UC IrvineUC Irvine Previously Published Works

TitleHybridization among the ancient mariners: characterization of marine turtle hybrids with molecular genetic assays.

Permalinkhttps://escholarship.org/uc/item/6720n5bp

JournalThe Journal of heredity, 86(4)

ISSN0022-1503

AuthorsKarl, SABowen, BWAvise, JC

Publication Date1995-07-01

DOI10.1093/oxfordjournals.jhered.a111579

Licensehttps://creativecommons.org/licenses/by/4.0/ 4.0 Peer reviewed

eScholarship.org Powered by the California Digital LibraryUniversity of California

Hybridization Among the Ancient Mariners:Characterization of Marine Turtle HybridsWith Molecular Genetic AssaysS. A. Karl, B. W. Bowen, and J. C. Avise

From the Department of Biology, University of SouthFlorida, 4202 E. Fowler Ave, LIF 136, Tampa, FL 33620-5150 (Karl), BEECS Genetic Analysis Core, Universityof Florida, Gainesville, Florida (Bowen), and Depart-ment of Genetics, University of Georgia, Athens, Geor-gia (Avise). For providing samples, we thank L. Ehr-hart, S. Johnson, J. Keinath, M. Marcovaldi, J. Musick,J. Richardson, T. Wilmer, F. Wood, Cayman TurtleFarm, LTD, Fundacao Pro-Tamar, and Sea World of Or-lando, Florida. For insightful discussions, we thank J.Frazier, C. Limpus, P. Pritchard, J. Richardson, and F.Wood. We thank W. Nelson and K. Scribner for labo-ratory assistance. Our work has been supported byN1H Training Grants in Genetics to B.W.B. and S.A.K.and NSF grants to J.C.A., B.W.B., and S.A.K.

Journal of Heredity 1995;86:262-268; 0022-1503/95/S5.00

Reports of hybridization between marine turtle species (family Cheloniidae) havebeen difficult to authenticate based solely on morphological evidence. Here weemploy molecular genetic assays to document the sporadic, natural occurrence ofviable interspecific hybrids between species representing four of the five generaof cheloniid sea turtles. Using multiple DNA markers from single-copy nuclear loci,eight suspected hybrids (based on morphology) were confirmed to be the productsof matings involving the loggerhead turtle (Caretta caretta) x Kemp's ridley (Lepi-dochelys kempii) (N = 1 specimen), loggerhead turtle x hawksbill (Eretmochelysimbricata) (N = 2), loggerhead turtle x green turtle (Chelonia mydas) (N = 4), andgreen turtle x hawksbill (N = 1). Molecular markers from mitochondrial DNA per-mitted identification of the maternal parental species in each cross. The speciesinvolved in these hybridization events represent evolutionary lineages thought tohave separated 10-75 million years ago (mya) and thus may be among the oldestvertebrate lineages capable of producing viable hybrids in nature. In some cases,human intervention with the life cycles of marine turtles (e.g., through habitat al-teration, captive rearing, or attempts to establish new breeding sites) may haveincreased the opportunities for interspecific hybridization.

Anecdotal reports of marine turtle hybrid-ization have circulated for decades, butscientific documentation has been elusivedue in part to the overall morphologicalconservatism of this taxonomic group.Perhaps the earliest report of marineturtle hybridization involved the "Mc-Queggie," alleged by Caribbean fishermento be a hybrid between the loggerheadturtle (Caretta caretta) and the hawksbillturtle (Eretmochelys imbricata') (Garman1888).

In the century since Garman's (1888) re-port, several sea turtle hybrids have beendescribed, including offspring of Cheloniamydas x E. imbricata (Wood et al. 1983),Ca. caretta X E. imbricata (Frazier 1988;Kamezaki 1983), and Ch. mydas x Ca. ca-retta (Limpus C, personal communica-tion). These appraisals were generally theresult of serendipitous observation ratherthan systematic survey and were based onintermediate features in otherwise diag-nostic morphologic characters. The firstbiochemical assay employed to identifyhybrid marine turtles was protein electro-phoresis, applied by Wood et al. (1983) toconfirm the Chelonia X Eretmochelys crossnoted above, and by Conceicao et al.

(1990) to confirm a diagnosis of Caretta XEretmochelys hybrids in Brazil.

Cases of marine turtle hybridization areremarkable because fossil and geneticdata indicate that these species divergedfrom a common ancestor at least 10 mil-lion years ago (mya). In particular, thetribes Carettini (represented by Caretta,Lepidochelys, and Eretmochelys) and Che-lonini (represented by Chelonia) are be-lieved to reflect an ancient phylogeneticdivision within the family Cheloniidae, dat-ing to perhaps 50+ mya (Bowen et al.1993, and references therein).

