Quick viewing(Text Mode)

Size and Locomotion in Teratorns (Aves: Teratornithidae)

Size and Locomotion in Teratorns (Aves: Teratornithidae)

SIZE AND LOCOMOTION IN TERATORNS (AVES: )

KENNETHE. CAMPBELL,JR. 1 AND EDUARDOP. TONNI2 'NaturalHistory Museum of LosAngeles County, 900 ExpositionBoulevard, Los Angeles, California90007 USA, and Divisi6n PaleontologfaVertebrados, Facultad de CienciasNaturales y Museo,1900---La Plata,

ABSTRACT.--Theextinct family Teratornithidaecontains the world'slargest known flying . A new methodof determiningbody weightsof extinctbirds, basedon the size of their tibiotarsi,facilitates the estimationof the wing dimensionsof thesegiant birds. An analysisof the bonesof the teratornwing showsthat they closelyresemble those of , suggestingthat teratornsflew in a manner similar to theselarge New World . The bonesof the pelvicgirdle and hindlimbsindicate that teratornswere probablyagile on the ground,though better adapted for walkingand stalkingthan running. We estimatethat the largestteratorn, magnificens, weighed 80 kg and had a wingspanof 6• m. It probablybecame airborne by spreadingits hugewings into the strong,continuous, westerly winds that blew across southern before the elevation of the Mountains and, once aloft, flew in the manner of condors.Received 26 April 1982, accepted18 October 1982.

THE extinct avian family TeratornithidaeL. viduals of T. rnerriarnifrom the asphalt depos- Miller 1909 is presently known from four its of Rancho La Brea, , where re- : rnerriarni L. Miller 1909, T. mains of raptorsand vultures abound (Howard incredibilisHoward 1952, Cathartornisgracilis L. 1962b). It was assumed that the teratorns were Miller 1910, and Argentavismagnificens Camp- entrapped in oil seepsat RanchoLa Brea while bell and Tonni 1980. The known teratorns feeding on carcassesof other trapped . rangedfrom very large to truly giganticin size, Consequently,all reconstructionsof teratoms with A. magnificensbeing the largestknown have pictured them as slightly larger versions flying . The single known specimenof A. c•fcondors, usually sitting in a treewaiting for magnificens,an associatedpartial skeletonfrom a trapped to die or feeding in groups (late ) deposits of Argen- on large carcassesas vultures are wont to do. tina, is nearly 21/2times as large, in linear di- As discussedherein, teratornsprobably did mensions, as living condors.Argentavis rnag- fly in the mannerof condors,but an analysis nificensis also the oldest known teratorn, and, of the functional morphology of the teratorn with the exceptionof an indeterminate speci- skull (Campbell and Tonni 1981) indicates that men from late Pleistocenedeposits of south- thesebirds were not .The structure westernEcuador (Campbell 1976, Campbell and of their maxillary rostrum and mandibles and Tonni 1980),it is the only teratornknown from their ability to spreadtheir mandiblesproxi- outside . mally, as evidencedby the planeof rotationof Interpretationsof teratornshave alwayspic- their quadrates, indicate that teratorns were tured them as scavengers,flying and feeding predaceouscamivores that consumedtheir prey in a manner like that of living vultures (Fisher whole. They were functionally incapable of 1945; Howard 1950, 1962a; Stock 1956). When feeding by tearing piecesof flesh from car- the first speciesof teratorn, Teratornisrnerriarni, cassesas vultures do. Appropriatelysized prey was described(Miller 1909), it was placed with for Teratornis rnerriarni would have included the New World vultures in the family Vulturi- frogs, lizards, nestlingor fledglingbirds, and dae (= Cathartidaeauct.), probablybecause of .With a skull over 55 cm long and 15 its very elongated, highly vaulted, raptorial- cm wide, however, Argentavismagnificens was appearing, hooked beak and the dose resem- capableof swallowingeven -sizedanimals blance of its postcranialskeleton to that of con- whole. dors. These interpretationswere reinforced by Although the postcranialskeleton of tera- the recoveryof the remains of over 100 indi- torns alwaysimpressed previous workers with

39O The Auk 100: 390-403. April 1983 April 1983] TeratornSize and Locomotion 391 its similarity to that of condors,major differ- tween weight and wing area by the equation enceswere alsonoted (Miller 1909, 1910;Fisher 1945).A detailed comparativeanalysis (in pro- W = •z'S •', (1) gressby KEC) of the bones of Teratornismer- where W is weight in g and S is surfacearea riami indicates that, while teratorns are related in cm2. Data were fitted by least squaresto the to New World vultures, they may be just as logarithmic form of this and the following closelyrelated to (Ciconiidae). The post- equations.The data for the (= cranial skeletonof teratornsdisplays a mosaic Falconiformesauct.) yielded the following es- of vulturid, ciconiid, and unique characters. timation equation: The most strikingdifferences among the three families are found in their skulls, the result of W = 0.05338'S •'27• adaptations for very different feeding meth- ods. New inquiries and synthesesof available Similarly,he suggesteda relation,ship between data (Ligon1967, Olson 1978, Rea in press)have wing area (S) and wingspan(b, in ½m), questioned the traditionally perceived rela- tionships of storks and New World vultures, b = c•2'S•2. (2) and an ever-increasingbody of evidence is The values fitted for his passeriform model being accumulatedthat supports the hypoth- were: esis of their common origin. The recognition that teratornsare related to both of thesegroups D = 2.221'$ ø'•. strengthensthis hypothesis. The proposedrelationship between wingspan SIZE (b) and aspectratio (AR) was:

Estimation of body mass and dimensions AR = c•'b •3. (3) of extinct animals is often difficult and frus- For the passeriformmodel he obtained the fol- trating. The presentwork is no exception.Dif- lowing estimation equation: ferent methods produce different results, and the lack of critical data for modern birds seri- AR = 4.49'b ø'•8. ously impairs present efforts. Data are partic- ularly sparsefor large birds, especiallythose Using this seriesof equations,each fitted from threatened with , such as condors, empirical data, one can predict the wing-sur- the active collectionof scientificspecimens of face area,wingspan, and aspectratio of a which ceased before the importance of such bird that conformsto the passeriformmodel if simple information as weight data was recog- an accuratefigure for its weight can be ob- nized. tained. Greenewalt (1975a) used previously pub- Fisher (1945) was the first to estimate the lished dimensional data from over 1,400 indi- weight of a teratorn,that of Teratornismerria- vidual birds to analyze the relationshipsbe- mi. He measured the areas of the and tween (a) live body weight and wing area, (b) synsacrumin the Bald (Haliaeetusleuco- wing areaand wingspan,and (c) wingspanand cephalus)and the California (Gymno- aspectratio. He divided the flying birds (ex- californianus)and comparedthose areas cluding hummingbirds) into three groups: a with the known weights of the individuals passeriformmodel, a shorebird model, and a measuredin an attempt to determine an index duck model. Each group differs substantially to body weight for birds of different sizes but from the others in wing loading (weight sup- with similar body build. His estimatefor T. ported per unit wing area) at a given body merriarniwas "about 50 lbs [23 kg]." weight. The New World vultures fall into his For a bone's dimensions to correlate well with "passeriformmodel," and, becausewing load- body mass,the bone must function to support ing in teratornswas probablyvery similar to the body. In birds, these are the bones of the that in condors (see below), the dimensional wing or the leg. Althoughhigh correlationsbe- relationshipsobserved within New World vul- tween somedimensions of certain wing bones tures probably hold for teratornsas well. and body weight undoubtedlyexist within the Greenewalt expressed the relationship be- numerousmorphofunctional groups of birds, 392 CAMPBELLAND TONNI [Auk, Vol. 100

5.0-

4.5-

4.0-

o, ß

0.5 1.0 1.5 2'.0 log Tibiotarsus Circumference {mml Mean St. Dev. Regression Line Res. Meal1 Sq. N=324 X .90334 .37234 X = .38272 •Y+.08727 .00388 COR =.986 Y 2.1845 .95922 Y= 2.5400 *X-.10996 .02574

Fig. 1. Computer-generatedplot demonstratingthe correlationbetween the live bodyweight (g) and the least circumference of the tibiotarsus (mm) in birds. their widely disparateflying abilities preclude correlation coefficient is 0.986 (data from 324 any single wing-bone measurement from individuals of 19 orders and 45 families, in- maintaininga high degreeof correlationto body duding such disparate taxonomic and mor- weight throughoutthe ClassAves. A high cor- phofunctionalgroups as penguins, ostriches, relation, first noted by John Anderson (pers. falcons,shorebirds, owls, hummingbirds, and comm., Anderson et al. 1979), does exist, how- passerines).By using the regressionequation ever, between the least shaft circumference of log Y = 2.54-1ogX - 0.10996, the leg bones of birds and their live body weights (Fig. 1). Betweenthe logarithm of the where Y is live body weight (g) and X is least least shaft circumference of a bird's tibiotarsus shaft circumference (mm) of the tibiotarsus and the logarithmof its live body weight, the (log = commonlogarithm), we can predict a April 1983] TeratornSize and Locomotion 393

TAnrE1. Predictedmean dimensionsof Teratornismerriami, Argentavis magnificens, and Gymnogypscali- fornianus;and observedmean dimensionsof G. californianus.

