Multiple mating strategies explain unexpected genetic mixing of New Zealand fur seals with two congenerics in a recently recolonized population

Molecular Ecology (2007) 16, 5267–5276 doi: 10.1111/j.1365-294X.2007.03586.x

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

Blackwell Publishing LtdMultiple mating strategies explain unexpected genetic mixing of New Zealand fur seals with two congenerics in a recently recolonized population

MELANIE L . LANCASTER,* S IMON D. GOLDSWORTHY*† and PAUL SUNNUCKS*‡*Zoology Department, La Trobe University, Bundoora, Vic. 3083, Australia, †South Australian Research and Development Institute, Aquatic Sciences, West Beach, SA 5024, Australia, ‡Australian Centre for Biodiversity: Analysis, Policy & Management, Biological Sciences, Monash University, Clayton, Vic. 3800, Australia

Abstract

Human impacts on natural systems can cause local population extinctions, which may promoteredistribution of taxa and secondary contact between divergent lineages. In mammalianpopulations that have mating systems shaped by polygyny and sexual selection, the potentialfor hybridization to ensue and persist depends on individual and demographic factors. AtMacquarie Island, a recently formed fur seal population is comprised of both sexes ofbreeding Antarctic (Arctocephalus gazella) and subantarctic (A. tropicalis) fur seals, and anitinerant collection of male New Zealand fur seals (A. forsteri), presumed to be non-breedersdue to their absence from principle breeding areas. The mating system of the three speciesis described as resource-defence polygyny: males defend beach territories containingbreeding females for exclusive mating rights. A recent genetic study identified a high levelof hybridization in the population (17–30%), unexpectedly involving all three species. Thisstudy examined the source of involvement in breeding by A. forsteri with respect to matingstrategies operating in the population. Ninety-five (10%) pups born from 1992 to 2003 weregenetically identified as New Zealand hybrids. Most resulted from reproduction withinterritories by New Zealand hybrids of both sexes, although some were conceived extra-territorially, indicating that males successfully utilize strategies other than territory holdingto achieve paternities. Female reproductive status influenced mating partner and matinglocation, and females without pups were more likely to conceive extra-territorially and withA. forsteri males. This study illustrates an important consequence of low heterospecificdiscrimination in a sympatric population of long-lived mammals.

Keywords: alternative mating strategies, Arctocephalus, hybridization, mate choice, microsatellite,paternal haplotype

Received 22 July 2007; revision accepted 14 September 2007

Introduction

Human disturbance to natural systems can result insecondary contact between formerly isolated species. Ifreproductive isolation is incomplete, hybridization and/or genetic introgression may ensue, but its extent andpersistence is dependent on a number of factors. The

foremost of these is the fitness of hybrids relative toparental species, but in systems under sexual selectionwhere individuals of one sex exhibit mate choice, whetheror not interspecific mating occurs at all may be influen-ced by demographic and individual factors. These factorscan affect the level of heterospecific discriminationexhibited when choosing a suitable mate (Grant & Grant1997; Wirtz 1999).

Under conditions where allocation of parental care isunequal and the fitness of members of one sex can beincreased by controlling access to the other, the potentialfor polygamy is high (Emlen & Oring 1977; Boness 1991).

Correspondence: Melanie Lancaster, School of Earth andEnvironmental Sciences, University of Adelaide, North Terrace,Adelaide, SA 5005, Australia. Fax: +61 88303 6222; E-mail:[email protected]

5268 M . L . L A N C A S T E R , S . D . G O L D S W O RT H Y and P. S U N N U C K S

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

In mammals, females that assume all parental care throughlactation should be selective in their choice of mate, whilemales are free to maximize their reproductive success bymating with multiple females (Trivers 1972; Clutton-Brock1991). Female fur seals (family Otariidae) are sole investorsof parental care, alternating bouts of onshore nursingwith foraging at sea until their pup is weaned. Polygyny infur seals is further enhanced by sexual size dimorphism(males are two to 10 times heavier than females), mod-erate synchrony of oestrus, and clustered female spatialdistributions as a result of the relative scarcity of suitablepupping sites on oceanic islands and coastal beaches(Boness 1991). The mating system is described as resource-defence polygyny, where males defend areas of islandor coastal beaches (territories) that are pupping sites offemales (Boness 1991). An assumption of the matingsystem is that females are passive recipients of mating toresident territory males, but mounting evidence for femalemate choice (Goldsworthy et al. 1999; Hoffman et al. 2007;Lancaster et al. 2007) as well as the finding that femalesselect mates based on genetic diversity and comparativerelatedness (Hoffman et al. 2007) suggest that sexual selec-tion is not always biased towards competitively superiormales. Furthermore, female mate choice and ability todiscriminate against heterospecific males may be influencedby timing of breeding, age and experience (e.g. Grant &Grant 1997; Wirtz 1999).

Recent genetic studies of two fur seal populations sup-port the notion that territory holding is the most successfulreproductive strategy for males during the breeding season(Hoffman et al. 2003; Lancaster et al. 2007); however, lowassignment of paternity to key males in some pinnipedsystems suggests the use of alternative mating strategies.‘Sneaky’ copulations by smaller males behaving likefemales, individual pursuit by males of females as theyleave their territory to forage at sea, and attacks on femalesin breeding areas by raiding parties of males have beenobserved in some colonies (Cox & Le Boeuf 1977; Campagnaet al. 1988). Previous studies of grey seal (Halichoerus grypus)and Antarctic fur seal (Arctocephalus gazella) mating systemshave found high levels of extra-territory paternities withmany sires not identified, indicating that true fathers arenot observed on breeding beaches and instead matingmay occur at peripheral locations or even aquatically(Worthington Wilmer et al. 1999; Gemmell et al. 2001).Subsequent studies of these systems have revealed thatmany unassigned paternities may be partially due tosampling bias rather than extra-territory mating, thushighlighting the need for detailed behavioural observationsof colonies during the breeding season to identify potentialfathers (Hoffman et al. 2003; Twiss et al. 2006).