In the course of population and evolu-tionary genetic analyses of marine turtles(Bowen et al. 1992, 1993, in press, and un-published data; Karl et al. 1992), we de-veloped a series of mitochondrial (mt)DNA and single copy nuclear (sen) DNAmarkers that can distinguish extant ma-rine turtle species using restriction sitepolymorphisms (RSPs) and/or DNA se-quence assays. The diploid scnDNA mark-ers can identify the parental species ofsuspected hybrids and indicate whethersuch hybrids are probable first, or later,generation products. Under the assump-tion of maternal inheritance, the mtDNAassays indicate which species is the ma-

262

Table 1. Summary of genetic analyses ofputative sea turtle hybrids

Table 2. Summary of nuclear DNA restriction profiles observed in four species and five putativehybrids of marine turtles

Hybridindividual

Collectionsite

mtDNA scnDNAgenotype genotype

Specimen

Hyl and Hy2«Hy3-Hy6"Hy?Hy8"

U.S.A.BrazilU.S.A.Suriname

CCCCLKCM

EI/CCCM/CCCC/LKEl/CM

° Clutch-mates Irom Southern Florida (L Ehrhart and S.Johnson).

"Clutch-mates from the state of Bahia (Marcovaldi M,

personal communication).c Collected as a juvenile in Chesapeake Bay (Keinath J,

personal communication).

* Collected as a single clutch by the Cayman Turtle Farm(Wood et al. 1983).Species designations of the respective genotypes areas follows: CC = Caretta caretta; LK = Lepidochefyskempii; CM = Chelonia mydas; El = Eretmochelys im-bricata.

ternal parent. Taken together, nuclear andmitochondrial assays afford a relativelythorough genetic assessment of putativehybrids.

Materials and MethodsSpecimen CollectionSamples for this study were either tissuesfrom hatchlings or blood taken from ju-veniles that appeared to be intermediatein putative parental species characteristics.Total cell DNA was isolated with a standardphenol-chloroform protocol (Herrman andFrischauf 1987). All DNAs were stored at4°C in TE buffer (10 mM Tris-HCl, pH 8.0,1 mM EDTA). Table 1 provides informationon the collection sites of the individualsthat proved to be interspecific hybrids.

Mitochondrial Genotype DeterminationFor each specimen, either RSP patterns orcytochrome b sequences from mitochon-drial DNA were used to assign a maternalparent species. When fresh tissue sampleswere available (from live eggs or hatch-lings), RSP analyses were conducted onpurified mtDNA isolated by CsCl densitygradient centrifugation (Lansman et al.1981). Aliquots of purified mtDNA were di-gested with at least 11 of the 17 informa-tive restriction enzymes listed in Bowen etal. (1992). Digestion fragments were end-labeled with 35S nucleotides and separatedon 1.0% agarose gels. Restriction frag-ments were visualized with autoradiogra-phy and assigned molecular weights bycomparison to a 1-kb size standard (Gib-co-BRL Inc.). Restriction profiles werecompared to known fragment patternsfrom the six Cheloniid species (Bowen1992, unpublished data).

In cases where the available tissues

Locus

Cm-12

Cm-14

Cm-28

Cm-39

Cm-45

Enzyme

Bg/IIDde 1Dra 1Rsa\

Alu\Dra IHae III

Ava IIBst NlC/blDde\

Bsl EHDde I

NsiiSac\

CM

AAAA

AAA

AAAA

AA

AA

El

ABBA

BBA

BBBB

—

BB

LK°

BBAB

CCB

—

—

—

CC

BBAC

BBC

BCBC

BB

BB

Hyl-2

A&BBA&BA&C

BBA&C

—

—

—

Hy3-6

A&BA&BAA&C

A&BA&BA&C

A&BA&CA&BA&C

A&BA&B

A&BA&B

Hy7

BBAB&C

B&CB&CB&C

—

—

—

Hy8

AA&BA&BA

A&BA&BA

AAAA

—

A&BA&B

" Lepidochefys oliuacea and L kempii produced identical banding patterns for all enzymes at all nuclear loci surveyed.Each identifiable restriction profile is assigned a different letter designation. A dash indicates that the enzyme wasnot surveyed. Species designations are as in Table 1.

were frozen or partially degraded, we usedmtDNA cytochrome b sequences to iden-tify the maternal parent. Biotinylated ver-sions of the primers described by Kocheret al. (1989) were used to amplify mtDNAsequences via the polymerase chain reac-tion (PCR) (Innes et al. 1990). To facilitatethe automated sequencing protocol, M13oligonucleotides complementary to stan-dard sequencing primers were appendedto the 5' ends of these primers. Standardprecautions, including the use of negativecontrols (template-free PCR reactions),were taken to guard against template con-tamination and related problems. PCRproducts were purified with streptavidin-coated magnetic particles (Promega). Sin-gle-stranded sequencing reactions wereconducted with fluorescently labeled M13primers in a robotic work station (AppliedBiosystems Model 800), and the labeledextension products were analyzed with anautomated DNA sequencer (Applied Bio-systems Model 373A) in the DNA Sequenc-ing Core at the University of Florida. Ap-proximately 200 bps of mtDNA cyto-chrome b sequence from the putative hy-brids were compared to known sequencesfrom the six Cheloniid species (Bowen etal. 1993). In all cases, the species assign-ments were unambiguous, because boththe RSP and cytochrome b assays ofmtDNA included a minimum of five spe-cies-diagnostic characters.

