REPRODUCTIVE MECHANISM OF UNISEXUAL AND BISEXUAL STRAINS OF THE VIVIPAROUS FISH POECILIOPSISl

R. JACK SCHULTZ Museum of Zoology, The University of Michigan

Received September 12, 1960

The fishes of the order Cyprinodonti­ one or two uneven rows that follow the formes have evolved a variety of systems heart-shaped curvature of the lips and the of sex-determination, some of which are outer row of teeth. The second type of unique among vertebrate animals. One dentition is characteristic of an aberrant such anomalous condition was recently re­ strain of "F" females which are designated ported by Miller and Schultz (1959) in the as "Fx." The inner teeth of this strain viviparous genus Poeciliopsis Regan, where­ consist of a fine sandpaper-like patch on in certain strains gave birth to only female either side of the midline. No other con­ offspring. Poeciliopsis comprises approxi­ sistent morphological differences have been mately 16 species that range from southern found between the two strains, yet repro­ Arizona to Colombia, with at least two ductively they are radically different. Mat­ endemics in the Atlantic drainage of the ings of "F" females having normal denti­ Isthmus of Tehuantepec (fig. 1). Western tion to "F" males result in young of both

Mexico harbors the largest number of sexes, but the "Fx " females with aberrant species. dentition mated to the same male produce This study deals principally with two only female progeny. species of Poecdiopsis tentatively assigned The same aberrant dental character was to the species group Leptorhaphis Regan later found in samples of strain "c." In (Miller, 1960). Since neither species has these, however, dentition proved to be less been described.f they are here assigned the reliable for distinguishing the unisexual symbolic designations "C" and "F." Poe­ strain, designated "C x ," from the normal ciliopsis "C" is found in southern Sonora females. Offspring from "C x " individuals and northern Sinaloa, where it is confined showed a spectrum of variation from the to the Mocorito, Sinaloa, and Fuerte river nearly normal to the aberrant types of drainages. The southern limit of "F" is dentition. These progeny were nevertheless less than 50 miles to the north; it extends all females, as were the succeeding genera­ from the Rio Mayo northward to the tions. Yaqui, Matape, Sonora, and Concepcion This "all-female" phenomenon appears rivers in Sonora. to be similar to that discovered by Hubbs From preserved material collected within and Hubbs (1932) in Mollienesia formosa, the range of strain "F," Miller was able a member of the same family as Poe­ to identify females with two types of denti­ ciliopsis. tion. One of these is considered to be the It is the purpose of this study to find normal type in that the minute inner teeth the sex-determining mechanism responsible are like those of the male. They consist of for unisexuality in Poeciliopsis. Aspects of

1 Part of a thesis submitted to The University the reproductive biology and genetics of of Michigan in partial fulfillment of the require­ both the bisexual and the unisexual types, ments for the degree of Doctor of Philosophy, therefore, have been studied and compari­ June, 1960. Cost of extra pages paid for by the author. sons have been made with other animals 2 Since this was written, the species here re­ known to have unusual sex-ratios. The fol­ ferred to "C" and "Cx" has been described as lowing mechanisms have been considered: Poeciliopsis lucida; the one designated "F" and lethal systems, polyploidy, parthenogenesis, "Fx" may be identical with P. occidentalis (Miller, 1960) . gynogenesis, sex-inversion, cytoplasmic in-

EVOLUTION 15: 302-325. September, 1961 302 REPRODUCTIVE MECHANISM OF POECILIOPSIS 303

20' /

II POEClLlOPS1S ·c-cx Ea POECILIOPSIS 'f- f~ [J ALL OTHERS

..'

FIG. 1. Distribution of Poeciliopsis in part of North America, showing localities from which live stocks were obtained. heritance, unbalanced autosomal system, cated in fig. 1 by collecting station num­ and selective maturation. bers. Specimens to be used for genetic analysis MATERIALS AND METHODS or to be dissected for gonadal studies were Live stocks used in this investigation fixed in 10% formalin and preserved in were collected in northwestern Mexico dur­ 70% ethyl alcohol. Examination of live ing expeditions by Miller and Greenbank material was facilitated by the use of an in 1955, Miller and Miller in 1957, and anesthetic, MS-222 (Tricaine methane­ Miller and Schultz in 1959. The localities sulfonate). Immature fish used in histo­ from which they were obtained are indi- logical studies were killed and fixed in 304 R. JACK SCHULTZ

