ResearchReports Sexing Little Penguins Eudyptula minor using bill measurements Rebecca Overeenv. Robert Wallis’ and Scott Salzman1 1SchoolofLifeandEnvironmentalSciences,DeakinUniversity,Warmambool,Victoria3280 2OfficeoftheProVice-Chancellor(RuralandRegional),DeakinUniversity,Warmambool,Victoria3280 SchoolofInformationSystems,DeakinUniversity,Warmambool,Victoria3280 Abstract InLittlePenguinsEudyptulaminortherearenoreliableplumageorbodysizedifferencesthatcan be used visually to distinguish the sex ol individuals. However, sexual dimorphism ofmorphometric measures has been noted, withmales always being a little largerthan females. In this study, differ- ences between h. minorsexes at eightcolonies in south-eastern Australia were determined statisti- callyviadiscriminant functionanalysis (DFA)andthrough the utilization ofDNA-basedtechniques developedlornon-ratite birds. The DFAcorrectlydeterminedgenderin91.1%ofcasesand molecu- larmethodswere 100%accurate.OurDFAsuccessraleofclassification issimilartothatpreviously published lor Little Penguins in Victoria. In this study statistically significant differences in mean GbiallbodepIstlhasnda,ndcolmepnagrthesdwteorceolfoonuineds batetKwaenegnarLoitotleIslPaenndg.uGirnanciotleoniIselsanadt,hiMtiKdidlldea,IsPlhainlldipanIdslaLnodndaonnd Bridge. As birds in eastern populations (St Kilda, Phillip Island, Gabo Island) exhibit statistically significantlysmallerbeaks (bill depth and bill length), separate discriminant functionswere investi- gatedloreachphenotypicallydistinctgeo-spatial cohort. Interestingly,clusteranalysisforbill length identified three groups: western (Kangaroo Island and Granite Island), eastern (St Kilda, Phillip rIastleangdroaunpd.MWieddcloencIlsuladned)thaatndwhtihleeLtohnedreonisBarsildiggehtLiitntclreeaPseenginuiDnFcoploowneyr, wfohriccohlocnoinesstiwteutsetdoafsCeappae- Otway and for some specific colonies, colony-specific DFA is not required to identify the sex of LittlePenguinsinsouth-easternAustralia.(TheVictorianNaturalist123(6)2006,390-395). Introduction The Little Penguin Eudyptula minor is Phillip Island and Gibson Steps in Victoria the smallestpenguin species and is endem- (Arnould et al. 2004). Gales (1988) was ic to temperate seas in Australia and New the first to use a discriminant Function Zealand. Australia has one sub-species (E. (DF) that uses bill depth and length. The minor novaehoilandiae), found from New Zealand work also developed DFs Fremantle (WA) in the west to northern that demonstrated each sub-species NSW and Tasmania inthe southeast.There required different functions. Arnould et al. are five sub-species in New Zealand (2004) found the DFs derived by workers (Kinksy and Falla 1976). However, Banks for New Zealand populations ofE. minor et al. (2002) recently demonstrated that were poor predictors ofgender for Little molecular results subdivide E. minor into Penguins at Phillip Island and Gibson only two clades: 1) the majority ofNew Steps. Gales’ (1988) DF derived from a Zealand colonies and 2) Australia (sample Tasmanian population yielded a reliability from Phillip Island)and Otago E. minor„ of 89.3% and 92.2% for birds at the two In order to manage Little Penguin popula- Victorian sites (Arnould et al. 2004) com- tions effectively, demographic analyses paredto94% in Tasmania. requiretheaccuratedetermination ofgender Preliminary results from our studies at ofthe animals in the field (Caughley and seven colonies ofLittle Penguins in south- Gunn 1996). However, Little Penguinsshow easternAustraliasuggestedbill dimensions no differences in plumage between genders, ofadults varied among colonies. Arnould