The Auk 116(2):504-512, 1999

WHY DO APTENODYTES HAVE HIGH DIVORCE RATES?

JOi•LBRIED, 1 FRI2DI2RICJIGUET, AND PIERREJOUVENTIN CentreNational de la RechercheScientifique, Centre d'Etudes Biologiques de Chizd, 79360 Beauvoir sur Niort, France

ABSTRACT.--Inlong-lived , monogamy is thoughtto enablebreeders to retainthe

samemate from year to year,but exceptionsoccur. For example, King Penguins(Aptenodytes Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 patagonicus)and Emperor Penguins (A. forsteri) have much lower mate fidelity than do small- er speciesof penguins,despite their high rates of survival.Recently, Olsson (1998) suggested that divorcein King Penguinscould be adaptive.Although Olsson was the first to propose an adaptivefunction for divorcein this ,he wasunable to assessthe relationships amongindividual quality, date of arrival,mate choice, and breeding success. Accordingly, we studiedKing Penguins and Emperor Penguins to furtherexamine the determinantsand consequencesof divorce. Mate retentionwas not affectedby breedingperformance in the previousyear or by experience,and neithermate retention nor divorcehad significantcon- sequenceson chick production the following year. King Penguins were more likely to divorce asarrival asynchrony of previouspartners increased. In EmperorPenguins, this tendency to divorceoccurred only when females returned earlier than their previousmates. Most Em- perorPenguin pairs formed within 24 hoursafter the arrivalof males,which were outnum- beredby females.King Penguins that had nestedsuccessfully in theirnext to lastattempt werefavored as mates for onesthat had been unsuccessful, and individuals of bothspecies probablychose the best mates available. Most of ourresults for KingPenguins and Emperor Penguinssupported Olsson's (1998) conclusions in thatdivorce appears to beadaptive. Mate retentionin the absenceof a true nestingsite (neither species builds a nest)would be mal- adaptivefor thesespecies, which face strong time constraintsfor breeding.Therefore, di- vorceoccurs because costs of materetention are high.Aptenodytes penguins appear to have adoptedan optimaldivorce strategy in orderto adaptto their longbreeding cycle in a de- mandingenvironment. Received 27 June1998, accepted 30 October1998.

SOCIALMONOGAMY is the predominantmat- theirchick on top of theirfeet. Only twospecies ing systemamong birds (Lack1968), and many occurin thisgenus, the King (Apteno- long-livedspecies have a longreproductive life dytespatagonicus) and the EmperorPenguin (A. span(Stearns 1992). High survivalmay enable forsteri).Being the largestspecies of penguins, monogamousspecies to retain the samemate theyhave a longbreeding cycle, part of which between successivebreeding attempts more takesplace during the australwinter (Stone- frequentlythan do specieswith highermortal- house1953, 1960). Consequently, they must op- ity (Rowley 1983). In addition, breeding suc- timize their reproductiveoutput during severe cessoften is enhancedby mate retention(e.g. conditions that involve a decline in food avail- Rowley 1983, Ens et al. 1996). ability and a four-monthwinter fastfor chicks High survival rates and long reproductive of the subantarcticKing Penguin(Stonehouse life spans are common in seabirds (Ricklefs 1960, Weimerskirchet al. 1992), and low tem- 1990),but mate fidelity varies amonggroups. peraturesand ice for the antarcticEmperor For example,albatrosses are very faithful to Penguin(Stonehouse 1953). In addition,both their partners (Rowley 1983, Warham 1990), speciesface strongtime constraintsfor breed- whereas frigatebirdsregularly divorce (see ing. Thus,laying after the endof Januaryat Iles Black 1996) between successivebreeding at- Crozetor after 1 Januaryon SouthGeorgia Is- tempts(Nelson 1976). This variabilityalso oc- land invariably results in breeding failure in curswithin the samefamily, as is the casefor King Penguins(Weimerskirch et al. 1992,Jou- penguins(Williams 1996). The genusApteno- ventinand Lagarde1995, Olsson 1996) because dytesis unique in that adults do not build a late-hatched chicks have not been able to store nest,incubating their singleegg andbrooding sufficient fat reserves before the austral winter (Stonehouse1960, Van Heezik et al. 1994, Jou- E-mail: [email protected] ventinand Lagarde1995). In themore synchro-