Wing- surface Wingspan(cm) c Weight area Aspect (kg)a (cm•)b A B C D E ratioa

Predicted dimensions Teratornis merriami 13.7 e 17,497 293.8 317.1 398.9 317.0 337.9 8.9 Argentavismagnificens 79.8f 69,609 586.0 639.2 830.7 570 + 10 607 + l0 t 9.6 Gymnogypscalifornianus 10.2b 13,915 262.0 282.2 353.2 -- -- 8.7 Observed Dimensions Gymnogypscalifornianus 9.5i 13,160k 274.0j 6.0k a Predictedfrom the equationlog Y = 2.54'1ogX - 0.10996,where Y is weight(g) andX is leastcircumference of tibiotarsus(ram). bThe predicted surface areas, wingspans, and aspect ratios are based on the predictedmean weights. The surfacE areaswere predicted by use of the equation W = 0.05338'S •'27• of Greenewalt(1975a), where W is weight (g) and S is surfacearea (cm2). c The wingspanwas predictedusing differentmethods, as follows:A, derivedfrom the theoreticallypredicted equation for dimensional similarity,b = c•.S ø.•øø, where b is wingspan(cm), and S is surfacearea (cm•), and c•is fromGreenewalt (1975a), or 2.221;B, derivedfrom the equationb = c•.S O •0•, which predictsthe observedwingspan of G. californianusfrom the surfacearea (13,119.5 cm •) predictedfor that species from its observedweight; C, derivedfrom the equation b = c•-S ø.•'• of Greenewalt(1975a); D, estimatedfrom the ratio "lengthof humerus: wingspan"of G. californianus,or 1:10;E, estimatedfrom the ratio "total wing-bonelength: wingspan"of G. californianus,or 1:3.3. a Predictedfrom the equationAR = 4.49-bø'n• of Greenewalt(1975a), where AR is aspectratio and b is wingspan(cm), using the estimated wingspanfrom columnB. • The valuefort - 2sis 11.1kg and thatfor• + 2sis 16.7kg. The 95% confidenceintervals for the valuesare: fc, (12.5-15.0 kg); .• - 2s, (10.1- 12.1 kg); andre+ 2s, (15.2-18.4);n = 37. t The 95% confidenceinterval for the predictedvalue as representative of an individualis 38.3-166.2kg; the 95%confidence interval for the predictedvalue as representativeof a populationis 70.6-90.2kg; n = 1. Equationsfor determinationof 95% confidenceintervals from Simpson et al. (1960). • Calculatedby assuminghumerus represents 31% of totalwing-bone length, as observed for T. merr•ami. • The 95%confidence interval for the predictedvalue is 9.4-11.1kg; n = 3. I FromKoford (1953), with rangeof 7.7-14.1kg, s = 1.8kg, andn = 14. • FromKoford (1953), with rangeof 250.0-290.0cm; n = 33. k Measuredon an LACM mount.The predictedsurface area for G. californianus,using the equationin b aboveand basedon the mean observedweight, is 13,141cm 2. bird's weight with a fair degree of accuracyif biotarsal measurement of three Recent speci- we have its tibiotarsus. The details of this re- menswas used in the equation,yielding a pre- lationship are still being pursued, but the im- dicted mean weight of 10.2 kg, with a 95% portant point of concern here is the high de- confidenceinterval of 9.4-11.1 kg. Thesevalues gree of correlationthroughout the ClassAves. compare favorably with the mean observed This is significant when working with extinct weight of 14 specimenslisted by Koford(1953), forms whose meansof locomotionare still hy- 9.5 kg (range, 7.7-14.1 kg). The close corre- pothetical. spondence between predicted and observed The mean least circumference of 37 tibiotarsi weights for G. californianussuggests that the of Teratornis rnerriarni in the collections of the weights predicted for the teratornsare fairly Natural History Museum of Los Angeles Coun- accurateand can be used in the equations of ty is 47.0 mm (range,42.5--49.3, s = 1.9), yield- Greenewalt. ing a mean weight for this extinct speciesof Three predicted values for wingspan, result- 13.7 kg, with a 95% confidenceinterval of 12.5- ing from the use of different values for /• in 15.0 kg. equation(2), were derivedfor eachspecies. The The least shaft circumferenceof the only smallestwas derived using the value required known tibiotarsusof Argentavismagnificens is for dimensional similarity, or /• = 0.50. The 94.0 mm, double the mean value of Teratornis largestresulted from using the value of/• fitted rnerriami.The predictedweight is 78.8 kg, with by Greenewalt (1975a), 0.5313. The interme- a 95% confidenceinterval of 36.8-166.2 kg. If diate value came from using a value of 0.5078 the single specimenis treated as the mean of for/•. This value was calculatedfor a population,however, the 95% confidencein- californianusby predicting that species'wing- terval for the predicted value becomes 70.5- surfacearea from its observedweight and solv- 90.2 kg. ing the equation above for /• using the pre- This procedurewas applied to the California dicted wing-surface area, the mean observed Condor as a test of its validity. The mean ti- wingspan obtained from the data of Koford 394 CAMPBELLAND TONNI [Auk,Vol. 100

(1953), and Greenewalt's fitted value of 2.221 membralproportions disagrees with the find- for c•. The wingspanspredicted by using this ings of Greenewalt (1975b) that "... the hu- value for/• most closelyagree with the wing- merus will become a greater fraction of the spans predicted by using wing bone propor- wingspan as size increases."It appears,how- tions (see below and Table 1). ever, that he included in his calculations data Greenewalt's fitted value for/• was altered, fromseveral different morphofunctional groups as opposedto that of c•, because,in the caseat of birds, which may have given misleadingre- hand, a changein the fitted value of/• of one suits.Because intramembral proportions do not standard deviation (0.00529)alters the predict- changesignificantly in New World vulturesof ed wingspan value by 4.5%, while a corre- various sizes, the humerus would have to-in- sponding changein c• (s = 0.01688)alters the creasein lengthat a ratefaster than the primary predictedwingspan value by only0.75%. If the for his conclusion to hold for them. data used by Greenewalthad producedan es- Data to supportor disprovethis hypothesisare timation equation with a fitted value of 0.5078 unavailable. If the relationship between hu- for/•, the fittedvalue for c•would undoubtedly meruslength and the length of the primary have varied slightlyfrom 2.221, but, by ignor- featherswithin this morphofunctionalgroup ing this variation,we believewe introduceonly were known, it should be possibleto predict a minor error into the results. the wingspanof an extinctspecies belonging It is clearfrom this exercise,granting its po- to this group from the length of its humerus. tential for error, that the value for/• in the above At present we can only assumethat the ratio equation is closerto that required for dimen- of primaryfeather length to wing lengthwithin sional similarity than Greenewalt estimated a morphofunctionalgroup is constant. from the data available to him and that his es- The ratio "mean humerus length:mean timation equation cannot be used to predict wingspan"for Gymnogypscalifornianus is 1:10, the wingspansof large birds. The data he used based on four Recent humeri and the mean ob- were heavily weighted toward small- to mod- served wingspan calculated from the data of erate-sized birds, as are the data presented in Koford (1953).The ratio "mean wing skeleton Fig. 1. The lack of data on wing-surfacearea length: mean wingspan" for the samespecies for large birds may have skewed his results, is 1:3.3, based on three completeRecent skel- giving too large a fitted value for/•. A possible etonsand the samemean observedwingspan. similar effectis noted with the computationof Wing skeletonlength is the sumof the lengths the estimationequation relating log weight to of all the bonesof one wing. Wingspanspre- log least tibiotarsuscircumference. Removal of dicted for Teratornis merriami from these two data for larger birds gives an equation that ratios(Table 1) are virtually identicalto the in- underestimatesthe weights of large birds. In termediatevalue predicted by the two-stage fact, if more data for large birds suchas rheas, prediction from tibiotarsal circumference to emus, and cassowaries were available, we be- body weight to wingspan, and they are close lieve the predicted weight of Argentavismag- to the value predictedby dimensionalsimilar- nificenswould be greater than that presented ity. above. The wingspanpredicted for Argentavismag- A secondway of estimatingthe wingspan of nificensby the first ratio is uncertain, because teratorns uses wing-bone proportions (Table the single known humerus of that speciesis 2). The relative lengths of the different bones incomplete (Fig. 2). Use of the secondratio to of the wing, i.e. the intramembralproportions, predict the wingspan of A. magnificensis ten- vary only slightly from the smallestvulturid, uous, becausethe wing-skeletonlength for this Coragypsatratus, to the largest,Vultur gryphus. species is unknown. The percentage of the There appears a slight trend with increasing wing-skeletonlength representedby the hu- size toward an increase in the length of the merusdoes not changesignificantly from Cor- ulna relativeto the more distalwing bones,but agyps atratus to Teratornismerriami (Table 2), no suchtrend is apparentfor the humeri. These where there is an overall size increase of 142%, intramembral proportions in Teratornismerria- but we cannotbe certainthat this relationship mi do not differ significantlyfrom thoseof the will continueto hold when the wing-skeleton vulturid species. This consistencyin intra- lengthis doubledonce again. Nevertheless, the April1983] TeratornSizeand Locomotion 395 396 CA• AN• Tokai [Auk, Vol. 100