Before historic seal harvesting, subantarctic MacquarieIsland accommodated a large population of approximately200 000 fur seals (Cumpston 1968; Ling 1999). The population

(species unknown) was hunted to extinction by sealinggangs in the early 19th century and recolonization has beenslow and recent (1940s, Csordas 1958), by three species: theAntarctic, subantarctic (Arctocephalus tropicalis) and NewZealand fur seal (Arctocephalus forsteri, Shaughnessy et al.1988). Based on mitochondrial control region and cyto-chrome b data, and in agreement with morphologicalassessments, the three species are genetically distinct,with no evidence of past hybridization (Wynen et al. 2000;Wynen et al. 2001). Males of A. forsteri and females of A. gazellahave been most abundant on the island since recoloniza-tion, and A. forsteri numbers have increased annually byabout 3.6% per year since 1950 (Shaughnessy & Goldsworthy1993; S.D. Goldsworthy, P.D. Shaughnessy, L. Wynen, S.Robinson, unpublished data). The A. forsteri males presentare mostly juveniles and subadults that do not contest foror hold breeding territories and instead haul out in covesaway from breeding beaches (as close as 200 m) in numbersthat peak one to two months after the breeding season ofA. gazella and A. tropicalis (Goldsworthy et al. unpublisheddata; M.L.L., personal observation). Since females ofA. forsteri are rarely seen on Macquarie Island, the twomain breeding species were thought to be A. gazella andA. tropicalis.

Genetic analysis of the population has identified a highlevel of hybridization (Goldsworthy et al. 1999; Wynen 2001;Lancaster et al. 2006). The most recent study, encompassinga 12-year period (1992–2003), found that the majority ofpups were pure (53–63% A. gazella, 12–22% A. tropicalis) but17–30% of pups born were hybrids, yet surprisingly notonly of the expected tropicalis/gazella type (Lancaster et al.2006). Rather, a proportion of hybrid pups had geneticcharacteristics of A. forsteri, despite the fact that this specieswas believed to be represented overwhelmingly by non-breeding males. Mitochondrial DNA analysis of A. forsteri(NZ) hybrid pups found that all had either A. gazella orA. tropicalis matrilines (Lancaster et al. 2006), thus did notoriginate from A. forsteri females. The production of NZ hybridpups must therefore be through A. gazella or A. tropicalisfemales mating extra-territorially with A. forsteri males (toproduce F1 hybrid pups), or within territories with NZhybrid males (to produce post-F1 hybrid pups). Pups bornto NZ hybrid females would also be post-F1 hybrids. Sinceover 50% of NZ hybrid pups born from 1992 to 2003 werepost-F1 hybrids (Lancaster et al. 2006) and NZ hybridmales are known to have held territories over that period(Lancaster et al. 2007), any or all of the above scenarios arepossible. To clarify the breeding contribution by A. forsteriand in doing so, gain a better understanding of the matingsystem operating in this population, we aimed to (i)quantify the couplings that produced NZ hybrid pups from1992 to 2003, and (ii) identify extra-territory paternities thatmay imply the use of alternative mating strategies bymales or females to produce NZ hybrid pups.

M AT I N G S T R AT E G I E S I N S Y M PAT R I C S E A L S 5269

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

Materials and methods

Study site and sample collection

Macquarie Island (54°30′S, 158°56′E) is situated in theSouthern Ocean, approximately 1500 km southeast ofmainland Australia and north of the Antarctic PolarFront (Fig. 1). Fur seals breed every austral summer almostexclusively on the northern tip of the island, North HeadPeninsula, in three bays: Secluded Beach and Goat Bayon the east coast and Aerial Cove on the west coast (Fig. 1).Minor breeding localities include Garden Cove (North Head),Brother’s Point (approximately 12 km south of North Head)and Hurd Point, the southernmost point of the island(Fig. 1). Nonbreeding seals (mostly A. forsteri males)haul out to undergo their annual moult along the island’seastern and southern coasts and around North Head,away from breeding beaches (Fig. 1). As well as there beingspatial separation of A. forsteri from the two main breedingspecies, A. gazella and A. tropicalis commonly select differentbeach types for breeding (M.L. Lancaster, S.D. Goldsworthy,P. Sunnucks, unpublished data). Observations indicate thatprincipal breeding areas of A. gazella are characterizedby shingled beaches in Aerial Cove and Secluded Beach,whereas boulder coves in Goat Bay and southern SecludedBeach are typical breeding areas of A. tropicalis.

Every pup born over eight years (1007 individuals) wassampled from Macquarie Island between 1992 and 2003 asdescribed by Lancaster et al. (2006). Pups were markedand sampled opportunistically when their mothers left to

forage at sea, and mother–pup pairs were matched in thefield (mothers only nurse their own pup) and confirmedgenetically (Lancaster et al. 2007). Multiple-year comparisonswere possible as breeding adults of both sexes have beentagged and sampled opportunistically since the early 1990sas part of a long-term monitoring program on MacquarieIsland. Sampling of adult males was restricted to territory-holding males and males observed actively contestingterritories (challenger males). All breeding territories wereobserved intensively at least once daily during the matingseason to obtain birthing and mating information on alltagged females present. Movements and locations of territoryand challenger males were also recorded daily. Femaleswere only recorded as absent from the colony in a givenyear if they were sighted in subsequent years with readableand intact flipper tags, since tag loss is common in fur seals.