5O4 April 1999] Divorcein Penguins 505 nouslybreeding (Stonehouse posedthe "expensivefat-storing" hypothesis, 1953, 1960; Isenmann 1971; Weimerskirch et al. suggestingthat individualswould facea trade- 1992),only the earliest-hatchedchicks (i.e. be- off between the costs of divorce and the costs fore 10 August;Isenmann 1971) are able to de- of building up reserveswhile waiting for their part to seawhen the ice breaksin December, previouspartners so that materetention would havingcompleted their molt but attainingonly occur.Although Olsson believed that divorce 50% of adult body mass(Isenmann 1971). was adaptivein King Penguins,his studystill Aptenodytespenguins are long lived, with a neededto be supplementedin somerespects. mean annual survival rate of 0.91 to 0.95 (Jou- In particular,he did not evaluateaccurately the Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 ventin and Weimerskirch 1991, Weimerskirchet eventual benefits (in terms of offspring pro- al. 1992).Yet, they show low mate fidelity be- duction) of extensivedivorce, and he did not tween years (15% in Emperor Penguin [Isen- analyze the relationshipsamong individual mann 1971,Jouventin 1971]; 19 to 29% in King quality,date of arrival,mate choice, and breed- Penguin[Barrat 1976, Olsson 1998]) compared ing success.Despite a lack of data on Emperor with otherpenguins (œ = 84%;Williams 1996). Penguins,Olsson (1998) also suspected that his For divorceto be adaptive,individuals that di- hypotheseswere relevantfor this species;like vorce should obtain some benefit; i.e. it would the ,Emperor Penguins have be "optimal" for an individual to divorce if its part-time pair bonds, and fat reservesplay a reproductivesuccess from breeding with a pre- major role during courtshipand incubation vious partner in the next year is lower than its (Pr•vost 1961, Isenmann 1971). Here we pro- averagefuture success(i.e. the averagefor all vide resultsfrom our long-termstudy of King future breedingattempts until the individual Penguins,supplemented with data from Em- dies; McNamara and Forslund 1996). peror Penguins,that enabledus to further as- Severalhypotheses have been put forward to sessthe adaptivevalue of divorcein thesespe- explain divorce in birds (e.g. Ens et al. 1993, cies. Choudhury 1995), including penguins (Wil- liams 1996).The "incompatibility"hypothesis STUDY AREA AND METHODS predictsthat pairs having low breedingsuccess are morelikely to divorce,and that eachof the From September1990 to March 1994,we conduct- ed a demographicstudy of King Penguinsin a small previouspartners should be more successful colony (750 pairs) on PossessionIsland, Crozet Ar- with its new mate (Coulson 1966, Rowley chipelago(46ø25'S, 51ø45'E). We attachedflipper 1983).In specieswith part-timepair bonds(i.e. bandsto 989 birds and checkedfor their presenceev- during part of the year only), the main reason ery otherday. To assessthe influenceof previousre- for pairs to split might be asynchronousre- productive successon divorce,we consideredpairs turns by previousmates (Coulson1972, Boe- to be successfulif at least one mate was seenfeeding kelheide and Ainley 1989, Davis and Speirs a chickafter September,or wasmolting after Novem- 1990). In his study of King Penguins,Barrat ber If the egg did not hatch,or if oneof the parents (1976)also hypothesized that divorceis caused was molting before Decemberor displaying in early December,the pair wasassumed to havefailed (Jou- by the asynchronousreturn of previouspart- ventin and Lagarde 1995). At best, King Penguins ners to the colony.However, his assumption can breed successfullyevery other year (Weimer- wasbased on a smallsample (three pairs), and skirchet al. 1992, Olsson1996). Therefore, individual his study was conducted during only one quality at the onsetof breedingcycle n was defined breeding season. Recently, Olsson (1998) from the reproductiveperformance at breedingcycle showed that divorce rates in King Penguins n - 2 (Olsson 1996). from South Georgia Island increasedas the Data on Emperor Penguinswere from studiescon- previousmates returned more asynchronously ducted in 1968 by P. Isenmann(pers. comm.)and in to the colony,confirming Barrat's (1976) hy- 1969by P. Jouventin.Field work was carried out at a pothesis.He alsoshowed that King Penguins colony(12,500 pairs) on PointeG•ologie, Terre Ad•- lie, (66ø40'S,140øE), where flipper tags cameashore at the onsetof the breedingcycle were attached to 855 individuals between 1965 and with only half of their body reserves.Because 1967. From arrival until laying, daily observations they fast when on land, their fat storeswere wereperformed on bandedbirds that wereknown to closeto exhaustionat the end of their first stay have bred in 1967 and 1968. Thus, dates of arrival at in the colony.Accordingly, Olsson (1998) pro- the colony(in 1968 and 1969) and pair formation (in 506 BRIED,JIGUET, AND JOUVENTIN [Auk, Vol. 116