o 21oCM

Fig. 2. Wing bonesof (A) Argentavismagnificens (humerus only); (B) Gymnogypscalifornianus; (C) Tera- tornismerriami; and (D) Diomedeaexulans, showing general structure, stoutness,and proportions. wingspanspredicted for A. magnificensby these maining speciesof teratornsthat size estimates two ratios are reasonablyclose to the smaller are very uncertain. Teratornis incredibilishas of the values calculated from the predicted been estimatedas being approximately40% weight. larger than T. merriami, with a wingspan of The small but almost continuous increase of about 5 m (Howard 1952). This estimate is based 4.4% in the proportionatelength of the ulna on only three fragmentaryfossil specimens(a from the smallest vulturid to Teratornis merria- cuneiform, partial beak, and the distal end of mi must also be considered. If this trend con- a radius)so lackingin diagnosticcharacters that tinued in largerforms, then, unlessthe lengths they may not all represent the same species. of the bones distal to the ulna decreased rela- Cathartornisgracilis is known from only two tive to that of the ulna, the wingspanestimates tarsometatarsi from Rancho La Brea, Califor- given above would be too small. There is also nia. Their length approachesthat of the tar- the question of whether or not the primary sometatarsiof T. merriami,but they are ap- feathersincreased proportionately with the rest proximately one-third narrower, suggestinga of the wing. If the primaries represented the lighter body build. These specieswill be omit- same percentageof the wing as in Gymnogyps ted from further discussions. californianus,the larger primariesof Argentavis magnificenswould have been approximately18- LOCOMOTION 20 cm wide and 140-150 cm long. The aspectratio of a bird's wing is given by Adaptations for both aerial and terrestriallo- dividing the wingspan by the mean chord comotion determine a bird's overall morpho- (width) of the wing, which is the same as the . We think that teratornsflew in a manner square of the wingspan divided by the wing very similar to condors, while their terrestrial area. Birds with a high aspectratio fly differ- locomotion was more similar to that of storks ently from those with low aspect ratios. Be- and turkeys. The evidence for these interpre- causethe aspectratio of 8.7 given for Gymno- tations comes from Teratornis merriami limb gyps californianusby Greenewalt'sestimation bones from Rancho La Brea. equation (Table 1) is much higher than the ob- The largestliving birds capableof sustained served value of 6, it appears that this formula flight are albatrossesand condors.These two doesnot predictthe aspectratios of largebirds. groupsuse the two different flight strategies, The reasonfor this is probablythe lack of large dynamic and up-current soaring,open to very bird data mentioned above in connection with large birds. Beyond a certain size, sustained the estimationequation relating wing surface flapping flight becomesprohibitive becauseof area to wingspan. energy considerationsand physical limits of There are so few specimensof the two re- bone and muscle (Tucker 1977). are April1983] TeratornSize and Locomotion 397

c

o •5c•

Fig. 3. Terminalright wing bonesof (A) Teratornismerriami; (B) Gymnogypscalifornianus; and (C) Di- omedeaexulans, showing the differencein structureof thesebones between dynamic and upcurrentsoarers. Arrowindicates shelf of digit II, phalanx1, that helpssupport some of the primaryfeathers.

