ORIGINALARTICLE
Sexualsizedimorphismandsexdeterminationbymorphometricmeasurements intheCoscorobaSwan
CeciliaPerezCalabuiga∗, AndyJ.Greenb,MiguelFerrera,RobertoMuriela HedenMoreirac
aDepartmentofBiodiversityConservationandAppliedBiology,EstaciónBiológicadeDoñana-CSIC,Sevilla,Spain;bDepartment of Wetland Ecology, Estación BiológicadeDoñana-CSIC,Sevilla, Spain; cDepartment of GeneticsandZoology, Federal UniversityofPelotas,Pelotas,RS,Brazil
Theaccuracy ofmorphologicalsexingand theoccurrence ofsexualdimorphism wereanalyzed inmature and immature CoscorobaSwans (Coscorobacoscoroba,Anatidae) near the Estação Ecológica do Taim, southern Brazil.Onthebasisofweightand10linearmeasurements ofexternalmorphology,multivariateanalysisofvari- anceshowedthatmaleswere consistentlylargerthanfemales(sexconfirmedviageneticmarkers)andmaturebirds wereconsistentlylargerthanimmatures.Overall,38%ofimmaturesand14%ofmaturebirdsweresexedincorrectly bycloacalexaminationwhencompared togeneticdata.Therefore,weperformed adiscriminant functionanalysis ofdifferentageclassesbasedonmorphometricmeasurements.Mature birdsweresexedwith96%accuracyusing headandtarsuslengthsaspredictor variables,whereasimmatures weresexedwith90%accuracybasedonhead andforearm lengths.Method validation conducted withdata foradditionalmature sampled inadifferentyear showedthattheuseofheadlengthalonewasasaccurateforsexing(92%correctclassification)thandiscriminant functionsbasedontwocharacteristics (91%).
Keywords:Brazil; cloacal examination; Coscoroba Swan; genetic sexing; morphological sexing; sexual size dimorphism
Introduction
Accurateandeasymethodstodeterminethesex of individualsarevaluableforstudiesof avianevolu- tionaryecologyandgenetics,population dynamics, behavior, migration andconservationmanagementof speciesandpopulations(Clutton-Brock1986;Newton
1998).Inspecieslackingobvioussexualdimorphism, cloacal sexing is often used but can be unreliable (Brown et al. 2003; Odwyer et al. 2006, Volodin etal.2009). Molecularsexing,basedontheamplifica- tionofthechromo-helicase-DNA-binding1(CHD1) gene,isthemostreliablemethod(EllegrenSheldon
1997;Griffithsetal.1998),butisrelativelyexpensive. However, molecular sexing can complement meth- ods based on discriminant function analysis (DFA) inspecieswithweaksizedimorphism (Dumbell etal.
1988; Ackerman etal.2008; Hart etal.2009).In waterbirds,DFAhasoftenbeen used tocombinethe discriminatory powerofsingle charactersintoone formula thatbestdiscriminatesbetweensexes (e.g. WeidingervanFraneker1998;SvageljQuintana
2007;Ackermanet al.2008;Quintanaet al.2008;Hart etal.2009).DFA canbeareliable,fastandinexpen- sivemethod fordiscriminating thesexofindividuals
duringnon-breeding seasonswhensexuallydimorphic characters arenotexpressed(e.g.Zavalaga Paredes
1997;Bourgeoisetal.2007).
The Coscoroba(Coscorobacoscoroba)isan unusualmemberoftheAnatidaefoundin South America,fromtheFalkland IslandsandTierradel Fuego, north through ChileandArgentina, Uruguay and southern Brazil, and as far north as Paraguay (Kear2005).InBrazil,C.coscorobaoccursyear-round inRioGrande doSul(Belton2000) withirregular records in other states (Bornschein et al. 1997). It istraditionallyconsidered to beaswan, but molec- ularstudiessuggestitis morecloselyrelatedtothe aberrantCapeBarrenGooseCereopsis novaehollan- diae(Donne-Gousséetal.2002).Verylittleisknown about itsecology(seeKear 2005forareview).It is amonochromatic species,andpreviousattempts to separatesexesbasedonmorphologyhavebeenunsuc- cessful(Nascimentoetal.2001).DFA haspreviously beenapplied insomeother swanspecies(e.g.Miller etal.1988; Brownetal.2003).Otherstudiesofswans havecompared differentsizemeasures betweensexes without applying DFA (e.g.Scott 1972;Mathiasson
1981;2005;Limpertetal.1987).
