgland within Phalacrocoracidaeis phylogenetichistory. Received 28 February1989, accepted1 August1989.

THE FIRSTinvestigations of the nasal glands mon (e.g.Technau 1936, Zaks and Sokolova1961, of water indicated that theseglands were Thomson and Morley 1966), and only a few more developed in living in marine studies have focused on the cranial structure than in species living in freshwater associatedwith the nasal gland (Marpies 1932; habitats (Heinroth and Heinroth 1927, Marpies Bock 1958, 1963; Staaland 1967; Watson and Di- 1932). Schildmacher (1932), Technau (1936), and voky 1971; Lavery 1972). othersshowed that the degree of development Unlike most other birds, have among specieswas associatedwith habitat. Lat- nasal glands situated in depressionsfound in er experimental studies (reviewed by Holmes the anteromedialroof of the orbit (Siegel-Cau- and Phillips 1985) established the role of the sey 1988). In specieswith extra- or supraorbital nasal gland as an extrarenal excretory organ nasal glands, glandular tissue can enlarge by that maintainselectrolyte homeostasis and water growth outward from the skull (Owen and Kear balanceduring conditionsof salt-loading. 1972)and by increasein width (Bock1958), but The size of the nasalgland and its degree of rarely by increase in length (Staaland 1967, function varies alsowithin species.Marine pop- Peaker and Linzell 1975). Becausecranial bone ulations have larger glands and greater func- is membranous, mesenchymal, and nonintus- tional activity than do freshwater populations susceptive,and becauseit grows very slowly of a given species(Schmidt-Nielsen 1959, 1960, if at all after ossification and maturation (de 1965; Peaker and Linzell 1975). Experimental Beer 1937,Bellairs and Jenkin 1960),change in studies with captive wild and domestic birds bony outline of the gland depressionmust oc- indicate that gland size (i.e. mass)and function cur slowly. Thus, functionally inducedgrowth increasedquickly in individuals of somespecies of the nasal gland in the orbit is constrainedin when subjectedto high electrolytestress or low Pelecaniformesbecause substantial thickening water intake (Holmes et al. 1961, Ellis et al. 1963, of the gland without osteologicalaccommoda- Peaker and Linzell 1975). tion could distort the eye. In other species,accommodation to salt-load- I have shown qualitatively (Siegel-Causey ing is less effective (Skadhauge 1981). Devel- 1988) that the size and shape of the nasal gland opment and growth rate in captive Mallards depression is variable within shags and cor- (Anasplatyrhynchos) that were fed 1%solutions morants (Phalacrocoracidae).In this study, I of sodiumchloride were reducedcompared with surveyedthe variation in size and shape of the development and growth of birds fed fresh nasalgland depressionin the orbit of selected water (Schmidt-Nielsen and Kim 1964). Data members of this . I examined its pre- from field populationsof wild birds are uncom- sumptive associationwith gland size relative to

110 The 107: 110-118. January 1990 January1990] PhalacrocoracidaeNasal Glands 111

Fig. 1. Outlinesof the ventral view of the frontal bone and nasalgland depressionin selectedPelecani- formes.All views representthe area indicated in Figure 2: lower; shadedareas represent the nasal gland depression:(a) GreatFrigatebird; (b) White ;(c) Red-footedBooby; (d) Northern ;(e) ; (f) Little Pied ;(g) OlivaceousCormorant (left, freshwater;right, marine); (h) Double-crested Cormorant (left, freshwater;right, marine); (i) Pied Cormorant;(j) ;(k) Brandt'sCor- morant;(1) Black-facedCormorant; (m) SocotraShag; (n) CapeShag; (o) ,marine (left, e; right, •); (p) Imperial Shag,freshwater (left, e; right, •); (q) RockShag; (r) PelagicShag; (s) Red-leggedShag. collectionhabitat, age (measuredby size of bur- I measuredcranium length and frontal width by sa of Fabricius),size, and speciesidentity. dial calipersto a 95% replicated accuracy(RA) of 0.1 mm. The least frontal width was the minimum width