dynamicsoarers, adapted for nonflappingflight occasionallyreaching a weight over 15 kg and in oceanic regions with strong, continuous, a wingspan near 3.2 m. No really accuratese- unidirectionalwinds. They havewings with a ries of measurementsis available for V. gry- low camber(nearly flat), a very high aspectra- phus. tio, relatively weak flight musculature,and The role of primary feathersin maintaining short, unemarginatedprimary feathers. The flight, which is not the samein dynamic soar- wing bones are very elongatedand slender, ers and up-current soarers,is reflectedin both particularlythe more distalbones. The humer- the shapeof the primaries and the structureof us and ulna tend to be of equallength and have the distal wing bones. Up-current soarersde- straight,relatively flattened leading edges (Fig. pend upon rising air currents,which are often 2; Table 2). The largestalbatross is the Wan- weak or very localizedand may changedirec- deringAlbatross (Diomedea exulans), which may tion or strengthrapidly, as opposedto the con- weigh more than 10 kg and have a wingspan tinuous strong winds exploited by dynamic of over 3.4 m. soarers.They must be able to react quickly to Condorsare up-currentsoarers, adapted for slight changesin wind directionand strength nonflappingflight over land. They dependfor in order to maximize lift and control for sus- sustainedflight upon rising air currentspro- tained flight. Up-current soarersaccomplish duced by the heating of the earth's surface this with their large, emarginated primaries, (thermals), passing storm fronts, or surface which act as independent airfoils when sepa- winds deflectedupward by physiographicfea- rated to provide more lift than if they were tures suchas ridges or cliffs. They have wings unemarginatedand incapableof independent of high camber,a low aspectratio, flight mus- movementas in dynamicsoarers (Oehme 1977). cles that are weak (but stronger than those The primary feathersare anchoredto and rest found in the WanderingAlbatross), and long, upon different portionsof the carpometacarpus emarginatedprimary feathers. All of their wing and carpal phalanges,which increasepropor- bones are very stout. The humerus is much tionatelyin size in formswith largerprimaries. shorter than the ulna; both are rounded with Each distal wing joint is flexed or extendedas no tendencytoward flattened leading edges, needed, via tendons of the middle wing mus- and they are more curvedthan thoseof alba- culature, to separateor bring togetherthe pri- trosses.The (Vultur gryphus) maries. The shape of the articular facets and is slightly larger than the , the largersize of the tendonanchorages of the 398 CAMPBELLA•D To• [Auk, Vol. 100

TABLE3. Ratios involving the carpometacarpusand digit II, phalanx 1 of the five speciesof New World vultures, Teratornismerriami, and the Wandering (Diomedea exulans).

Length of carpometa- carpus: Mean wing-skeleton wing-skeleton Digit II, Phalanx 1 length (mm) length width: length Coragypsatratus 423 1:5.6 1:2.5 (n = 7) aura 462 1:5.7 1:2.5 (n = 13) Sarcoramphuspapa 528 1:6.0 1:2.3 (n = 2) Gymnogypscalifornianus 834 1:6.0 1:2.3 (n = 3) Vultur gryphus 878 1:6.1 1:2.5 (n = 4) Teratornis merriami 1,024 a 1:6.1 1:2.3 Diomedea exulans 1,083 1:7.8 1:4.9 (n = •)

From Fisher (1945).

distal wing bones of up-current soarersreflect T. merriamiwas much larger than ours (23 kg this movement, as opposedto the structure of vs. 13.7 kg). Becausethe overallwing-skeleton the albatrosses,where the primaries are held length of T. merriamiwas not muchgreater than togetherin an extendedposition for long pe- that of Gymnogypscalifornianus, it was reason- riods of time. able to assumethat more flapping would be A clue to the relative size of primary feathers necessaryto get the bird airborne and that its is the "width: length" ratio of carpaldigit II, greater wing loading would reduce its soaring phalanx I (Table 3). Two or three primaries may capabilities. The lower weight estimate re- anchor to this bone, and the width of the shelf duces this problem. supportingthem is a rough indicationof their Second,Fisher thought the more distal wing size. This ratio and the shapeof the bone (Fig. bonesof Teratornismerriami, especially the car- 3) are very similar among all speciesof New pometacarpus, were relatively weaker than World vultures and Teratornis merriami, but those of condorsand incapableof supporting quite differentfrom that of the WanderingAl- largeprimaries. The absenceof largeprimaries batross. Teratornis merriami probably pos- would have restrictedthe teratorn'sability to sessedprimary feathers similar in form and soar. The shaftsof metacarpalII and III are rel- function to those of condors. Confirmation atively smaller in T. merriamithan in Gymno- comes from the similarities in both relative size gypscalifornianus, but the overallwidth (meta- and form between the carpometacarpusof T. carpalII plus metacarpalIII) at the centerof the merriami and those of condors. In addition to carpometacarpusis proportionatelythe same the intramembral proportions, the general (Fig. 3). Fisher also interpreted the areas of structureof all the wing bones of T. merriami muscleorigin and insertionon T. merriamiwing is very similar to that of the wing bones of bones to mean that teratornshad a greater de- condors.These data suggestthat the wings of velopment of the proximal wing musculature T. merriami looked and functioned like those and relatively reduced distal musculaturecom- of condors. pared to condors,thus indicatinggreater pow- Fisher(1945) thought the flight of Teratornis er for flapping flight. This may be true, but, as merriami involved more flapping than that of Fisher noted, the power potential of musdes condors, perhaps resembling more the slow, may not be reflectedin the size of their origin steadyflapping observedin heronsand peli- and insertion, and it is possiblethat the larger cans. Our conclusions differ from his for two size of the teratornwings may have resultedin reasons.First, his estimationof the weight of a different physicalconfiguration of the flight April 1983] TeratornSize and Locomotion 399

T^BL• 4. Ratiosof the hindlimb elementsof Gymnogypscalifornianus, Vultur gryphus,and Teratornismer- riami. n's refer to values for this study.