Bouin's solution and preserved in 70% completion of yolk deposition, it migrates alcohol. The entire fish was embedded in to the periphery, where maturation division paraffin and sections were cut at 10 mi­ occurs. crons. These were stained in Ehrlich's After this series of eggs is fertilized, haematoxylin and counterstained with eosin, yolk accumulates in a second series. By and were examined at magnifications of the time the first embryos are 7 to 10 days 430x and 980x. Cytological preparations old and have reached the halfway point in were made from embryos one or two days their development, the second series of ova old and from testes of adult males. The is fertilized. In strains "C" and "Cx" the tissue was killed and fixed in Newcomer's blastula or streak stage of a third brood of solution, stained in aceto-carmine and embryos is sometimes found before the squashed between cover slip and slide. All birth of the first or oldest stage of embryos, fish measurements are the total length in but in "F" and "Fx" no more than two mm estimated to the nearest tenth. stages of embryos occur at one time. If sperm are not available to fertilize the REPRODUCTIVE BIOLOGY newly matured ova the condition does not Insemination.-Poeciliopsis, being a vi­ remain static but degeneration of the ova viparous fish, has internal fertilization. ensues. Degenerating ova and embryos are Spermatophores or sperm packets are common in females living under subopti­ placed in the genital pore of the female mum conditions or in those used in certain by means of the male's modified anal fin, hybrid crosses. Frequently one to several the gonopodium. Here, the packets break ova of a clutch fail to become fertilized or up, releasing the spermatozoa. These mi­ an embryo dies in the course of its develop­ grate up the gonaduct to the ovary where ment. The age of these can be correlated they remain viable for months. As a re­ with the brood of embryos of which they sult of sperm storage, Poeciliopsis females normally would have been a part. Degen­ may produce as many as ten broods of erating elements in the ovary can be recog­ young after they have been isolated from nized for more than two weeks after their their mates. death. Ovarian Cycle.-Most viviparous fishes In dealing with unisexual animals, one have only one stage of embryos in the of the concerns is whether or not the one ovary at a time: the birth of these must sex dies before maturity. A more complete take place before the next series of ova is evaluation of lethals as they relate to the fertilized. Poeciliopsis, however, has super­ unisexual strains of Poeciliopsis will be fetation, wherein two or more stages of considered later, but certain aspects that embryos develop simultaneously (Turner, involve the ovary are treated appropriately 1937) . here. From the brood records and from the In a reproductive system such as that of examination of more than 400 females, a viviparous fishes, a lethal mechanism acting generalized sequence of events occurring during the intraovarian cycle would reveal within the ovary has been reconstructed. itself by the presence of dead embryos, Except for minor deviations it seems ap­ eggs, or oocytes. In "Cx" and "Fx," a suf­ plicable to all species of Poeciliopsis. ficient number of failures to account for Ovaries of submature females contain unisexuality might be expected at any of several strings of tiny, white oocytes, .25 the following stages of the ovarian cycle: to .30 mm in diameter. As the female ( 1) maturation division; (2) fertilization; matures, one or two of these oocytes begin (3) early cleavage; and (4) advanced em­ to accumulate yolk, which gives them an bryos. In "C" and "F" no more than a few amber color. The nucleus is centrally lo­ prenatal deaths theoretically should occur. cated in these submature eggs; but upon From the 38 "C" and "Cx" females dis- REPRODUCTIVE MECHANISM OF POECILIOPSIS 305

TABLE 1. Comparison of intraovarian mortality TABLE 2. Minimum and average brood interval between strains "C" and "Cs" in days of strains "C;" "Cs" "F," and "E,"

Ova and Per cent Minimum Average Strain of Number embryos of Strain of Sample brood brood female dissected observed mortality female size interval interval