METHODS measuredon the frontal (Fig. 2: upper). Length and width of the bursa of Fabricius were measured in the All skeletal measurements were done on museum field to nearestmm with dial calipers,but determi- specimens.Scientific names and sample sizes of species nations of RA were not made. usedin this studyare in parentheses: I classified collection localities as freshwater and (Microcarboafricanus; 15), (M. marine (saltwater) habitats; ambiguous caseswere melanoleucos;21), Brandt's Cormorant (Compsohalieus classifiedas missing. Species identity, gender,and age penicillatus;4), Black-facedCormorant (C. fuscescens;were determined at the time of collection or, in the 2), OlivaceousCormorant (Hypoleucus olivaceus; 50), caseof museumspecimens, from the label. Unknown Double-crested Cormorant (H. auritus;57), Little Black genderwas coded as missing. I codedthose specimens Cormorant (H. sulcirostris;3), Pied Cormorant (H. var- without agedeterminations as adults unless there was ius; 4), (Phalacrocoraxcarbo; 11), So- evidence(e.g. partly ossifiedbone) to indicatejuve- cotra Shag (Leucocarbonigrogularis; 3), Cape Shag (L. nile status. capensis;4), Imperial Shag(Notocarbo atriceps; 12), Ant- I approximatedthe areaof the nasalglands in species arcticShag (N. bransfieldensis;19), (N. with fusedlobes by the productof the greatestlength verrucosus;2), South GeorgianShag (N. georgianus;2), and width of the nasalgland depressionin the cra- (Stictocarbo magellanicus; 73), PelagicShag nium. For specieswith bilobed nasal glands, I used (S. pelagicus;10), and Red-legged Shag (S. gaimardi; the product of the greatestlength of the medial lobe 30). See Siegel-Causey(1988) for details on system- and the greatestcombined width of both lobes.Mea- atics and nomenclature. surements(see Fig. 2: lower) were made by dial cal- For comparativepurposes, depression outlines for ipers(+ 0.2 mm RA). Theseapproximations may over- Great (Fregataminor; 2), American White estimatethe actualarea of the nasalgland depression, Pelican (Pelecanuserythrorhyncos; 4), Red-looted Boo- especiallyin cormorants,which have lessrectangular by (Sulasula; 2), (Morusbassanus; 2), depressionsthan do shags.All analyseswere done and Anhinga (Anhingaanhinga; 5) are illustrated(Fig. using log-transformeddata. 1). The nasalglands of (Phaethon spp.) do ! performedmultivariate analyses using the BMDP not occur within the orbit and are not illustrated. A statisticalprograms (Dixon [Ed.] 1988). Becauseof the list of specimensand data are available from the au- sensitivity of the analysisof covariance(ANCOVA) thor. program to small samples,I excluded cells with sam- 112 DOUGLAS$IEGEL-CAUSEY [Auk, Vol. 107