Mean total length (mm) of femur + Femur: tibio- Tarsometa- tibiotarsus + tarsus: tarso- tarsus: tibio- tarsometatarsus metatarsus tarsus

Gymnogypscalifornianus 480 (464a) 1:1.58:0.86 1:1.83 (n = 3) Vultur gryphus 526 (526a) 1:1.59:0.88 1:1.87 (n = 2) Teratornis merriami 516 a 1:1.50:0.91 1:1.64

From Fisher (1945).

muscles than that found in condors. In other faces. The longitudinal axis of the tarsometa- words, the larger size of teratornsmay have tarsusis more in line with digit II than it is in been more important than differencesin flight condors, resulting in greater stability. The mode in producing the differencesin muscle metatarsalfacet shows that the hallux was po- attachments that Fisher observed. sitioned farther distally and located closer to Available wing bonesof Argentavismagnifi- the longitudinalaxis of the tarsometatarsusthan cens(the humerus and partial ulna, radius, and in condors. The toes of Teratornis merriami were carpometacarpus)are of similar shape but al- rather long, as in condors,and their terminal most twice the size of those of Teratornis mer- phalangeswere long, moderatelycurved, and riami. On the ulnar fragment the papillae for blunt at their ends,indicating that the feet were the attachmentof secondaryfeathers are almost not used for catchingor holding prey as in twice as far apart as those of T. merriami (30 , ,and owls. The long toesand the mm vs. 15-18mm), indicatingthat the second- posteriorposition of the halluxprovided great- ary featherswere about twice as broad as in T. er stability but may have reduced the bird's merriami.Large secondary feathers and a prop- ability to run. erly proportionedcarpometacarpus indicate that The pelvis revealsmuch about how the leg the wings of A. magnificenswere fully adapted of Teratornismerriami functioned. The angle for flight. This, plus the following leg-bone between the long axesof the preacetabularil- data, indicates that A. magnificenswas simply ium and the postacetabularilium in T. merria- a larger version of the La Brea teratorn mor- mi is about 165ø, or nearly the same as in the photype. storks (Fig. 4) and other birds that walk a lot The morphologyof their leg bonesand pel- searchingfor food. The same angle in Gym- vis suggeststhat teratornsmay have been ca- nogypscalifornianus is 155ø and in the Golden pable of extendedwalking and were more agile Eagle (Aquilachrysaetos) 140 ø. The downward on the ground than are condors.As with the tilt of the postacetabularportion of the pelvis wing, the lack of articulatedspecimens of Ter- is an adaptation for bringing the postacetab- atornis merriami renders uncertain the exact ular pelvic musculaturemore nearly parallel to proportionsamong the three major leg bones. the vector force required to pull the femur The meanvalue for eachbone suggests that the backward, as noted by Fisher (1945). This leg of T. merriamiwas shorterthan that of Vul- adaptationis generallyfound only in birds of tur gryphusand longer than that of Gymnogyps prey, , and other birds that use their californianus(Table 4). The ratios between the hindlimbs for specializedgrasping or preda- femur, tibiotarsus, and tarsometatarsussug- tion. The moderate downward tilt of the post- gestthat, in comparisonto condors,teratorns acetabularpelvis in condors and other vultures had a slightlyshorter tibiotarsus relative to the may result from the use of their feet to hold tarsometatarsus,but this small difference may food itemswhile they use their beak to tear off be an artifactof sampling. piecessmall enough to swallow. The leg bonesof teratornsare very stout and While the depression of the postacetabular columnar,lack large crestsfor muscleattach- pelvis increasesthe leg's graspingforce, it also ments, and have relatively flat articular sur- limits the anteroposteriormovement of the leg 400 CAMPBELLAUO Tom•x [Auk, Vol. 100

bones, and the structureof the phalangessug- gest that T. merriamidid not use its feet for grasping, that its stride was as long as its rel- atively short legs would permit, that it proba- bly was not a goodrunner, and that its terres- trial locomotion was not awkward. The hindlimb of Argentavismagnificens is known only from the shaftsof a tibiotarsusand a tarsometatarsus, both of which repeat the stout, columnar structure found in Teratornis merriami. The position of the metatarsal facet indicates that the hallux was also directed pos- teriorly in A. magnificens.Terrestrial locomo- tion was probably similar in the two forms. There has never been any seriousdoubt that Teratornismerriami could fly, althoughhow it took off has been debated. Fisher (1945) con- sideredT. merriamimore capableof getting air- borne by flapping than Gymnogypscaliforni- anus and thought that it neither required nor was capableof a long run for takeoff.The pres- ent, lower weight estimate suggeststhat the takeoffof T. merriamimay have been quite like Fig. 4. The pelvis of (A) Ephippiorhynchussene- that of condors.Following the proceduresout- galensis,Saddle-billed ; (B) Teratornismerriami; and (C) Gymnogypscalifornianus, showing the angle lined above for predicting weight and wing- between the preacetabularilium and the postacetab- span, one can predict wing loading for T. mer- ular ilium. Dashed lines indicate approximateposi- riami, although the use of Greenewalt's tion for edgesof bone, which have been erodedfrom estimationequation for relatingweight to wing- all specimensof T. merriami, although most of the surfacearea may lead to erroneousresults for pelvis is available. large birds for the same reason discussedin connection with his other estimation equa- tions. The predicted value is closeenough to by shortening the musclesthat pull it back- that of G. californianus(Table 5) to suggestthat ward and by physically obstructingposterior a teratorn could have taken off simply by movement of the femur. In birds that are spe- spreading its wings into the wind, as condors cialized for swift terrestrial locomotion, such often do. The less restrictive pelvis in T. mer- as ostrichesand roadrunners,or thosethat may riami may have increasedits running ability walk a lot searchingfor food, such as storks, sufficientlyover that of condorsto enableit to herons, and turkeys, the axesof the preacetab- become airborne with greater ease. As de- ular ilium and the postacetabularilium are vir- scribed above, the available osteologicalevi- tually parallel. dence suggeststhat once airbone, T. merriami The generalproportions of the pelvis of Ter- flew like condors. atornis merriami are very similar to those of It is difficult to conceivethat Argentavismag- storks, as is the lack of any great depressionof nificensbecame airborne solely by flapping its the postacetabularilium. Although nearly the wings, or without wind assistance.With each samelength, the width of this teratorn'spelvis wing over 3 m long, the bird would have to be is about 40% greater than that of a condor. high into the air before it could get in a full Fisher (1945) incorrectlystated that in T. mer- wingbeat. With a weight of 80 kg, it is doubtful riami the postacetabularilium was depressed thatA. magnificenscould jump sufficientlyhigh as in Gymnogypscalifornianus and concluded to give itself enoughtime to flap its wings and that terrestrial locomotion in teratorns was become airborne, even if it were capable of probably as awkward as in condors. This is running with a bounding mode in synchrony clearly not the case.The featuresof the pelvis, with flapping,because of the problemof wing- the stout, columnar nature of the hindlimb beat frequency:the larger the bird, the lower April 1983] TeratornSize and Locomotion 401