C 11 102 21.7 P.C 182 4 11.5 Cx 27 319 5.6 P. c, 124 7 11.7 P.F 26 5 13.8 P.Fx 76 9 13.2 sected for this phase of the study, only a few dead embryos could be found. The is retained in the ovary and is not born degenerating elements observed were for until one or two days after the others. Fe­ the most part identified as ova. Many of males in poor condition have irregular in­ these, however, were undoubtedly e~rly em­ tervals, but even healthy individuals ex­ bryos undetected because of their small perience interruption or "rest periods" after size and degenerate condition. When the a long sequence of births. Although more total intraovarian mortality of "C" and typical of winter months, long brood inter­ "Cx" was compared (table 1), the uni­ vals also may occur during the summer. sexual strain actually had fewer degenerat­ The above irregularities were disposed of ing elements (5.6%) than did the bisexual by considering only those intervals that strain (21.7%). A study of the aquarium fell between 4 and 20 days (table 2). Com­ history of each female revealed that all of parison of the average brood intervals .of those containing more than one dead ovum the four strains of females revealed no sig­ or embryo had been subjected to subopti­ nificant difference between the 11.5 day mal rearing conditions such as inadequate interval of "C" and the 11.7 day interval illumination or overcrowding. Since strain of "Cx," nor was there any difference be­ "C" is in general less hardy than "Cx," tween the 13.8 and 13.2 day intervals of the higher mortality rate in the ovaries of "F" and "F.." The occasional presence of these fish is accountable. When individuals three stages of embryos in the ovaries of of both strains are maintained under ideal "C" and "Cx" is reflected in the shorter conditions, intraovarian failures are almost brood intervals of these two strains. negligible. It must, therefore, be assumed Brood size data are difficult to evaluate. that if differential mortality is responsible Not only does the number of young in for unisexuality in this genus, it takes place broods from a single female sometimes vary during a later stage of development. for no apparent reason, but environmental Brood Intervals and Brood Sizes.-De­ factors also affect the reproductive condi­ velopment of superfetation in poecil~id tion of the fish. In Poeciliopsis, inadequate fishes has significantly reduced brood In­ illumination, even during early life, may tervals and brood sizes. In Lebistes and reduce the reproductive capacity of fe­ similar forms that lack superfetation, as males. Artificial light sufficient to produce many as 75 young are born at intervals of a plankton bloom does not compare with 22 to 30 days, whereas species with two natural light for rearing Poeciliopsis young. or three stages of embryos seldom have more than 25 young per brood. These are Overcrowding and peck-order dominance born at intervals of 7 to 14 days. result in stunted fish which, upon reaching Normally there is little fluctuation in adulthood, may have reduced fecundity. the brood interval of healthy Poeciliopsis These factors are so complex as to con­ females' but in compiling extensive data stitute a separate study and will not be for com~aring species or strains, some dis­ pursued here. . torting features must be taken into account. In order to obtain at least some notion Occasionally a single young from a brood of the reproductive potential of Poeciliop- 306 R. JACK SCHULTZ 35 - Imttl AVERAGE BROOD SIZE OF' C ~ 30 - MAXIMUM BROOD SIZE OF' C mIIIIill AVERAGE BROOD SIZE OF Cx D MAXIMUM BROOD SIZE OF' CX

10

5

~Irnll I 'S 12 30-34 34-38 38-42 46-50 TOTAL LENGTH IN FIG. 2. Average and maximum brood sizes of "C" and "C," females in relation to the total length of the female. The sample size of each class is indicated in the square on each bar. sis, females of each type were anesthetized that "C" reaches its maximum brood size and measured after the birth of young. The and stops reproducing at about 41 mm, average and maximum brood for females whereas "Cx" continues to have young until of various size classes are presented in fig­ about 50 mm long. In these size ranges ures 2 and 3. Although certain size classes "Cx" has produced broods with far more are poorly represented, particularly for "F" young than any other species of Poeciiiop­ and "Fx ," several interesting tendencies sis reared in this laboratory. Four such are apparent. In comparing strains "C" broods contained 34, 23, 21, and 19 off­ and "Cx ," the expected trend of larger fe­ spring. Only "C" has approached this rec­ males to have larger broods is evident. The ord, with broods of 16, 14, 12, and 12. two strains behave differently, however, in Since only two size classes of "F" are rep- REPRODUCTIVE MECHANISM OF POECILIOPSIS 307