Nuclear Genotype DeterminationGenotypes from single copy nuclear DNAwere determined following the procedureof Karl et al. (1992; see also Karl and Avise1993). Five of the anonymous scnDNA

primers (Cm-12, Cm-14, Cm-28, Cm-39, andCm-45) previously developed for the greenturtle (Ch. mydas) also amplified homolo-gous regions from six of the seven marineturtle species. The exception was theleatherback turtle, Dermochelys coriacea(sole member of the family Dermochely-idae), whose DNA failed to amplify withthese five primers under a variety of con-ditions (Karl, unpublished data). Usingthese primers, the DNA from one individ-ual of each of the six species was ampli-fied and screened with up to 40 endonu-cleases (4-, 5-, and 6-base pair recognitionsequences) to identify species-character-istic restriction profiles. Once suitableprimer/enzyme combinations were deter-mined, the target loci from suspected hy-brids were amplified and digested with thetaxonomically informative enzymes. Frag-ment digestion profiles were assayed us-ing 2% agarose gels stained with ethidiumbromide.

Initially, only one individual from eachspecies was assayed for informative mark-ers (with the exceptions noted below), sothere was a possibility that intraspecificpolymorphisms could compromise thespecies assignments. To circumvent thisproblem, several putative species-charac-teristic restriction sites were surveyed ineach hybrid. Agreement in genotypic as-signment at most or all enzyme sites pro-vided an internal control for species iden-tification (in other words, a species diag-nosis would unlikely be compromised bysimultaneous polymorphisms at morethan one restriction site, particularly forgenetically distinct species and genera[Bowen et al. 1993]). Furthermore, for

Karl et al • Marine Turtle Hybrids 263

green turtles and loggerheads (wheremuch larger sample sizes were available),the degree of intraspecific polymorphismat the restriction sites employed as mark-ers was known to be highly constrained(Karl et al. 1992, unpublished data), an ob-servation that further supports our confi-dence in the species assignments in thecurrent study.

Results

Due to the diversity of genetic assays andthe occasional complexities of interpreta-tions, results will be presented separatelyfor each hybrid class identified. The overallresults are summarized in Tables 1 and 2.

Chelonia x CarettaDuring the course of global population ge-netic surveys of green and loggerhead tur-tles (>175 individuals assayed from eachspecies; Bowen et al. 1992, in press), fourhatchling clutch-mates from Brazil (hence-forth Hy3-Hy6), originally thought to begreen turtles, unexpectedly displayed log-gerhead mtDNA restriction profiles upondigestion with 11 informative enzymes(Figure 1). Subsequently, —200 bps of thecytochrome b gene from Hy3 were as-sayed, and these included five nucleotidesnormally diagnostic for loggerheads.

All four individuals also were screenedat several scnDNA loci (Figure 2), usingtwo to five restriction enzymes per locus,which when considered in combinationproduced species-characteristic gel pat-terns [Cm-12 (Bgl II, Dde I, Dra I, and RsaI); Cm-14 (Alu I, Dra I, and Hae III); Cm-28(Ava II, Bst NI, Cfo I, and Dde I); Cm-39 (BstEII and Dde I); and Cm-45 (Nsi I and SacI)]. Although no single enzyme producedunique restriction profiles for all species,the multiple enzyme patterns allowed thespecies identity to be determined un-equivocally (Table 2). For example, the di-gestion of locus Cm-12 with Rsa I pro-duced a restriction profile unique for log-gerhead turtles, but a shared pattern forgreen turtles and hawksbills (the Lepido-chelys banding pattern is also unique withthis enzyme and was not observed in Hy3-Hy6), whereas at this same locus, diges-tion with Dde I produced a restriction pro-file unique for green turtles, but patternsotherwise nondiagnostic for the remainingturtle species. Thus, by considering theprofiles for both enzyme digests jointly,one allele at the Cm-12 locus must havecome from a loggerhead turtle and theother from a green turtle. Using similarreasoning, digestion profiles from Cm-14

Caretta Hybrids CheloniaM/ v V \ M

1 .0 -

Pvu IIFigure 1. Mitochondrial DNA restriction fragment profiles produced by digestion with Pvu II for four specimenseach of Caretta caretta (lanes 2-5), Chelonia mydas (lanes 10-13), and presumptive hybrids between these speciesfrom Brazil (Hy3-Hy6, Table 1). Lanes 1 and 14 are molecular weight standards with the sizes (in kb) of selectedfragments indicated.