lO 8 8 28 1969) were known accuratelyfor each individual. lOO Pairsin which onemate was seenfeeding a chickaf- ter 15 November were considered successful. 80 Forboth species,observations lasted for half a day so that the same individual could be observed sev- 60 eral timeswhile performingdifferent activities. Sex was determined from measurements and behavior (Pr•vost1961, Jouventin 1982) whenever possible. In- dividuals that lost their band in the course of the study, or that paired with an unbandedpartner Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 o while their previous(banded) partner was not ob- o-3 4-7 8-1o >1o served(the rate of band lossis 22.3%in the first year afterbanding for King Penguins;Weimerskirch et al. Asynchrony of returns (deys) 1992),were excludedfrom our data set.To assessthe effectof the asynchronyof returnsof previouspart- 10 10 10 nerson matefidelity in King Penguins,we controlled 1 oo for the effectsof previousbreeding performance .__.6o (successfulvs. unsuccessful)and breeding experi- ence,because these parameters are known to influ- • 6o ence mate retention in a number of species(Green- wood and Harvey1982, Coulson and Thomas1983). We performedstepwise logistic regressions using PROC CATMOD (maximum-likelihoodanalysis) of • 2o SAS(1988). Only thebest-fitting (i.e. most parsimo- nious)models are presented in theResults. It wasnot 0-5 6-15 >15 possibleto control for previous breeding perfor- mance in Emperor Penguinsbecause we did not Asynchrony of returns (days) know whether1967 breeding pairs were successful FIG. 1. Mate retentionand asynchronyof return or failedprior to divorceor remating.Breeding suc- of partnersfrom the previous year in KingPenguins cesswas known for somepairs that bred in 1968,but (top) and Emperor Penguins(bottom). Successful we did not take it into accountbecause our sample and unsuccessfulpairs were pooled. Blackbars = size was too small for reliableanalyses. When con- mate-fidelityrates, gray bars = divorcerates; sample trolling for breedingexperience, we consideredonly sizes indicated above bars. theknown experience of individuals(i.e. the number of breeding attempts since the beginning of our study, not the number of breeding attemptsper- formed by an individual sinceit recruitedinto the Potentialfactors in divorce.--AmongKing Pen- breedingpopulation). If the samebird wasobserved guin pairsfor whichreturn dateswere known during severalconsecutive years, we avoidedpseu- for both mates,mates of the 39 pairs that di- doreplicationby calculatingthe meanvalue for each vorced were more asynchronousin returning parameterbefore conductingour analyses.G-tests to the colony(• = 15.3 +_SD of 9.8 days)than were performedusing Williams' correction.All tests were thoseof the 16pairs that reunited(•? = 5.9 were consideredsignificant at P < 0.05. - 5.0 days). Likewise, return asynchronyof previouspartners was significantlyhigher in RESULTS unsuccessfulpairs (• = 13.6 -- 9.8 days,n = 48) than in successfulpairs (•?= 5.3 - 6.0 days,n Pair-bondduration and mate fidelity.--The du- = 6; Kolmogorov-Smirnovtest, D = 0.60, P = rationof pair bondsin King Penguinsnever ex- 0.04). However,preliminary modelsgiven by ceededtwo successiveyears, and the probabil- the CATMODprocedure failed to find a signif- ity of divorcewas 78% (n = 76 pair years).Mate icant effect of previous reproductiveperfor- fidelity (22%) tended to be higher than that mance and breeding experienceof pairs on found in two consecutiveyears for Emperor mate retention,which decreasedsignificantly Penguins(15%) by Isenmann(1971; 41 pairs) as the secondpartner returnedto the colony and Jouventin(this study;21 pairs),although later (best-fittingmodel; asynchrony, X2 = 9.06, the differencewas not significant(G = 1.37,df df = 1, P = 0.026, n = 54 pairs; see Fig. 1). = 1, P > 0.2). Pair bondslasted up to four years Amongpairs that divorced, asynchrony of pre- in Emperor Penguins. viouspartners was similar regardless of which April 1999] Divorcein Penguins 507

lOO Males: n= 116 = 1, P = 0.097; see Fig. 1). Among pairs in Females: n= 94 8o whichthe female returned to thecolony earlier than the male (n = 18), divorcetended to occur 6o more frequentlyas asynchronyincreased, but this tendencywas not significant(asynchrony, 40 X2 = 3.33, df = 1, P = 0.068; experience,X 2 = 0.22, df = 1, P = 0.63). FemaleEmperor Pen- 20 guinsreturned earlier than malesboth in 1968 Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 o and 1969, the differencebeing significantin o-1 2-10 >10 1969 (Wilcoxon'srank-sum test, z = 2.15, n• = Days after arrival 92, n2 = 86, P = 0.031).Females also spent sig- nificantly more time between returning and pairing(œ = 4.4 + 7.3 days,n = 70)in 1969than