and body size is also similar for males and et al (2004) found similar differences in females(Agnewand Kerry 1995). bill depth in their studies that prompted Bill depth has been a useful sexually dis- them topropose colony-specific DFsmight tinguishing morphological feature in Little be needed in ordertodeterminethe gender Penguins from Tasmania (Gales 1988), ofbirds accurately, rather thanjust using New Zealand (Renner and Davis 1999, theone DFfortheAustralian sub-species. Hocken and Russell 2002) and also from 390 The Victorian Naturalist Research Reports The aim ofthis paper is to see whether it Results is possible to derive a single DF that can Examination ofthe seven sites for statis- be used to determine accurately the gender tically significant differences (Table 1) in of Little Penguins at eight sites in south- bill length asafunction oflocation (Fig. 1) easternAustralia. revealed that E. minor from the Kangaroo Island, Granite Island, Middle Island and Methods London Bridge colonies have significantly Bill depth (vertical thickness at the nos- longer bills when compared to birds from trils) and length (length at exposed culmen the more eastern colonies (Fig. 2a), totip ofbill) were measured (±0.1mm)on Analysis ofthe seven sites for statistically 50 adult E. minor at seven of the sites significantdifferences in bill depthderived shown in Fig. 1. To minimize inter-opera- a similar result, with E. minor from torvariation the same person took all mea- Kangaroo Island, Granite Island and surements. Data provided in Amould el al. London Bridge having significantly deeper (2004) for known-gender birds at Gibson bills compared to E. minor from the more Steps were used to test both Gales' (1988) easterncolonies(Fig. 2b). andouroverall discriminantfunction. Cluster analysis was also performed on Blood samples were collected from the the variables Bill Length and Bill Depth as birds using standard techniques (Ellegren a function oflocation. Bill Length proved 1996) and gender determined genetically the more interesting variable with three using the methods of Fridolfsson and groups identified: western (Kangaroo and Ellegren (1999) that rely on intron length Granite Island), eastern (St Kilda, Phillip tvCeaHirniD)atgi(eocnnherionlmoctouh-seh.esTlehixicscahsarelo-lmDooNwseAodmbeiu-nsdstiponegckinpfroiowc- BsIesrpliaadnrgdaetaeLnigdtrtoMlueipdPde(lFneiggu.Iis3ln)a.ncdoG)alloaennsyd’,t(wh1eh9i8Lc8oh)ndihsoanda the gender ofthe birds that had previously DF = -83.10 + (10.06 In BL) + (17.99 In hadtheirbill measurementstaken. BD), where DS is the discriminant score The molecular method of gender deter- and In the natural logarithm. When we mLiitntalteioPnewnagsuivnericfairecdabsyseasppflircoamtioMnidtodl4e0 applied this DF, we found it produced dif- ferences in the success rate ofclassifica- Island, Warrnambool, that had been killed tion for predicting the gender of Little by foxes and subjected to gender determi- Penguins (Table 1). The DF that Gales nationbydissection. (1988) derived was most reliable for birds Geographic variation in sexual dimor- intheeastofVictoria. ppahriasmmetwraisctAesNteOdVuAs.ingThKreusMcaanl-nW-aWllhiistnneony inAgrnDoFulfdoretLiatlt,le(P20e0n4g)uidnesriavtePdhitlhleipfoIlslloawn-d test was used to assess forpost hoc differ- andGibson Steps: DS= 1.242 BD- 16.774 ences and a P-value of<0.05 was consid- The DF model derived by Arnould et al, ered statistically significant. Discriminant (2004) from Phillip Island and Gibson bgfouentnchdteibroilnlofadneiaplntydhsiivs(idBwuDaa,lsmpumesne)gduaitnnodsi.dbieWlnlteilfyeunsgtehtdeh