Species and NZ hybrid identification

Genotypes and species assignments of all pups and adults(228 females, 54 males) from Lancaster et al. (2006, 2007)were used in this study. Individuals were genotyped atnine microsatellite loci and a 417-bp fragment of the mito-chondrial tRNAthr-control region [RFLP (restriction fragmentlength polymorphism)] and assigned to a pure speciesor hybrid class using the Bayesian clustering softwarestructure version 2.0 (Pritchard et al. 2000). Q was usedas a metric of individual assignment to each of the threespecies, where individuals with Q ≥ 0.9 for a particularspecies were considered to be pure members of that species

Fig. 1 Macquarie Island (1 cm: 3 km),illustrating minor fur seal breeding localities(Brother’s and Hurd Point) and NorthHead Peninsula, the primary site of breedingon the island (enlarged). Main breedingbeaches on North Head are highlighted ingrey and New Zealand fur seal haul-outsare in black.

5270 M . L . L A N C A S T E R , S . D . G O L D S W O RT H Y and P. S U N N U C K S

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

if their mtDNA profile matched their nuclear profile, whilethose with Q < 0.9 or with mismatching nuclear and mito-chondrial profiles were classed as hybrids. Individualswith Q > 0.1 for belonging to A. forsteri were classed as NZhybrids. No pure A. forsteri males or females were detectedin the breeding population but NZ hybrid territory malesand breeding hybrid females were present and genotyped(Lancaster et al. 2007).

Quantifying breeding involvement by NZ hybrid males

To quantify the proportion of pups produced by breedingof females with NZ hybrid males, we attempted to assigngenetic paternity to all NZ hybrid pups born from 1992 to2003 using cervus version 2.0 (Marshall et al. 1998). Weincluded pups with known mothers, which were pairedin the field and confirmed genetically, as well as pupswhose mothers were unsampled (both parents unknown).Parameters described in Lancaster et al. (2007) were usedfor paternity analysis of NZ hybrid pups born from 1992 to1999; however, because of less comprehensive sampling ofterritory and challenger males in the population from 2000to 2003, paternity analysis of pups born in 2001 and 2003used a much smaller proportion of sampled candidate males(0.3 cf. 0.6–0.8). All other parameters remained the same.

Detecting mating strategies of unsampled males

Some NZ hybrid pups did not have a father assignedgenetically, implying that for these pups, true fathers werenot sampled. This may have been because mothers of thesepups mated with A. forsteri males outside breeding areas orbecause the sampling shortfall for territory males increasedafter 1999. To gain some insight into the mating strategiesthat produced these pups (within-territory or extra-territory),we used the Monte Carlo randomization program dadshare(Hoffman et al. 2003). dadshare estimates relatednessamong unsampled fathers using paternal haplotypes(half-genotypes deduced from pups that have knownmothers with different multilocus genotypes) and finds thebest-fit level of shared paternity among pups. In a systemwhere territory males have much higher reproductive successthan challengers, siring of multiple pups by a single malemay imply that the father was a territory male, while siringof only a single pup may indicate that the father was aperipheral or challenger male. Previous paternity analysisof pups at Macquarie Island indicated that territory malesdid indeed sire more pups on average than non-territorymales (Lancaster et al. 2007), but unusually high reproductivesuccess of two challenger males (each fathered multiplepups) meant that a threshold number of paternities eitherside of which a male was known to be a territory male ora challenger could not be applied. We therefore coupleddadshare results with observed locations of tagged mothers

during the year of conception to determine whether pupswere produced within or outside a territory. Pups wereinferred to have been sired extra-territorially if the followingcriteria were met: (i) the pup’s mother was not observedwithin a territory, or (ii) the pup’s mother was within aterritory of a sampled territory male and the pup mis-matched with the sampled territory male at multiple loci.Pups could not be excluded from being sired withinterritories if the territory male was not sampled.

Species membership of unsampled fathers

After determining whether NZ hybrid pups with unsampledfathers were produced within or outside breeding territories,we sought to find out whether these fathers were hybridmales or pure A. forsteri males. Since pure A. forsteri malesare absent from breeding areas, pups with unsampledbut discernibly pure A. forsteri fathers must have beenproduced away from main breeding beaches, implyingunconventional mating strategies by males and/or females.We assigned unsampled fathers to a species or hybridclass using their paternal haplotypes and the assignmentprogram whichrun version 4.0 (Banks & Eichert 1999).Baseline allele frequencies from pure representatives ofall three species across their geographical ranges were usedin the analysis (Lancaster et al. 2006). This included allelefrequencies of A. forsteri individuals from colonies aroundNew Zealand that are source populations for the MacquarieIsland A. forsteri males. To enable the program to accept thepaternal haplotypes of unsampled fathers, a second ‘pseudo’allele not found in the baseline data was included at eachlocus. The pseudo-allele was given a low, equal frequencyof occurrence in each species and thus had no net effect onindividual assignment.

whichrun assigned unsampled fathers a statisticallikelihood of belonging to one of the three species (log ofodds ratio, LOD) based on their haplotypes. To interpretwhether the LOD scores categorized these males as pureA. forsteri or hybrids, we generated LOD scores fromhaplotypes of 200 pure A. forsteri individuals (simulatedusing hybridlab version 0.9, Nielson et al. 2001), and allknown Macquarie Island NZ hybrids for comparison.Diploid genotypes were reduced by half to obtain haplo-types by randomly subtracting one allele from each locus.We also explored the effect of variation in an individual’shaplotype (as with gamete production) on LOD scoresassigned by generating 10 haplotypes from an individual’sfull genotype for 10 each of pure A. forsteri and NZ hybridgenotypes.