Mill AVERAGE BROOD SIZE OF' F ~ MAXIMUM BROOD SIZE OF' F m AVERAGE BROOD SIZE OF' FX 0 MAXIMUM BROOD SIZE OF' FX

I 9 10 I

26-30 30-34 34-38 46-50 TOTAL LENGTH FIG. 3. Average and maximum brood size of "F" and "F;" females in relation to the total length of the female. The sample size of each class is indicated in the square on each bar. resented, all that can be said of the repro­ yolk, was interpreted as evidence of im­ ductive potential of this species is that the maturity. maximum brood consisted of 10 young­ The various species and strains of Poe­ these by a 36.4 mm female (fig. 3). Al­ ciliopsis differ in the size and age at which though the data for "Fx" are somewhat they mature. Of the four types considered scanty, this strain apparently produces here, "C" is the smallest. Females of this fewer young in the laboratory than either strain have mature ova when 20 to 22 mm "c" or "Cx." "F,' females do, however, long and at an age of 49 days. The young­ attain nearly the same maximum length as est to reproduce were 69 days old and 21 the "Cx" females. It is also possible that to 23 mm long. Strain "Cx" matures at "Fx" produces fewer young than "F," but 21 to 24 mm and at a minimum age of 70 more information is required to clarify this days-20 days older than "c." point. Females of strains "F" and "Fx" ma­ Maturation and Senescence.-The attain­ tured at a minimum size of 23 mm and at ment of sexual maturity in Poeciliopsis fe­ an age of more than two months; no young males is not accompanied by apparent ex­ were born prior to the attainment of 29.9 ternal changes. Therefore, dissection was mm. The variations and discrepancies required in order to determine the age or among individuals, however, indicate that size at which females become adults. Ova­ in these strains, particularly in "F," fac­ ries from aquarium-reared individuals of tors influencing gonad development are various ages and sizes were examined. more complex than those involved in either Those with at least one mature or degen­ "c" or "C x ." Five of the "F" females erating ovum were considered fully de­ examined exceeded 25 mm; yet only one veloped, whereas the presence of only of these contained ova. Certainly their size oocytes, whether or not they contained should have been sufficient for gonadal 308 R. JACK SCHULTZ

B "

K er

FIG. 4. Descendants of five wild "Cx" females from M57-69. Letter = female designation; sub­ script = laboratory generation; number = offspring from each female that were raised to maturity. All young that matured were females. maturation. There is some evidence that The number of days required for males females born during the summer do not to mature, as evidenced by gonopodial de­ mature until early spring of the following velopment, varied considerably among in­ year. dividuals and species. "C" males on the Strains "C" and "Cx" differ not only in average reached adulthood in approxi­ the size and length at which they mature mately half the time required for "F"-68 but also in the maximum length attained. versus 142. Once mature, the males stop Females of "C" reach their maximum growing. No significant change in length length at about 43 mm, whereas "Cx" fe­ could be demonstrated over a period of five males not uncommonly exceed this by more to nine months for five adult males of Poe­ than 5 mm. "F" females were smaller than ciliopsis "C." The total life span for males those of the other three strains. Their con­ of each species is about one year, but a few dition, however, probably reflected the ef­ individuals have lived more than a year fects of aquarium life, since specimens col­ and a half. lected in nature and placed in the aquarium had lengths of 40 to 43 mm, approaching EXPERIMENTAL RESULTS the maximum for "F, ." The largest of the Intraspecific Matings and Sex-Ratios of "Fx" females had total lengths of about 46 Offspring.-Although it is known that cer­ mm. tain Poeciliopsis females give birth only Females of most species of Poeciliopsis to female offspring whereas others, mated reared under optimum conditions live for to the same male, produce young of both about a year and a half; 678 days was the sexes, several questions arise regarding the maximum age attained by any individual. nature of such reproduction: (1) Do the No great difference was found in the lon­ unisexual strains consistently produce fe­ gevity of the four strains. male offspring or does an occasional male REPRODUCTIVE MECHANISM OF POECILIOPSIS 309

U5511 "'57-6'

AI 25U 15d'd'

FIG. 5. Descendants of five wild "C" females. Letter = female designation; subscript = laboratory generation; number = offspring from each female that were raised to maturity. result? (2) In broods from unisexual parents, for only those "Cx" females which mothers, do bisexual type females ever had nine or more female offspring can arise? (3) Do bisexual females sometimes randomness be discarded as a factor in the give birth to the unisexual type offspring? unisexuality of the progeny. From the five (4) What is the sex-ratio among progeny original "Cx" females, 25 descendants ful­ from the bisexual parents? filled these requirements, whereas in not a In an attempt to answer these questions, single one of the 36 F1-F 4 generations females of all four strains were mated and could a transfer to the bisexual type be the resulting young raised to maturity. The demonstrated. hardiness of "C" and "Cx" made possible The same 1957 collection from which the the backcrossing, outcrossing, and inbreed­ "Cx" aquarium stock was derived con­ ing of various lines, but matings of "F" tained three females that ultimately were and "Fx" were somewhat limited by the shown to represent the bisexual strain. Two reduced fecundity of "F" under aquarium other females, one from the 1955 collection conditions. and one from the 1959 collection, were also From the live material collected in 1957, of the "C" type. From these five females five females proved to be of the "Cx" and their descendants, 744 young were strain. These and their descendants (fig. born, some of which constituted an F 5 4) gave birth to 923 young. After 50 died generation (fig. 5). The number of these of natural causes and 276 were either dis­ offspring that reached maturity was 592 carded or preserved, 597 attained matur­ (an additional 97 died of natural causes ity; not one of these was a male. Although and 55 were discarded or preserved as offspring were produced by 41 different young). Of those that matured, 313 were 310 R. JACK SCHULTZ

