THE AIR-SAC SYSTEMS AND BUOYANCY OF THE ANHINGA AND DOUBLE-CRESTED

CLARK L. CASLER

BOTI-Ithe Anhinga,Anhinga anhinga, and the Double-crestedCormorant, Phalacrocoraxauritus, secureprey while swimmingsubmerged, but they vary considerablyin their buoyancy (Owre 1967: 60-63). Anhingas usually swim at the surface with only the head and part of the neck emergent. ,when at the surface,always swim with the back and shouldersexposed. Anhingas seem to expendlittle effort in remaining submergedeven when moving at low speed. In fact, the Anhinga may use its feet to overcomea tendency to sink slowly (Owre 1967: 107). In contrast,cormorants must paddle with their feet to remain submerged. A numberof factorsmay operatein producingthis differencein buoy- ancy. Water absorptionby the feathersof the Anhingaand cormorantis consideredan adaptation to reduce buoyancy (Owre 1967: 61). The Anhinga'sbreast feathersare completelyvoid of hookletsfor interlocking the barbules. In the cormorantonly the distal portions of the barbs are devoidof hooklets(Casler MS). The structureof Anhingafeathers allows water to penetrateany air spacesnext to the skin thus reducingbuoyancy. As only the distal portionsof cormorant feathers are modified to hold water, the feathers apparently enclosesome air next to the skin in the thicker layer of down feathersand thereby afford somebuoyancy. In addition,the cormorant'sskin is thicker and spongierand appears morepneumatic. Structural differences of the feathersand skin undoubt- edly contributeto differencesin buoyancybetween the two . The Anhingamay useballast in the form of water. Owre (pers. comm.) has often noted what appearsto be copiousdrinking before submerging. This would be helpful, especiallyas the specificgravity of the 's body is normally less than that of water. The air-sac systemmay also contribute to the differencesin buoyancy, but little is known about air-sac systemsin diving . The Anhinga's air sacsmight be permanentlyand greatly reducedin volumeand in extent of bonepenetration, thus making the bird lessbuoyant. Alsothe Anhinga's aquatic habits, unlike those of the cormorant, suggestthat the Anhinga controlsits specificgravity; this may be accomplishedby regulatingthe volume of the air-sac system. This paper describesand illustrates the structureof the air-sacsystems of the Anhinga and Double-crestedCormo- rant and correlates modifications of their structure and behavior with dif- ferencesin buoyancy.

324 The Auk 90: 324-340. April 1973 April 1973] Air-Sac Systems 325

TABLE 1 PARTSOF Tt:IEAIR-SAC SYSTE1VJ[ IN Tt:IE ANtIINGA AND DOUBLE-CRESTEDCOR1VJ[ORA1N'T

Air sac Cormorant Anhinga Medial interclavicular Single,no Same diverticula Lateral interclavicular Singleor paired Single A. Medial chamber D. subcordale Present Present B. Lateral chambers D. subpectorale Present Present D. subscapulare Present Present D. cervicale Present Absent D. propatagiale Present Absent D. suprahumerale Present Present (?) Cervical Paired, no Same diverticula Lateral cervical Paired, no Same diverticula Anterior thoracic Single,right Paired,no side only diverticula Posterior thoracic Paired, no Same diverticula Abdominal Paired,no Same diverticula

METHODS A•D MATERIALS

In past years, various techniqueshave been developedto study air sacsby injecting them with different materials to form casts that can be handled. This study used the recently improved technique of Hamlet and Fisher (1967). Fully discussedin their paper, this method usescarbon dioxide to absorb the liquid latex as it fills the minute diverticula and passagewaysof the air-sac system. Adult specimenswere collected in southern Florida in March 1968. Four Anhingas (two of each sex) and three female cormorants were used. To reduce nomenclatural problems, I have used the terminology of Hamlet and Fisher (1967) who mostly followed that of Ulrich (1904) and Miiller (1908). Table 1 lists parts of the air-sac systemsin the Anhinga and Double-crested Cormorant. The air-sac system and lungs of the Anhinga and Cormorant are similar, and the descriptionspertain to both speciesexcept as differencesare noted. The casts of lungs and air-sacs showed little intraspecific variation except in the diverticulum subcordale.