lOO did males (• = 1.9 _+5.0 days, n = 73; Kolmo- gorov-Smirnovtest, D = 0.26, P = 0.013). Con- sequently, 82% of males and 56% of females pairedwithin 24 hoursafter their arrival (Fig. 2). Fordivorced pairs in whichpairing date of the first-arrived mate was known, the second partnerreturned to the colonyafter its previ- ous mate was paired in 7 casesand while its previousmate was unpaired in 12 cases. Data on breedingsuccess in the previous o-t 2-1o >to year were availablefor 74 pairs of King Pen- Days after arrival guins.For Emperor Penguins, data on breeding FIc. 2. Time elapsedbetween arrival and court- attemptsin the previousyear were available for ing in King Penguins(top) and between arrival and 36 pairs,but we knewthe outcomeof thoseat- pairing in EmperorPenguins (bottom). Gray bars = tempts for only six pairs. Consideringrepro- males, black bars = females. ductiveperformance per capita(i.e. over the to- tal numberof individualsor pairsstudied), in sexreturned the earliest.Independent of fidel- both speciespairs that divorcedproduced as ity, femalesbegan courtingsignificantly faster many chicksas thosethat reunitedin the next after returningto the colony(• = 1.9 _+4.3 year (data for King Penguins;divorce, X2 = days, n = 94) than did males (œ= 4.0 _+6.3 0.06, df = 1, P = 0.8, experience,X 2 = 0.01, df days, n = 116; D = 0.25, P = 0.003). Conse- = 1, P = 0.91, n = 74). However,high-quality quently,63% of females,but only 38% of males, King Penguins(i.e. successfulbreeders in next were seen courting potential partnersduring to lastbreeding cycle) tended to divorceless of- the 24 hoursafter their arrival (Fig. 2). In di- ten than poor-qualityones, although the dif- vorcedpairs, the partnerthat arrived earliest ferencewas not significant(quality, X 2 = 3.12, was already performing courtship display df = 1, P = 0.077;experience, X 2 = 0.63, df = 1, when its previousmate returnedin 37 cases, P = 0.43, n = 69). and was still alone in one case. Olsson(1995) found a slightsurplus of males In EmperorPenguins (n = 26 pairs),neither in theKing Penguin colony he studiedon South asynchronyof previouspartners, nor the sexof Georgia,although he did not assessits impor- the partnerthat returned first, nor experience tance.In our Iles Crozetcolony, 428 of the 989 of breedingpairs (one breedingattempt vs. bandedKing Penguinscould be sexed,yielding morethan one attempt) affected the probability a ratio of 1.53 malesper female (259 males,169 of divorce. Yet, the latter factor tended to de- females).Yet, malesare easierto identify than pendon theinteraction between the former two femalesbecause they performdisplays more factors,although not significantlyso (asyn- often.Consequently, we mayhave overestimat- chrony,X 2 = 0.51, df = 1, P = 0.47; sex, X2 = ed the proportionof malesin theKing Penguin 2.20, df = 1, P = 0.14;experience, X 2 = 1.90,df population.In theEmperor Penguin, Isenmann = 1, P = 0.17;asynchrony x sex,X 2 = 2.74, df (1971) and Jouventin(1971) showed that re- 508 BRIED,JIGUET, AND JOUVENTIN [Auk,Vol. 116 males outnumberedmales by about 10% at theyreturned to thecolony later (date of return, PointeG•ologie. For both species, reproductive X2 = 4.97, df = 1, P = 0.026, n = 44). We knew performancein theyear before divorcing or re- the qualityof the new andthe formermates for uniting did not differ significantlyfrom that in five individualsthat divorced,and they were the yearfollowing a divorceor rematingwith similarin all cases.For EmperorPenguins, the a previous partner very small number of individuals for which re- Mate retentionvs. divorce:Costs and benefits.- productiveperformance was known during the Retaining one'spartner or divorcinghad no previousbreeding season precluded the analy- significanteffect on reproductiveperformance sesthat we performedfor KingPenguins. Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 in the nextyear in King Penguins,even if we controlledfor previousbreeding experience. In DISCUSSION Emperor Penguins,divorced individuals tend- ed to producemore offspring (œ = 0.47 chicks) Factorsin divorce:'14synchrony of return" and than thosethat remated(œ = 0.17chicks) with 'fat-storage"hypotheses.--Materetention in our their previouspartners, but the differencewas King Penguincolony occurred more frequently not significant(preliminary model; experience, if bothpartners of the previousyear returned P = 0.14; best-fitting model; mate retention,X 2 synchronouslyat the startof thebreeding sea- = 3.28, df = 1, P = 0.07; sex, X2= 0.47, df = 1, son,confirming Barrat's (1976) hypothesis for P = 0.49,n = 63). In neitherspecies did dateof King Penguinsin anothercolony on Iles Crozet arrival (earlieror later than previouspartner) andOlsson's (1998) results from South Georgia or sexhave a significanteffect on breeding sta- Island.In EmperorPenguins, asynchrony also tus (breederor nonbreeder)during the year fol- might play a role,but only in thosepairs in lowing a divorce. whichfemales (i.e. the mostrepresented sex) Individualquality and matechoice.