J. Field Ornithol., 63(3):359-364

SEX IDENTIFICATION IN THE SPANISH IMPERIAL

MIGUEL FERRER AND CLAUDINE DE LE COURT Estacidr•Bioldgica de Dor•ar•a,CSIC Aver•ida Marœa Luisa, Pabelldr• del Perd dlOl$ Sevilla,

Abstract.--An external morphometric criterion for sex discrimination in the Spanish Im- perial Eagle, adalberti,was investigated.Seven measurementswere taken on 38 museumskins. StepwiseDiscriminant Analysis selectedforearm length, which classified correctly94.7% of the . Tarsus length was also a useful criterion for sexing . Repeatability,accuracy for sexingeagles in different conditions(alive, museumskins and skeleton),and agewere tested using tarsus and forearmlength. Forearm length was proposed as the bestmeasurement for sexingSpanish Imperial Eaglesand perhapsother dimorphic raptors.

IDENTIFICACiON DEL SEXO EN AGUILA ADALBERTI Sinopsis.--En este artlculo se investigancriterios morfom•tricos externos para la discri- minaci0ndel sexo en el •guila ImperialIbmfica Aquila adalberti. Para ello se tomaron siete medidasen 38 pielesde museo.E1 Anfilisis Discriminantepor Pasosseleccion6 una sola variable,la longituddel antebrazo(longitud del cubito-radio),la cualclasific6 correctamente al 94.7% de los individuos.La longitud del tarso tambi•n result6 comoun criterio vfilido para determinarel sexoen figuilas.Se estudi6la repetibilidady la precisi6nde la 1ongitud del tarsoy del antebrazopara el sexadode figuilasde diferentescondiciones (vivas, pieles de museoy esqueletos)y edades.La ,longituddel antebrazoes propuestacomo la mejor medidapara determinarel sexoen Aguilas ImperialesIb•ricas y quizilsotras rapaces dim6rficas.

Studiesof many aspectsof the biology of a necessitatean easy, reliable methodof sexdetermination. Ideally, a techniquefor sexingwould be valid for individuals of all agesand under different conditions. As sexual dimorphismin size is widespreadin birds, morphometrics oftenprovide a reliablemethod (Green 1989). Despitethe potentialvalue of morphometrics,reliable criteria for sex determinationusing external characteristicshave been developedfor only a few raptors (Bortolotti 1984a,b,c; Edwards and Kochert 1987; GraceIon et al. 1985). No criteria enablingreliable sexingfrom external characteristicshave been described for the SpanishImperial Eagle, Aquila adalberti,although the sexesre- portedlydiffer in size(Cramp and Simmons1979). With the development of captive breeding and reintroductionprograms for this endangered species(Collar and Andrews 1988), accuratesexing is particularly im- portant. The goal of our study was to find an accurateand versatile morphometriccriterion for sexdetermination in SpanishImperial Eagles valid for different statesof preservation(alive, museumskins and skel- etons) and for different age classes.

MATERIAL AND METHODS We took sevenmeasurements of each of the 38 skins of the Spanish Imperial Eagles from the collectionof the Biological Station of Dofiana.

359 360] M. Ferrerand C. dele Court j. FieldOrnithol. Summer 1992 These eaglesdied between 1960 and 1988, and all were found in An- dalucia, south west Spain, mainly in Dofiana National Park. Sex was determinedby internal examinationby the collectors.Individuals were classifiedby plumage as either immature (lessthan 5 yr old) or adult (Ferrer and Calderon 1990). Using callipers,we took the following measurements(to the nearest 0.1 mm) on the left sideof eachskin: tarsus length (TARS), bill including cere length (BCER), length of exposedculmen without cere (BCUL), and lengthof hallux claw (HALC) (Bortolotti1984a,c). Wing (WING) and tail length (TAIL) were measuredwith a metal ruler to the nearest mm (Bortolotti 1984b). We also measuredthe forearm length (FORE) which is the length from the front of the folded wrist to the proximal extremityof the ulna. FORE is measuredby laying a ruler alongthe external side of the bone (Fig. 1). Mean, standarddeviation, and range for all variableswere calculatedfor eachsex and eachage class.All the variableswere normally distributed.Length of hallux claw had a large number (18) of missingvalues and was excludedfrom the analysis.The 38 individuals in the analysiswere 22 females (17 immatures and five adults), and 16 males (12 immatures and four adults). Adults and immaturesmay differ in size (e.g., Bortolotti1984b,c), so multivariateanalysis of variance(MANOVA) and univariateanalysis of variance (ANOVA) of differencesbetween adults and immatures were performed.Next, in order to detectoverall sexual differences, MANOVA wasconducted, followed by an ANOVA for eachof the six morphological characters.We estimatedthe power of ANOVA for each variable (Zar 1984). Sexeswere discriminatedwith a StepwiseDiscriminant Analysis usinga jackknifeprocedure (Dixon et al. 1983) to calculatethe percentage of misclassification. The following stepswere only performed on TARS and FORE, the two best predictorsof sex. Three observersmeasured one study skin 10 times. Coefficients of variation for both variables were calculated for each person.A test amongobserver variances (Levene test) was performedfor bothmeasurements. To evaluatethe adequacyof measurementsfor sexing in the nest, we comparedTARS and FORE using the data from nine eaglesmeasured and banded at the nest (52-72 d old) and later found deadafter fiedging(110-370 d old) and sexedinternally (six femalesand three males). We also comparedTARS and FORE on skeletalremains of 11 freshlydead eagles (after ligamentsand tissueswere removed)sexed internally (six femalesand five males).