THE LUNGS M½sobronchus.--Themes0br0nchus first givesrise to five ventrobronchi (Figure 1). It then curves posteriorly toward and appears upon the dorsolateralsurface, where it gives rise to a variable number of dorso- bronchi and laterobronchi (Figure 2). Caudally the diameter of the mesobronchusnarrows and, at the posterioredge of the lung, opensinto the abdominalair sac (Figures 1 and 5). The main openingis accompanied by secondaryostia (not shownin figures) only in the Anhinga. In the regionof the ventrobronchi,the mesobronchushas no connecting 326 CLARXL. CASZ• [Auk, Vol. 90

TO INTERCLAY. MED. and ANT. THORA.

TO INTERCLAMMED. and ANT. THORA.

• PRI.BR. • /TO ABe. anterior*-•• 1cm. • TOPOST. THORA.

Figure 1. Ventral view of right lung showingrelationship between mesobronchus and ventrobronchi.Top, Anhinga;bottom, cormorant. Abbreviations: Lmaos--Br., bronchus;D., dorsobronchi;L., laterobronchi;Mesobr., mesobronchus;Parabr., parabronchus;Pri. br., primary bronchus;V., ventrobronchi.Am S^cs--Abd., abdominalsac; Ant. thora., anteriorthoracic sac; Cerv., cervicalsac; Cerv. lat., lateralcervical sac; Interclay.lat., lateralinterclavicular sac; Interclay.med., medial interclavicularsac; Post. thora., posteriorthoracic sac. DrVERT•C•JL^--D.cerv., D. cervicale;D. prop.,D. propatagiale;D. subcor.,D. subcordale;D. subpect.,D. sub- pectorale;D. subscap.,D., subscapulare;D. suprah.,D. suprahumerale. parabronchi,but somesmall bronchi connectto the mesobronchusin the regionof the dorso-and latero.bronchi.These may correspondto the caudolateralsecondary bronchi of King (1966). First ventrobro.nchus.--V.1 is the largestin diameterand its largest branchextends to the anterioredge of the lungand connectsto. the cervical sac (Figure1). The mainopening to the cervicalsac is accompaniedby April 1973] Air-Sac Systems 327

[212 0.3 0.4 0.5

Figure 2. Anhinga. Dorsolateral view of left lung showing relationshipsbetween mesobronchus,dorsobronchi and laterobronchi. Abbreviations as in Figure 1. secondaryostia. Another branch, usually the next lateral one, goesto the lateral cervicalsac. The main o.stiumof this sac is not accompaniedby secondaryostia, but parabronchial(also calledtertiary bronchial) openings are just insidethe lung. The lateromostof the anteriorprojecting branches of V. 1 is the anterior openingto the lateral interclavicularsac. In one Anhinga two branchesopened into this sac anteriorly. Secondaryostia may be present near the major opening, and parabronchiconnect to. the branch. The cormorant has no secondary ostia into the three air sacs. Only three anteriorpro•ecting branches of V. 1 exist; each opensinto an air sac (Figure 1, bottom). Secondventrobronchus.--V. 2 has two main branches(Figures 1 and 3). A largebranch extends medially on the ventral surfaceof the lung and gives

I D'34 I MESOBR. TO INTERCLAM MED. I 2 õ6Z8 TOABD. andANT. THORA.• anterior•V.1 V.2• • ,O,T.T,,*.

PRI.BR•'1•"" \ TO INTERCLAV. LAT.

Figure 3. Anhinga. Mediodorsal view of mesobronchus with ventrobronchi and dorsobronchi.Natural size. Abbreviationsas in Figure 1. 328 CLAR• L. CASLV.R [Auk, Vol. 90

TO CERV.