--In King return the earliest. However, and like in Ols- Penguins,high-quality individuals (see above) son's(1998) King Penguincolony, the incidence tendedto returnto the colonyearlier than birds of divorcewas high (50%in CrozetKing Pen- thathad failed, but the differencewas not sig- guins and 60% in EmperorPenguins; Fig. 1) nificant (Wilcoxon'srank-sum test, z = 1.84, n• evenwhen asynchronywas low, confirming = 18, n2 = 96, P = 0.065).The probabilityof that this factor alone was not sufficient to ex- successfullyraising a chicktended to behigher plain high divorcerates in Aptenodytespen- in pairs with at least one high-qualitymate guins. (20%,n = 15)than in pairsformed by twolow- Olsson(1998) also suggested that mateshift- qualitymates (0%, n = 18),but again,the dif- ing in King Penguinsdepended on the amount ferencewas not quitesignificant (Fisher's exact of fat stored.According to the "expensivefat- test,P = 0.08).Because high-quality individu- storing"hypothesis, attaining fat reservesbe- alsappeared to be at a reproductiveadvantage, fore matingis costlyowing to decreasedma- we examinedwhether mate quality was an im- neuverabilityin the waterand decreased fight- portant criterionin mate choice.Low-quality ing ability,and because storing fat to ensurere- individualswere significantlymore likely to mating with one'spartner from the previous obtaina high-qualitypartner than werehigh- yearcan imply latebreeding. Thus, individuals quality ones, and the interactionbetween in- would face a tradeoff between the costs of di- dividualquality and return date also had a sig- vorceand thoseof materetention. In Emperor nificanteffect on the qualityof themate chosen Penguins,females return to thecolony in April (individual quality,X 2 = 4.48, df = 1, P = 0.034, with lowerfat reservesthan do males(Pr•vost date x individual quality,X2 = 5.54, df = 1, P 1961).Once ashore, they fast until egglaying, = 0.018,n = 54). Preliminarymodels failed to whichoccurs in May. Thereafter,they under- reveala significanteffect of experience.Mate take a long foragingtrip at sea to replenish quality did not seemto dependon return date their body reserves.Meanwhile, males take in high-qualityindividuals (sex, X 2 = 0.13,df = chargeof incubation(Pr•vost 1961). Their fast 1, P = 0.72; date, X2 = 0.74, df = 1, P = 0.39, n will end at the beginningof chickrearing, = 10). In contrast,low-quality individuals (i.e. when femalesrelieve them (Isenmann1971). failed breederstwo cyclesearlier) were more Wetherefore suggest that female Emperor Pen- likely to obtain a low-qualitypartner when guins face a tradeoffbetween replenishing April 1999] Divorcein Penguins 509 theirbody reserves before fasting during court- constraintsthan the latter two speciesthat live ship,and the potentialcosts of divorceand late at lower latitudes (Marchant and Higgins breeding,or of notbreeding at all if theycannot 1990). find a male. Storing and carrying fat also is Absenceof a nest site.--Penguinsgenerally costlyfor adult males,which have lower sur- live in huge colonies(Marchant and Higgins vival rates in this species(Jouventin and Wei- 1990). Locating one'sprevious partner among merskirch1991), partly becausethey aremore several thousand individuals apparently is vulnerableto predationat seathan are females mucheasier for penguinsthat build nestsor dig

(Jouventin1974). While at sea during this pe- burrows (Jouventin1982), and theseterritorial Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 riod, their maneuverabilityis decreased,which speciestend to exhibitstrong nest-site tenacity alsois consistentwith the "expensivefat-stor- (60 to 98% for nine species)and high mate fi- ing" hypothesis.In both species,laying (and delity (up to 97%;Williams 1996). Nonetheless, hence,mating) as early aspossible should en- Williams (1996) found no significantcorrela- able the female to limit the duration of her for- tion between mate retention and nest-site fi- aging trip at sea before she relievesher incu- delity among14 speciesof penguins.In addi- bating mate. Simultaneously,the probability tion,if individualsdivorce simply because they that males would exhaust their fat reserves and areunable to locatetheir previous partner, then abandontheir eggs should decrease. Individu- divorce should occur less frequently in the als of both sexes should obtain a mate before smallestcolonies. Yet, Barrat (1976) and Olsson their body reservesare reduced to a critical (1998) obtained similar results to ours in King threshold;beyond this threshold, a breedingat- Penguincolonies of different sizes (29% mate tempt almostinvariably would resultin failure. retentionin a 56,000-paircolony and 19%in a Factorsin divorce:Sex ratio.