females and 279 were males-essentially a backcrossing were abnormal sex-ratios again 1:1 ratio (Chi-square = 1.95, P > 10). produced. From the original five wild females, 28 rep­ Information derived from matings of the resenting the F 1 through F <[ generations "F-Fx" group is similar to that of the were mated. All but one of these produced "C-Cx" group. All "Fx" individuals were at least some males and, therefore, must be derived from three wild females collected considered bisexual. Since from the one fe­ in 1955. These and their 11 F 1- through male only three offspring reached maturity, F<[-generation female progeny gave birth to she cannot be considered unisexual even 453 young, of which 290 females were though no males developed. When the sex­ reared to maturity (112 were either pre­ ratio of offspring from each female was served or discarded and 51 died of natural analyzed, it was seen that significant devia­ causes). As was true of "Cx ," not a single tions from the expected 1: 1 ratio occa­ male was produced.

sionally occur; examples are M57-69 <;? G2 The situation in the bisexual strain, "F," (with 25 <;? <;? :80' 0') and M57-69 <;? C2 is similar to that of "C." Offspring were (with 0 <;? <;? : 14 0' 0' ). A number of authors obtained from 17 wild females collected in (particularly Krumholz, 1948) have indi­ 1955, 1957, and 1959. Of the 243 young cated that among the viviparous fishes born, 70 died of natural causes and 45 males are less hardy than are females. In were discarded or preserved. The 128 that explanation, Essenberg (1923) suggested reached maturity consisted of 75 females that the higher metabolic rate of males is and 53 males. Here again, as in "C," the responsible. In strain "C," although a 1: 1 deviation from a 1: 1 ratio is not significant ratio was demonstrated, there is still an at the .05 level; yet, some differential mor­ apparent tendency for more females to tality is strongly suggested. reach maturity than males. It is also pos­ It is now possible to answer within cer­ sible that males are more delicate than fe­ tain limits the questions posed earlier re­ garding the nature of unisexual and bi­ males during the period between birth and sexual reproduction in Poeciliopsis. In view the attainment of adulthood. Since 14.2 of the fact that no males occurred among per cent of the offspring from "C" females 887 offspring of the two unisexual strains, died before they matured and only 7.7 per the likelihood of such an event taking place cent of the "Cx" young died during the is fairly remote. It is furthermore unlikely same period, differential mortality might that females of either the bisexual or uni­ account for the deviation of the <;? G2 prog­ sexual strains ever produce offspring of the eny from a 1: 1 ratio, and also cause the other type. The sex-ratio of the bisexual total number of "C" offspring to comprise strains is probably about 1: 1 at birth, and slightly more females than males. although differential mortality results in a The exceptional sex-ratio of <;? C2 , how­ few more females reaching adulthood than ever, cannot be accounted for on the basis males, the deviation from a 1: 1 ratio is of differential mortality or chance. Since not sufficient to call for the postulation of the probability of obtaining 14 male off­ a genetic system. spring by chance alone is 1 in 16,224 it Early Gonad and Sex-Inversion.-The might be expected that one parent or the occurrence of unbalanced sex-ratios among other carried genetic factors that were cyprinodont fishes has been subject to strongly male-determining. The pedigree numerous investigations. Notable among of the parents involved in this event, how­ these is the work of Essenberg (1923), who ever, did not suggest this. A series of mat­ showed that the preponderance of males in ings were conducted to determine whether aquarium stocks of Xiphophorus helleri or not this maleness was genetically trans­ was a result of sex-inversion. Conceivably, mittable, but neither by inbreeding nor inversion of sex could be carried suffi- REPRODUCTIVE MECHANISM OF POECILIOPSIS 311