IN

INTERCLAV. MED. • 1

anterior.--]& • r• j vefitral

FJõure4. Anhinõa. MedJoYentm!¾iew of riõht Junõ s•o•g co•ect[o•s bet•ee• roediM [•terc]•[cu]•r •d •ter[o• t•omcic s•cs. •tum] size. Abbreviations •s [• •[gu•e 1. rise to parabronchi. The smallerbranch curvesdorsal to V. 1 inside the lung and comesto the ventral surfacejust anterior to the entranceof the primary bronchus.This secondbranch gives rise to the posteriorof the two openingsinto the lateral interclavicularsac. Only in the Anhinga are secondaryostia present. Third ventrobronchus.--V.3 dividesinto two branches(Figure 1). A large branch extendingcaudomedially on the ventral surfacegives rise to parabronchionly. Close to the mesobronchus,at the ventral surface,a short branch turns sharply and bifurcates just outside the lung. The

D. SUPRAH2 INTERCLAM LAT. D. SUBSCAP. I / .::.,.:......

INTERCLAVLAT '--••../.::.-:•..•,..'• •." '•,"'•'...... 7

STERNAL PLATEAUS.''•...-".''-.....-'"' '•. ' '.."::..'•,?.' •1 • • cut edge

Figure 5. Anhinga. Lateral view showing relationships of air sacs and skeleton. D. subpectoralenot shown. Abbreviationsas in Figure 1. April 1973] Air-Sac Systems 329

/•?•'•__•'• ;. M.sternotrach. • •

I \'"•-furc.ulum •....,,,.•f...•./'•, Xx BR< A •1•'• 'TO LEFT

'..'. INTERCLAV.MED.,) ...... /

Figure6. Lateralview of medialinterclavicular sac. Top, Anhinga;bottom, cor- morant. Natural size. Abbreviations as in Figure 1. branchopens anteriorly into the medial interclavicular sac and posteriorly into the anteriorthoracic sac (Figure 4). Secondarybronchi may connect to the medial interclavicular sac. In theleft lungof thecormorant, the short anterior branch of V. 3 opens into the medialinterclavicular sac only. The left anteriorthoracic sac is absentand there are no secondarybronchi. Fourthventrobronchus.--V. 4 extends posteriorly on the ventralsurface, connectingto parabronchionly (Figure 1). Fifth ventrobronchus.---V.5 is much smaller than the others, connects to parabronchionly, and is notseen on theventral surface (Figure 1). Dorsobronchi.•Themeso.bronchus gives rise to 8 to 10 dorsobronchi (Figures2 and 3). They connectto parabronchionly. Laterobronchi.--Most laterobronchiare smallerthan the other secondary 330 CLARKL. CASLER [Auk,Vol. 90

HT LUNG

trachea ./r-• •?.-.,•..• anterior--J'•••TOLu•i•T

/ x••,,•" •.,•,/ • • • •. LEFT

LOBE BETWEEN .h .... • • • • TRACHEAa• ESOPHAGUSM. ster•rac

Figure 7. Dorsalview of medialinterclavicular sac. Top, Anhinga;bottom, cormorant. Natural size. Abbreviationsas in Figure 1. bronchi,but laterobronchus 1, which opens into the posterior thoracic sac, is aslarge or largerin diameterthan the primary bronchus (Figure 2). Thehomologies of the laterobronchi are unknown. In thechicken, Gallus domesticus(Locy and La.rsell 1916a, 1916b) and House Sparrow, Passer domesticus(Delphia 1961) the posteriorthoracic sac arises from latero- bronchus3. Thissame laterobronchus appears to benumber 1 in theadult Anhingaand cormorant. Parabronchial connections with laterobronchus1 areabsent except for one(in Anhingaonly) from the ventral edge where manyparabronchi converge (Figure 2). Posteriortolaterobronchus 1,five small laterobronchi areusually present on the dorsalsurface (Figure 2).