--Female Emperor colonyof 150individuals, respectively; data on Penguins,which seem to return earlierand are colonysize in Voisin[1971] and Olsson[1998]). slightly more numerousthan males, tend to Anotherconsequence is that divorcewould be monopolizeeach solitary male uponits arrival no more than a by-product,with divorcedin- at the colony(Isenmann 1971, Jouventin 1971) dividualsmaking the best of a bad situation. so that pair formationoccurs almost immedi- Divorce and mate choice.--We failed to find an ately after maleshave arrived (Isenmann1971, advantage of mate retention or a cost of di- this study).As a result,the earliera femalere- vorce. Furthermore, the combination of factors turns at the onsetof the breedingcycle (1) the notedabove was insufficient to explainthe high higher the probabilityshe will get a mate (Is- incidenceof divorcewhen asynchrony was low. enmann1971), and (2) the lowerthe probability Olsson (1998) explained this phenomenonby that her previousmate will alreadybe paired speculatingthat many birds previouslyhad with anotherfemale. Consequently, the unbal- mated with a lower-qualitypartner than they anced sex ratio in Emperor Penguins,com- couldhave obtained had they managedearlier bined with the body-reserveproblem noted to storeenough fat prior to the fastingperiod; above(plus the physiologicalsynchrony of in- under theseconditions, divorce would be adap- dividualsand the advantageof breedingas ear- tive. Olsson(1998) predictedthat, in the ab- ly aspossible), may be a factorin divorcein this senceof time constraintscaused by thefat-stor- species.Monopolizing partners has not been ageproblem, birds would choosea partnerof recorded in King Penguins.Moreover, Yellow- ashigh a quality aspossible in an "ideal-free" eyedPenguins ( antipodes) and Little manner (Fretwell and Lucas 1970). Olsson Penguins( minor) have high matefi- (1998)suggested that the absenceof territorial delity (82%;Richdale 1947, Reilly and Cullen behaviorin King Penguinswould enableindi- 1981,Dann and Cullen 1990)despite biased sex vidualsto assesspotential partners rapidly and ratios.Therefore, a biasedsex ratio is unlikely with low costwhen movingthrough the colo- to be a generalfactor in divorceamong sphen- ny. Thus, they couldchoose the most suitable isciforms.The Emperor Penguin seems to be an mate as long as they could afford to sample exception,but onlybecause the differentphas- mates,because of fat-storingconstraints. These es of its breedingcycle occur more synchro- predictionssupported the "better-option"hy- nously than in King Penguins,and Emperor pothesis(Ens et al. 1993, Choudhury 1995), Penguinsare under much more severetime whichstates that onemember of a pair initiates 510 BRIED,JIGUET, AND JOUVENTIN [Auk, VoL 116 divorceif it hasthe opportunityto improveits patibility" hypothesis(Coulson 1966, Rowley reproductivesuccess by obtaininga higher- 1983).Yet, it shouldbe moreinteresting to test quality mate. this hypothesisusing the true breedingexpe- Our resultssuggest that mate quality was rience(i.e. that sinceindividuals have entered not linkedto returndate for high-qualityKing the breedingpopulation) combined with qual- Penguins,although the latter tendedto arrive ity. The behavioralplasticity of Aptenodytes first and to havethe highestbreeding success. penguinsis in accordancewith predictionsof The fact that these tendencies were not statis- the "better-option"hypothesis, as previously tically significantmay be explainedby small suggestedfor King Penguinsby Olsson(1998). Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 samplesize. In contrast,high-quality individ- Our resultsconcerning the latest-arrivingin- uals were particularly attractivefor low-quali- dividualscan be explainedby the "musical- ty ones,which had more difficulty obtaining chairs"hypothesis, which statesthat individ- them as partnersas time elapsed.Yet, late ualssettle on the best territory available; a King breedingincreases the probability of failure,in- Penguinreturning too late will find its previ- dependentof quality (Olsson1996). Therefore, ous breedingarea occupiedand will have to our results are consistent with the "ideal-free" settle elsewherein the colony (Barrat 1976). mate-choicehypothesis. However, the propor- Thishypothesis predicts that divorce"may be tion of high-qualityindividuals and evenly a side-effect of differential arrival of the sexes" matchedpartners must not vary greatly be- (Choudhury1995). In the EmperorPenguin, tween different colonies,as suggestedby di- maleswould replaceterritories as "chairs,"at vorcerates similar to oursin SouthGeorgia and least in some cases. in Barrat's(1976) colony. In the moresynchro- In most cases,however, access to potential nouslybreeding Emperor Penguin, the lack of partners doesnot seemvery difficult nor time data on individual quality precludesfirm con- consumingfor King Penguinsand Emperor