Evolution, 34(5), 1980, pp. 917-922

INTERSPECIFIC HYBRIDIZATION AND THE EVOLUTIONARY ORIGIN OF A GYNOGENETIC FISH, POECILIA FORMOSA

1 BRUCE J. TURNER ,3 , BETTy-LOU H. BRETT"", AND ROBERT R. MILLER! 'Museum ofZoology and "School of Natural Resources, University of Michigan, Ann Arbor, Michigan 48109

Received November IS, 1979. Revised February 11, 1980

In recent years, largely through the ap­ Atlantic slope of Mexico and Guatemala. plication of allozyme techniques, several The stocks of shortfin mollies used by the unisexual (thelytokous) vertebrates have Hubbses (and by Meyer, who obtained his been shown to be genetically equivalent material from them) were descended to F 1 hybrids of certain related bisexual largely from progenitors collected on the species. The causal relationship, however, Pacific coast of Mexico near Acapulco (L. between interspecific hybridization and C. Hubbs, pers. comm.). They were the origin of unisexuality has not been therefore most likely P. butleri or P. sphe­ elucidated, and the role of hybridization nops (both sensu Schultz and Miller, per se is controversial (Cuellar, 1974, 1971), or perhaps inadvertent laboratory 1977, 1978; Cole, 1978; Wright, 1978). hybrids of one of these species with the The amazon molly, Poeciliaformosa (Gi­ other or with P. mexicana. In other rard), an ovoviviparous gynogentic teleost words, the attempts to produce gynoge­ (family Poeciliidae) was the first known netic P. formosa in the laboratory by in­ unisexual vertebrate (Hubbs and Hubbs, terspecific hybridization were probably 1932). The discoverers of its unisexuality done with what later turned out to be at recognized early that it was probably of least one "wrong" species. hybrid origin, as it was an almost exact To our knowledge, there have been no morphological intermediate between the subsequent attempts to produce gynogens sailfin molly (Poecilia latipinna) and the by interspecific hybridization of Poecilia shortfin mollies (which were then regarded species. Prompted by this, and by the suc­ as a single species, P. sphenops). Labo­ cessful laboratory "synthesis" of hybrido­ ratory hybridization of the presumptive genetic unisexual Poeciliopsis by direct parental species, however, produced only hybridization of bisexual progenitor species bisexual progeny, including fertile males (Schultz, 1973, 1977), we hypothesized that (Hubbs, 1933, 1955, 1961; Meyer, 1938; hybridization of the "correct" Poecilia Hubbs and Hubbs, 19400, 1946b). Sub­ species should produce at least some gyn­ sequent to those early hybridization ex­ ogenetic progeny. Moreover, if gynogens periments, P. "sphenops ," as it was then could be easily synthesized, the interfer­ recognized, has been shown to be an as­ tility of other shortfin mollies in laboratory semblage of morphologically similar but crosses would provide a ready-made sys­ genetically quite distinct species (Hubbs, tem (complete with allozyme markers) in 1961; Schultz and Miller, 1971; Miller, which the genetic bases of hybrid unisex­ 1975; Brett et al., unpubl.). A variety of uality could be studied in detail. The re­ zoogeographic, morphological (Darnell sults presented here, however, indicate and Abramoff, 1968), chromosomal (Prehn that our hypothesis was incorrect or at and Rasch, 1969), and biochemical genetic least incomplete; laboratory hybrid prog­ (Abramoff et aI., 1968; Turner et aI., eny of P. latipinna and P. mexicana are 1980) data suggest that the shortfin species not gynogenetic. involved in the ancestry of P.formosa was P. mexicana, a species restricted to the MATERIALS AND METHODS Our experiments took advantage of the relative ease with which mollies can be 3 Present address: Department of Biology, Virgin­ ia Polytechnic Institute and State University, cultivated and interbred in laboratory Blacksburg, Virginia 24061. aquaria. Hybridization was accomplished 917 918 B. J. TURNER ET AL.