∗Correspondingauthor.Email:
The present study describes sexual dimorphism ofimmatureandmatureCoscoroba inBrazil,and compares theaccuracyofalternative sexingmethods. Themainobjectiveswereto:(a)determine theaccu- racyofcloacalexamination,bycomparing theresults withthosefrommolecular sexing;(b)present typical morphometricmeasuresofCoscoroba, anddetermine differencesbetweenthesexesandtwoageclassesiden- tifiable byplumagecharacteristics;(c)developdiscrim- inantmodelstofacilitatethesexingofbirdsbasedon morphometry;and(d)comparetheaccuracyofDFA withthatofsexingbycloacalexamination.
Materialsandmethods
Studyarea
RioGrande doSulStateinthesouthern tipofBrazil hasabig lagooncomplexinthesouthernportionof its coastalplain,formedby Patos,Mirim,andMangueira Lagoons andother smallerlakes.Allbirdswerecap- turednearthe “EstaçãoEcológicado Taim”,anarea located inthesouthern coastal plainofRS,between the counties of Rio Grande and Santa Vitória do
Palmar (∼32◦23 W,52◦32 W).This area wascho-
senbecauseitiswithinthemost importantarea for
breeding and molting ofCoscorobainBrazil(Seijas
2001).
Captureofbirdsandsampling
Birdswere capturedbyhandfromaboatduring flightless wingmoltbetweenAugustandJanuary (Nascimentoetal.2001;Seijas2001)inthreeconsecu- tiveseasons(2005,2006and2007).Birdsweredivided intotwogroups:immatures (fledgedbutlessthantwo yearsold)and matures (2yearsold)according to plumagecharacteristics (Johnsgard1978).Coscoroba Swanshavered eyes,legs,feetand billfrom imma- turetomature ageclassesbuttherearedifferencesin theplumagepattern. Theadult’sdefinitiveplumageis totally whitewiththeexceptionofablacktiponthe firstsixexternalprimaries.
Although plumageis mainlywhite,immature Coscoroba Swanshavebrownishdownfeathersand brightishbrownplumesall overthebodyandhave grayish brown stains on the tail, back and wings (in primaries,secondariesandupperwing coverts). Mature swans have gray down feathers only on the backandunderthewings (axillaries).Theyusually displaymorethansix primarieswithgrayishbrown fleckingsbutveryrarelyonthesecondaries(Calabuig etal.2010).
The immature group consisted of41individuals
(19,11and11for2005,2006and2007,respectively).
Themature group consisted of345individuals (120,
123and102for2005, 2006and2007, respectively).
Morphometricmeasurements
Birdswere weighedwithabalance(tothenearest20g) and10 biometricmeasuresweretakenbythesame person (CPC).Adigitalcaliperwasusedtomeasure thefollowing(tothenearest0.1 mm):headlength (length ofhead to the occipital-tip ofthe bill),bill height, billdepth (maximum width ofthebill),nos- tril(distaledgeofanostriltotheendofthebill),total culmenandtarsuslength(ontheleftsideofthebody). Arulerwasusedtomeasurethefollowing(tothenear- est1 mm):tail(fromthepreengland),winglength without feathers (metacarpophalangealarticulation), forearm(FerrerDeleCourt1992)andnecklength. Beforereleasing,allswanswerebanded withanum- beredmetalCEMAVE-IBAMAring.Recaptureswere readilyidentified,anddatawere onlyusedforeach individualonthefirstcapture.
Cloacal andgeneticsexdetermination
Genital identification
CloacalexaminationwascarriedoutbyCPC,identify- ing“males”bythepresenceofavisibleerectilegroove penis on the ventral wallof the cloaca (Proctor Lynch1993;Brownetal.2003;Mathiasson2005).
DNAidentification
Blood(3ml)wastakenfromeachbirdfromthewing veinandsampleswerestoredinVacutainertubeswith EDTAandkeptcoolinice, untilprocessing.Blood sampleswere analyzedintheBiotechnologyCenter (CenBiot),FederalUniversityofPelotas,Brazil.DNA extraction wasperformedaccordingtotheprocedures ofLahiri and Nurnberger(1991)and DNA samples were amplifiedusingprimersdescribedbyHeetal. (2005)for the Tundra Swan (Cygnus atratus). The PCRamplificationswere carriedoutinthereaction mixture ofIto etal.(2003).After amplification, the PCRproducts wereseparated ona1% agarosegel (synthesizedbySangonCo.,Shangai,China),stained with ethidiumbromide(SangonCo.)andvisualized underUVlight.
Dataanalyses
We usedmultivariateanalysisofvariance(MANOVA) to compare mean differences between age and sex groups for morphological measurements and body mass.Also,one-wayanalysisofvariancewasusedto
determinewhetherindividualmeasuresvariedwithsex ineachagegroup(SokalRohlf1995).Forboth analyses,we usedonlythe225matureCoscorobas sampledduringthe2006and2007seasons(thoseused forDFA, seebelow).Differenceswereconsideredsig-
nificant atp≤0.05.Alldata satisfiedLillieforsand
Levenetestsofnormality.