moSbtisenepersvdEa.tsimeoinxnsoffrorrocom8l8soon.ui3te%hs-seouafcsctetesrhsnefAuEul.sltyrmadileintaeo.rr- (BL, mm) inourDF. Assumptionsassociat- Testing each of the seven E. minor ed with discriminant function analysis were colonies separately for the DF model onont3v5i0olabtierdd.s.TWhielkD’Fs Lwaemdbedraivteedstwwaassteussteedd d(e2r0i0v4e)drbeysuGlatleedsi(n19v8a8r)yainndg Asmucocuelssd.etThale. to determine whether both variables (BD accuracy of both DF models decreased at and BL) contributed significantly to the the Kangaroo Island, Granite Island, model. Canonical discriminant function Middle Island and London Bridge cloiesfhfitchieentlsinweearrefduenrcitviedonin(oGradleersto1e9s8t8a)b.- cPohliolnliieps,IswlhainldeaintdinGcraebaosedIsaltatnhdeES.tmKiilndoar, Cluster analysis was undertaken with colonies(Table2). respect to location to determine ifthere The DF we derived from all samples wereanydistincthomogeneoussubsets. from all locations is: DS = -18.710 + (1.292 BD)+(0.015 BL). Vol. 123 (6) 2006 391 Research Reports Table 1. Mann-Whitneypost hoc one-tailed differences for bill length ofEudyptula minor from sevencoloniestabulated(n= 50 individuals/colony, * statistically significantatthe0.05 level) Z= z-scores, P=p-value. Kangaroo Granite Middle London StKilda Phillip Island Island Island Bridge Island GraniteIsland Z 0.574 1.110 0.707 -3.149 3.623 P 0.566 0.267 0.480 0.001* 0.000* Middle Island z -0.622 1.110 0.565 -1.968 2.385 p 0.534 0.267 0.572 0.025* 0.009* LondonBridge z -0.299 0.707 0.565 -2.907 3.167 p 0.765 0.480 0.572 0.002* 0.001* StKilda z -3.051 -3.149 -1.968 -2.907 0.128 p 0.002* 0.002* 0.049* 0.004* 0.898 K>StK GrI>StK MI>StK LB>StK Phillip Island z 3.414 3.623 2.385 3.167 0.128 p 0.001* 0.000* 0.017* 0.002* 0.898 K>PI GrI>PI MI>PI LB>PI GaboIsland z -3.112 3.240 2.197 2.914 -0.073 0.130 p 0.002* 0.001* 0.028* 0.004* 0.941 0.897 K>GI GrI>GI MI>GI LB>GI Fig. 1.ThelocationoftheeightcoloniesofLittlePenguinsusedinthisstudy. 392 The Victorian Naturalist Research Reports 60 50 B 30 - g -=l 20 ' 03 * 10 - 0 , ! Kangaroo Is Granite Is Middle Is London Br St Kilda Phillip Is Gabo Is Fig. 2a. Box plot ofbill length for the seven Eudyptula minorcolonies in south-eastern Australia (themedianisidentifiedwithinthebox,thedataspreadis identifiedasthedistancebetweentheends ofthebox,andthelinesextendtoextremevalues.Xrepresentsoutliers). 20 -i o 10 - cq 8 - 6 - 4- 2 - 0 \ 1 T T 1 - T KangarooIs GraniteIs MiddleIs LondonBr StKilda PhillipIs GaboIs Site Fig.2b. Boxplotofbill depthforthesevenEudyptulaminorcoloniesin south-easternAustralia(the median is identified within thebox,thedata spreadisidentifiedasthedistancebetweentheends of thebox,andthelinesextendtoextremevalues). RescaledDistanceClusterCombine CASE 0 —5 10 15 20 25 Label Num + + + + + + KI 66666660 1 GI 2 ch- 11666660 SK 5 666Q606 6 PI 6 666-^-116-^ 11666666666666666666666666666666666660 MI 3 66666^- 6 GA 7 6666666666666^- LB 4 6666666666666666666666666666666666666666666666666^ Fig.3.ClusterAnalysisofEudyptulaminorbilllengthforsevencoloniesinsouth-easternAustralia. (KI: KangarooIsland,GI:Granite Island, SK: StKildaBreakwater, PI: PhillipIsland, MI:Middle Island,GA: GaboIsland, LB: London Bridge). Vol. 123 (6) 2006 393 Research Reports Table 2. Geographically grouped and colony-specific DFs for the Eudyptula minor breeding in south-easternAustralia. Location DiscriminantFunction Successrateofclassification ThisStudy Gales(1988) Amouldet TasmaniaDF at.