for each of these individuals appeared tobe a composite of green turtle and logger-head turtle alleles. Three other scnDNAloci (Cm-28, Cm-39, and Cm-45) were sur-veyed in green turtles and loggerheads.The restriction profiles for green turtleswere different from the loggerheads withall enzymes surveyed. Because not all ma-rine turtle species were assayed, exactspecies assignments could not be made atthese loci. Nonetheless, all digestion pro-files were consistent with the results fromCm-12 and Cm-14. These findings, in con-

junction with the mtDNA evidence, indi-cate that these hatchlings were F, hybridsbetween a loggerhead female and a greenturtle male.

Caretta x EretmochelysBased on morphological appearance, twoclutch-mates (henceforth Hyl and Hy2,collected in Florida and maintained in cap-tivity at Sea World in Orlando, Florida)were suspected of being hybrid deriva-tives of a cross between a loggerheadturtle and a hawksbill. Cytochrome b se-

264 The Journal of Heredity 1995:86(4)

1000 bp

506 bp.

298 bp

Figure 2. Nuclear DNA restriction fragment profiles (Cm-45 cut with Sac I) for four specimens each of Carettacaretta (lanes 10-13), Chelonia mydas (lanes 2-5), and presumptive hybrids involving Caretta caretta X Cheloniamydas (Hy3-Hy6, Table 1). Lanes 1 and 14 are molecular weight standards with the sizes (in bp) of selectedfragments indicated.

quences from these individuals includedsix nucleotides normally diagnostic forloggerhead turtles, thus indicating thatthe maternal lineage was Ca. caretta.

Nuclear DNA from these individuals wasof marginal quality and did not amplifywell for several of the scnDNA loci. None-theless, Hae III digestions at the Cm-14 lo-cus indicated that these individuals werehybrids between a loggerhead turtle andeither a green turtle or a hawksbill (butnot a ridley—Lepidochelys olivacea or L.kempii) and a Dra I digestion eliminatedthe green turtle as a possible parent. In asimilar fashion, digestions at the Cm-12 lo-cus with two enzymes (Dra I and Rsa I)unequivocally identified both Hyl and Hy2as loggerhead x hawksbill hybrids. Takentogether, the genetic and morphologicaldata indicate that these two specimensare F, hybrids between a loggerhead fe-male and a hawksbill male.

Caretta x LepidochelysDuring the summer of 1992, J. Keinath andJ. Musick (Virginia Institute of Marine Sci-ence) identified a suspected loggerhead xKemp's ridley hybrid turtle (Hy7) in Ches-apeake Bay. Blood drawn from this indi-vidual was the source for cytochrome bsequence analysis of mtDNA and for RSPanalysis of scnDNA. The cytochrome b se-quence revealed four nucleotides diagnos-tic for the genus Lepidochelys, and two ad-ditional sites unequivocally identified thespecimen as having a Kemp's ridley (Lkempii) maternal lineage.

Only two nuclear loci were screened,both indicating that Hy7 was a ridley Xloggerhead hybrid. Restriction digests at

the Cm-12 (Bgl II, Dde I, Dra I, and Rsa I)and Cm-14 (Alu I, Dra I, and Hae HI) lociproduced gel profiles that combined frag-ments otherwise characteristic for sam-ples of Caretta versus Lepidochelys (in-cluding a 25-bp size polymorphism at lo-cus Cm-14, unique in our samples to thetwo species of ridley turtles). In summary,Hy7 appeared to be the F, product of across involving a Kemp's ridley female anda loggerhead male.

Chelonia x EretmochelysSince 1977, Cayman Turtle Farm (GrandCayman Island) has maintained 37 turtles(originally collected in Suriname) thatwere suspected of being hybrids betweengreen turtles and hawksbills. In captivity,these individuals have been observed tomate with green turtles and are known topossess motile sperm. Blood was collect-ed from one of these specimens (Hy8) andsubjected to genetic analysis. Cytochromeb sequencing revealed five nucleotidescharacteristic of green turtle mtDNA.

Analyses with scnDNA loci revealed acomplex pattern. For two loci, Cm-12 andCm-14, enzyme digestions (Bgl II, Dde I,Dra I, and Rsa 1; Alu 1, Dra I, and Hae HI,respectively) produced restriction frag-ment profiles indicating that this speci-men was unequivocally a green turtle xhawksbill hybrid. Digestion profiles forCm-45 (Nsi I and Sac I) were consistentwith one of the parents being a greenturtle and the other parent being either ahawksbill or a loggerhead. The inability inthis case to distinguish between the alter-natives for the second parent is due to alack of a known species-specific restric-

tion profile, which can distinguish hawks-bills from loggerheads at this locus. Onelocus (Cm-28) produced only the greenturtle restriction pattern after digestion byfour enzymes (Aua II, Bst Nl, Cfo I, and DdeI), and Cm-39 was not screened. The fail-ure of Cm-28 to confirm the hybrid statusfor this individual raises the possibilitythat the specimen is a second generation(or later) hybrid derived from back-crosses) to green turtles. Alternatively,the primers could have failed to amplifyadequately from the heterologous logger-head template.