Forallvariables,we calculatedasexualsize dimorphism index as: SSD ={(xm −x¯f)/x¯f}×100 (WeidingervanFraneker1998;SvageljQuintana
2007);wherex¯m andx¯f arethemeanvaluesofdiffer- entage malesandfemales,respectively.Thecoefficient ofvariation (CV =(SD/x¯)×100)wascalculated for each sex and averaged between them (Fletcher
Hamer 2003)toindicate thedegreeofvariability of eachmeasurement (SokalRohlf1995).
DFAsweredevelopedseparatelyforimmatureand maturebirds.Weexcludedbodymassfromtheseanal- yses sinceitvariesgreatlyovertime(Croxall1995; SvageljQuintana2007).Theperformanceofeach singlemeasurement asadiscriminating variable(uni- variate DFA) was evaluated. Forward DFAs were appliedtoobtaincombinationsofcharacteristics (dis- criminant functions) that bestdistinguished thesexes (seeTabachnickFidell 1996;Phillips Furness
1997).
Forimmaturebirds,theDFAwas appliedtoall individuals. For mature birds, the DFA wasapplied to 225Coscorobassampled in2006and 2007.For both ageclasses,theeffectivenessoftheanalyseswas assessed,firstinterms oftheproportionofbirds of knownsexthatwereclassifiedcorrectly,andsecondby jackknife validation. Correct classification rates tend
immature birds;overall,38%ofimmature Coscoroba weresexedincorrectly. Among mature birds, overall,
14%ofmatureCoscoroba weresexedincorrectly.
Morphometricdifferencesaccordingtosexandage
Analysisof thewholedatasetshowedthatmature Coscorobawerelargerthanimmaturesandmaleswere bigger than females, with no age×sexinteraction
(MANOVA: age:F11,242=4.6,p0.01, Wilks=0.82;
sex:F=29,p0.01,Wilks=0.43;andage×sex:
F11,354 =1.5,p=0.1,Wilks=0.93).Thus, ineach
agecohort, maleswerelargerthanfemalesinallmea-
surements, except in tail length for immature birds
(Tables1and2).
Inimmature birds,characteristicsthatshowedthe highestsexualsize dimorphism weremassandtail, whereasbillheight,wingandtailwerelessdimorphic. Mass showed thehighest within-sexvariation whilst headlengthhadthelowest(Table1).Inmature birds, mass,tarsusandneck lengthshowedthehighestsexual sizedimorphisms whereasbilldepthandtailwereless dimorphic.Massandneck lengthshowedthehighest within-sexvariation whilstbilldepthandheadlength hadthelowest(Table2).
Table 1.Body measurements (mean ±SD and range, inmm),coefficientsofvariation (CV) andsexualsize dimorphismindex (SSD) forimmaturemaleandfemale CoscorobaSwans(Coscorobacoscoroba)fromsouthern Rio
Grande doSul,Brazil.Allmeasuredcharacteristicsdiffered betweenthesexes(p0.05)excepttaillength.CVwasfirst calculatedforeachsexandthenaveraged.
One-way
tobeoverestimated whenDFAsarevalidatedwiththe
same sample used to generate them (Tabachnick
MaleFemale
Body
ANOVA(%)
Fidell1996).Thejackknife validation isaprocessin which eachindividualcaseisclassifiedusingafunction obtained fromthetotalsample,excludingtheindivid- ualcasetobeclassified(TabachnickFidell1996). Furthermore,theaccuracyofDFAs formature birds wasconfirmedbyapplyingtheresultingfunctionstoa noveldataset, composed of120birdscaptured inthe
2005season.
Results
Cloacal andgeneticsexdetermination
Genetic sexingshowedthat therewere14malesand
27femalesamongyoungbirds,and189malesand156 femalesamong mature birds.Somebirdswereincor- rectlyclassifiedbycloacalexamination. Although geneticfemaleswererarelymisclassified as“males”, males were often wrongly identified as “female”. Cloacal inspection was particularly unreliable for
measurementn=14n=27F1.39CVSSD
Mass(g)3760±4603230±40014.414.416.1
2650–44002250–4100
Totalculmen70.6±4.665.4±2.422.76.18.0
62.5–79.262–73
Nostril52.7±2.848.7±2.621.46.68.2
48.6–56.641.3–56.3
Billheight25.7±1.024.5±1.56.764.9
23.6–2722.2–28.8
Billdepth30.2±1.428.5±1.313.75.26.0
27.2–32.724–30.1
Headlength138±4.9129.6±3.146.64.26.5
129–146.3 123.1–135
Tarsus101.6±5.695.4±4.414.65.76.5
87.8–108.285–104
Wing173.8±7.1164.1±9.710.965.9
162–185145–185
Forearm213.1±8.3195.9±8.837.75.98.8
200–225175–215
Tail183.1±14.7 173.2±21.92.511.45.7
161–205107–220
Necklength345.4±24.7315±21.816.18.39.6
305–390260–370
Table 2.Body measurements (mean ±SD and range, in mm),coefficientsofvariation (CV)and sexualsizedimor- phismindex(SSD)formaturemaleandfemale Coscoroba Swans(Coscorobacoscoroba)sampled in2006and2007in
southern RioGrande doSul,Brazil.Allmeasured charac- teristicsdifferedbetweenthesexes(p0.05).CVwasfirst calculatedforeachsexandthenaveraged.