by combining males and females in the 1. Some hybrid females were bred to same aquarium and removing broods of "tester" male commercial "black mollies." young as they appeared. Virgin (sperm­ ("Black mollies" are melanistic Poecilia free) females, from established laboratory hybrids that are bred in large numbers by stocks, P. mexicana Veracruz and P. the ornamental fish industry. They have sphenops Veracruz, were used. The field­ had a complex history [e.g., Hubbs, 1936, caught females of other stocks we used p. 247, pI. 8], as separate all-black strains were assumed to have stored homo specific were derived by selective breeding from sperm before capture. In these cases, the partially black P. latipinna, P. velifera first two broods that each female delivered and possibly from P. mexicana and/or P. after being placed with a heterospecific sphenops. These were subsequently hy­ male were considered to be contaminated bridized in various combinations. The by homospecific progeny and discarded; males we used resembled those of P. la­ the third and subsequent broods were pre­ tipinna, and were all from the same sumed to consist of hybrids. All hybrid source. Shortfin-like strains [sometimes broods were reared in individual aquaria. called "Yucatan mollies"] are also avail­ Males were removed as they became evi­ able commercially.) When mated to such dent by their developing fin colors and/or a tester, nongynogenetic ("bisexual") hy­ gonopodia, and reared separately. brid females should produce offspring that Sources of parental specimens used in express some of the paternal genes for hybridization experiments were as fol­ black pigment (see Schroder, 1964, for the lows: P. latipinna Naples-Stock B77-2, genetics of black body color in Poecilia). canal along Florida Hwy 451; 19 km SE Gynogenetic females, however, should Naples, Florida; collected January, 1977. produce progeny that never express the P. latipinna Nueces-Stock T77-1, Nueces "paternal" genes. This method of detect­ R., boat launching inlet at ranger head­ ing potential gynogenesis requires that all quarters, Hazel Bazemore State Park, test broods be reared until black color Robstown, Texas; collected February genes can be expressed (in some cases, up 1977. P. formosa have also been taken at to 16 wk). this locality (W. S. Moore, pers. comm.) 2. Some hybrid females were bred to but not by us. P. mexicana Monterrey­ tester males of the more distantly related Stock M77-39, Ojo de Agua de Apodaca, insular species, Poecilia (Limia) vittata. 32 km W of Hwy 54, near Monterrey, In the laboratory, males of P. vittata are Nuevo Leon, Mexico; collected June, efficient "fathers" of pseudogamous P. for­ 1977. This stock is presumably referable mosa broods (Hubbs and Hubbs, 1946a, to the subspecies P. m. limantouri (Menzel 1946b;]. S. Balsano, pers. comm.). How­ and Darnell, 1973). P. mexicana Vera­ ever, the hybrid cross of P. vittata males cruz-Stock M66-29; trib. to a lagoon on with females of several bisexual molly the Rancho San Gabriel, 38.3 km N of species is apparently associated with sig­ San Jose Cardel, Veracruz, Mexico; line­ nificant zygote mortality (Schroder, 1965). bred laboratory stock, progenitors collect­ We therefore reasoned that in an aquari­ ed 1966. This stock is presumably refer­ um containing F 1 hybrid females and a able to the subspecies P. m. mexicana male P. vittata of known fertility, any fe­ (Menzel and Darnell, 1973). P. sphenops males which became gravid and carried Veracruz-Stock M67-2, pond 1 km S of broods to term were likely to be gynoge­ La Piedra bridge, approx. 35 km S of Cd. netic; the broods of nongynogenetic fe­ Veracruz, Veracruz, Mexico; line-bred lab­ males would fail before parturition, and oratory stock, progenitors collected 1967. most likely be resorbed. This method is Tests for gynogenesis.-We used three potentially very efficient, for, unless gy­ separate testcross procedures in attempts nogens are very common among the hy­ to detect gynogenetic reproduction among brid progeny, relatively few test broods F 1 hybrid females. need be reared. HYBRIDIZATI ON IN POECILI A 919