Species-specific alleles and Q values were used inconjunction with LOD scores from paternal haplotypes tofurther exclude unsampled fathers from being pure A. forsterimales. If paternal haplotypes contained ≥ 2 alleles specificto either pure A. gazella or pure A. tropicalis, fathers were

M AT I N G S T R AT E G I E S I N S Y M PAT R I C S E A L S 5271

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

considered not to be pure A. forsteri. If Q of hybrid pups felloutside the range of F1 hybrids (0.2–0.78, based on simu-lated data, Lancaster et al. 2006) and they had pure mothers,pups were likely to be backcrosses, in which case fatherswere NZ hybrids.

Results

Hybrid composition of pups and breeding adults

Approximately 10% (95/1007) of all pups born over 8 yearsfrom 1992 to 2003 were one of three types of NZ hybrid:A. gazella-A. forsteri (A-NZ), A. tropicalis-A. forsteri (S-NZ)or hybrids formed from past breeding among all threespecies (A-S-NZ) (Table 1, Fig. 2). A-NZ hybrid pups weremost abundant across all years (71.5% of all NZ hybridsand 6.8% of all pups born) whereas S-NZ and A-S-NZhybrids were less common (1.2% and 1.5% of all pups born,respectively) (A-NZ, S-NZ χ2 = 71.22, d.f. = 1, P < 0.0001,A-NZ, A-S-NZ χ2 = 63.64, d.f. = 1, P < 0.0001). Nearly 17%

(9 of 54) of adult males sampled were NZ hybrids,compared with only 7.8% of adult females (χ2 = 3.7,d.f. = 1, P = 0.05) (Table 1).

Maternal contribution to NZ hybrid pup production

Nineteen hybrid females (4 A-NZ, 8 A-S, 4 S-NZ and3 A-S-NZ) produced 29% (28/95) of all NZ hybrid pups inthe population. Seven of these females returned to MacquarieIsland and produced hybrid pups for two or more years,with the highest recorded number of pups produced by asingle hybrid female being five in five years. Nineteen NZhybrid pups had unsampled mothers (from field observa-tions) and fathers (mismatched with most likely father atthree or more loci), so the direction of hybridization couldnot be analysed further than examining Q values andmitochondrial profiles. Sixteen of these had Q within therange of F1 hybrids, and mothers more commonly boreA. gazella mtDNA haplotypes (70% A-NZ pups) than A. tropi-calis ones (30% S-NZ pups). A similar matrilinear trend wasfound in NZ hybrid pups with known mothers: of the 32pure females that produced 42 hybrid pups, significantlymore were pure A. gazella (N = 31) than A. tropicalis (N = 1)(χ2 = 176.72, d.f. = 1, P < 0.0001). This may be partially dueto a lower encounter rate of NZ hybrid males by A. tropicalisfemales, as most NZ hybrid males held territories in areascommonly used by A. gazella females (16 territories held byNZ hybrid males on pebbled beaches cf. four in boulderedcoves, χ2 = 72.0, d.f. = 1, P < 0.0001).

Mating behaviour in relation to female reproductive history

Since reproductive status can affect mate choice by females,the reproductive history of female fur seals that matedwith territory males rather than challengers or unsampledmales was explored. Almost half (9/20; 45%) of all femalesthat mated extra-territorially to produce NZ hybrid pupshad not pupped in the year they conceived their hybridoffspring, while only 7% (3/41) of females that mated withterritory males had not pupped that year. This differencewas significant (χ2 = 37.53, d.f. = 1, P < 0.0001) and suggeststhat female choice concerning the status of the father oftheir offspring (territory holder or challenger) may dependin part on a female’s own recent reproductive history: ifshe has not reproduced the year before, she is more likelyto mate with a male who is not a territory holder.

Production of NZ hybrid pups by sampled males

Fifty-one (54%) NZ hybrid pups were assigned fathersgenetically using cervus (39 with 95% confidence, 13 with80% confidence). Territory males sired most (45) of these, ofwhich 40 (89%) were fathered by four NZ hybrids. Two NZ

Fig. 2 Distribution of Q in NZ hybrid pups (grey) and adults(black) across all years, where Qforsteri is the posterior probability ofmembership to Arctocephalus forsteri (individuals with Qforsteri > 0.1are classed as NZ hybrids). The skew in the distribution of Q forpups indicates that many pups are post-F1 hybrids.

Table 1 Arctocephalus forsteri hybrid types observed among pupsand breeding adults (percentages of hybrids in all age classes inparentheses)

Hybrid type

Pups n = 1007

Adult females n = 228

Adult males n = 54

Total n = 1289

A-NZ 68 (6.8) 5 (2.2) 4 (7.4) 77 (6.0)S-NZ 12 (1.2) 3 (1.3) 1 (1.9) 16 (1.2)A-S-NZ 15 (1.5) 3 (1.3) 4 (7.4) 22 (1.7)Total 95 (9.3) 11 (4.8) 9 (16.7) 115 (8.9)

A, A. gazella; S, A. tropicalis; and NZ, A. forsteri.

5272 M . L . L A N C A S T E R , S . D . G O L D S W O RT H Y and P. S U N N U C K S

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

hybrid territory males alone were responsible for siring36% (34/95) of all NZ hybrid pups detected in the popu-lation. These males held territories for at least five and sixconsecutive years. The remaining five pups with territorialmales as fathers (either A. gazella or A. tropicalis) had NZhybrid mothers. Alternative mating strategies operated inthe population, with six hybrid pups genetically assignedto non-territory males that were observed in breedingareas actively challenging territories (5 with 95% confidence,1 with 80% confidence). Two of the six males were NZhybrids and four were pure A. gazella or A. tropicalis males.