ciently far to result in the unisexual strains eighth day the ovary is filled with germ of Poeciliopsis. Here, however, the inver­ cells of various sizes; the two halves of the sion must be reciprocal to that of Xipho­ gonad are almost completely fused and its phorus, so that all individuals genetically shape has become quite irregular. oriented to maleness are converted to fe­ Four of the eight "C" young that were males. If such a reversal does take place, sectioned were males. Three characteristics examination of gonads from a sufficient render the early testis distinct from the number of "Cx" or "Fx" young at the time early ovary: (1) the germ cells are all that sex is first distinguishable should re­ approximately the same size, (2) most veal individuals with rudiments of testes. of the stroma cells are located medially, Because "C" and "C," young were more whereas the germ cells are pushed laterally, readily available, they were chosen for and, (3) some of the stroma cells have this study. Offspring of different ages were clustered together at the center of the testis sectioned and their gonads examined. in the formation of sperm ducts. The gonads of Poeciliopsis "Cx" young Having determined the differences in killed at birth are sexually undifferentiated. structure of the early testis and the early They are widely paired and are suspended ovary, 13 "Cx" females, 6 days old, were beneath the swimbladder in the peritoneal examined for evidence of maleness. The sac that later becomes the permanent wall mother of these young was an F 1 female of the gland. Internally they are charac­ that had produced 33 other offspring, all of terized by two cell types: (1) the small which matured as females. All 13 of the stroma cell that is indistinguishable from sectioned gonads were unmistakably fe­ peritoneal cells and has an irregular, darkly male. Since the probability is 1 in 8,112 staining nucleus, and (2) a series of larger that 13 females would be selected by cells representing the initial stages of trans­ chance alone, it is safe to conclude that at formation from the stroma to the primor­ the time sex is first established all "Cx" dial germ cell. In the course of this trans­ young are females. Hence, the premise that formation, the nucleus and cytoplasm of sex-inversion is responsible for broods of the stroma cell increase in size, the chro­ all-female offspring has no histological sup­ matin becomes coarse and starts to clump port. The genetic evidence, considered and the once irregularly shaped nucleus later, further refutes this possibility. rounds out, and loses its affinity for stain. Cytological Results.-Chromosome num­ Both the stroma cells and the series of bers have thus far been determined for 27 transforming germ cells are evenly distrib­ species of cyprinodont fishes (Wickbom, uted throughout the indifferent gonad. 1943; Makino, 1951; and Oztan, 1954). By the sixth day several changes have With few exceptions the haploid number is taken place and it is possible to distinguish 23 or 24 and the diploid, 46 or 48. In all the sexes cytologically. In females, the two of these studies the small size of the chro­ halves of the ovary have begun to fuse mosomes prevented detailed morphological anteriorly. The number of transforming analysis; consequently, little is known be­ germ cells has significantly increased, and yond the fact that most are rod-shaped, some have more than doubled in size. The with terminal centromeres. nuclei of the largest have lost their affinity In the present study counts were made for stain and, except for the strands of from spermatogonial divisions of strains chromatin, are now lighter than the cyto­ "C" and "F" and from somatic tissue of plasm. The germ cells of the early ovary the embryos of "C," -c, ," and "Fx." The remain evenly distributed throughout the haploid number for both "C" and "F" structure; but most of the stroma cells males is clearly 24 (fig. 6) . No deviation have become pressed about them in the from this number was found in cells from initial formation of the follicle. By the approximately 12 gonads that were pre- 312 R. JACK SCHULTZ