(2004) EastofCape =(BDX 1.425)+(BLX0.048)-21.312 91.3% Otway WestofCape =(BDX 1.328)+(BLX0.011)- 19.692 90.5% Otway KangarooIsland =(BDX 1.443)+(BL X0.042)-22.148 88.2% 73.0% 74.0% GraniteIsland -(BDX 1.362)+(BL X0.107)-24.115 88.0% 80.0% 80.0% MiddleIsland - (BDX 1.452)+(BLX0.049)-21.889 96.0% 76.0% 76.0% SLtonKdilodnaBridge =-((BBDDXX 11..3273)6)++(B(LBLXX-00..013389))--1282..700523 9848..09%% 8786..90%% 7868..00%% Phillip Island -(BDX 1.687)+(BL X0.169)-29.578 94.3% 94.3% 93.0% Gabo Island -(BDX 1.341)+(BLX-0.039)-16.939 94.1% 84.3% 84.0% Usingall samples the DFcorrectly predict- cases. Further, its reliability in determining ed gender in 91.1% of birds tested, with gender of birds in the eastern colonies DS values >0 as male and those <0 (91.3%) compared with those in the west females. This DF was almost as reliable as (90.5%) was also high. The colony-specific the one we derived for determining gender DF success ratesvaried,with ahigh success ofbirds in the eastern colonies (91.3%) of96% for the Middle Island colony and a compared with those in the west (90.5%). lowof88%fortheGranite Islandbirds. A discriminant function was developed for The DF wc have derived is thus more each ofthe seven sites, for sites clustered reliable at predicting the gender of pen- both east and west ofCape Otway, and guinsin south-eastAustralia in comparison overall, forall sites(Table 1), with other DFs that have been published forthe samespecieselsewhere. Discussion We found mean adult Little Penguin bill Acknowledgements depth and length varied between the eight We thank several field assistants who assisted sites sampled in this study. This supports wMiutrhrdaayt,acColleMcetiColnu(sNkyOv,erSeemW,ilAliOavmesroenc,m,AT the observations reported by Amould etaL Chiaradiaand L Renwick)andcolleagues atthe (2004) who also found differences in bill Molecular Ecologyand Biodiversity Laboratory depth in birds from different colonies; at Deakin University. The project could not these differencespromptedthem tosuggest have been undertaken without the financial theremightbea needforadifferent DF for aRsessiesatarnccheoAfwaarDdeaakinnd UtnhievelrlsoiltsywoProtsht-GWrialddulaitfee each colony in Australia. Fig. 2 indicates Research Fund. The research was carried out that birds from our sites clustered into two with permission from the various governing groups - one east ofCape Otway and one bodies: Deakin University Animal Ethics to the west. In all sites, the males had sta- Committee (permit number: A10/2003) the Department of Natural Resources and tistically significantly larger bills than the Environment/Sustainability and Environment females. approved fieldwork to be undertaken at Middle When the DF derived by Gales (1988) in Island, London Bridge, St Kilda and Gabo Tasmania was applied for the eight sites Island (permit number: 10002229) while in (the seven sites used in this study and South Australiathe Department ofEnvironment Gibson Steps), gender was successfully approvedwork at Kangaroo andGranite Islands determined in 81.7% of cases. When permitnumber: 10/2003)andthe Department ol Environment and Heritage (Z24663). In addi- Gales’ DF was applied separately to the tion, the Warrnambool City Council (through data from the seven colonies studied, the the Coast and Rivers Advisory Committee) and reliability was lowest for the western the St Kilda Penguin Study Group approved colonies and higher for the colonies at St fieldworkatMiddleIslandandStKilda. Kilda, Phillip IslandandGaboIsland. Our DF correctly determined gender in birds from the eight colonies in 91.1% of 394 TheVictorian Naturalist 2 Research Reports References versal method for molecular sexing ofnon-ratitie AgnewDJandKerryKR(1995)Sexualdimorphismin birds.JournalofBiology’30, 116-121. penguins. In The Penguins: Ecology and