Discussion

The hybridization events documentedhere involve four of the five Cheloniid gen-era and four of the six Cheloniid species.Of the remaining species, the olive ridley(Lepidochelys olivacea) is reported to hy-bridize with green turtles in Brazil (Mar-colvaldi M, personal communication), butspecimens were unavailable for geneticanalysis at the time of this writing. An ab-sence of documented hybrids involvingthe flatback turtle (Natator depressus) maybe due in part to its limited range in Aus-tralia and New Guinea.

Ages of Hybridizing LineagesMarine turtles (superfamily Chelonioidea)apparently diverged from an ancestralform —150 mya, and the family Cheloni-idae may have separated from the Der-mochelyidae (extant genus Dermochelys)at least 100 mya (Weems 1988; Zangerland Sloan I960). Within Cheloniidae, thetwo tribes Chelonini (Chelonia) and Caret-tini (Caretta, Eretmochelys, and Lepidoche-lys) have been separated for perhaps 50-75 million years (Bowen et al. 1993; Ernstand Barbour 1989; but see Zangerl 1980),whereas lineages leading to the latterthree genera probably separated on theorder of 10-20 mya (Dodd and Morgan1992; Zangerl 1980). Based on these pro-visional dates as determined primarilyfrom fossil evidence, the average time ofseparation for the hybridizing lineages inthis report is at least 30 mya, and the old-est estimates involve species separatedfor more than 50 million years.

How do these values compare to otherage records for vertebrate hybridization?From surveys of species known to be ca-pable of producing viable hybrids, Wilsonet al. (1974) reported that the oldest suchhybridizing lineages among mammaliantaxa involved species separated for ~6million years. From similar genetic evi-

Karl el al • Marine Turtle Hybrids 265

Table 3. Estimated times of phylogenetic separation for various vertebrate species known to becapable of producing viable hybrids

Lineages involved in hybrid cross

Time sinceseparation(mya)°

Number ofindividuals Reference

Caretta caretta/Eretmochelys imbricataChelonia mydas/Eretmochelys imbricataChelonia mydas/Caretta carettaLepidochelys kempii/Caretta carettaVarious mammalsVarious birdsVarious amphibians

10-2050+50+10-203 (max. 6)

20 (max. 57)20 (max. 54)

2"14*1———

Current studyCurrent studyCurrent studyCurrent studyWilson et al. 1974Wilson et al. 1974Prager and Wilson 1975

" For the marine turtles in this report, estimated dates come primarily from fossil evidence, but are in part corrob-orated by molecular data (Bowen et al. 1993, and references therein). Separation times for the mammals, birds,and amphibians are from the references indicated and citations therein.

"Turtles from the same clutch.

dence, the average age of hybridization-ca-pable lineages within both birds and frogswas estimated to be —20-25 million years(Prager and Wilson 1975; Wilson et al.1974), with a few taxa reportedly havingretained the evolutionary capacity for vi-able hybrid production for slightly morethan 50 million years (Table 3). Thus, avi-an and amphibian taxa apparently lose thecapacity for hybrid production muchmore slowly over evolutionary time thando mammals. To our knowledge, the mostancient lineages known to be capable ofproducing viable hybrids involve frogs ofthe genera Hyla and Pseudacris (Wilson etal. 1974), which may have been isolatedfor perhaps 80 million years by the sepa-ration of Africa from South America.

However, many of the above-mentionedhybridizations involve crosses conductedunder artificial or forced conditions in thelaboratory. Therefore, the Carettini andChelonini (Chelonia x Caretta, and Chelo-nia x Eretmochelys) crosses may be theoldest vertebrate lineages known to hy-bridize in nature.

What accounts for the capacity of theseancient mariners to produce viable hy-brids? Several studies have suggested thatturtle mtDNA evolves slowly relative tothe conventional mtDNA clock for verte-brates (Avise et al. 1992), and this conclu-sion may apply to scnDNA (Karl and Avise1993; Karl et al. 1992) and microsatelliteloci as well (FitzSimmons et al. 1995).Chromosomal evolution is known to be ex-tremely slow in marine turtles (Bickham1981), and the reduced pace of anatomicalevolution of turtles is a proverbial featureof the group. These observations promptthe conclusion that a low evolutionaryrate (including the loss of interspecific hy-bridization potential) is a pervasive fea-ture of marine turtle genetics and evolu-tion.