latipinna x P. sphenops hybrids were pro­ duced (Table 1). The following consider­ ations are note worthy. 1. There is little evidence, if any, of sig­ nificant zygote mortality associated with a any of the hybrid combinations. Brood size was lowest with the two crosses in­ volving P. latipinna Nueces females, but, lacking data on the perform ance of equ iv­ alent females when mated to conspecifics, we do not know if these small broods are characteristic of the particular femal e, the population, the cross, or other factors. b 2. All of the crosses produced repro­ ductively competent males as well as fe­ male s. Two male hybrid P. mexicana Ve­ racruz female x P. latipinna Nueces males, chosen at random, were individ­ ually backcrossed to female P. mexicana ; they produced ten broods totaling 98 prog­ eny. Similarly, three males of the recip­ c rocal cross, when backcrossed to female P. latipinna , produced 19 broods with 122 progeny. T wo P. sphenops x P. latipinna hybrid males produced more than 30 youn g when crossed to P. latipinna fe­ male s. Males wer e less frequent in the d progeny of all the P. latipinna x P. mex­ icana crosses than would be expected on the basis on 1:1 sex ratios (Table 1 "G;" FIG. 1. a. F, hybrid, sexually mature male, 32 values); interbrood heterogeneities (Table mm Sta ndard Length (S.L.); P. latipinna Nueces 1, "G;" values) were not significant. The male x P. mexicana Veracruz female . b. Hybrid as combined apparent sex ratio for all of the in a female, 37 mm S.L. Progeny from the other interspecific crosses are very similar to these in gen­ P . latipinna x P. mex icana hybrid prog­ eral appear an ce. c. Backcross progeny, female 36 eny was 106 males to 231 females or mm S.L.; F, hybrid reciprocal of that in a X male rou ghly 1:2.2. F rom the pooled brood P. mexicana Veracruz. Extern al morphology very data, the sex ratio of the progeny of the similar to that of P. mexi cana, d. Backcross progeny, P. sphenops x P. latipinna cross is not female, 27 mm S.L.; F, hybrid as in a X male P. latipinna Nu eces. External morphology like that of different from 1:1, but interbrood hetero­ P. latipin na. Parental species and P. f ormosa are geneity of sex ratios in this cross was sta­ figured in Turner et al. (1980). tistically significant (P = .03) and the dif­ ference between it and the three other crosses cannot be evaluated without ad­ 3. Some hybrid females we re back­ diti onal data. The sex ratio data should be crossed to males of one of the tw o parent al treated with some caution, as we lack an

species or mated to their hybrid brothers. estim ate of the proportion of F I "females" The phenotypes of the resultant young that were in fact sexuall y immature mal es

were compared to the F 1 hybrids. without secondary sexual characteristics; the latter phenomenon is common in our RESULTS AND DISCUSSION laborat ory stocks of P. mexicana and in A total of 337 F 1 P. latipinna x P. other members of the P . sphenops com­ mexicana (both reciprocals) and 83 F 1 P. plex. '0 IV 0 TABLE 1. Details of Poecilia hybridization experiments and tests for gynogenesis.

Sex ratios No. F 1 '( '( tested for gynogenesis" (G test) Backcross No. Among Pooled P. parentals Brood size broods versus 1:1 Black (Limia) P. P. Hybrid Total N N molly vittata lat. mex. F, 0 Total combination 00 22 N J ± SE F, 00 22 G" P G,. P o cross o cross 0 0 cross tested P. mexicana

Veracruz <;> x 21 2 7 19.6 ± 4.9 137 48 89 9.92 0.16 12.4 <.005 59 15 3 5 - 82 P. latipinna Nueces <3 P. latipinna ~ ':-< Nueces <;> x 1 2 9 12.4 ± 1.9 112 37 75 6.47 0.59 13.1 <.005 33 15 6 - 54 >-3 P. mexicana d Veracruz <3 ~ t':I P. mexicana ;tI t':I Monterrey <;> >-3 x 1 1 1 (8) 8 3 5" -- - 1 -- - 1 2 ;l> P. latipinna r-' Nueces <3 P. latipinna

Nueces <;> x 2 2 9 8.9 ± 2.2 80 18 62 8.59 0.42 25.6 <.005 20 8 - 1 - 28 P. mexicana Monterrey <3 P. sphenops