Mating strategies of unsampled fathers

Forty-four NZ hybrid pups were sired by unsampled males:they mismatched with the most likely genetically inferredfather at three or more loci. Seventeen of these pups hadknown mothers, so paternal haplotypes were used to analysepaternal species and mating strategy. Using relatednessestimates and Monte Carlo simulations in dadshare, these17 pups were likely to have been fathered by 11 differentmales, with a degree of shared paternity involving 11 pupsamong five males (Table 2). Eight pups were likely to havebeen fathered extra-territorially by unsampled males,based on maternal locations in the year of conception(Table 2), making the total proportion of NZ hybrid pupsproduced by extra-territory mating at least 14/95, or 15%.The remaining nine pups could not be excluded from beingsired within territories by unsampled territory males,

making the total estimate of NZ hybrid pup production bywithin-territory mating up to 57% (54/95).

Usefulness of paternal haplotypes for inferring species/hybrid class

Based on LOD scores of A. forsteri and NZ hybrid half-genotypes compared to their respective full genotypes(Fig. 3), pure-species individuals were more accuratelyassigned to their correct species class than were hybrid

Table 2 Allocation of 11 fathers to NZ hybrid pups based on paternal haplotypes (dadshare), and inferred status and species of unsampledfathers deduced from maternal locations in the year of conception for each pup

Pup no.Father no.

Year pup born

Mother in territory

Mother pup prev yr

TM sampled

TM-pup mismatch

Father status

Father species

1 1 1995 No UK — — ET NZ2 2 1995 No UK — — ET A-S3 3 1996 No* UK — — ET NZ4 4 1998 No No — — ET A-NZ5 5 1999 Yes Yes Yes Yes ET UK6 6 2001 Yes Yes No — UK7 7 1998 Yes† No Yes Yes ET A-NZ8 7 2003 Yes Yes Yes Yes ET9 8 1998 Yes No No — TM A-NZ10 8 2001 Yes Yes No — TM11 8 2003 Yes Yes No — TM12 9 1998 No No — — UK NZ13 9 1998 UK UK — — UK14 10 1995 UK UK — — UK A-NZ15 10 2001 No No — — ET16 11 1995 Yes Yes No — TM A-S-NZ17 11 2001 No No — — UK

TM, territory male; ET, extra-territory male; UK, unresolved; A, Arctocephalus gazella; S, Arctocephalus tropicalis; and NZ, Arctocephalus forsteri. Combinations, for example A-S, represent hybrids. *pup born at Hurd Point; †first observed in late January, towards end of mating season. Arctocephalus forsteri fathers are in bold type.

Fig. 3 Relationship between LOD scores assigned to individualsbased on their haplotypes and complete genotypes for 200 generatedA. forsteri individuals (black, R2 = 0.42) and NZ hybrids (grey,R2 = 0.17). Variances within individuals are presented in Fig. 4.

M AT I N G S T R AT E G I E S I N S Y M PAT R I C S E A L S 5273

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

individuals when only half of their genetic informationwas analysed. The 95% CI of LOD scores of NZ hybrids(0.1–5.0) overlapped that of A. forsteri (2.5–8.8), but meanLOD of hybrid haplotypes was significantly lower andwith a considerable range of diagnostic nonoverlap, indic-ating lower statistical confidence of assignment in hybridsand a useful means for distinguishing hybrids from pureA. forsteri individuals (mean LOD NZ hybrids = 2.33,mean LOD A. forsteri = 5.63, t = 14.88, d.f. = 270, P < 0.0001).LOD scores assigned to the 10 haplotypes permuted foreach individual (to emulate gamete production) were morevariable in NZ hybrids than in A. forsteri individuals(Fig. 4). A number of haplotypes deduced from full hybridgenotypes actually had negative likelihoods of being pureA. forsteri individuals because they bore alleles morecommon to the alternative parental species (A. gazella orA. tropicalis).

Species/hybrid class of unsampled fathers

Having found that LOD scores of paternal haplotypes wereuseful in distinguishing A. forsteri individuals from hybridseven with some individual variation, we used the paternal

haplotypes of the 17 NZ hybrid pups with known mothersto determine the species/hybrid status of unsampledfathers and relate this information to their inferred matingstrategies. Based on their LOD scores and/or the presenceof ≥ 2 alleles specific to either A. gazella or A. tropicalis, eightof the 11 missing fathers were either NZ hybrids, A. gazellaor A. tropicalis. Two fathers that sired one pup each andone father that sired two pups in the same season wereidentified as pure A. forsteri: Q values of pups were withinthe range expected for F1 hybridization, their paternalhaplotypes had high likelihoods of being pure A. forsteriindividuals (LOD scores above 3.0) and they containedno alleles specific to the other two species (Table 3). Whenrelated back to female location in the year of conceptionand mating strategies inferred from analysis in dadshare,three of the females conceived pups outside territories(to father 1 and 3, Table 2), while the location of thefourth female (father 9) was unknown because she was nottagged before pupping in 1998. This suggests that aminimum of 4.2% of NZ hybrid pups resulted from pureA. gazella or A. tropicalis females mating with A. forsterimales outside territories and away from main breedingareas.

Discussion

Disturbed environments can predispose animal speciesto hybridize in nature (Mayr 1963; Templeton 1981). Ina recently recolonized population of fur seals where twospecies hybridize (A. gazella and A. tropicalis), the unexpectedcontribution to interbreeding by males of a third species(A. forsteri) was explored using genetic techniques. Sporadichybridization through low heterospecific discriminationby females has produced fertile hybrids that have had along-lasting effect on population composition.