Table3.Accuracyofsexingofimmature CoscorobaSwans usingsinglemeasurements or combinationsinadiscrimi- nant function (DF1 =headlengthandforearm). Identical resultswereobtained bythejackknifemethod(seetext).
Originalsample(n=41)(%correct classification)
Body
measurementMalesFemales
One-way
ANOVA(%)
Variables
Wilks’
lambdaMales Females Overall
Cut-off point (mm)
2006–2007n=125n=100F1.223CVSSD
Mass(g)4000±3353340±360197.212.919.8
3050–5140 2400–4500
Totalculmen70.6±2.665.5±2.3241.55.17.8
62.8–78.859.8–71.8
Nostril52.8±1.649±2.2221.75.37.8
48.4–58.840–59.2
Billheight26.3±1.124.6±1.1136.35.57.1
23.7–2922.2–27.5
Billdepth30.7±0.829.2±0.9152.23.74.9
29–33.326.4–31.3
Headlength140.1±2.8131.2±2.8548.93.96.8
132.2–147123.8–138.8
Tarsus105.8±3.896.1±3.4397.3610.1
93.3–117.2 84.6–104.5
Wing178.7±7.5165±8.6162.86.18.3
152–200130–186
Forearm212.6±8196.4±7.7235.35.58.2
177.0–230168–217
Totalculmen 0.63 64.3 92.6 83 69.3
Nostril 0.65 71.4 89 83 51.9
Billheight 0.85 64.3 74.1 70.7 26.2
Billdepth 0.74 78.6 70.4 73.2 30.1
Headlength 0.45 86 89 88 135.0
Tarsus 0.73 78.6 66.7 70.7 101.0
Wing 0.78 78.6 70.4 73.2 173.8
Forearm 0.51 78.6 81.5 80.5 207.1
Tail 0.94 64.3 66.7 66 202.4
Necklength 0.74 78.6 70.4 73.2 339.0
DF1 0.42 78.6 96.3 90.2
onlytwoCoscorobas differedbetweenclassifications (changing thecorrect classification to86%formales and 85%overall). Validation withanovelsampleof birdsprovidedslightly differentclassificationswhen compared with DFA, decreasing the accuracy for
Tail175.4±6.9166.4±8.576.25.15.4
145–190141–187
all measures except for bill height and
bill
depth
Necklength 355.9±18.6 317.3±20.2 221.1 8 12.2
275–400265–365
Discriminantanalysis
Immatureswans
Head length wasthe most accurate singleindicator ofsex,correctlyclassifying88% ofbirds(Table3). Coscorobaswith head lengths longer than 135mm wereclassifiedasmales.Jackknifevalidation provided exactlythesameclassificationsas thoseproducedby DFAsforallsinglemeasurements.DFAretainedhead lengthandforearmasthebestpredictor variablesand excludedothers. Thismodel correctly classified79% offemalesand96% ofmales(Table3).Onlythree femalesandonemaleweremisclassified.Thediscrimi- nantfunctionobtained forimmatureCoscorobaswas:
DF1 =0.43(head length) +0.13(forearm) –84.6.
Values of DF1 0identified males and values of
DF10identifiedfemales.
Matureswans
Head length was the most accurate single indica- torofsex,correctlyclassifying93% ofCoscorobas (Table 4).Jackknife validation provided exactly the sameclassificationsasthoseproduced byDFAsforall singlemeasurements exceptfornostril length, where
(Table4).DFA retainedheadlengthandtarsusasthe
bestpredictors. Thismodelcorrectlyclassified96% of femalesand95%ofmales(overallsuccess,96%,three femalesandfourmalesmisclassified)withalowvalue for Wilks’lambda (Table 4).TheDF1 obtained for
mature Coscorobaswas:DF1 =0.84(head length)
+0.38(tarsus) –152.6.Thismodel wasrepresented
inFigure 1wheremature Coscorobaswereclassified
accordingto headlengthandtarsusmeasurements. Misclassificationswere unusuallylargefemalesmis- classified asmales.