spring had the dorsal-fin spot; but the males outnumbered the females by about 3 to 1. When the same cross was made using the unisexual female, "Cx," instead of the bisexual type (Hybrid Mating II), the fin spot was again transmitted to all of the progeny; but in keeping with the char­ FIG. 6. First meiotic prophase (spermato­ acteristic of the "Cx" strain, all of the off­ genesis), strain "C," N: 24, 1300 x. spring were females. The cross, "Fx" X "C" (Hybrid Mating III), as expected, pared for study. The diploid chromosome also produced only females, but all of these number determined from the somatic tissue had clear fins. Hence, in the first three is 48 for the bisexual and the two unisexual crosses only the phenotype of the male was strains (figs. 7 and 8). transmitted to the F 1 generation. When, Interspecific Matings and Genetic Analy­ however, the F 1 hybrid male ("C" X "F") sis.-In the absence of intraspecific marker which had the dorsal-fin spot was mated to genes suitable for tracing the movement of "Cx" "Fs ," or the F 1 ("Fx" X "C") fe­ genetic material through the unisexual male, half of the offspring were spot-fin strains, hybrid matings of "C" and "F" and half were clear-fin (again, no males were attempted. Both sexes of "F," in­ appeared). cluding the "Fx" strain, have a spot on the At this point, transmission of the dorsal­ dorsal fin (fig. 9). This aggregate of fin spot would appear to be best explained melanophores is basally located on the by a system of sex-linkage. Although the posterior, interradial membranes, for the existence of sex chromosomes has not been most part between rays 4 and 6. In large cytologically verified in cyprinodont fishes, specimens melanophores become more nu­ the few species for which their presence has merous and in some fish cover the entire been genetically established females, with lower third of the posterior rays; at mini­ one exception, were indicated to be homo­ mum development the melanin is always gametic, XX; the males, heterogametic, sufficiently concentrated to form a spot. XV. In "C" and "C," this character is lacking When a sex-linked system with homo­ (fig. 10). Although large females have a gametic XX females was applied to the few melanophores scattered along the base spot-fin transmission data, a fit could not of the dorsal fin, they are never concen­ be obtained, regardless of whether the spot trated into a spot. character was considered dominant or re­ In the first series of hybrid matings be­ cessive. If the female, however, were het- tween "C-Cx" and "F-F-" a clear-fin "c" female was mated to a spot-fin "F" male (Hybrid Mating I, table 3). All of the off-

FIG. 7. Mitotic metaphase, strain "C," 2 N: 48, FIG. 8. Mitotic metaphase, strain -c.,: 2 N: 1300 x. 48, 1300 x. REPRODUCTIVE MECHANISM OF POECILIOPSIS 313

FIG. 9. Strain "F" female (upper, 27.0 mm), "F," female (middle, 30.2 mm), and "F" male (lower, 22.1 mm), showing dorsal-fin spot characteristic of this group.

FIG. 10. Strain "C" female (upper, 22.5 mm), "C,' female (middle, 25.1 mm), and "C" male (lower, 17.1 mm), showing clear condition of dorsal fin of this group. 314 R. JACK SCHULTZ

TABLE 3. First series of hybrid matings between "C-Cz" and "F-Fz," showing transmission of the dorsal-fin spot

Phenotype of offspring Hyhrid Female Male Dorsal-fin spot No dorsal-fin spot mating Female Male Female Male

I C F 21 57 0 0 II Cx F 23 0 0 0 III Fx C 0 0 39 0 IV Cx (C X F) 39 0 46 0 V Fx (C X F) 4 0 2 0 VI (Fx X C) (C X F) 12 0 12 0 VII (Fx X C) (C X F) C 0 0 23 0 VIII (Fx X C) (C X F) C 0 0 10 0

erogametic, the male homogametic, and the parents and progeny in terms of this hy­ dorsal-fin spot dominant to its allele, no pothesis. Following the method of Gordon spot, the phenotypes of the offspring from (1952), the sex chromosomes of the het­ all six of the hybrid matings fulfill the re­ erogametic female are designated "WY" quirements of the theoretical genotypes. and those of the homogametic male "YY." Hybrid Matings I, III, and VI (fig. 11) In Hybrid Mating III, in which a spot-fin show the genotype and phenotype of the "F,.' female, WrY", was mated to a clear-

STRAIN C STRAIN F STRAIN Fx STRAIN C

HYBRID MATING I HYBRID MATING m

F:f. .,.,_-@-;-el ...., x {IV i I I '-<--,: ~==.;;- - __I

HYBRID MATING:m:

FIG. 11. Inheritance of dorsal-fin spot in hybrid crosses between strains "C-Cx" and "F-Fx," as­ suming females to be heterogametic (WY) and males homogametic (YY) and the spot-fin character to be dominant (S). Unisexual strains are indicated by stippling; males by presence of gonopodium. Had males been born to unisexual females, their genotypes and phenotypes would have been as repre­ sented in broken-line figures. REPRODUCTIVE MECHANISM OF POEClLIOPSIS 315