Disagreement between observed and actual matingsuccess is not a new concept in studies of reproductivesuccess in polygynous mammals (Coltman et al. 1999;Worthington Wilmer et al. 1999; Gottelli et al. 2007), but attimes can be due to sampling bias rather than the use ofunidentified mating strategies (as shown by Twiss et al.2006). Although the unexpected involvement in breeding

Fig. 4 Individual variation in LOD scores assigned to 10randomly generated haplotypes for 10 Arctocephalus forsteri (black)and 10 NZ hybrid (grey) individuals. LOD scores indicatelikelihoods of haplotypes being A. forsteri (mean ± 2 SE).

Pup no. Pup QMother species

Paternal haplotype

Father species

Sp. specific alleles

Assignment test

LOD score

1 0.463 A 2 NZ NZ 10.44 NZ3 0.273 A 1 NZ NZ 7.09 NZ12 0.677 A 1 NZ NZ 6.84 NZ13 0.26 S — NZ 9.1 NZ

A, A. gazelle; S, A. tropicalis; NZ, A. forsteri.

Table 3 Details of unsampled fathersidentified as A. forsteri males based onpaternal haplotypes deduced from pupswith known mothers

5274 M . L . L A N C A S T E R , S . D . G O L D S W O RT H Y and P. S U N N U C K S

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

by A. forsteri prompted us to scrutinize the mating systemoperating at Macquarie Island, our genetic findings largelysupport observations that terrestrial mating dominates,since the majority of NZ hybrid pups were produced bywithin-territory mating. Somewhat surprising however,was the inability of observers to recognize the hybrid statusof the NZ hybrid territory males that fathered over half ofall NZ hybrid pups. In particular, two successful hybridmales with Q values in the range of first generation hybridswere thought to be pure A. gazella males by experiencedfield observers. At Macquarie Island, hybrid males are shownto have lower reproductive success than pure-species malesas a result of active avoidance by females (Lancaster et al.2007), but it is still unclear which traits females use toselect mates. The phenotypic similarity of these two malesto A. gazella (at least by human perceptions) may havecontributed to their relatively high reproductive success.Alternatively, different hybrid types may vary in theirattractiveness to females. Genetically, A. gazella and A. forsteriare more closely related to one another than either is toA. tropicalis (Wynen et al. 2001), and males of the formertwo species have vocalizations that are more alike (Page et al.2002). Arctocephalus tropicalis males have a very distinctivepelage sharply divided by a pale yellow chest, neck andface, and a crest on the forehead (Condy 1978; Shaughnessyet al. 1988), whereas both A. gazella and A. forsteri males lacka crest and have a uniform pelage. If traits under sexualselection reflect phylogenetic distance among the threespecies and females use phenotype to assess male suitability(as suggested by Goldsworthy et al. 1999), A. tropicalis femalesmay be better than A. gazella females at recognizing heter-ospecific or hybrid males because of the distinctiveness ofconspecifics. Our results support this, as significantlyfewer A. tropicalis than A. gazella females mated withheterospecific or hybrid males, either within or outsideterritories, to produce NZ hybrid pups.

Interspecific mating propensities have been shown todiffer with age, experience and recent reproductive history(reviewed by Wirtz 1999). A significant finding of thisstudy in terms of the role extra-territory mating played inA. forsteri hybridization was the relationship between thereproductive status of females and their propensity to mateextra-territorially: nearly half of all females that matedwith unsampled males outside territories to produce NZhybrid pups did not give birth in the year of conception,and so were not induced to postpartum oestrus. Femalereproductive status or recent reproductive history has beenobserved to influence mating partner in monogamousmammals like the allied rock-wallaby (Spencer et al. 1998)but also in Antarctic fur seals at South Georgia, wherefemales observed without pups had a significantly lowerchance of conceiving with a territory male than femaleswho did pup (Hoffman et al. 2003). While this may reflectgreater freedom in mate choice by females that are not con-

strained to a male’s territory by their pup, at MacquarieIsland, a female that deliberately mates away from breedingbeaches has a high probability of mating with a hetero-specific (A. forsteri) male. Although mate choice by female furseals in an A. gazella colony at South Georgia has been shownto favour highly heterozygous, unrelated territory males(Hoffman et al. 2007), females at Macquarie Island areknown to discriminate against heterospecific and hybridmales (Goldsworthy et al. 1999; Lancaster et al. 2007). Sincefemales that do mate with such males may incur fitnesscosts, it seems unlikely that NZ hybrid pups are producedat Macquarie Island as a result of females preferentiallymating with subordinate NZ hybrid or heterospecificmales over conspecific territory holders. Instead, we mightpropose that females without pups (nulliparous females)mate with A. forsteri or NZ hybrid subordinate males fortwo reasons. First, they miss out on mating with territorymales because they come into oestrus after the peak of thebreeding season when territories have disbanded (Table 2,and see Bartholomew & Hoel 1953; Campagna & Le Boeuf1988 for anecdotal evidence of timing of oestrus in nullipa-rous females) or second, they are young and inexperiencedand haul out to breed in areas away from main haremswhere A. forsteri males are abundant. This latter theory issupported by several observations of young females puppingnorth of Aerial Cove and in Garden Cove but moving intoterritories when their pups are a few weeks old (M.L.L.,personal observation). Such patterns indicate that repro-ductive behaviour and mate choice in females, includingselection of birthing sites and mates, may develop with ageand experience, and the ontogeny of reproductive behaviourappears to be closely linked with the occurrence ofhybridization at Macquarie Island.