T•E Am SAcs TheAnhinga and cormorant appear to havetwo interclavicular sacs, contrary to the situationin other birds studied. I refer to the sac on the April 1973] Air-Sac Systems 331

To,.su,PEcT. •. s•.• • X• X •. s•co•. anterior'"

V. 1

D. PROP. • TOV.2 TOHUMERUS TOD.SUBPECT. TO STERNUM • SUPRAH. •'-INTE• • MED. D, SUBCORD.

Figure 8. Lateral view of lateral interclavicular sac. Top, Anhinga; bottom, cormo- rant. Natural size. Abbreviations as in Figure 1. 332 CLARKL. CASLER [Auk,Vol. 90

openings for blood vessels

'rOD,

• MIDLINE NTERCLAV.•----•-• -- • _---__.•k•

I 2cm. • ) STERN•JMpost.bor•eer

Figure 9. Ventral view of lateral interclavicularsac. Top, Anhinga;bottom, cormorant: D. subcordaleshown unfused at the midline. D. subscapularenot shown. Abbreviationsas in Figure 1. midline (enclosingthe syrinx) as the medial interclavicularsac and the other (possessingdiverticulum subcordale) as the lateral interclavicular sac (Figure 5). In the Anhingathe air-sacsystem does not penetrateany bones. Only thehumerus, sternum, and centra of a fewcervical vertebrae (nos. 15-17) of the cormorantare penetrated.In both speciesonly the lateral inter- clavicular sac has diverticula. Medialinterclavicular sac.--The medial interclavicular sac is unpaired andcentrally located immediately ventral to theesophagus (Figures 6 and 7). The sac consistsof two flattened lateral lobes that arise from the lungsvia V. 3. Thelobes partially envelop the bronchi and join dorsally at the midlineto forma centralchamber (Figure 7). This centralchamber enclosesthe syrinxand extendsanteriorly between the coracoldsand into theinterclavicular fossa (Figure 6). Laterallythe sac is surrounded by the lateral interclavicular sac. Anteriorlythe centralchamber is greatlywidened and V-shaped in the cormorant(Figures 6 and 7, bottom)and within the V, a smallmedian lobelies dorsal to the trachea(Figure 7, bottom). April 1973] Air-Sac Systems 333

Lateral interclavicular sac.--The lateral interclavicular sac connects to eachlung via V. 1 and V. 2 (Figure 8). From eachlung, laterally flattened lobesextend ventrally and anteriorlyand join at the midline,giving rise to a mediandiverticulum--D. subcordale(Figures 5, 8, 9). The anterior portion of the sac consistingof D. subscapulare,D. suprahumerale(?), and D. subpectoralelies lateral to the ribs and posteriorto the coracoid. The caudal edge of the sac fits tightly against the posterior thoracic sac (Figure 5). Anteriorly M. supracoracoideusand M. subscapularisare lateral to the sac. External to the midportion of the sac are M. coraco- brachialis posterior, M. serratus anterior, and M. sternocoracoideus. Posteriorlycervical ribs 1 and 2 are lateral to the sac. The saclies external to the medial interclavicularsac, heart, neck muscles,and esophagus. Blood vesselsconnecting to M. pectoralissuperficialis are surroundedby this sac. The lateral interclavicularsac of the cormorantis larger in relation to body size and extendsanterior to the furculum (Figures 8 and 9, bottom). It has two additional diverticula--D. cervicaleand D. propatagiale. Diverticulum subcordale.•D. subcordale is a medium lobe extending posteriorlyfrom the junction of the paired portions of the lateral inter- clavicularsac (Figures 5, 8, 9). It lies between the heart and sternum with the posteriorapex sometimessurrounding the caudal tip of the heart. Anteriorly lies a seriesof small, tightly fitting lobes (esp. in Anhinga); the foldsbetween them may causethe diverticulumto be greatly constricted in certain areas (Figures 8 and 9, top). In the cormorant,D. subcordalecovers only the anterior half of the sternum,but extendsposteriorly to coverpart of the ventral surfaceof the heart (Figures 8 and 9, bottom). The diverticulumat its anterior edge pneumatizesthe sternumbut not extensively. In two cormorantsD. sub- cordalewas not fusedat the midline (Figure 9, bottom), producinga paired rather than a singlelateral interclavicularsac. Diverticulum subpectorale.--D. subpectoralearises by one small con- nection(1-3 mm diameter) from the lateral interclavicularsac immediately posteriorto the coracoidand just ventral to the suprahumerallobe (Fig- ure 5). Removal of M. pectoralissuperficialis exposes the sac laterally (Figure 10). The air-sacwall adheresto M. triceps,external head, as well as to other muscles. Thus configuration of the sac is changed as the humerusis moved in flight. A constrictionwhere blood vesselsextend laterally dividesD. subpectoraleinto anterior and posteriorparts. The anterior portion is laterally flattened and usually borders the furculum. Its medial wall is against M. supracoracoideusand M. coracobrachialis posterior.The openingof the diverticulumis next to the insertionof the latter muscle. The posteriorportion ends as a blunt apex dorsal to the 334 CLARK L. CASLER [Auk, Vol. 90