TAs•,E5. Totalweight, wing area,and wing loadingfor severalspecies of largeflying birds.

Total weight Wing area Wing loading (S) (,c;m) (s/cm2) Cygnusolor a 11,602 6,808 1.70 Mute Otis tardab 8,950 5,728 1.56 Great Diomedea exulansb 8,502 6,206 1.37 Gypsfulvus b 7,269 10,540 0.69 Griffon Leptoptiluscrumeniferus b 7,030 8,225 0.86 Branta canadensis• 5,662 2,820 2.01 Canada Goose Aquilachrysaetos 4,664a 6,520 0.72 GoldenEagle 3,712b 5,382 0.69 Meleagrisgallopavo a 3,897 3,752 1.04 Turkey Tetraourogallus b 3,361 1,412 2.38 Capercaille Gymnogypscalifornianus c 9,500 13,160 0.72 California Condor 10,200 13,915 0.73 Teratornismerriami ½ 13,700 17,497 0.78 Argentavismagnificens c 79,800 69,609 1.15 Data from Poole (1938). Data from Magnan(1922). Data from Table 1.

the wing beat frequency (Greenewalt1962, the bird could become stranded in a food-poor Goldspink1977). The reasonsfor this may lie area. in the mechanicalproperties of muscle and Chartsof presentglobal wind patternsshow tendon tissues,and the ability of these tissues that SouthAmerica south of approximately40øS to respondto the forcesimposed upon them latitude lies within a band of globe-encircling (Goldspink1977). Inertia, elasticity,and elastic westerlywinds. North of thislatitude the winds rebound of the tissuesand bonesof suchlarge are controlledby high and low pressuresys- wingsmust also have presentedsevere prob- tems. North of 31øS,the South PacificHigh ef- lemsfor activeflapping flight. fectivelyeliminates the westerliesfrom the The wing loading predictedfor Argentavis lower levelsof the atmospherethroughout the magnificensis less than that of severalliving (Miller 1976). During winter in the south- speciesof largeflying birds (Table 5). Although ern hemisphere,the strengthof the SouthPa- the wing loading of A. magnificensis greater cific High increases,and the westerliesare than that of Gymnogypscalifornianus, it is less forcedfarther southwardthan during the sum- than that of Diomedea exulans, both of which mer. The height and width of the Andes canbecome airborne by spreadingtheir wings Mountainsnorth of approximately30øS are ma- into the wind. The higher wing-loadingvalue, jor factorsgoverning the locationof the South over that of Teratornismerriami, may have made PacificHigh and its effecton the westerlies. it more difficult for A. magnificensto take off, North of that latitude, the high altitude of the but high wing loading is more advantageous Andes alone would be sufficient to block the for cross-countrysoaring (Pennycuick 1972). If westerlies. A. magnificenscould not takeoff withoutassis- The major orogenicmovements leading to tance, it either flew only during periods of the elevationof the Andes Mountainsbegan in windy weatheror it lived where winds blew the Miocene. Before the elevation of the Andes, constantly.The former seems unlikely, because and the arrival of the South American conti- 402 CAMeBELLAND TONNI [Auk, Vol. 100 nent at its present position on the surfaceof fauna of La Carolina, . Smithsonian the earth, the wind patterns acrossthe conti- Contrib. Paleobiol. 27: 155-168. nent would have differed from thoseof today. ß & E. P. TONNI. 1980. A new of ter- An important differenceis that the northward atom from the Huayquerianof Argentina (Aves: Teratornithidae). Contrib. Sci. Nat. Hist. Mus. extent of the southern Testerlies would not have Los Angeles County 330: 59-68. beenlimited by the blockingaction of the South ß & -- 1981. Preliminary observations PacificHigh or the Andes, and much more of on the paleobiologyand evolution of teratorns the continentwould have experiencedthe con- (Aves: Teratornithidae). J. Vert. Paleontol. 1: 265- tinual westerlywinds. The singlespecimen of 272. Argentavismagnificens was found in late Mio- FISHER, H. 1945. Locomotion in the fossil vulture cenedeposits approximately 37øS, just north of Teratornis. Amer. Midland Natur. 33: 725-742. the present limit of the Testerlies. GOLDSeINIC,G. 1977. Mechanics and energeticsof In the regions of dominated by the muscle in animals of different sizes, with par- ticular referenceto the musclefibre composition Testerlies, "The wind... hardly ever lets up: of vertebratemuscle. Pp. 37-55 in Scaleeffects exceptfor protectedinterior valleys, calmssel- in animal locomotion(T. J. Pedley, Ed.). New dom occur" (Miller 1976: 115). And, as noted York, Academic Press. by Prohaska(1976: 14) in referenceto parts of GREENEWALT,C. 1962. Dimensional relationships Argentina under the influenceof the Tester- for flying animals.Smith. Misc. Coil. 144:1-46. lies, "In few parts of the world is the climate 1975a. The flight of birds. Trans. Amer. of the region and its life so determined by a Phil. Soc., n.s. 65: 167. single meteorologicalelement, as is the climate 1975b. Letter to editor. Science 188: 676. of Patagoniaby the constancyand strengthof HOWARD,H. 1950. Wonder bird of the . the winds." If similar continual, strong west- Los Angeles County Mus. Leaflet Ser., Sci. 3: 1-3. erly winds were presentthroughout southern 1952. The prehistoric avifauna of Smith South America until the latter portion of the Creek Cave, , with a descriptionof a new Tertiary, they would havebeen more than suf- giganticraptor. Bull. SouthernCalifornia Acad. ficient to carry Argentavismagnificens aloft Sci. 51: 50--54. whenever it spread its huge wings. The evo- 1962a. Fossil birds. Los Angeles County lution of this teratom to its gigantic propor- Mus. Sci. Ser. No. 17, Paleontol. 10: 1-44. tions was probablydirectly tied to the presence 1962b. A comparisonof prehistoric avian of these continuouswesterly winds. assemblagesfrom individual pits at RanchoLa Brea, California. Los Angeles County Mus., Contrib. Sci. 58: 1-24. ACKNOWLEDGMENTS KOFORD, C. 1953. The California Condor. New We thank Marion Mengel, StorrsOlson, Amadeo Yorkß Dover Publ. Rea, and Robert Storer for providing data on bird LIcoN, J. 1967. Relationshipsof the cathartid vul- weights and tibiotarsal dimensions and Charles tures. Occ. Pap. Mus. Zool. Univ. Michigan 651: Hogue for advice on computerusage. Hildegarde 1-26. Howardß William Akersten, Les Marcus, Storrs Ol- MAGNAN,A. 1922. Les characteristiquesdes oiseaux son, Ralph Schreiber,and JohnWright criticallyre- suivant le mode de vol. Ann. Sci. Nat. Ser. 10, viewedan earlydraft of this paperand offeredmany 5: 125-334. valuablesuggestions. We are especiallygrateful to MILLER,A. 1976. The climate of Chile. Pp. 113-145 John Andersonfor allowing us to pursuehis idea in Climates of Central and South America. World about the relationship of body mass and leg-bone surveyof climatology,vol. 12 (W. Schwerdtfeg- dimensions and to Jill Littlewood for volunteering er, Ed.). New York, Elsevier Sci. Publ. Co. her time and talents to do the illustrations. A Na- MILLERßL. 1909. Teratornis,a new avian genusfrom tional Geographic Society grant to KEC was instru- Rancho La Brea. Univ. California Publ., Bull. mental in initiating this study. Dept. Geol. 5: 305-317. 1910. The condor-like vultures of Rancho