ADFA withonlyhead length aspredictor vari- able classified93%of cases correctly, with a value ofWilks’ lambdaclosetothebestmodelwithhead length and tarsus together. This alternative discrim-
inant function (DF2) was:DF2 =1.1(head length)
–150.DF2performed slightlybetterthan DF1when
validatedagainstanewsampleofbirds(Table4),with
92%ofindividualssexedcorrectly.
Discussion
Sexdeterminationbycloacalinspection
Althoughcloacalexaminationisawidelyusedmethod forsexingwaterfowl(e.g.Green2000;GrayHamer
2001;Nascimento etal.2001),ithasseveraldisadvan- tages.Thismethod requires observers withconsider- ableexperience,isonlypossible during thebreeding
Table4.AccuracyofsexingofadultCoscorobaSwansusingsinglemeasurements andadiscriminant function (DF1=head lengthandtarsus)for2006and2007(originalsampleincludedindiscriminantanalyses)and2005(usedforvalidation,seetext). Ajackknifemethod produced identicalresultstothoseshownherefortheoriginalsample,withtheexceptionofnostril(see text).
Originalsample(n=225)(%correctclassification)Newsamplecross-validation(n=120)
Variables / Wilks’ lambda / Males / Females / Overall / Males / Females / Cut-offpoint(mm)Totalculmen / 0.48 / 88.8 / 87.9 / 88.4 / 92.2 / 75 / 67.8
Nostril / 0.5 / 88 / 83.8 / 86.2 / 98.4 / 54 / 48.8
Billheight / 0.62 / 75 / 81 / 77.7 / 67.2 / 90.2 / 25.3
Billdepth / 0.59 / 80 / 77 / 78.7 / 100 / 78.4 / 19.4
Headlength / 0.29 / 94.4 / 92 / 93.3 / 91 / 93 / 136.3
Tarsus / 0.36 / 90.4 / 89.9 / 90.1 / 86 / 87.5 / 100.4
Wing / 0.58 / 83.2 / 78 / 80.9 / 87.5 / 60.3 / 173.3
Forearm / 0.49 / 87.9 / 84.8 / 86.5 / 97 / 66.1 / 203.4
Tail / 0.74 / 76.6 / 69.8 / 73.6 / 98.4 / 48 / 169.2
Necklength / 0.5 / 87.2 / 82 / 84.9 / 73.5 / 80.4 / 330.8
DF1 / 0.26 / 95.2 / 95.9 / 95.5 / 91 / 91.1
Figure1.Plotof 329matureCoscorobaSwansfromsouthernRioGrandedoSul,Brazil,accordingtoheadlengthandforearm length.Malesandfemaleswereidentifiedbygeneticsexing.Thestraight linerepresents 50%probabilityofsexclassification accordingtothediscriminant function.
periodforsomespecies,andcancauseinternal dam- agetothebirdduetothepressureappliedduringpenis visualization (e.g.SaxHoi 1998;Lombardo2001; Oliveiraetal.2004).Wefound ittobeanunreliable methodfortheCoscoroba,owingtothedifficultiesof observingthepenis,especially inyoungbirdswhere penisdevelopment isincomplete(Odwyeretal.2006). Similar levelsof inaccuracy have been recorded in petrels(Odwyeretal.2006) andinwhistlingducks (genusDendrocygna,Volodin etal.2009).Wedonot believethat the error insexingcan beattributedto an observer effectin our study because the person who sexed the swans was an experienced specialist whohasbeenringingwaterbirdsfor20years.Previous studiesrelyingexclusivelyoncloacalinspectiontosex Anatidae (e.g.Green2000)mayalsocontainerrors.
Sexualsizedimorphismandageeffects
Fewpreviousdata wereavailableonthemorphomet- ricsofCoscoroba (Kear2005). Ourresultsindicate significantsize dimorphism,withmaleCoscoroba generallybeinglargerthanfemales,bothforimmature andmaturebirds.Ingeneral,largesizemaybeadvan- tageousinmaleswansandgeese becausetheyare responsiblefornestprotection(Scott1972;Veselovsky
1973;Hawkins 1986;Whitehead Tschirner 1990) or may need to defend females against males seek- ingextra-pair copulations (Mineau& Cooke1979; McKinney et al. 1983; Welsh Sedinger 1990; GauthierTardif1991;ChoinièreGauthier 1995). AsintheCapeBarren Goose, both Coscorobasexes care for their cygnets and maintain long-term pair bonds,butthemaleisprimarilyaguardian andhelps withnestbuildingbutnotincubation.