M. PECT. SUPERR cut M. TRICEPS,EXT. HD. M. CORACOBRACH. ANT. M. SERR. POST., SUPERE cut NECK

M. PECT. su[

2 cm.

• I 2cm. I Figure10. Anhinga.Lateral vie•v of D. subpectorale.Top: M. pectoralissuper- ficialisremoved. Bottom: shows dorsal lobes of posteriorportion; ostium to Interclav. lat. is on medial•vall. Abbreviations:1, lobe bet•veenM. dorsalisscapulae and M. tricepsexternal head; 2, lobemedial to M. dorsalisscapulae; 3, impressionof humerus. Other abbreviationsas in Figure 1. caudaledge of M. pectoralissuperficialis, sometimes just underthe skin. M. serratusposterior, superficial layer, lies along the posterodorsalsurface of the sac. Dorsally,the posteriorportion is revealedlateral to M. dorsalis scapulaeby removingM. latissimusdorsi anterior and posterior (Figure 11, top). Diverticulumsubscapulare.--D. subscapulare is a flattenedlobe ventral andmedial to the scapula(Figures 5 and 8, top). The caudaledge lies April 1973] Air-Sac Systems 335

M. SERR. PROE ANT. M. RHOMB. PROF.

M. RHOMB. M. SERR. POST.

M. TENS. PAT.BREV. and LONG.

M. SERR. M. TRICEPS,SCAP. POST., DEEP 2 cm. anterior - M. LAT. DORSI,

M. LAT. DORSI, I M. RHOMB. NECK RHOMB. SUPERE, POST,

M. SERR. PROF.,

M. SCAP. M. PECT. SUPERF. M. TENS. PAT. BREV. LONG. anterior,2 cm. I M.TRICEPS, SCAR •1.LAT. DORSI• Figure 11. Dorsolateralview. Top, Anhinga: showsD. subscapulareand D. subpectorale;muscles attaching to the medialside of the scapulaare cut; M. subscapu- laris not shown. Bottom, cormorant: showsD. cervlcale,D. subscapulare,and D. propatagiale;D. subpectoralenot shown. Abbreviationsas in Figure 1. lateral to cervicalrib 1. M. subscapularis(not shownin Figure 11, top) lies lateral to the diverticulum,and M. serratusprofundus anterior is medial and dorsal (Figure 11, top). In the cormorantD. subscapularedoes not form the anterodorsalborder of the lateral interclavicularsac but lies posteriorto D. cervicale(Figures 8 and 11, bottom). Diverticulumcervicale.--It is absentin Anhinga (Figure 8, top). In the cormorantD. cervicaleis a broad lobe formingthe anterior portion of the lateralinterclavicular sac (Figure 8, bottom). It is medialto the junction of the coracold and furculum and extends anterior to the furculum. The medialwall liesnext to the neckmuscles. Dorsally D. cervicaleis exposed by removingM. rhomboideussuperficialis, but anteriorto thismuscle the saclies just beneaththe skin (Figure 11, bottom). 336 Cz•x L. C•szz• [Auk, Vol. 90

ANT. BRANCH OF, V. 1

x';"•J•NcERVCALRIB1/ ' • Figure 12. Cormorant. Dorsal view of cervical sacs. Penetration of centra of vertebrae 15-17 not shown. Natural size. Abbreviations as in Figure 1.