Veracruz <;> x 1 1 4 20.8 ± 2.3 83 47 36 9.04 0.03 1.44 0.33 4 - 2 6 P. latipinna Naples <3

I Both pairs were housed in the same aquarium. Since both males were sexually active, and successful multiple fertilization is common among laboratory poeciliids, this arrangement probably maximizes the number of genetic combinations involved in the cross. 2: 3 '( F I specimens died of undetermined causes before they could be tested. 3 None of the females tested was gynogenetic. HYBRIDIZATION IN POECILIA 921

3. The morphology (shape, color pat­ and remove them. Limited fertility of test­ terns, meristics) of the F j hybrids (Fig. 1) er males may also have been a problem. was roughly intermediate between the pa­ c. Backcross progeny were inter­ rental extremes. Female hybrids of all mediate in phenotype between the F i hy­ crosses were impossible for us to distin­ brids and the parental species (Fig. 1). guish from laboratory-reared P. formosa The apparent lack of gynogenesis in any of comparable size. The inner jaw teeth of the P. latipinna X P. mexicana hybrid of all hybrids we examined are conical and females tested is surprising in view of all slightly blunted (as they are in P. formosa the other data which suggest that these and P. latipinna). The dentitions of the P. two species are involved in the ancestry latipinna X P. mexicana hybrids are in­ of P. formosa. Our failure to produce gy­ distinguishable from those of the P. lati­ nogenetic progeny by laboratory hybrid­ pinna x P. sphenops hybrids. The genes ization of the putative parental species encoding the distinctly tricuspid inner jaw could be reasonably explained by either of of the P. sphenops parent are evidently the following two hypothesis. recessive to those encoding the conical 1. Despite the wealth of morphological, teeth of P. latipinna, a finding seemingly allozyme, and other data, at least one pa­ at variance with the partial dominance of rental species has been misidentified. Our conical teeth in other P. sphenops hybrids allozyme surveys of the Mexican compo­ (Schultz and Miller, 1971). The morphol­ nents of the P. sphenops complex (Brett ogy of the hybrids and of their backcross et al., unpubl.; Brett, unpubl.) have now progeny will be dealt with in more detail become extensive, and it is clear that P. elsewhere. mexicana provides by far the best "match" 4. No gynogens could be identified with the genome of P. formosa of any among the F, hybrid females. member of that complex. Evaluation of a. None of the 38 females bred to this hypothesis should therefore focus pri­ male P. (Limia) vittata males produced marily on members of the P. latipinna any progeny. complex, most especially P. petenensis. b. Of the 121 F j females bred to The latter species, unknown genetically, "tester" male black mollies, 25 produced is morphologically rather similar to P. la­ broods; each of these was reared separate­ tipinna. ly. All tester progeny, without exception, 2. There are particular "gynogenetic showed definite expression of paternal genotypes" among the genomes of at least genes for black body coloration. Consid­ one of the parental species that result in erable variation was noted among broods gynogenesis upon hybridization. The fre­ in the time at which the paternal genes quency of these genotypes varies geo­ were expressed; in some broods the black graphically and perhaps temporally. They markings were evident within a week of are rare or absent in the populations we birth, but in others, sometimes with the tested (we thus obtained only bisexual same father, expression was delayed, in progeny), but are (or were) more common one case up to about 16 weeks after birth. in others, including the actual progenitors The test progeny have blotchy or reticu­ of P. formosa. If this hypothesis is correct, lated color patterns with little bilateral it follows that the differences between symmetry; progeny