Massandneck lengthshowedthehighestdegree of sexualsizedimorphism inCoscoroba, butheadlength wasthemostusefulsingle variabletodistinguish betweenmalesandfemales,becauseit hadaninterme- diatelevelofdimorphism combined withparticularly lowCV.Inmostbirdspecies,intraspecificvariation is markedly lowerinthebillandotherbodypartsasso- ciatedwithfoodintakethanin therestofthebody, perhaps asaresultofadaptationtoparticularforag- ingbehavioranddiet(Milleretal.1988). However, differentCVsbetweenmeasurescanalsoreflect dif- ferencesinmeasurement error(Yezerinacetal.1992), andnecklengthislikelytohavehadarelativelyhigh measurement errorin ourstudy.Ourfindingswere consistentwithpreviousstudiesthatshowedthathead (e.g.Veselovsky1973;Milleretal.1988;Brownetal.
2003),tarsus(e.g.Veselovsky1973;Brownetal.2003; Mathiasson2005)andforearm(e.g. Mathiasson2005) lengthsarekeycharacteristicstodifferentiate sexes in Anseriformes.
CoscorobaSwans reach maturity when they are two years old and start breeding at three to four years old (Wilmore 1979; Silva-García Brewer
2007).Although birdsaregenerallyassumed tohave determinategrowth,ourresultsshowed thatmature Coscorobadifferedsignificantlyinsizefromswansless thantwoyearsofage,implyingthatgrowthcontinues inthisspeciesforanextendedperiod prior toreach- ingsexualmaturity (Kirkwoodetal.1989;Carrier Auriemma 1992; Tumarkin-Deratzianetal.2006). ANOVAs ofindividual measures showed that mass, billdepth, head length, tarsus, tail and necklength were all significant greater in mature birds (results not shown). Growth rates oflarge, precocious birds suchasswansareparticularly slow (Ricklefs1973; CarrierAuriemma 1992).
Sexdeterminationbydiscriminantfunctionanalysis
Discriminantfunctionsweredevelopedusingonlytwo morphometricvariables: head and forearm lengths for immaturebirds,andheadandtarsuslengthsfor mature birds, resulting in 90%and 96%of correct sexual classification,respectively.Theseclassification ratesaremuchhigherthanthosebasedon cloacal inspection.However,cross-validationwithanewsam- pleofmatureswanssuggestedthatasingle measure (head length) wasthe most reliable sexingmethod, with92% correctclassification.Whilejackknifeval- idationrevealsinfluentialobservations thatcanbias DFA, cross-validationis amorerigorousvalidation processandshouldideallyinvolveanew sampleof individualsmeasured atdifferenttimes,locations and bydifferentobservers(TabachnickFidell1996). We suspect that discriminant functions would provide a betteralternative tocloacalsexinginmanyotherbird species.
More research isrequired tocalibrate theuseof ourdiscriminant functions inotherCoscorobapopu- lationswithotherobservers.Giventheresultsof the validation with noveldata, sexdeterminationbased onheadlengthis likelytobethemostrobust dis- criminant function method. Whencloacalinspection iscarried out without geneticsexing,thesedata can beplotted together withheadlengthdatatoestablish thebestcut-offpoint forsexdeterminationbasedon morphometrics.Forexample,inapopulationinwhich Coscorobawerephysicallysmaller,thecut-offpoints ofTables3and4forheadlengthmightbetoohigh.
Acknowledgments
WewouldliketothankS. Schererforhisvaluable assistanceinthefield.WethankDr.B.Vaz, R.Tavares, J.Camacho and L.Bassinifrom theDepartmentof
GeneticsandZoology, Federal University ofPelotas, RS,Brazilfortechnicalhelpin carryingoutDNA extraction.
References
Ackerman JT,Takekawa JY,BlusoJD,YeeJL,Eagles-SmithCA.
2008. Genderidentification ofCaspianTernsusingexternalmor- phology and discriminant function analysis.WilsonJOrnithol.
120(2):378–383.
BeltonW.2000.AvesdoRioGrandedoSul:Distribuiçãoebiologia.
2nded.SãoLeopoldo (Brazil):Editora UNISINOS.
BornscheinMR, Reinart BL,PichorinM.1997.Notas sobrealgu- masavesnovasoupoucoconhecidasnosuldoBrasil.Ararajuba.
5(1):53–59.
Bourgeois K,CuréC,Legrand J,Gómez-DíazE,VidalE,Aubin T,MathevonN.2007. Morphologicalversusacousticanalysis: whatisthemostefficientmethodforsexing yelkouanshearwaters Puffinusyelkouan?JOrnithol. 148:261–269.
Brown AW, Brown LM, Stevick PT. 2003.Sexing Mute Swans
Cygnus olor by discriminant analysis. Ring & Migr. 21:
174–180.