clusions.Yet, the absenceof territorialityreach- Penguins,which have to face two conflicting es its highestlevel in this species(Stonehouse constraints:(1) theyhave long breeding cycles 1953);breeders walk with theiregg or chickon relativeto theirlarge body size (Stearns 1992); their feet and congregatein "huddles"when and(2) theirbreeding schedule is severelycon- the weatherbecomes especially harsh (Pr•vost strainedsuch that theymust save as much time 1961, Isenmann1971). Thus, searchingfor a aspossible to rear their chickwithin the limits new partnershould not be very costlyfor this setby environmentalconditions. In light of this specieseither Moreover,the numberof male scenario,Olsson (1998) may not have devoted partners available per unpaired female de- sufficient attention to the role of the absence of creasesas time elapses,which enabledIsen- nestingsites and territorial behavioron mate mann and Jouventin(1970) to attract up to 20 retention.Hence, we suggestthat divorceis unpairedfemales simultaneously by usingthe commonin Aptenodytespenguins, not only be- playbackof a male'scalls during the last days causeit is notcostly, but becausein the absence of the courtshipperiod. At this time, thesefe- of nestsites that serveas meetingpoints (e.g. males probablywould have paired with the Hinde 1956, Morse and Kress 1984), mate re- first available male. Consequently,Emperor tentioncould be maladaptive.Aptenodytes pen- Penguinsmay alsoperform mate choice in an guinsseem to have adoptedan "optimal di- "ideal-fred' manner Our conclusions are con- vorce" strategy (McNamara and Forslund sistentwith Olsson's(1998). In both speciesof 1996)that enables them to adapttheir breeding Aptenodytespenguins, only some individuals cyclesto seasonalchanges in their environ- (probablythe latest arrivals) would seek a part- ment. ner, not to improve their breeding success (whichhas a greaterchance to decreaseas time ACKNOWLEDGMENTS elapses),but to limit the risk of not breeding We thank the Institut Fran•aispour la Recherche altogether. et la TechnologiePolaires and the TerresAustrales et Conclusion:Why is divorceso common?--We AntarctiquesFran•aises for financial and logistical failed to discovera link amongdivorce, breed- support, respectively.We are indebted to P. Isen- ing experience,and breeding failure in thepre- mannfor kindly providingdata from his 1968study. vious year, giving no supportto the "incom- We alsothank M. Salamolard,E Lagarde,F. Cu•not- April 1999] Divorcein Penguins 511

Chaillet,C. Boiteau,E. Cam,and P.Lys for fieldwork GREENWOOD,P. J., AND P. H. HARVEY.1982. The natal and dataprocessing. E. Guinardhelped with draw- and breeding dispersalof birds. Annual Review ing the figures,and J. Ekman,B. Ens,M. Kersten,J. of Ecologyand Systematics13:1-21. Marks, O. Olsson,and two anonymousreviewers HINDE,R. A. 1956.The biologicalsignificance of the greatly improved the manuscriptby their critical territories in birds. Ibis 98:340-369. commentsand helpwith the English. ISENMANN,P. 1971. Contribution i l'•thologie et i l'•cologie du Manchot empereur (Aptenodytes LITERATURE CITED forsteri Gray) 3 la colonie de Pointe G•ologie (TerreAd•lie). L'Oiseauet R.EO. 41:9-64.

13ARRAT,A. 1976.Quelques aspects de la biologieet ISENMANN,P., AND P. JOUVENTIN. 1970. Eco-•thologie Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 de l'•cologiedu Manchotroyal (Aptenodytespa- du Manchotempereur (Aptenodytes forsteri) en tagonicus)des ties Crozet. Comit• NationalFran- comparaisonavec le ManchotAd•lie ( gaisde la Rechercheen Antarctique40:9-51. adeliae)et le Manchotroyal (Aptenodytes patagon- 13LACK,J.M. 1996.Introduction: Pair bonds and part- ica).L'Oiseau et R.EO. 40:136-159. nerships.Pages 3-20 in Partnershipsin birds: JOUVENTIN,P. 1971. Comportement et structure so- The study of monogamy(J. M. Black,Ed.). Ox- ciale chezle Manchotempereur. La Terre et la ford UniversityPress, Oxford. Vie 25:510-586. 13OEKELHEIDE,R. J., AND D. G. AINLEY.1989. Age, re- JOUVENTIN,P. 1974.Mortality parametersin Emper- source availability and breeding effort in or PenguinsAptenodytes forsteri. Pages434-446 Brandt's Cormorant. Auk 106:389-401. in The biologyof penguins(B. Stonehouse,Ed.). CHOUDHUR¾, S. 1995. Divorce in birds: A review of Macmillan, London. the hypotheses. Behaviour50:413-429. JOUVENTIN,P. 1982.Visual and vocal signalsin pen- COULSON,J. C. 1966.The influenceof the pair-bond guins, their evolutionand adaptivecharacters. and ageon thebreeding biology of the Kittiwake Advancesin Ethology24:1-19. Gull Rissatridactyla. Journal of Animal Ecology JOUVENTIN,P., AND E LAGARDE.1995. Evolutionary 35:269-279. ecologyof theKing Penguin Aptenodytes patagon- COULSON,J. C. 1972. The significanceof the pair- icus:The self-regulationof the breeding cycle. bond