Hybrid adult females exhibited a level of viability andfertility sufficient to considerably increase the proportionof NZ hybrids in the population over the study period.A comprehensive analysis of lifetime reproductive successin hybrid and pure-species females is required beforeconclusions can be drawn regarding costs of hybridization,but we found that several females produced hybrid pupsover multiple years. Because A. gazella females wean theirpups after 4 months compared with 8 to 10 months inA. tropicalis and A. forsteri (Kerley 1983; Doidge et al. 1986),a major cost of hybridization suggested for A. gazella femalesis the increase in lactation length that a hybrid pup mightrequire (Goldsworthy et al. 1999; Lancaster et al. 2006).Maternal expenditure has been linked to pupping successin grey seals, whereby greater expenditure resulted in afailure to return and pup the following year (Pomeroy et al.1999). In this study, three gazella/forsteri females and oneA. gazella female produced pups over consecutive years,indicating that the consequences of having a hybrid pupmay be more subtle than simply not giving birth thefollowing year. Further work is required to fully understand

M AT I N G S T R AT E G I E S I N S Y M PAT R I C S E A L S 5275

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

the effect of hybridization on female fitness, but as withother mammals that hybridize infrequently because ofdifferences in life-history traits related to parental care(e.g. Vilà & Wayne 1999), lactation requirements of pupsmay be a major limiting factor for female hybridizationin this population.

In a system where male size and aggression as well asvarious potential mating tactics make all candidate fathersdifficult to sample, we have used a novel combination ofgenetic analyses to illustrate how female reproductivehistory as well as the production of a small number ofviable and fertile hybrid offspring have contributed tounexpected hybridization. Macquarie Island is the onlysource of appreciable numbers of A. forsteri hybrids andfar-ranging dispersal patterns of individuals may facilitatethe spread of hybrid genotypes into other populations.Two factors reduce the potential for this to pose a majorthreat to pure lineages at allopatric sites; the first is thatMacquarie Island comprises the smallest of all fur sealbreeding colonies in the Southern Ocean because it hasbeen the slowest to recover from historic seal harvesting.Thus, occasional immigration of hybrids into other popu-lations would probably be diluted by large numbers of pure-species individuals. Second, the remoteness of MacquarieIsland, more than 5000 km from the nearest A. gazella andA. tropicalis colonies, may discourage dispersal into thesebreeding populations by hybrid fur seals. This is consistentwith the high degree of male philopatry observed in thepopulation (Lancaster et al. 2007). However, MacquarieIsland is much closer to A. forsteri colonies (~1000 km) andfur seals from Macquarie Island, including an adult maleNZ hybrid, have been observed at breeding colonies andholding territories around New Zealand. Further study intoidentifying these emigrants will allow us to make moreaccurate predictions of the regional impacts of hybridiza-tion at Macquarie Island, and detailed analyses of mateselection and hybrid discrimination by female A. forsteriwill be useful for examining the evolution of mating pre-ferences in these species. This study highlights the potentialfor long-term impacts of historic human disturbance oncommunity composition and species purity in long-lived,far ranging and polygynous mammals.

Acknowledgements

We thank all field volunteers for data collection across the studyperiod. Thanks also to L. Wynen, N. Gemmell and S. Negro forproviding genotypes for many individuals from allopatric popu-lations and to Bill Amos and three anonymous reviewers forproviding valuable comments. Tissue samples were collected andretained under permits issued by the Tasmanian Parks andWildlife Service and all work was done with the approval of theAntarctic and La Trobe University Animal Ethics Committees(Australian Government Antarctic Division). This study wasfunded by the Australian Antarctic Division’s Antarctic Science

Advisory Group and Australian Antarctic Science grantsschemes.

References

Banks MA, Eichert W (1999) whichrun (version 3.2): a computerprogram for population assignment of individuals based onmultilocus genotype data. The Journal of Heredity, 91, 87–88.

Bartholomew GA, Hoel PG (1953) Reproductive behaviour of theAlaska fur seal, Callorhinus ursinus. Journal of Mammalogy, 34,417–436.

Boness DJ (1991) Determinants of mating systems in the Otariidae(Pinnipedia). In: Behaviour of Pinnipeds (ed. Renouf D), pp. 1–35.Chapman & Hall, New York.

Campagna C, Le Boeuf BJ (1988) Reproductive behaviour of south-ern sea lions. Behaviour, 104, 233–261.

Clutton-Brock TH (1991) The Evolution of Parental Care. PrincetonUniversity Press, Princeton, New Jersey.

Coltman DW, Bowen WD, Wright JM (1999) A multivariateanalysis of phenotype and paternity in male harbor seals, Phocavitulina, at Sable Island, Nova Scotia. Behavioural Ecology, 10,169–177.

Condy PR (1978) Distribution, abundance and annual cycle of furseals (Arctocephalus spp.) on the Prince Edward Islands. SouthAfrican Journal of Wildlife Research, 8, 159–168.

Cox CR, Le Boeuf BJ (1977) Female incitation of male competition: amechanism of mate selection. American Naturalist, 111, 317–335.

Csordas SE (1958) Breeding of the fur seal, Arctocephalus forsteri Lesson,at Macquarie Island. The Australian Journal of Science, 87–88.

Cumpston JS (1968) Macquarie Island. Antarctic Division, Depart-ment of External Affairs, Melbourne, Australia.

Doidge DW, McCann TS, Croxall JP (1986) Attendance behavior ofAntarctic fur seals. In: Fur Seals: Maternal Strategies on Land andat Sea (eds Gentry RL, Kooyman GL). Princeton UniversityPress, Princeton, New Jersey.

Emlen ST, Oring LW (1977) Ecology, sexual selection and theevolution of mating systems. Science, 197, 215–223.

Gemmell NJ, Burg TM, Boyd IL, Amos W (2001) Low reproductivesuccess in territorial male Antarctic fur seals (Arctocephalusgazella) suggests the existence of alternative mating strategies.Molecular Ecology, 10, 451–460.

Goldsworthy SD, Boness DJ, Fleischer RC (1999) Mate choiceamong sympatric fur seals: female preference for conphenotypicmales. Behavioral Ecology and Sociobiology, 45, 253–267.