Calabuig CP,Ferrer M,MurielR,TilgarV.2010.Plasmaalkaline phosphataseasasensitiveindicator ofageandskeletaldevelop- mentinwildCoscoroba Swans.WildlifeRes.37(6):504–511.
Carrier DR,Auriemma J.1992.Adevelopmental constraintonthe fledging timeofbirds.BiolJLinnSoc.47(1):61–77.
ChoinièreL,GauthierG.1995.Energeticsofreproductioninfemale andmaleGreater SnowGeese.Oecologia.103(3):379–389.
Clutton-Brock TH. 1986. Sex ratio variation in birds. Ibis.
128(3):317–329.
Croxall JP. 1995. Sexual size dimorphism in seabirds. Oikos.
73(3):399–403.
Donne-GousséC, Laudet V,Hanni C. 2002.A molecular phy- logenyofAnseriformes based onmitochondrialDNA analysis. MolPhylogenetEvol.23(3):339–356.
DumbellGS,McArdleBH,CraigJL.1988.Discriminatingthesex ofbirds:astudyofbrownteal(Anasaucklandicachlorotis).New ZealJZool.15(1):47–53.
EllegrenH,SheldonBC.1997.Newtoolsforsexidentification and thestudyofsexallocation inbirds.Tree.12(7):255–259.
Ferrer M,DeLeCourt C.1992.SexdeterminationintheSpanish
ImperialEagle.JFieldOrnithol. 63(3):359–364.
Fletcher KL, Hamer KC. 2003. Sexing terns using biometrics:
the advantage of within-pair comparisons. Bird Study. 50(1):
78–83.
GauthierG, Tardif J. 1991. Female feeding and male vigilance duringnestingingreatersnowgeese.Condor. 93(3):701–711.
Gray CM,Hamer KC.2001.Food provisioning behaviour ofmale and female Manx Shearwaters Puffinus puffinus. Anim Behav.
62(1):117–121.
GreenAJ.2000. Thescalingandselectionofsexuallydimorphic characters: an example using the Marbled Teal. JAvian Biol.
31(3):345–350.
GriffithsR,Double MC,OrrK,Dawson RJG. 1998.ADNA test tosexmostbirds.MolEcol.7(8):1071–1075.
Hart T,Fitzcharles E,TrathanPN,Coulson T,Rogers AD.2009. Testing and improving the accuracy of discriminant function tests:acomparison betweenmorphometricandmolecularsexing inMacaroniPenguins.Waterbirds32(3):437–443.
HawkinsLL.1986.Nestingbehaviour ofmaleandfemalewhistling swansandimplicationsofmaleincubation.Wildfowl.37:5–27.
HeP-J,YuJ-Q,Fang S-G.2005.Sexidentification ofBlackSwan (Cygnus atratus)using the Locus/specific and Implications for thisReproduction.Reprod DomestAnim.40(3):196–198.
ItoH,Sudo-YamajiA,AbeM,MuraseT,Tsubota T.2003. Sex identification byalternative polymerase chainreaction methods inFalconiformes.ZoolSci.20(3):339–344.
JohnsgardP.1978.Ducks, GeeseandSwansoftheWorld. 1sted. Lincoln(NE):UniversityofNebraska Press.
KearJ.editor.2005.Ducks,GeeseandSwans.Vol1.Oxford(UK): OxfordUniversityPress.
KirkwoodJK, Duignan PJ, Kember NF, Bennett PM, PriceDJ.
1989.The growth rate ofthe tarsometatarsusbone inbirds. J Zool.217(3):403–416.
LahiriDK,Nurnberger JIJr.1991.Arapidnonenzymaticmethod forthepreparationofHMWDNAfrombloodforRFLPstudies. NucleicAcidsRes.19(19):5444.
Limpert RJ, AllenHA,SladenWJL.1987.Weightsandmeasure- mentsofwinteringTundra Swans.Wildfowl.38:108–113.
LombardoMP.2001.Individual andseasonalvariation inexternal genitaliaofmaleTreeSwallows.Auk.118(3):789–795.
Mathiasson S.1981.Weightandgrowthratesofmorphological characters ofCygnusolor. Paperpresentedat:IIISS1981. Proceedingsofthe2ndInternationalSwanSymposium;1981Feb
21–23;Sapporo, Japan.
MathiassonS.2005.Biometricsandstructures oftheMute Swan, Cygnus olor – parameters and technique used in a Swedish project.GöteborgsNaturhistoriskaMuseumÅrstryck.77–86.
McKinneyF,DerricksonSR,MineauP.1983.Forcedcopulationin waterfowl.Behaviour.86(3):250–294.