Perhaps a conservative pattern of ge-

nomic evolution is a necessary prerequi-site to hybridization between relativelyancient turtle lineages. The maintenanceof chromosomal number and structure(Bickham 1981) is one indication of poten-tial genomic compatibility, but additionalgenetic features no doubt are also in-volved in the development of viable hy-brids. Prager and Wilson (1975) suggestthat evolutionary changes in genetic reg-ulatory systems are the primary basis forloss of hybridization potential. In otherwords, differences in the pattern of geneexpression (rather than in the makeup ofstructural genes) may contribute most tothe physiological and ontogenetic basis ofhybrid inviability. Data presented here areconsistent with such conclusions but donot permit a critical appraisal.

Frequencies of InterspecificHybridizationThe nesting habitats of marine turtle spe-cies overlap extensively in every tropicaland temperate ocean basin, occasionallywith three or more species sharing a nest-ing beach. Hence, opportunities for hy-bridization may abound, yet reported in-stances of hybrids are few. One possibilityis that such cases are underreported.When preliminary molecular data on hy-brid turtles was presented at the 12th an-nual marine turtle symposium in 1992(Richardson and Richardson 1995), manyparticipants responded with anecdotal re-ports of "unusual hatchlings," which inretrospect may have been hybrids. For ex-ample, some captive-hatched "green tur-tles" at the Cayman Turtle Farm have "Er-etmochelys-\ike" morphology (Wood F, per-sonal communication), and, hatchlingsrepresenting a presumed Chelonia x Ca-retta cross (at a Queensland, Australiarookery during the 1990-1991 nesting sea-son) have been maintained in captivity for

several years (Limpus C, personal com-munication).

In discussions of hybridization frequen-cy, an important point is that a clear dis-tinction must be made between interspe-cific matings themselves and successfulhybridization events that result in produc-tion of viable progeny. For example, ourdata cannot rule out the possibility thatinterspecific matings are relatively com-mon in areas of overlap, but that viableoffspring are produced only occasionally.The fate of marine turtles hybrids in thewild is extremely difficult to document.However, some clues may be assembledfrom observations on captive specimensat the Cayman Turtle Farm. Hybrid off-spring from the Chelonia x Eretmochelyscross described by Wood et al. (1983, per-sonal communication) had high mortalityand were more susceptible to lung infec-tion than captive green turtles, suggestingthat crosses between species representingChelonini and Carettini have reduced fit-ness. On the other hand, this intertribecross produced at least one male who sur-vived to maturity, had motile sperm, andwas observed to mate with resident greenturtle females (Wood F, personal commu-nication).

Gender-Based Aspects of HybridizationThe mating behaviors of marine turtlesare poorly understood (but see Limpus1993). Most species appear to have dis-tinct courting areas in the vicinity of nest-ing beaches. From commonalities amongthe four cases of hybridization document-ed here, two generalities may be postulat-ed. First, the marine turtle hybrids all oc-curred in regions of shared nesting range.Therefore, a temporal and spatial overlapin mating areas probably facilitates andmay even be necessary for hybridization(Conceicao et al. 1990; Wood et al. 1983).Second, all documented instances of hy-bridization occurred in locations whereone species is abundant and the other rel-atively rare. Perhaps a scarcity of matesfor one species enhances the likelihood ofinterspecific couplings, as, for example,has previously been suggested for geneti-cally documented hybridizations amongspecies of sunfish in Georgia (Avise andSaunders 1984). However, in both cases(sunfish and marine turtles), this latter ob-servation may be an artifact of the rela-tively small numbers of hybrids discov-ered and of the fact that areas of speciesoverlap typically are characterized byhighly unequal species densities.

In another respect, the hybridization re-

266 The Journal of Heredity 1995:86(4)

suits for the sunfish and the marine turtlesdiffered. Among the former, there was astrong tendency (six of the seven reportedinstances) for the rare species in a hybridcross to provide the female parent. Thisobservation suggests that an absence ofconspecific pairing partners and/or spawn-ing stimuli for females of rarer speciesmight be important factors increasing thelikelihood of interspecific hybridization(Avise and Saunders 1984). By contrast, inthree of the four hybrid crosses presentlyreported for marine turtles (E. imbricata 6x Ca. caretta 9, Ch. mydas 6 x Ca. caretta$, and E. imbricata S X Ch. mydas 5), thefemale parent came from the locally abun-dant species. In the fourth instance of ma-rine turtle hybridization, the female parentL. kempii was exceedingly rare relative tothe male parent Ca. caretta, but this maybe a special case because the reciprocalcross may be prohibited for mechanicalreasons (see below). Notably, Ca. carettaserved as either a male or a female parentin these hybridizations, so no inherentgender bias in the directionality of crosseswas apparent.