Diverticulum propatagiale.--D. propatagialeis absent in Anhinga. In the cormorant (Figure 11, bottom) it arisesfrom D. cervicale. It is a small lobe about 15 mm in length and fills the triossealcanal (Figures 8 and 9, bottom). Diverticulumsuprahumerale.--The existence of D. suprahumeralein the Anhingais questionable.A small lobe is presentthat correspondsin loca- tion to I). suprahumeralein the cormorant,but it doesnot pneumatizethe humerusor evenreach the pneumaticfossa (Figures 8 and 9, top). In the cormorantD. suprahumeralearises from the lateral interclavicular sac slightly posteriorto the coracoid(Figures 8 and 9, bottom). It pneu- matizes the humerusvia the foramen in the pneumatic fossa. Cervical sacs.--The paired cervical sacs arise from V. 1 near cervical vertebra18 and extendcranially through only five or six vertebrae(Figures 5 and 12). Thesesmall, tubular sacspenetrate the vertebrarterialcanals, the intervertebralspaces (where left and right sacsfuse), and the neural canal. The sacsare mostextensive near cervicals17 and 18, but immediately diminishin diameter anteriorly and lie only in the neural canal. A septum separatesthe sacsfrom the nerve cord. The septumis especiallydistinct in the cormorant where it divides the diameter of the neural canal into three sections.The most veneral sectioncontains the nerve cord, and the two dorsal sectionsthe left and right cervical sacs. In the cormorant the cervical sacs extend anteriorly to approximately cervical number 7. The centra of vertebrae 15-17 are penetrated. Lateral cervical sacs.--The paired, lateral cervical sacs arise from V. 1 just lateral to the connectionwith but are not fused with the cervical sacs (Figure 4). These are small sacsin the Anhinga; in the cormorant they are larger in relation to body size and lie medial to D. cervicale. The sacsvary in size from bird to bird but not in number and origin. The terminologyof Juillet (1912: 350) is applied to thesesacs. April 1973] Air-Sac Systems 337

Anterior thoracic sacs.--A small anterior thoracic sac arises from each lung via V. 3. Each sacextends ventrally around the esophagusand may cover the side of the heart. Each is medial to the posteriorthoracic sac and variablein volume (Figure 5). Usually the sac is laterally flattened with the distal end thicker, but sometimesis flattened into a thin slip lateral to the esophagus(Figure 4). A distendedesophagus seems to pre- vent filling of thesesacs. In the cormorantonly the right anterior thoracicsac exists. It extends from the right lung to the midlinewhere it lies betweenthe esophagusand posteriorthoracic sacs. The enlargeddistal end partly coversthe caudal end of the heart. Posterior thoracic sacs.---The paired posterior thoracic sacs are the largest. Each communicateswith the lung by a singleostium latero- bronchus1 (Figure 5). The sacsare just ventral to the lungsand almost meet at the midline. Anteriorlythey fit againstthe lateral interclavicular sac,laterally against the ribs and sternum,and mediallyagainst the heart, anterior thoracic sacs, the liver and/or esophagus,and sometimesthe abdominalsacs. D. subcordalelies betweenthe ventral edgesof this pair of sacs (Figure 9, top). Thus the esophagusand part of the liver are encircled.The sacsare laterally and differentiallycompressed, depending upon the degreeof distentionof the esophagus. In the cormorantthe connectionwith the lung is farther posterioron the lateral side, and D. subcordale(Figure 9, bottom) separatesonly the anteroventraledges of the sacs. Abdominal sacs.--The paired abdominalsacs arise from the posterior end of the mesobronchusand lie againstthe dorsalwall of the abdominal cavity (Figure 5). The ventral surfacefits tightly againstthe intestineon the right side of the body and the stomachon the left side. The sacs partially envelopthe testes; in femalesthe left abdominalsac is ventral to the ovary. Abdominal sacs vary in volume. When the stomachand intestine are distendedwith food, the sacsare greatly compressedand may have less volumethan the lungs. In one cormorantwith esophagusand stomach completelycontracted, the large left sac was 43 ml whereasthe right sac was only 6 ml in volume. There seemsto be little space for expansion eventhough the sacsare capableof it.