MillerSL,GreggMA,KuritsuboAR,CombsSM,MurdockMK, Nilsson JA,Noon BR,BotzlerRG. 1988.Morphometricvaria- tioninTundra Swans:relationshipsamong sexandageclasses. Condor. 90:802–815.
Mineau P,CookeF.1979.TerritorialityinSnowGeeseorthepro- tectionof parenthood-Ryder’sandInglis’shypothesisre-assessed. Wildfowl.30:16–19.
Nascimento JLX, Flores JM, Ataguile BS,Koch M, Barbosa S, Parreira DosSantosPJ.2001. BiologicalaspectsoftheBlack- neckedSwan(Cygnusmelancoryphus)andCoscoroba Swan (Coscorobacoscoroba) inRioGrande doSulstate,Brazil. Melopsittacus.4(1):31–38.
NewtonI.1998. Populationlimitation inbirds.SanDiego(CA): AcademicPress.
OdwyerTW,PriddelD,CarlileN,BartleJA,Buttemer WA.2006. Anevaluation ofthreefieldtechniquesforsexingGould’sPetrels (Pterodromaleucoptera)(Procellariidae). Emu.106(3):245–252.
OliveiraCA,SilvaRM,SantosMM,Mahecha AB.2004.Location of the ureteral openings in the cloacas of tinamous, some ratite birds,andcrocodilians: Aprimitivecharacter. JMorphol.
260(2):234–246.
Phillips RA, Furness RW. 1997. Predicting the sex of para- siticJaegers bydiscriminant analysis. Colon Waterbird.20(1):
14–23.
Proctor NS, Lynch PJ. 1993. Manual of Ornithology: Avian Structure andFunction.NewHaven(CT):YaleUniversityPress. QuintanaF,LópezG,SomozaG.2008.Acheapandquickmethod
forDNA-basedsexingofbirds.Waterbirds.31(3):485–488. RicklefsRE.1973.Patterns ofgrowthinbirds.II.Growth rateand
modeofdevelopment. Ibis.115(2):177–201.
SaxA,HoiH.1998.Individual andtemporal variation incloacal protuberance sizeofmaleBeardedTits(Panurusbiarmicus).Auk.
115(4):964–969.
Scott P. 1972. The swans. Boston (MA): Houghton Mifflin
Company.
SeijasSM.2001.CensoNeotropicaldeCisnes,período1998–2000. BuenosAires(Argentina): L.O.L.A.
Silva-García CM,BrewerGL.2007. Breedingbehaviorofthe CoscorobaSwan(Coscorobacoscoroba) inElYaliwetland, cen- tralChile.Ornitol.Neotrop.18:573–585.
SokalRR,RohlfFJ.1995.Biometry.3rded.NewYork(NY):W.H. Freeman andCompany.
SvageljSW,QuintanaF.2007.Sexualsizedimorphism anddeter- mination bymorphometricmeasurements inbreeding Imperial Shags(Phalacrocoraxatriceps).Waterbirds.30(1):97–102.
TabachnickBG,FidellLS.1996.Usingmultivariatestatistics. 3rd ed.NewYork(NY):Harper CollinsPublishers.
Tumarkin-DeratzianAR,VannDR,Dodson P.2006.Bonesurface textureasanontogenetic indicator inlongbonesoftheCanada gooseBrantacanadensis(Anseriformes: Anatidae). ZoolJLinn Soc.148(2):133–168.
Veselovsky Z. 1973.The breeding biology of Cape Barren geeseCereopsisnovaehollandiae.InternationalZoo Yb. 13(1):
48–55.
VolodinI,KaiserM,MatrosovaV,VolodinaE,KlenovaA, Filatova O,KholodovaM.2009.Thetechniqueofnoninvasivedistantsex- ingforfourmonomorphicDendrocygnawhistlingduckspeciesby theirloudwhistles.Bioacoustics.18(3):277–290.
WeidingerK,vanFraneker JA.1998.Applicabilityofexternal measurements tosexingoftheCapePetrelDaptioncapense at within-pair, within-populationandbetween-populationscales.J Zool.245:473–482.
WelshD,SedingerJS.1990.Extra-paircopulationsinblackbrant. Condor. 92(1):242–244.
Whitehead PJ, Tschirner K. 1990. Eggs and hatchlings of the Magpie Goose Anseranas semipalmata. Emu. 90(3):
154–160.
Wilmore SB. 1979. Swans of the World. Taplinger Publishing
Company, NewYork,USA.
Yezerinac SM, Lougheed SC, HandfordP. 1992. Measurement error and morphometricstudies:statistical power and observer experience.SystBiol.41(4):471–482.
Zavalaga CB, Paredes R. 1997. Sex determination of adult Humboldt penguinsusingmorphometriccharacters. JField Ornithol. 68(1):102–112.