sion surfacesof all specimensI examined were composedof smooth, densebone. There was no osteologicalevidence of vascularization or in- complete ossificationas is seen in speciesthat have extra- or supraorbitalglands (e.g. Procel- lariiformes and Charadriiformes). Phalacrocoracidaeare unique within Pele- caniformes and show intrafamiliar variation in depression shape. Four qualitative characters varied within the cormorantsand shags(char- acters 12-15 in Siegel-Causey1988), but only the shape of the lateral edge of the depression Fig. 2. Generalizedphalacrocoracid cranium and showed intraspecificvariation. This feature is gland depression-relatedmeasurements. Upper: dor- sal view (LFW = least frontal width, CL = cranium influenced by the width of the frontal and the length).Lower: ventral view with palatinesremoved length of the cranium (e.g. the lateral curvature (rectanglerepresents view in Figure 1, DW = depres- increases as the least frontal width decreases sion "width," DL = depression"length"). relative to cranium length). This variation in lateral curvature was evident in Double-crested and Olivaceous cormorants collected from dif- ple sizesof <3. I excludedall caseswith missingdata ferent habitats (Fig. 1: g, h), but it was most for the variables under analysis. Equality of slopes pronounced in comparisonsbetween freshwa- (parallelism)of the regressionsof the dependentvari- ables on the covariatecranium length was testedby ter and marine Imperial Shags(Fig. 1: o, p). The ANOVA, and no significant differences (P < 0.05) relation between area of the gland depression were detected.All slopeswere significantly(P < 0.05) and frontal curvature differed in . different from zero. For Phalacrocoracinae,depression area did not correlatewith leastfrontal width, although oth- RESULTS er cranial measureswere significant (Double- crested Cormorant: cranium length, anterior Typesof nasalgland depression of the cranium.- frontal width, F = 14.87, df = 2, r = 0.596, P < Three types of depressionswere recognizable 0.01;Olivaceous Cormorant: cranium length and within the Pelecaniformes(Fig. 1). Most of these width, anterior frontal width, F = 10.93, df = were of two main categories,either bilobed or 3, r = 0.615, P < 0.01; Reed Cormorant: cranium single-lobed (fused). Macroscopicexamination length, anterior frontal width, F = 7.23, df = 2, of the nasal glands in spirit specimensof Dou- r = 0.802, P < 0.001; Little Pied Cormorant: ble-crestedCormorants, Rock Shags,and Pelag- cranium length, F = 5.56, df = 1, r = 0.426, P ic Shagsrevealed that they lie completelywith- < 0.05). For Leucocarboninae,depression area in the depression,they are flushwith the interior correlated most significantly with frontal cur- surfaceof the orbit, and they extend somewhat vature (Imperial Shag: cranium length, least anteriorly into the nasal cavity. The external frontal width, F = 78.17, df = 2, r = 0.771, P < morphology of the gland within the orbit is 0.001; Rock Shag: least frontal width, F = 6.17, reflectedprecisely by the shapeand size of the df = 1, r = 0.283, P < 0.01). nasal gland depression, and therefore gland Generalmodel.--The size of the nasal gland is shape and size can be studied in skeletal spec- correlated with body size (fig. 11.1 in Peaker imens. Only Leucocarboninae(shags) have sin- and Linzell 1975). For the analysesthat follow, gle-lobed (fused) nasal glands (Fig. 1: m-s); I used cranium length as the size covariatebe- Phalacrocoracinae (cormorants) and the re- causeit was highly correlated with limb length maining Pelecaniformeshave small bilobed and other estimates of size (P < 0.01 for hu- glands(Fig. 1: a-k). The Gal•pagos,Bank, and merus, ulna, femur, tibiotarsus length for all Black-facedcormorants have very broaddepres- species)and highly correlatedwith area of the sionsof unusual shape.The nasalgland depres- nasal gland depression and the least frontal sionsof these speciesare anteriorly bilobed as width (P < 0.01 for area and width for all in other cormorants(Fig. 1: 1), but the posterior species). edgesof the depressionsare scallopedand all Analysis of covariance (ANCOVA) revealed marginsare shallow and sloping. The depres- that the significantsimultaneous effects on area Janua•,y1990] PhalacrocoracidaeNasal Glands 113 of the nasalgland depressionamong five species (Imperial Shags and Double-crested, Oliva- ceous, Reed, and Little Pied cormorants) were speciesidentity (F = 452.14, df = 4 and 248, P < 0.001), gender (F = 9.73, df = 1 and 248, P = 0.002), and habitat. The significant effects of habitat were complex and in interaction with speciesidentity alone (F = 13.91, df = 4 and 248, P < 0.001) and with speciesidentity and gender in a second-orderinteraction (F = 3.23, df = 4 and 248, P = 0.013). No other interaction or variables (i.e. age class)had any significant effects. The results were similar for least frontal width except that no second-orderinteraction was significant (species:F = 4.55, df = 4 and 248, P = 0.014;gender: F = 7.47, df = 1 and 248, P = 0.007; species-habitat:F = 21.74, df = 4 and 248, P < 0.001). The underlying pattern of these effectscan be visualizedby plotting the regressionmeans of the area of the nasal gland depressionand least frontal width on cranium length for each categoryof birds(Fig. 3). The greatestdifference among groups for depressionarea and frontal width was speciesidentity. The next consistent pattern was sexuallydimorphic (i.e. femalesal- ways had smaller depressionareas and frontal \ widths than their male counterparts).Both of these results are predicted by the first-order •,.6 •8 4.0 4.2 ANCOVA effects. The relation between freshwater and marine forms within a speciesis more complicated.For Fig. 3. Habitat and gender patternsin depression nasalgland depressionarea, there appear to be area and frontal width in five speciesof phalacrocor- acids.Upper: relationof gland depressionarea to the two habitat-relatedpatterns (Fig. 3: upper).The covariate cranium length. Lower: relation of least first is that, depending upon the species,the frontal width to the covariatecranium length. Open mean regressionvalues of marine birds vary symbolsindicate mean values from freshwater hab- relative to their freshwater counterparts. For itats, closedsymbols are mean values from marine Reed, Little Pied, and Olivaceous cormorants, habitats. Speciesare indicated by name; genders are for example,the marine forms generally have marked. smaller nasal gland depressionsand longer heads,whereas the oppositeis true for Imperial Shags.Double-crested Cormorants have no ap- change in mean scorebetween freshwater and parent differences between populations. The marine groups varied with the species,but it secondpattern is related to the first but involves appearedsimilar for each sex within a species. a gender componentin addition. For example, In other words, there is only a first-order effect the change in mean score between freshwater between habitat and speciesidentity. Thesere- and marine Reed and Little Pied cormorant fe- suits are also predicted by the ANCOVA and, males was greater than the change in males, in addition to the patterns observedfor depres- and it was greater than the changein male and sion area, prompted a closerexamination of the female Double-crested Cormorants. These hab- effectsof habitat on depressionarea and frontal itat patternsare predicted by the first- and sec- width. ond-order ANCOVA effects. Habitat-relatedeffects.--Intraspecific compar- The patterns related to habitat for least fron- isons by habitat of cranial length, of the nasal tal width were simpler(Fig. 3: lower). Here, the gland depressionarea, and of leastfrontal width 114 DOUGLAS$IEGEL-CAusEY [Auk,Vol. 107