A possible tendency for hybridizing fe-males to represent common as opposed torare species in marine turtles may be at-tributable in part to the general matingproclivities of the two genders. Male ma-rine turtles are somewhat indiscriminatein mating preferences. At some breedinglocations, fishermen routinely capturemale green turtles by placing barrels orother crude decoys in the water (Carr1956). Males will mount decoys in stereo-typical breeding behavior and remain at-tached while the decoy is retrieved. As istrue for many other animal species, femalemarine turtles may be the more discrimi-nating gender in mate selection and wouldthereby represent the limiting factor in in-terspecific couplings. With more femalesavailable from common species, a con-stant "error" rate in mate choice per fe-male (heterospecific matings) could trans-late into a numerical predominance of hy-brids with mothers from the more com-mon species.

Nonetheless, additional gender-basedconsiderations likely come into play. Forexample, some interspecific matings maybe difficult or impossible for simple me-chanical reasons. Male turtles are equippedwith a large claw on each front flipper andmust firmly grasp the anterior margin ofthe female's shell for copulation. Maleswho are missing front flippers or other-wise cannot secure the female are gener-ally unable to mate (Limpus 1993). Hence,

males of a smaller species (such as L kem-pii) may be unable to copulate with fe-males from a larger species (such as Ca.caretta or Ch. mydas) for purely mechani-cal reasons. In the single case of hybrid-ization in this study involving species dif-fering greatly in size (Ca. caretta x L kem-pii), the female parent was indeed fromthe smaller (Kemp's ridley) species (Table

Hybridization Issues and Marine TurtleConservationAll species of marine turtles are formallylisted as threatened or endangered (IUCN1993). In recent decades, many manage-ment programs for marine turtles have at-tempted transplantations and captiverearing to enhance natural populations(Mrosovsky 1983). In the case of Kemp'sridley, in the years 1978-1992, > 18,000eggs were transplanted from Tamaulipas,Mexico to Padre Island, Texas, and thehatchlings were reared ("headstarted") incaptivity before release. In recent years,some unusual behaviors have been ob-served for this species in nature, includingreproductive activity on the Atlanticcoast, some 2,000 km outside the histori-cal nesting range of L. kempii (Bowen etal. 1994). The possibility exists that thispreviously unnoticed nesting behaviormay be attributable to modification of ear-ly life-history stages via the captive rear-ing program at Padre Island (Bowen et al.1994). In this regard, it is noteworthy thatthe Kemp's ridley x loggerhead hybrididentified in the current study was recov-ered on the mid-Atlantic coast in 1992(Frank 1992).

Generalizations about the impact of hy-bridization on natural populations aredifficult to make, but other cases existin which human-directed transplantationschemes have led to decline and extinc-tion of endangered taxa (Templeton 1994).It is therefore imperative that wildlife man-agers consider such possibilities beforemanipulating the natural histories of en-dangered species.

References

Avise JC and Saunders NC, 1984. Hybridization and in-trogression among species of sunfish (Lepomis): anal-ysis by mitochondrial DNA and allozyme markers. Ge-netics 108:237-255.

Avise JC, Bowen BW, Bermingham E, Meylan AB, andLamb T, 1992. Mitochondrial DNA evolution at a turtle'space: evidence for low genetic variability and reducedmicroevolutionary rate in the Testudines. Mol Biol Evol9:457-473.

Bickham JW, 1981. Two-hundred-million-year-old chro-

mosomes: deceleration of the rate of karyotypic evo-lution in turtles. Science 212:1291-1293.

Bowen BW, 1992. Evolutionary genetics and natural his-tory of marine turtles (PhD dissertation). Athens: Uni-versity of Georgia.

Bowen BW, Conant TA, and Hopkins-Murphy SR, 1994.Where are they now? The Kemp's ridley headstart proj-ect. Conserv Biol 8:853-856.

Bowen BW, Nelson WS, and Avise JC, 1993. A molecularphylogeny for marine turtles: trait mapping, rate as-sessment, and conservation relevance. Proc Natl AcadSci USA 90:5574-5577.

Bowen BW, Kamezaki N, Limpus CJ, Hughes GH, MeylanAB, and Avise JC, in press. Global phylogeography ofthe loggerhead turtle (Caretta caretta) as indicated bymitochondrial DNA haplotypes. Evolution.

Bowen BW, Meylan AB, Ross JP, Limpus CJ, Balazs GH,and Avise JC, 1992. Global population structure andnatural history of the green turtle (Chelonia mydas') interms of matriarchal phylogeny. Evolution 46:865-881.

Carr A, 1956. The Windward Road. New York: Knopf.

Conceicao MB, Levy JA, Marins LF, and Marcovaldi MA,1990. Electrophoretic characterization of a hybrid be-tween Eretmochelys imbricata and Caretta caretta (Che-loniidae). Comp Biochem Physiol 97B:275-278.