RESULTS MANOVA showeddifferences between immaturesand adults (Ho- telling T 2 = 19.52, df --- 6, 31, P = 0.0024). ANOVA revealedthat TAIL was greater in immaturesthan in adults (F = 10.70, df -- 36, P = 0.0024). We thereforedid not includeTAIL in the multivariateanalysis. MANO- VA for the five measurementstaken togethershowed highly significant differencesbetween the sexesfor all measurements(Hotelling T 2 = 123.38, Vol.63, No. 3 SexDetermination in Eagles [361

FIGURE 1. Measurementof forearmlength (FORE) in SpanishImperial Eagle. df = 5, 32, P -- 0.0001). ANOVA revealed that FORE, TARS and WING are the mostdimorphic variables (Table 1). With only one pre- dictor variable, FORE, in a discriminantfunction analysis,the correct sex was assignedto individuals 94.7% of the time, and this percentage was maintainedusing a jackknife procedure.One male and one female of the 38 birds were misclassified. The discriminant function was: female-- 6.96 + FORE - 843.37; male = 6.41 + FORE - 716.91. A FORE of 232.7 mm separatesmales, which are smaller,and females. The frequencydistribution of males and femalesalong the discriminant axis is representedin Figure 2. TARS alone in a StepwiseDiscriminant Functionanalysis also correctly classified 94.7% of the birds. Coefficients of variation for each of the three observers for 10 FORE measurementswere 0.4%, 0.24% and 0.24%, respectively.For the observer with the leastexperience the variationwas highest.For TARS, CV values were 3.5%, 1.0% and 3.3%. The CV for FORE was nearly an order of magnitudeless than for tarsus. Significantdifferences between observers for tarsus (P (0.05), but not for forearm measurements(P • 0.05) were shown by the Levene test for comparisonof variances.The correlation 362] M. Ferrerand C. dele Court J.Field Ornithol. Summer 1992

Tt•LE 1. Sex discrimination of Spanish Imperial Eagle: mean (•), standard deviation (SD) and range of measurementsand analysisof variance(ANOVA). Values of •5are indicated for a = 0.05.

Female (n = 22) Male (n = 16) ANOVA SD Range • SD Range F P

FORE 242.2 5.53 232-255 223.3 6.37 213-235 95.19 0.0001 <0.01 TARS 102.1 3.15 98.9-110.0 93.8 3.14 87.8-99.3 63.03 0.0001 <0.01 WING 613.3 12.81 588-632 565.1 24.28 525-617 62.93 0.0001 <0.01 BCUL 60.6 1.99 56.2-65.0 56.1 4.98 41.0-64.8 15.35 0.001 0.04 BCER 46.0 2.86 39.8-49.4 43.1 3.47 35.6-49.6 7.55 0.01 0.60

Immature (n = 29) Adult (n = 9) ANOVA • SD Range • SD Range F P t5

TAIL 319.4 17.39 276-370 296.7 20.87 270-335 10.7 0.002 <0.01

between TARS measurementsof nestlingsand their study skins was significant(r = 0.791, P = 0.009, n = 9). The correlationwas stronger with FORE measurements(r = 0.993, P = 0.001, n = 9). There was a strong correlationbetween the FORE measurementsof 11 fresh dead birds and the correspondingbones after tissueremoval (r = 0.986, P = 0.001, n = 11), the same was true for TARS measurementsbefore and after tissue removal (r = 0.888, P = 0.001, n = 11). In all casesFORE and TARS from femaleswere greater than for males.