TABLE1. Comparisonof cranial measurements(œ +_ SD) between freshwaterand marine populationsa for six speciesof Phalacrocoracidae.

Cranium length Area of gland Leastfrontal Species Habitat n (ram) depression(mm 2) width (mm) Reed Cormorant Freshwater 7 41.03 + 1.55 12.64 + 4.33 11.06 + 1.61 Marine 8 42.84 + 1.37 12.75 + 0.98 8.20 + 0.42*** Little Pied Cormorant Freshwater 10 44.57 + 1.79 18.74 + 2.86 11.34 + 1.28 Marine 11 43.87 + 1.85 11.97 + 3.13'** 10.98 + 2.89 Olivaceous Cormorant Freshwater 24 52.25 + 1.31 34.78 + 6.66 13.00 + 1.24 Marine 26 56.34 + 3.43*** 27.36 + 5.28*** 13.68 + 2.09 Double-crested Cormorant Freshwater 32 61.21 + 2.06 38.69 + 7.57 14.59 + 1.61 Marine 38 58.32 + 3.75*** 38.94 + 10.76 13.78 + 1.50 European Cormorant Freshwater 4 59.65 + 3.37 40.64 + 4.56 14.70 + 1.78 Marine 5 63.75 + 2.47 42.88 + 2.55 14.96 + 1.77 Imperial Shag Freshwater 16 55.54 + 2.05 106.23 + 11.21 13.05 + 1.08 Marine 96 58.69 + 2.64*** 160.51 + 15.76'** 15.82 + 1.61'**

Significancelevels were assessedusing the sequentialBonferroni inequality (Holm 1967);*** = P < 0.001, two-tailed. revealedthat 5 of 6 species(not EuropeanCor- field measurements of the bursa of Fabricius. morant) demonstratedsignificant habitat-relat- Size of the bursa of Fabricius has been used as ed variation in at least one of these variables an index of age becauseits growth occursearly (Table 1). There were no differences in cranium in life, and its involution is completeby sexual length between freshwaterand marine popu- maturity (cf. Davis 1947,Johnson 1956, McNeil lations of European,Reed, and Little Pied cor- and Burton1972, Siegel-Causey 1989). In all cases morants. Imperial Shags and Olivaceous and in which I examined bursae, bursa area was Double-crestedcormorants all showedstrongly uncorrelatedwith body mass,cranium length, significantdifferences in cranium length be- or any other length measurement.Bursa area tween habitats. was similarly uncorrelatedwith area of the na- The areaof the nasalgland depressionvaried salgland depression or leastfrontal width when significantlybetween habitatswithin Imperial the log-linear effectsof cranium length were Shagsand Olivaceous and Little Pied cormo- removed. rants.Freshwater populations of the Olivaceous and Little Pied cormorantshad larger depres- DISCUSSION sionareas than did the marine populations.The source of this difference in Olivaceous Cor- Speciesidentity and genderwere highly sig- morants was the greater depressionwidth of nificanteffects that determinedsize of the gland freshwater birds (t = 6.68, df = 48, P < 0.001). depressionand leastfrontal width. Habitat (ma- The depressionlength, however,was greater in rine vs. freshwater) was significantonly as a marine birds (t = 2.43, df = 48, P < 0.05). As higher-order effect. Most other interactionef- expectedfrom the general model analysis,the fects were nonsignificant. Of these variables, least width of the frontal varied significantly habitatshowed the most complex pattern among between habitatsfor Imperial Shagsand Reed species.In general, marine cormorantshad nar- Cormorants(Table 1), but the patternswere dif- rower frontalsand smallernasal gland depres- ferent between subfamilies. Marine popula- sionsthan did freshwaterspecies, but for Im- tions of Reed Cormorants had narrow frontals, perial Shags,marine forms had wider frontals whereasmarine populationsof Imperial Shags and larger depressions.The gender-relatedef- had wider frontals.For the other speciesex- fectson the areaof nasalgland depressionsand amined, I found no differences in least frontal least frontal width were predictable. For all width between habitats. species,females had relatively smaller depres- Age-relatedeffects.--For most specimensused sions and frontals. in the ANCOVA, age determinationswere made These differencesdid not have an age com- from the specimenor its label;I found no sig- ponent independent of size, although young nificant effects. For selected Olivaceous Cor- birds had smaller depressionsand narrower morantsand Imperial, Rock, and Red-legged frontals than adults. Young are smaller than shags,I made indirect age determinationsby adults,and gland and frontal dimensionswere January1990] PhalacrocoracidaeNasal Glands 115