in the Kittiwake. Pages424-433 in Pro- Pages80-95 in The penguins(P. Dann, I. Nor- ceedingsXV InternationalOrnithological Con- man, and E Reilly,Eds.). Surrey Beatty and Sons, gress(K. H. Voous,Ed.). The Hague, 1970.E. J. Chipping Norton, Australia. Brill, Leiden, The Netherlands. JOUVENTIN,P., AND H. WEIMERSKIRCH.1991. Chang- COULSON,J. C., AND C. S. THOMAS.1983. Mate choice es in the populationsize and demographyof in the Kittiwake Gull. Pages361-376 in Mate southernseabirds: Management implications. choice(E Bateson,Ed.). CambridgeUniversity Pages297-314 in populationsstudies: Rel- Press,Cambridge, United Kingdom. evanceto conservationand management (C. M. DANN, P. J., AND J. M. CULLEN.1990. Survival, pat- Perrins, J.-D. Lebreton, and G. J. M. Hirons, ternsof reproductionand lifetime reproductive Eds.).Oxford UniversityPress, Oxford. output in Little Penguins(Eudyptula minor) on LACK,D. 1968.Ecological adaptations for breeding in PhilipsIsland, Victoria, Australia. Pages 63-84 birds. Methuen, London. in Penguinbiology (L. S. Davis and J. T. Darby, MARCHANT,S., AND P.J. HIGGINS(Eds.). 1990. Hand- Eds.).Academic Press, San Diego, California. book of Australian, New Zealand and Antarctic DAVIS, L. S., AND E. A. H. SPEIRS.1990. Mate choice birds, vol. 1A. Oxford University Press, Mel- in penguins.Pages 377-397 in Penguinbiology bourne, Australia. (L. S. Davis and J. T. Darby, Eds.). Academic MCNAMARA, J. M., AND P. FORSLUND.1996. Divorce Press,San Diego, California. ratesin birds: Predictionsfrom an optimization ENS, B. J., S. CHOUDHURY,AND J. M. BLACK.1996. model. American Naturalist 147:609-640. Mate fidelity and divorcein monogamousbirds. MORSE,D. H., AND S. W. KRESS.1984. The effect of Pages 344-401 in Partnershipsin birds: The burrow losses on mate choice in Leach's Storm- study of monogamy(J. M. Black,Ed.). Oxford Petrel. Auk 101:158-160. University Press,Oxford. NELSON,J. 13.1976. The breedingbiology of frigate- ENS, B. J., U. N. SAFRIEL,AND M. E HARRIS. 1993. Di- birds. A comparative review. Living Bird 14: vorcein the long-livedand monogamousOys- 113-155. tercatcherHaematopus ostralegus: Incompatibility OLSSON,O. 1995. Timing and body-reserveadjust- or choosingthe better option?Animal Behav- mentsin King Penguinreproduction. Ph.D. the- iour 45:1199-1217. sis,Uppsala University, Sweden. FRETWELL,S. D., AND H. L. LUCAS,JR. 1970. On ter- OLSSON,O. 1996.Seasonal effects of timing and re- ritorial behavior and other factorsinfluencing production in the King Penguin: A unique habitat distribution in birds. I. Theoretical de- breedingcycle. Journal of Avian Biology27:7-14. velopment.Acta Biotheoretica19:16-36. OLSSON,O. 1998. Divorce in King Penguins:Asyn- 512 BRIED,JIGUET, AND JOUVENTIN [Auk, Vol. 116

chrony,expensive fat storingand idealfree mate STONEHOUSE,B. 1960. The King PenguinAptenodytes choice. Oikos 83:574-581. patagonicaof SouthGeorgia. I. Breedingbehav- PREVOST,J. 1961. Ecologiedu Manchot empereur. iour and development.Falkland Islands Depen- Hermann, Paris. denciesSurvey ScientificReports 23:1-83. REILLY,P. N., AND J. M. CULLEN.1981. The Little Pen- VAN HEEZIK, Y. M., P. J. SEDDON,J. COOPER,AND A. guin Eudyptulaminor in Victoria, II: Breeding. L. PLUS.1994. Interrelationships between breed- Emu 81:1-19. ing frequency,timing and outcomein King Pen- RICHDALE,L. E. 1947.The pair bondin penguinsand guins Aptenodytespatagonicus: Are King Pen- petrels:A banding study. Bird-Banding18:107- guins biennial breeders?Ibis 136:279-284. 117. VOISIN,J.-E 1971. Note sur les Manchots royaux (Ap- Downloaded from https://academic.oup.com/auk/article/116/2/504/5168774 by guest on 29 September 2021 RICKLEFS,R. E. 1990. Seabird life histories and the tenodytespatagonica) de l'ile de la Possession. marine environment:Some speculations.Colo- L'Oiseau et R.EO. 41:176-180. nial Waterbirds 13:1-6. WARHAM,J. 1990. The petrels. Their ecologyand ROWLEY,I. 1983.Re-mating in birds. Pages331-360 breeding systems.Academic Press,London. in Mate choice(P. Bateson, Ed.). CambridgeUni- WEIMERSKIRCH,H., J.-C. STAHL, AND P. JOUvENTIN. versity Press,Cambridge, United Kingdom. 1992.The breedingbiology and populationdy- SASINSTITUTE. 1988. SAS/STAT user'sguide, version namicsof King PenguinsAptenodytes patagonica 6.03. SASInstitute, Inc., Cary,North Carolina. on the Crozet Islands. Ibis 134:107-117. STEARNS,S.C. 1992. The evolution of life histories. WILLIAMS,T. D. 1996.Mate fidelity in penguins.Pag- Oxford UniversityPress, Oxford. es268-285 in Partnershipsin birds:The studyof STONEHOUSE,B. 1953. The EmperorPenguin Apten- monogamy(J. M. Black,Ed.). Oxford University odytesforsteri Gray. I. Breedingbehaviour and Press, Oxford. development.Falkland Islands Dependencies Survey ScientificReports 6:1-33 AssociateEditor: J. Ekman