with similar markings gynogenetic and nongynogenetic geno­ are produced when black molly tester types are probably small in magnitude, males are mated to P. latipinna females. and, at the extreme, may involve allelic At present we do not know why only differences at but a single locus. The evo­ about a fifth of the F j females that were lutionary origin of parthenogenesis in P. bred to black molly tester males actually formosa and other unisexuals may, there­ produced progeny. We suspect that some fore, reside not in the wholesale interac- newly-born broods may have been lost to tion of the two divergent components of maternal predation before we could detect a hybrid genome (as seems widely held) 922 B. J. TURNER ET AL. but in the action of certain alleles at one ---. 1955. Natural hybridization in fishes. Syst, or a few loci that have been placed, by Zoo!' 4:1-20. hybridization, into a novel genetic envi­ ---. 1961. Isolation mechanisms in the specia­ tion of fishes, p. 5-23. In W. F. Blair (ed.), Ver­ ronment. tebrate Speciation. Univ. Texas. Symp., Austin. HUBBS, C. L., AND L. C. HUBBS. 1932. Apparent ACKNOWLEDGMENTS parthenogenesis in nature, in the form of a fish We thank R. M. Bailey and M. L. of hybrid origin. Science 76:628-630. Smith for collecting specimens in the field, ---. 19400. Breeding experiments with the in­ variably female, strictly matroclinous fish, M 01­ J. Humphries for helpful discussions, and lienisia formosa. Genetics 31:218. P. Monaco and R. Andrews for critical ---. 1946b. Experimental breeding of the Am­ readings of the manuscript. We also thank azon molly. Aquarium J. (San Francisco) 17:4-6. Mrs. L. C. Hubbs for her time-consuming MENZEL, B. W., AND R. M. DARNELL. 1973. Sys­ tematics of Poecilia mexicana (Pisces: Poecili­ plunge into old records and discussions of idae) in northern Mexico. Copeia 1973:225-237. her Poecilia hybridization work. Research MEYER, H. 1938. Investigations concerning the re­ supported in part by NSF grants DEB76­ productive behavior of M ollienisia formosa. J. 20958 (BJT) and DEB77-17315 (RRM). Genet. 36:329-366. PREHN, L. M., AND E. M. RASCH. 1969. Cytoge­ LITERATURE CITED netic studies of Poecilia (Pisces). I. Chromosome numbers of naturally occurring poeciliid species ABRAMOFF, P., R. M. DARNELL, AND J. S. BAL­ and their hybrids from eastern Mexico. Can. J. SANO. 1968. Electrophoretic demonstration of Genet. Cytol, 11:888-895. the hybrid origin of the gynogenetic teleost Poe­ SCHRODER, J. H. 1964. Genetische Untersuchungen cilia formosa. Amer. Natur. 102:555-558. an domestizierten Stammen der Gattung Molli­ COLE, C. J. 1978. Parthenogenetic lizards. Science enesia (Poeciliidae). Zoo!' Beitr. 10:369-463. 201:1154-1155. ---. 1965. Zur vererbund der Dorsal Flossen­ CUELLAR, O. 1974. On the origin of parthenogen­ strahlenzahl bei Mollienesia Bastarden. Z. Zoo!' esis in vertebrates: the cytogenetic factors. Amer. Syst. Evolut.-Forsch. 3:330-348. Natur. 108:625-648. SCHULTZ, R. J. 1973. Unisexual fish: laboratory ---. 1977. Animal parthenogenesis. Science synthesis of a species. Science 179:180-181. 197:837-843. ---. 1977. Evolution and ecology of unisexual ---. 1978. Parthenogenetic lizards. Science fishes. Evo!. Bio!. 10:277-331. 201:1155. SCHULTZ, R. J., AND R. R. MILLER. 1971. Species DARNELL, R. M., AND P. ABRAMOFF. 1968. Dis­ of the Poecilia sphenops complex in Mexico. Co­ tribution of the gynogenetic fish Poecilia formosa, peia 1971:282-290. with remarks on the evolution of the species. TURNER, B. J., B. L. BRETT, E. M. RASCH, AND Copeia 1968:354-361. J. S. BALSANO. 1980. Evolutionary genetics of HUBBS, C. L. 1933. Species and hybrids of Mol­ a gynogenetic fish, Poeciliaformosa, the Amazon lienisia. The Aquarium (Philadelphia) 1:263­ molly. Evolution 34:246-258. 268, 277. ---. 1936. Fishes of the Yucatan peninsula. Corresponding Editor: J. R. Powell Carnegie Inst. Wash. Pub!. 457:157-287.