TABLE2. Nasal gland mass(œ + SD) for marine and freshwater populationsof cormorants. Data usedwith permissionof Thomsonand Morley (1966).Sample size is in parentheses.

Nasal gland mass(mg/100 g body mass) Species Freshwater Marine

EuropeanCormorant 5.5 + 2.30 (16) P < 0.001 14.4 ___1.70 (13) Little Pied Cormorant 13.3 --_4.34 (26) P < 0.001 26.6 + 6.32 (21) Little Black Cormorant 10.9 + 2.80 (20) 29.4 (1) Pied Cormorant -- 26.7 + 1.82 (3) Black-faced Cormorant -- 30.2 + 4.73 (30)

simply related to body size. Moreover,age as mines the shapeof the nasalgland in the orbit, measured by bursa size had no effect on the and not the reverse. Although in most other dimensionsof the nasalgland depression. If age birds gland size seems related to body size influencesdepression size, it must occurquite (Peaker and Linzell 1975), the pattern within early in life. In a comparablestudy on the cra- the Phalacrocoracidaeis different. For example, nial development in Short-tailed EuropeanCormorants, four timesthe body mass (Puffinustenuirostris), the relative sizeand shape of Little Pied Cormorants,had nasalglands that of nasalgland depressions were the same among weighed only 50% as much (see also Table 2). birds aged 1 month to 5 (Sugimori et al. Phylogenetic patterns of size and shape di- 1985). versity of the gland depressionwithin Phala- The strongesteffect on the area of the nasal crocoracidaedo not appearto be causedby en- gland depressionand least frontal width de- vironmental effects. This is in contrast to tected by the ANCOVA was speciesidentity. Thomson and Morley's (1966) hypothesisthat The greatest contrast was found between diet determined the ultimate size and shape of subfamilies(e.g. Double-crested Cormorant vs. the nasal gland in Australian cormorants. Imperial Shag),but I detectedsignificant effects Schmidt-Nielsen et al. (1958) showed experi- betweenmore closelyrelated species (e.g. Im- mentally that Double-crestedCormorants, like perialvs. Rock shags). Overall, cormorants have other marine birds, obtained sufficient water small,bilobed glands, whereas shags have large, from an exclusive diet of marine , which are fused glands.I observeda similar, corroborat- hypotonic to the environment. In field condi- ing pattern of cranial pneumatization within tions, ingestion of seawater is inevitable, and the family (Siegel-Causey1989). Shags, which the diet could include invertebrates, which are have large glands, lacked the posterofrontal isotonic.Thomson and Morley (1966)proposed pneumatization center present in all cormo- that, becausefeeding in a marine environment rants. inducessalt loading, nasal gland sizeshould be The relationshipbetween lateral curvatureof determined by dietary proportions of fish, in- the frontal and depressionarea also differed vertebrates,and freshwater. They reasonedthat between subfamilies. For cormorants, there was the nasalglands of EuropeanCormorants were no correlation between least frontal width and (relatively) the smallest(Table 2) becausethey areaof the nasalgland depression.In shags,the had free access to fresh water or freshwater in- relationbetween frontal width and gland area vertebrates.The nasal glands of Black-faced was strongly significant. Least frontal width Cormorants were (relatively) the largest be- varied between habitats in Reed Cormorants cause of their exclusive diet of marine fish. Wat- and Imperial Shags,but for reasonsyet unclear son and Divoky (1971) invoked this notion in- the patterns were different. In Reed Cormo- dependentlyto accountfor differing widths of rants, marine populations had narrower fron- gland furrows in the Diomedea. tals than freshwaterpopulations, but in Impe- My results do not support this hypothesis. rial Shags,the marine birds had wider frontals. First,small bilobed gland depressionsare a fea- Depressionshape reflected exactly the shape ture of most of the extant speciesof Pelecani- of the nasalgland within the orbit. Given the formes (including the fossil pseudodontorns differencein tissuelability to salt-loading(i.e. [Harrison and Walker 1976,Olson 1985]).Only bone vs. glandular tissue), it seemsclear that shags(Leucocarboninae) have large, fused gland shape of the depressionin adult birds deter- depressions,yet this feature is unlikely to be 116 DOUGLASSIEGEL-CAUSEY [Auk, VoL 107 related to diet because most (85%) cormorants ing body size in order to minimize the phys- (Phalacrocoracinae)and all shagsare marine, iologicaleffects of salt-loading,and minimizing and they eat