GalesR(1988)SexingadultBluePenguinsbyexternal aMnadnaPgReemilelny.t(,Spuprr2e9y9B-e3a1t8t.y&EdSsonPs:DaSnyndn,eyI)Norman HomcekaesnurAemGenatnsd.NRoustsoerlnlis.13(52.07012-)75A.methodfordeter- Arnould JPY, Dann P and Cullen JM (2004) minationofgenderfrombill measurementsinOtago Determining the sex ofLittle Penguin (Eudyptula blue penguins (Eudyptula minor). New Zealand minor) in northern Bass Strait using morphometric JournalofZoology29,63-69. measures.Emu104,261-265. KinskyFCand FallaRA(1976)Asubspecificrevision BaAnnksunJ,exMpietcctheedllpaAt.t,erWnaoafsmo.1laecnudlaPratdeirvseorngeAnc(e2w0i0t2h)-. oinftthheeANuestwraZleasailaanndBlaureea.PeRngeucionrd(sEuodfypttheulNaatmiiononra)l Ninottohrenibsl4u9e,p2e9-n3g8u.in (Eudyptula minor) complex. ReMnunesreuMmaonfdNeDawvZiseaLlSan(d19I.991)05S-e1x2i6n.gLittlePenguins CaughleyG andGunn A (1996)ConservationBiology’ Eudyptula minor from Cook Strait, New Zealand, inTheoryandPractice.(Blackwell:Melbourne) usingdiscriminantfunctionanalysis.Emu99,74-79. EllegrenH(1996)Firstgeneonavian Wchromosome (CHD) provides a tag for universal sexing ofnon- ratite birds. Proceedings oftheRoyal Society of FrLidoonldfossnonBA26a3n.d1E6l3l5e-g1r6e4n1.II(1999)Asimpleanduni- Received20October2005;accepted16November2006 Is there always a bias towards young males in road kill samples? The case in Victorian Koalas Phascolarctos cinereus Natasha McLean1 'DepartmentofZoology,TheUniversityofMelbourne,Victoria3010 2CurrentAddress:TheDepartmentofSustainabilityandEnvironment,8NicholsonSt, EastMelbourne,Victoria3002.Email:[email protected] Abstract Mortality due to road trauma can have large negative impacts on some populations and often is biased towardsage/sex classes that have higherrates ofmovement: individuals during the breeding season andjuveniles while theyare dispersing. A bias towardsyoung males has been found in two previous studies ofroad kill Koalas in southeastQueensland. Such abias was not found in the pre- sentstudy ofKoalaskulls from populationsacross Victoria. This may bedue to thedifferent Koala population structuresand densities orroad types and characteristics. {The Victorian Naturalist 123 (6) 2006,395-399). Introduction Individuals ofmany species are killed on high dispersal rates or increased activity the roads (Trombulak and Frissell 2000; levels (Bonnet et al. 1999, Inbar and Taylor and Goldingay 2004) and this can Mayer 1999). For example, Coulson have large negative effects on populations (1989) found that 48% of road killed ofwild animals (Dufty 1994; Jones 2000; Eastern Grey Kangaroos Macropus gigan- Hebblewhite et al. 2003; Lopez et aL teuswere l to2yearsofage, theage when 2003). Road trauma is known to impact they were dispersing. A significant bias some Koala populations (Backhouse and towards road kills oftwo-year-oldmacrop- Crouch 1990: Lunney et at. 1996; ods was also found by Lee el al. (2004). Thompson 1996). The Phillip Island Koala Additionally, maleswere more likely to be population in Victoria decreased substan- hit than females in livespecies ofmacrop- tially between 1973 and 1988, mostly due ods, possibly becauseoftheirgreaterrang- to high mortality from road trauma (Every ing behaviour(Coulson 1997). A highpro- 1986; BackhouseandCrouch 1990). portion ofungulate road traumas are also Mortality rates due to road trauma have related to dispersal and breedingbehaviour been found to differbetweentemporal sea- (Groot Bruinderink and Hazebroek 1996). sons (Taylor and Goldingay 2004) and Similar patterns of male-biased mortality may be greater in age/sex classes thathave caused by road trauma have been found in Vol. 123 (6) 2006 395