Gottelli D, Wang J, Bashir S, Durant SM (2007) Genetic analysisreveals promiscuity among female cheetahs. Proceedings of theRoyal Society B: Biological Sciences, 274, 1993–2001.

Grant PR, Grant BR (1997) Hybridization, sexual imprinting, andmate choice. The American Naturalist, 149, 1–28.

Hoffman JI, Boyd IL, Amos W (2003) Male reproductive strategyand the importance of maternal status in the Antarctic fur sealArctocephalus gazella. Evolution, 57, 1917–1930.

Hoffman JI, Forcada J, Trathan PN, Amos W (2007) Female furseals show active choice for males that are heterozygous andunrelated. Nature, 445, 912–914.

Kerley GIH (1983) Relative population sizes and trends, hybridi-sation of fur seals Arctocephalus tropicalis and A. gazella at thePrince Edward Islands, Southern Ocean. South African Journal ofZoology, 18, 388–392.

Lancaster ML, Gemmell NJ, Negro S, Goldsworthy S, Sunnucks P(2006) Ménage à trois on Macquarie Island: hybridization

5276 M . L . L A N C A S T E R , S . D . G O L D S W O RT H Y and P. S U N N U C K S

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

among three species of fur seal (Arctocephalus spp.) followinghistorical population extinction. Molecular Ecology, 15, 3681–3692.

Lancaster ML, Bradshaw CJA, Goldsworthy SD, Sunnucks P (2007)Lower reproductive success in hybrid fur seal males indicatesfitness costs to hybridization. Molecular Ecology, 16, 3187–3197.

Ling JK (1999) Exploitation of fur seals and sea lions from Aus-tralian, New Zealand and adjacent subantarctic islands duringthe eighteenth, nineteenth and twentieth centuries. AustralianZoologist, 31, 323–350.

Marshall TC, Slate J, Kruuk LEB, Pemberton JM (1998) Statisticalconfidence for likelihood-based paternity inference in naturalpopulations. Molecular Ecology, 7, 639–655.

Mayr E (1963) Animal Species and Evolution. Belknap, Cambridge,Massachusetts.

Nielson EE, Hansen MM, Bach LA (2001) Looking for a needle ina haystack: discovery of indigenous Atlantic salmon (Salmo salarL.) in stocked populations. Conservation Genetics, 2, 219–232.

Page B, Goldsworthy SD, Hindell MA, McKenzie J (2002) Inter-specific differences in male vocalizations of three sympatricfur seals (Arctocephalus spp.). Journal of Zoology, 258, 49–56.

Pomeroy PP, Fedak MA, Rothery P, Anderson S (1999) Con-sequences of maternal size for reproductive expenditure andpupping success of grey seals at North Rona, Scotland. Journalof Animal Ecology, 68, 235–253.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of popu-lation structure using multilocus genotype data. Genetics,155, 945–959.

Shaughnessy PD, Shaughnessy GL, Fletcher L (1988) Recovery ofthe fur seal population at Macquarie Island. Papers and Proceedingsof the Royal Society of Tasmania, 122, 177–187.

Shaughnessy PD, Goldsworthy SD (1993) Feeding ecology offur seals and their management at Heard and MacquarieIslands. Proceedings of the NIPR Symposium on Polar Biology, 6,173–175.

Spencer PBS, Horsup AB, Marsh HD (1998) Enhancement of repro-ductive success through mate choice in a social rock-wallaby,Petrogale assimilis (Macropodidae) as revealed by microsatellitemarkers. Behavioural Ecology and Sociobiology, 43, 1–9.

Templeton AR (1981) Mechanisms of speciation — a populationgenetic approach. Annual Review of Ecology and Systematics, 12,23–48.

Trivers RL (1972) Parental investment and sexual selection. In:Sexual Selection and the Descent of Man, 1871–1971 (ed. CampbellB), pp. 136–179. Aldine-Atherton, Chicago, Illinois.

Twiss SD, Poland VF, Graves JA, Pomeroy PP (2006) Findingfathers: spatio-temporal analysis of paternity assignment ingrey seals (Halichoerus grypus). Molecular Ecology, 15, 1939–1953.

Vilà C, Wayne RK (1999) Hybridization between wolves and dogs.Conservation Biology, 13, 195–198.

Wirtz P (1999) Mother species-father species: unidirectionalhybridization in animals with female choice. Animal Behaviour,58, 1–12.

Worthington Wilmer JW, Allen PJ, Pomeroy PP, Twiss SD, Amos W(1999) Where have all the fathers gone? An extensive micro-satellite analysis of paternity in the grey seal (Halichoerus grypus).Molecular Ecology, 8, 1417–1429.

Wynen LP (2001) Phylogenetic relationships, population genetics andhybridization of two species of southern fur seal (Arctocephalus spp.).PhD Thesis, University of Tasmania, Tasmania, Australia.

Wynen LP, Goldsworthy SD, Guinet C et al. (2000) Postsealinggenetic variation and population structure of two species of furseal (Arctocephalus gazella and A. tropicalis). Molecular Ecology, 9,299–314.

Wynen LP, Goldsworthy SD, Insley SJ et al. (2001) Phylogeneticrelationships within the eared seals (Otariidae: Carnivora):implications for the historical biogeography of the family.Molecular Phylogenetics and Evolution, 21, 270–284.

This study forms part of Melanie Lancaster’s PhD, which exploredhybridization among fur seals at Macquarie Island. Paul Sunnuckscodirects the Molecular Ecology Research Group at MonashUniversity and Simon Goldsworthy manages the long-term furseal monitoring program at Macquarie Island. S.D.G. and P.S. wereM.L.L.’s PhD supervisors.