marine food. Second, functional foragingenergy. Their analysisindicated that responsesof the nasalgland to salt-loadingap- differentspecies of ducksmay use various means parently do not affect depressionsize or shape to minimize salt-loading,and that possibleevo- as much as the phylogenetic patterns. All pre- lutionary strategiesmight include increased vious physiologicalstudies (e.g. Schmidt-Niel- body size and larger food items. Curi- sen 1960, Thomson and Morley 1966) measured ously,they did not considerthe effectof nasal the size and massof the nasalgland only in the gland size or what interspecific differences orbit (i.e. that in the nasal gland depression), might existin gland sizeor function.These and and thus thoseresults may be compareddirectly other weaknessesnotwithstanding, it is signif- with this osteologicalstudy. icant to my studythat they concludedthat salt- The contrastbetween phylogenetic and func- loading may act as a constrainton habitat se- tional patterns is even more striking in com- lection and food choice, and not the reverse. parisonsmade within or genus, e.g. None of the resultspresented here support a Brandt's(Fig. 2: k) vs. Black-facedcormorants phenomenon of rapid, temporary cranial ac- (Fig. 2: 1), Imperial (Fig. 2: o, p) vs. Rock (Fig. commodationof changingnasal gland dimen- 2: q) or Red-leggedshags (Fig. 2: r). Further- sions,as proposedin other groups(Bock 1958, more,the Imperial, Rockand Red-leggedshags 1963). In all casesI examined within Phalacro- coexistthroughout the marine littoral of Fuego- coracidae, bony margins of the nasal gland (Humphrey et al. 1970;Siegel-Causey depressionhad the appearanceof mature bone, 1986, 1987), but they have diets (Siegel-Causey even in partially ossifiedor pneumatizedcran- unpubl. data) contrary to what Thomson and ia. This implies that depressionsize in Phala- Morley's (1966) diet hypothesispredicts. Over- crocoracidaeis set early in life, but it is not clear all, RockShags eat a greaterproportion of crus- whether temporary functional accommodation taceansthan eitherImperial or Red-leggedshags to salt-loadingmight involve higher gland ac- eat, and Rock Shagsare not known to drink tivity, denser glandular tissue, or increased freshwater (Siegel-Causey1986), yet their gland gland massoutside of the orbit (i.e. within the depressionsare intermediate in size in relation nasals). Bock (1958, 1963) concluded that the to the other two species,both of which eat a width of the supraorbitalgland furrows in plo- greater proportion of fish. vers(Charadriinae) was due solely to habitat,but Thomsonand Morley's (1966)diet hypothesis his evidence was interspecific comparisons.In requiresthat speciesthat eat isotonicfood from contrast,Staaland (1967) hypothesizedthat [su- the sea (e.g. ,) must in- praorbital] gland size in Charadriiformeswas gest a compensatoryamount of fresh water or the ultimate factorin their ecology,because the hypotonic food. Otherwise, the relatively small greatest range of habitats and diets were. ex- glands of Europeanand Little Pied cormorants perienced by specieswith the largestglands. (which eat marine invertebrates in amounts up The perplexing pattern of smallernasal glands to 37%of the diet) and the relativelylarge glands in marine cormorantsmight be an artifact of of Black-facedCormorants (which eat only fish) (e.g. the marine forms may be sibling cannot be explained (see Thomson and Morley species),or it may indicatea decouplingof gland 1966: tables 3-5). There is no evidence that Eu- size from gland function. It is clear that ai- ropean or Little Pied cormorantsactively seek though depressionsize indicates size of the out "freshwater" sourcesof food while feeding gland in the orbit, it may reflect little of its in marine habitats (cf. Serventy 1938, Falla and potential physiological activity. The ultimate Stokell 1945, van Dobben 1952, Madsen and factor affecting the size of the nasal gland Sparck1950, McNalley 1957,Coulson 1961, and depressionin Pelecaniformesis not related to others). functional demand but insteadto phylogenetic Nystr6m and Pehrsson(1988) cameto similar history. conclusions using an energetic model. They predictedthat diving duckscould minimize salt ACKNOWLEDGMENTS intake by habitat selection (i.e. low salinity), diet (i.e. eating smaller musselsor switching to I thank the following curatorsand museumsfor hypotonic food), drinking fresh water, increas- assistancein borrowing or examining specimens:G. January1990] PhalacrocoracidaeNasal Glands 117