DISCUSSION Our resultsindicate significant size differencesbetween the sexes,and onlyin TAIL betweenimmature and adults.The threebest measurements for discriminatingthe sexesof Spanish Imperial Eagles were FORE, TARS and WING. WING has obviousdisadvantages, however, because it cannotbe usedon individualswith incompletefeather growth, in molt or with worn feathers,or for skeletalmaterial. Using discriminantfunc- tion basedon either FORE or TARS, the sex classificationof only two individualsdisagreed with the sexdetermined from internal examinations. Use of FORE rather than TARS as a discriminatingcharacteristic has importantadvantages, however. FORE is easierto measureand repeated measurementsby the sameand by different observersare lessvariable than repeatedmeasurements of TARS. Tarsusis moredifficult to measure becausethe pointsof measurementare not as clearly definedas FORE, and is more affectedby the observer'sexperience. Relative to the most experiencedobserver, the CV of the least experiencedobserver was 1.6 timesgreater for FORE measurementsand 3.5 timesgreater for TARS. FORE also varies lessamong nestling,museum skin, skeletonand ob- server than does TARS, so FORE was the best sexing criterion for different agesand under different circumstances.Forearm measurement Vol.63, No. 3 SexDetermination in Eagles [363

MALES

freq. I

FEMALES

4

freq.

0 -4 -2 0 2 4

DISCRIMINANT SCORES FIGURE2. Distribution of museumspecimens of male and female SpanishImperial Eagles along the discriminant axis with forearm length as selectedvariable. enablesthe sex of the Spanish Imperial Eaglesto be easily and reliably determined from at least 52 d of age. Other morphometriccriteria for sexing large dimorphic raptors such as Golden Eagles (Aquila chrysaetos)and Bald Eagles (Haliaetusleuco- cephalus)that have been proposedare: bill depth, hallux claw, footpad lengthand bodymass (Bortolotti 1984a,b,c;GraceIon et al. 1985; Edwards and Kochert 1987). All thesemethods have the important drawbacksthat they are only applicable to well preservedskins or live individuals, and someof them differ accordingto the specimen'sage and stageof growth (Bortolotti 1984a,c).The frequentloss of the lining of the beak and claw in corpsesand their different degreeof wear in captiveindividuals makes measurementsof bill depth, hallux claw and culmen length of limited use in population studies.The other two measurementssuggested in the 364] M. Ferrerand C. dele Court J.Field Ornithol. Summer 1992 literature, footpadlength and body weight can only be measuredaccu- rately on live specimensand with boththere is a highvariance for repeated measurements(Edwards and Kochert 1987). Other sexing techniques applied to dimorphic raptors have been karyotype determinationand laparoscopy(GraceIon et al. 1985). Both of thesetechniques are exclu- sivelyapplicable to live individuals,and have seriousdrawbacks. All of thesefactors make a reliable morphometrictechnique, such as forearm length, extremelyuseful for dimorphicbirds.

ACKNOWLEDGMENTS

We thank C. Santosand M. de la Riva for assistancein the field. We are mostgrateful to M. Languy and P. Jordanofor statisticaladvice and to A. Lazo, E. Le Bouleng•,E. Schupp and G. R. Bortolotti for the improvementof the early versionsof the manuscript. This study was supportedby DGCYT, project number PB87-0405 and by a post-doctoral grant to M. Ferrer from ConsejoSuperior de InvestigacionesCientificas.

LITERATURE CITED

BORTOLOTTI,G.R. 1984a. Criteria for determining age and sex of nestlingBald Eagles. J. Field Ornithol. 55:467-481. 1984b. Age and sex variation in Golden Eagles.J. Field Ornithol. 55:54-66. 1984c. Sexual size dimorphismand age-relatedsize variation in Bald Eagles.J. Wiidl.Manage. 48:72-81. CRAMP,S., AND K. E. L. SIMMONS. 1979. Handbook of the birds of Europe, the Middle East, and North Africa. Vol. II. Oxford Univ. Press.Oxford, United Kingdom. COLLAR,N.J., AND P. ANDREWS. 1988. Birds to watch: the ICBP world checklistof threatenedbirds. ICBP Techn. Publ. 8, ICBP Cambridge,United Kingdom. DIXON, W. j., M. j. BROWN,L. ENGELMAN,j. w. FRANE,M. A. HILL, R. I. JENNRICH, ANDJ. D. TOPOREK. 1983. BMDP statisticalsoftware 1983 printing with addition. Univ. California Press,Berkeley, California. 735 pp. EDWARDS,T. C., ANDM. N. KOCHERT. 1987. Use of body weight and length of footpad as predictorsof sex in Golden Eagles. J. Field Ornithol. 58:144-147. FERRER,M., AND J. CALDERON. 1990. The Spanish Imperial eagle Aquila adalbertiin Dofiana National Park (South West Spain): a study of population dynamics. Biol. Conserv. 51:151-161. GRACELON,D. K., M. S. MARTELL,P. T. REDIG,AND L. C. BUOEN. 1985. Morphometric, karyotypicand laparoscopictechniques for determiningsex in Bald Eagles.J. Wildl. Manage. 49:593-599. GREEN,P. T. 1989. Sexing Rooks Corvusfrugilegus by discriminantanalysis. Ibis 124: 320-324. ZAR, J. H. 1984. Biostatisticalanalysis, 2nd Edn. Prentice-Hall, EnglewoodCliffs, New Jersey. 718 pp. Received17 Jun. 1991; accepted27 Nov. 1991.