DISCUSSION

DIFFERENCES IN BUOYANCY Regulationof volume of air sacs.--Behavioraldifferences between the Anhingaand cormorantmay affect the regulationof the volumeof the 338 C•^•I• L. C•s• [Auk, Vol. 90 air-sac system. As mentionedearlier, certain of the Anhinga'shabits sug- gest it may to a degreecontrol its specificgravity. Unlike the cormorant, the Anhingamay increaseits specificgravity by expellingair before it dives. The Anhinga may regulate the air content of the subpectoral diverticula. Thesediverticula are large and eachis connectedto the lateral interclavicular sac via one small tube that lies in a narrow passageway between the coracoidand the insertion of M. coracobrachialisposterior. Wing position (flexion) or musclecontraction may affect the size of the openingor evenclose it. The Anhinga'speculiar habit of holdingits wings slightly extendedduring submergenceis noteworthy (Owre 1967: 61). If the opening is closed, two "buoyancy tanks" are separated from the remainderof the respiratorysystem. Increasedvolume of the diverticula could compensatefor an increasedspecific gravity. This may explain the Anhinga'sability to remain submerged,even after feeding,with only the head and neck (or just the head) emergent. Becausethe wingsand flight musclesare not usedwhile the bird is underwater,the diverticula may not be neededfor possiblerespiratory functions. Speculationabout this behaviorstems from Richardson's(1939) study of the inflation of sacsin the California Brown , Pelecanuso.cciden- taliscalifornicus. can inflate the largesuperficial air system,which may be modifiedsubpectoral diverticula. This is consideredan adaptation for diving, to lessenthe force of impact with the water. When artificially inflated, air soonleaves the superficial air system when pressureforcing air in is reduced. Richardsonattempted to explain why air doesnot rush out when the bird strikes the water. He proposedthat tight and strong closureof the glottis kept air from escaping. I believe the pelican may closeoff the subpectoraldiverticula where they connectto the interclavicu- lar sac,as suggestedfor the Anhinga. Thus the D. subpectorale,not the entire respiratorysystem, would absorbmost of the shockof impact with the water. Air-sac volumeand bonepneumaticity.--Compared to nondivingspecies, the air-sacsystems of the Anhingaand cormorantexhibit few diverticula; those from the lateral interclavicular sac being exceptions. However the systemsare not greatlyreduced in volume. Structural differencesbetween the Anhinga and cormorantair-sac sys- tems are not striking, but the Anhinga'sair-sac volumeis less in relation to body size. The medialinterclavicular (Figure 6) and lateral cervical sacs(Figure 4) are smaller;and D. cervicaleand D. propatagialeare absent from the lateral interclavicularsac (see figures 8 and 11). The air sacsin the Anhingapenetrate no bones. The bonesare heavy and have small central cavities (especially in the humerus) filled with marrow. Bonesof the cormorantare pneumaticand air sacspenetrate the April 1973] Air-Sac Systems 339 humerus,sternum, and the centraof a few cervicalvertebrae. The humerus has a large central cavity. Pneumaticbones, of course,increase buoyancy by reducingdensity in relation to total body volume. However I think the effect of differencesin bone pneumaticity,combined with the differencesin volume of the air sacs,may be overshadowedby the effect of the degreeof distentionof the digestivesystem. If the ingestionof food is not correlatedwith a propor- tional increasein total body volume,the distendeddigestive tract must displacesome of the air in the air sacs,and thus increasethe bird's specific gravity. This apparentlyhappens in the Anhinga and cormorant. Speci- mens with stomachs distended had small abdominal sacs. The anterior and posteriorthoracic sacs in somewere also reducedin volume,mostly by esophagealexpansion. In both speciesfish taken into the stomachmay extendwell up into the esophagusand be carried there for somelength of time (Owre, pers. comm.). Owre (1967: 129) discussedthe size,quantity, and kinds of fisheseaten by the Anhingaand cormorant. Stomachcontents of one Anhinga consti- tuted more than 7 percent of the bird's total weight. A daily requirement for Phalacrocoraxauritus is 27 percent of the body weight (Wetmore 1927) and 20 percent for P. carbo sinensis(Van Dobben 1952: 23). A cormorant with a distended esophagusand stomach still floats, even thoughit may have less total volumein its air sacsin relation to body size than an Anhinga with an empty digestivesystem. In my opinion, static differencesin volumeof the air-sacsystems and bonesare not signifi- cant factors in determining differencesin buoyancy between the two species.