LITERATURE CITED La Brea. Univ. California Publ, Bull. Dept. Geol. 6: 1-19. ANDERSONßJ. F., H. RAHN, & H. D. PRANCE. 1979. OEHME,H. 1977. On the aerodynamicsof separat- Scalingof supportive tissue mass. Quart. Rev. ed primaries in the avian wing. Pp. 479-494 in Biol. 54: 139-148. Scaleeffects in animal locomotion(T. J. Pedley, CAMeBELL,K. E., JR. 1976. The late Pleistoceneavi- Ed.). New York, Academic Press. April 1983] TeratornSize and Locomotion 403

OLson,S. 1978. Multiple originsof the Ciconiifor- REA, A. In press. Cathartid phylogeny:a reexam- mes. Proc. Col. Waterbird Group 1978: 165-170. ination of the evidence. In Vultures of the world PE•¾cu•c•c, C. 1972. Soaring behavior and per- (S. Wilbur, Ed.). formance of some East African birds, observed S•aPSO•, G. G., A. RoE, & R. C. LEwo•T•. 1960. from a motor glider. Ibis 114: 178-218. Quantitativezoology. New York, Harcourt,Brace POOLE,E. 1938. Weights and wing area of 143 & World. speciesof North American birds. Auk 55: 511- SToc•c,C. 1956. Rancho La Brea. A record of Pleis- 517. tocene life in California, 6th ed. Nat. Hist. Mus. PROHASKA,F. 1976. The climateof Argentina,Par- LosAngeles County, Sci. Ser. No. 20, Paleontol. aguay and Uruguay. Pp. 13-112 in Climates of 11: 1-81. Central and SouthAmerica. World surveyof TVCKER,V. 1977. Scalingand avianflight. Pp. 497- climatology, vol. 12 (W. Schwerdtfeger,Ed.). 509 in Scale effects in animal locomotion (T. J. New York, Elsevier Publ. Co. Pedley, Ed.). New York, AcademicPress.

The AmericanOrnithologists' Union will hold its AnnualMeeting in New York from 26 Septemberto 1 October1983. The deadlinefor submissionof abstractsfor papersand postersessions is 15 June1983. For furtherinformation on the scientificprogram contact Dr. GeorgeBarrowdough and for generalinformation on the meetingcontact Dr. LesterL. Short, both at the Departmentof ,American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024.

Fellows and Elective Members are reminded that nominations for Vice President and three Elective Coun- cillorsmay be madein writingto the Secretaryat any time priorto the annualmeeting.

Top View