F. Barrowclough(American Museum of Natural His- ual.Birds of IslaGrande (Tierra del Fuego).Wash- tory), J. W. Fitzpatrick (Field Museum of Natural His- ington, D.C.ßSmithsonian Institution. tory),N. K. Johnson(Museum of VertebrateZoology, JOHNSONßD.W. 1956. The annual reproductivecycle Universityof California-Berkeley),and R. L. Zusi (U.S. of the California : I. Criteria of age and the National Museumof Natural History).I thank P.S. testis cycle. Condor 58: 134-162. HumphreyßH. F. James,R. F. Johnston,B.C. Livezey, LAVERY,H.J. 1972. The Grey Teal at saline drought- S. L. Olson,K. I. Warheit, and two anonymousre- refugesin North Queensland.Wildfowl 23: 56- viewersfor helpfulcomments. This research was par- 63. tially supportedby the Universityof KansasMuseum MADSEN,F. J., & R. SPARCK.1950. On the feeding of NaturalHistory and by the NationalScience Foun- habits of the Southern Cormorant ( dation grant BSR8407365. carbosinensis Shaw). Danish Rev. Game Biol. 1: 45-75. MARPLES,B.J. 1932. The structureand development of the nasalglands of birds.Proc. Zool. Soc.Lon- LITERATURE CITED don 1932: 829-844.