A CI( N 0Vv'I,ED GlVI ENT S

I am especiallyindebted to my major professor,Harvey I. Fisher, and to V•il]iam G. George and John V•. Voigt of Southern Illinois University for critically reviewing the manuscript and offering guidance throughout the study, which constituted part of the requirementsfor the Master of Arts Degree in Zoology at Southern Illinois University at Carbondale. Special thanks are due Oscar T. Owre of the University of Miami for his useful information about Anhinga and cormorant behavior and for reviewing the manuscript. The birds were securedwith the help of Robert T. Edwards of Southern Illinois Uni- versity and Arthur P. Kirk of Goodland, Florida.

Su•^Rv The Anhingaand Double-crestedCormorant vary considerablyin their buoyancybut static differencesin volume of the air-sac systemsand bones are not regardedas significantfactors in determininginterspecific differ- encesin buoyancy. The Anhinga'sair-sac volume is lessin relation to body size but the difference is not striking. The differences in air-sac volume 340 CLARKL. CASLER [Auk, Vol. 90 betweenthe two speciesmay be overshadowedby the effect of degreeof distentionof the digestivesystem. If the digestivesystem is distendedwith food,the abdominalsacs are greatlycompressed and may haveless volume than the lungs.The anteriorand posteriorthoracic sacs are alsoreduced in volume. A cormorant with a full digestivesystem still floats, even thoughit may have lesstotal volumein its air sacsin relationto body size than the Anhingawith an empty digestivesystem. The air-sacsystem and lungsof the Anhingaand cormorantare similar exceptfor onesignificant difference--the cormorant has only one anterior thoracicsac. Two separatesacs occur where the singleinterclavicular sac is found in most species.

LITERATURE CITED D•LP•IA, J. M. 1961. Early developmentof the secondarybronchi in the House Sparrow, Passerdomestlcus (Linnaeus). Amer. Midl. Naturalist 65: 44-59. HA•rL•T, M.P., Am) H. I. F•S•ER. 1967. Air sacsof respiratoryorigin in some procellariiform birds. Condor 69: 586-595. JuIrrET, M. A. 1912. Recherchesanatomiques, embryologiques, histologiques et comparativessur le poumondes oiseaux.Arch. Zool. Exp. et Gem 9: 207-371. Kr•o, A.S. 1966. Structural and functionalaspects of the avian lungs and air sacs. Intern. Rev. Gem Exp. Zool. 2: 171-267. Loc¾, W. A., Am) O. LARS•rr. 1916a. The embryologyof the bird's lung, part 1. Amer. J. Anat. 19: 447-504. Loc•r, W. A., Am) O. LARSELr. 1916b. The embryologyof the bird's lung, part 2. Amer. J. Anat. 20: 1-44. M/)•.LER,B. 1908. The air-sacsof the pigeon. SmithsonianMisc. Coll. 50: 365-414. OWR•, O.T. 1967. Adaptationsfor locomotionand feedingin the Anhinga and the Double-crested Cormorant. Ornithol. Monogr. No. 6. RicaAm)SO•,F. 1939. Functional aspectsof the pneumaticsystem of the California Brown Pelican. Condor 41: 13-17. UrRIC•, F. 1904. Zur Kenntnisder Lufts•ickebei Diomedeaexulans und Diomedea fuliginosa. Wiss. Ergeb. Deutsch.Tiefsee-Exped. Valdivia 7: 319-342. VAn DOBBE•,W. H. 1952. The food of the cormorant in the Netherlands. Ardea 40: 1-63. W•r•roR•, A. 1927. The amount of food consumedby cormorants. Condor 29: 273-274.

Departmentof Zoology,Southern Illinois University,Carbondale, Illi- nois 62901. Accepted 1 May 1972.