Dodd CK Jr and Morgan GS, 1992. Fossil sea turtlesfrom the early Pliocene bone valley formation, centralFlorida. J Herpetol 26:1-8.

Ernst CH and Barbour RW, 1989. Turtles of the world.Washington, D.C.: Smithsonian Institution Press.

FitzSimmons NN, Moritz C, and Moore SS, 1995. Con-servation and dynamics of microsatellite loci over 300million years of marine turtle evolution. Mol Biol Evol12:432-440.

Frank J, 1992. Hybrid sea turtle found in net: discoveryby fisherman called "1 in a billion." Virginian Pilot TuesJuly 21.

Frazier J, 1988. Sea turtles in the land of the dragon.Sanctuary (Asia) 8:15-23.

Garman S, 1888. Reptiles and batrachians from the Cay-mans and the Bahamas. Bull Essex Inst 20:1-13.

Herrman B and Frischauf A, 1987. Isolation of genomicDNA. Methods Enzymol 152:180-183.

Innes MA, Gelfand DH, Sninsky JJ, and White TJ, 1990.PCR protocols: a guide to methods and applications.San Diego: Academic Press.

IUCN (World Conservation Union), 1993. 1994 IUCN redlist of threatened animals. Gland, Switzerland: IUCN.

Kamezaki N, 1983. The possibility of hybridization be-tween the loggerhead turtle, Caretta caretta, and thehawksbill turtle, Eretmochelys imbricata, in specimenshatched from eggs collected in Chita Peninsula. Jpn JHerpetol 10:52-53.

Karl SA and Avise JC, 1993. PCR-based assays of Men-delian polymorphisms from anonymous single copynuclear DNA: techniques and applications for popula-tion genetics. Mol Biol Evol 10:342-361.

Karl SA, Bowen BW, and Avise JC, 1992. Global popu-lation structure and male-mediated gene flow in thegreen turtle (Chelonia mydas): RFLP analysis of anon-ymous nuclear DNA loci. Genetics 131:163-173.

Kocher TD, Thomas WK, Meyer A, Edwards SV, PaaboS, Villablanca FX, and Wilson AC, 1989. Dynamics ofmitochondrial DNA evolution in animals: amplificationand sequencing with conserved primers. Proc NatlAcad Sci USA 86:6196-6200.

Lansman RA, Shade RO, Shapira JF, and Avise JC, 1981.The use of restriction endonucleases to measure mi-tochondrial DNA sequence relatedness in natural pop-ulations. III. Techniques and potential applications. JMol Evol 17:214-226.

Limpus C, 1993. The green turtle, Chelonia mydas, inQueensland: breeding males in the southern Great Bar-rier Reef. Wildl Res 20:513-523.

Karl et al • Marine Turtle Hybrids 267

Mrosovsky N, 1983. Conserving sea turtles. London: and outbreeding depression. In: Principles of conser- tion of Chelonia mydas and Erelmochelys imbricala. Co-British Herpetological Society. vation biology (Meffe GK and Carroll CR, eds). Sunder- peia 1983:839-842.

Prager EM and Wilson AC, 1975. Slow evolutionary loss ' a n d ' M a s s a c h u s e t t s : S i n a u e r : 1 5 2 " 1 5 3 - Zangerl R, 1980. Patterns of phylogenetic differentia-of the potential for interspecific hybridization in birds: Weems R, 1988. Paleocene turtles from the Aquia and t i o n i n t h e Toxochelyid and Chelonnd sea turtles. Ama manifestation of slow regulatory evolution. Proc Natl Brightseat formations, with a discussion of their bear- ^ o 0 ' 20:585-596.Acad Sci USA 72:200-204. ing on sea turtle evolution and phylogeny. Proc Biol Zangerl R and Sloan RE, 1960. A new specimen of Des-

Soc Wash 101:109-145. matochelys lowi Williston; a primitive Cheloniid seaRichardson jl and Richardson TH, 1995. Proceedings of turtle from the Cretaceous of South Dakota. Fieldianathe twelfth annual workshop on sea turtle biology and Wilson AC, Maxson LR, and Sanch VM, 1974. Two types Q • \A.-J_^Qconservation. NOAA technical memo (NMFS-SEFSC- of molecular evolution. Evidence from studies of inter-361). Springfield, Virginia: National Technical Informa- specific hybridization. Proc Natl Acad Sci USA 71:2843- Received November 1, 1994tion Service. 2847. Accepted February 28, 1995

Templeton AR, 1994. Coadaptation, local adaptation, Wood JR, Wood FE, and Critchley K, 1983. Hybridiza- Corresponding Editor: Stephen O'Brien

2 6 8 The Journal of Heredity 1995:86(4)