BELLAIRS,A. D'A., & C. R. JENKIN. 1960. The skeleton MCNALLEY,J. 1957. The feeding habitsof cormo- rants in Victoria. Fauna Contrib. 6. Victoria, Aus- of birdsßPp. 241-300in Biologyand comparative tralia, Fish. Game Dep. physiology of birds, vol. 1 (A. J. Marshal, Ed.). New York, Academic Pressß MCNEIL,R., & J. BURTON.1972. Cranial pneumatiza- BOCK,W. J. 1958. A genericreview of the plovers tion patterns and bursa of Fabricius in North American shorebirds. Wilson Bull. 84: 329-339. (Charadriiformes,Aves)ß Bull. Mus. Comp. Zool. 118: 27-97ß NYSTROM,K. G. K., & O. PEHRSSON.1988. Salinity as ß 1963. and phylogeny in morpho- a constraintaffecting food and habitatchoice of logicallyuniform groupsßAm. Nat. 97: 265-285ß mussel-feedingdiving ducks.Ibis 130: 94-110. OLSONßS.L. 1985. The fossil record of birds. Avian COULSON,J.C. 1961. Movements and seasonalvari- Biol. 8: 80-238. ation in mortality of shagsringed on the Farne Islands, Northumberlandß Brit. Birds 54: 225-235ß OWEN,M., & J.KEAR. 1972. Foodand feeding habits. DAVIS, D. E. 1947. Size of the bursa of Fabricius Pp. 58-77 in The swans(G. Atkinson-Wilkes, Ed.). comparedwith ossificationof skull and maturity London, Michael Joseph. of gonadsßJ. Wildl. Manageß11: 244-251. PEAKER,M., & J.L. LINZELL.1975ß Salt glandsin birds DE BEER,G. g. 1937. The development of the ver- and reptiles. London, Cambridge Univ. Press. tebrate skull. Oxford, Clarendon Press. SCHILDMACHER,H. 1932. Ober de Einflussdes Salz- DIXON, W. J. (Ed.). 1988. BMDP statistical software wassersauf die Entwicklung der Nasendriisen. 1988. BerkeleyßUniv. CalifßPress. J. Ornithol. 80: 293-299. ELLIS, R. A., C. C. GOERTMILLER,R. A. DELELLIS,& Y. SCHMiDT-NIELSEN,K. 1959. Saltglands. Sci. Am. 200: H. KABLOTSKY. 1963. The effect of a salt water 109-116. regimen on the developmentof the salt glands ß 1960. Salt-secretinggland of marinebirds. of domesticducklingsß Devel. Biol. 8: 286-308ß Circulation 21: 955-967. FALLA,R. A., & G. STOKELL.1945. Investigationof ß 1965. Physiologyof saltglands. Pp. 269-288 the stomach contents of New Zealand freshwater in Sekretion und Exkretion (K. Schmidt-Nielsen, Shags.Transß R. Soc.New Zealand 74: 320-331. Ed.). Berlin, Springer-Verlag. HARRISON,C. J. O., & C. A. WALKER. 1976. A review ß, C. B. JORGENSEN,& H. OSAKL 1958. Extrare- of the bony-toothed birds (Odontopterygi- nal salt excretionin birds. Am. J. Physiol.:101- formes) with descriptionsof some new species. 107. Tert. Res. Spec. Pap. 2: 1-72. ß & Y. T. KIM. 1964. The effect of salt intake HEINROTH,O., & M. HEINROTH.1927. Die Vogel Mit- on the sizeand functionof the saltgland in ducks. teleuropas.Bd. III. Berlin, Lichterfelde-Vermuch- Auk 81: 160-172. ler. SERVENTY,D. L. 1938. The feeding habits of cor- HOLM,S. 1979. A simple sequentially rejective mul- morants in southwestern . Emu 38: 293- tiple testprocedureß Scandinavian J. Statistß 6: 65- 316. 70. SIEGEL-CAUsEY,D. 1986. Behaviour and affinities of HOLMES,W. M., & J. G. PHILLIPS. 1985. The avian the Magellanic Cormorant (Phalacrocoraxmagel- salt glandßBiol. Rev. 60: 213-256. lanicus).Notornis 33: 249-257. --, D. G. BUTLER,& J. G. PHILLIPS. 1961. Obser- ß 1987. Behaviourof the Red-leggedCormo- vations on the effect of maintaining Glaucous- rant (Phalacrocoraxgaimardi). Notornis 34: 1-9. winged (Larusglaucescens) on fresh and sea- ß 1988. Phylogeny of the Phalacrocoracidae. water for long periods.J. Endocrin. 23: 53-61. Condor 90: 885-905. HUMPHREY, P. S., D. BRIDGE, P. W. REYNOLDSß& R. T. ß 1989. Cranial pneumatization in the Phal- PETERSONß1970. Preliminary Smithsonianman- acrocoracidae. Wilson Bull. 101: 108-112. 118 DOUGt.•SSIEGEL-CAUSEY [Auk, Vol. 107

SKADHAUGE,E. 1981. Osmoregulationin birds. Ber- THOMSON,J. A., & N. W. MORLEY.1966. Physiolog- lin, Springer-Verlag. ical correlates of habitat selection in Australian STAALAND,H. 1967. Anatomicaland physiological cormorants. Emu 66: 17-26. adaptationsof the nasal glands in Charadri- VAN DOBBEN,W.H. 1952. The food of the cormorant iformesbirds. Comp. Blochem.Physiol. 23: 933- in the Netherlands. 40: 1-63. 944. W^TSON, G. E., & G. J. DIVOKY. 1971. Identification SUGIMORI,F., N. OKA, & Y. ISHIBASHI. 1985. The de- of Diornedealeptorhyncha Coues 1866, an gree of ossificationas a means of aging Short- with remarkably small salt glands. Condor 73: tailed Shearwaters. J. Yamashina Inst. Ornithol. 487-489. 17: 159-165. ZAKS,M. G., & M. M. SOKOLOVA.1961. [Ontogenetic TECHNAU,G. 1936. Die Nasendri•se der V6gel. J. and specificfeatures of the nasalglands in some Ornithol. 84: 511-617. marine birds]. Fiziol. Zh. 47: 120-127.