fin "C" male, ysys, all offspring were tained, suggesting, rather than sex-linkage, clear-fin females. Under normal circum­ a simple Mendelian type of inheritance. In stances, males should be heterozygous ysys this hybrid experiment the spot character and have the fin spot. Since males are not in most of the offspring was less developed produced by unisexual females, however, than in the original "F" parent. Although no spot-fin progeny occurred-leaving only individuals of the "F-Fx" group vary some­ the W-ys genotype. When these W-ys what in the degree to which the dorsal-fin F 1 hybrid females were mated to the ysys spot is expressed, critical examination of heterozygous male (Hybrid Mating VI), the F 1 progeny from "C" X "F" and the half of the progeny were spot-fin females reciprocal cross left little doubt that the and half were clear-fin females and hence character is actually intermediate when were W-ys and »i-v», respectively. If hybrid matings involve bisexual females. males had been produced here their geno­ This type of inheritance is in keeping with types, ysys and ysys, would have resulted what is known of other fish hybrids, in half of them having the spot and half wherein the F 1 generations are almost ex­ having no spot. The two types of F 2 fe­ clusively intermediate between the parental males produced by this cross are the same species. In most of these hybrids, however, genotype as "Cx" (W-YS) and "Fx" expression of species characters is con­ (W-YS). It, therefore, follows that mat­ trolled or influenced by multiple factors, ing these females to "C" males (ysYS) such that in backcrossing to the original should result in the same type of off­ parents introgression is a gradual process. spring as would be obtained by mating Expression of the dorsal-fin spot, however, "Cx" or "F-" females to "c" males, does not depend on the cumulative effect that is, all clear-fin females. The progeny of a number of genes. Apparently only one from both of these crosses (W-ys X ysys allele is involved and the heterozygous con­ and W-ys X y"yS), in Hybrid Matings dition results in incomplete dominance, VII and VIII, were, as predicted, all clear­ such that for either sex: (1) SS = strong­ fin females (table 3). spot, (2) Ss = weak-spot, and (3) ss = no In the "Fx" X "C" cross, had a bisexual spot. In the backcrosses, then, "C" X ("C" "F" female been used instead of the uni­ X "F") becomes ss X Ss, which produced sexual strain, progeny of both sexes would both male and female offspring, half of have resulted. The F 1 males produced, un­ each with clear-fins, ss, and half with a like their clear-fin, W-ys sisters, would be weak-spot, Ss; ("C" X "F") X "C" be­ heterozygous ysys for the dorsal-fin spot. comes Ss X ss which again results in half It was not until 1959, however, after a new clear-fin and half spot-fin progeny, whereas stock of "F" was available, that this cross ("C" X "F") X "F" becomes Ss X SS, was successfully completed. The 13 off­ from which only spot-fin young are de­ spring that resulted consisted of eight fe­ rived, but of two types: Ss, weak-spot, and males and two males, but instead of all F 1 SS, strong-spot. females being clear-fin (W-Y") and all "C" and "F" males and females, con­ males spot-fin (ysyS), as the sex-linked sequently, have chromosomes with homol­ hypothesis demands, both sexes had the ogous loci, which act on the same character, dorsal-fin spot. but with opposite effect. In the unisexual In Hybrid Mating XII, in which an F 1 strains the corresponding locus is not found hybrid female ("C" X "F") was mated on the chromosome designated as "w" in to the clear-fin "C" male, the W-ys (clear fig. 11. The s-gene carried by either 'i? ), genotypes predicted by the sex-linked gamete of the "F" male is therefore hypothesis again were not obtained. In­ allowed full expression in combination with stead of the females being clear-fin and the the W-chromosome. Fertilization by either males spot-fin, a 1: 1: 1: 1 ratio was ob- gamete of the "C" male, however, in com- 316 R. JACK SCHULTZ

TABLE 4. Second series of hybrid matings between "C-Cx" and "F-Fx" showing transmission of fin-spot character

Phenotype of offspring Hybrid Female Male Dorsal-fin spot No dorsal-fin spot mating Female Male Female Male

IX F C 11 3 0 0 X C (C X F) 23 20 25 35 XI (C X F) F 4 23 0 0 XII (C X F) C 4 3 6 4 XIII (FX C) C 8 1 3 3 XIV (C X F) (C XF) 12 13 1 5 bination with the W-chromosome, results bryological data indicated that 50 to 80 in only the clear-fin condition. If males per cent of the would-be F2 progeny were were produced from "F," X "C," the spot stillborn. would then be present only in an inter­ A prominent feature in several of the mediate form, since the F 1 male would con­ hybrid matings involving females of the tain an S-gene from the "Fx" female and bisexual strains is the discrepancy from a an s-allele from the "C" male. In Hybrid 1: 1 sex-ratio. This is particularly notice­ Matings IV, V, and VI, wherein -c. ," able in the "C" X "F" and ("C" X "F") "Fx ," and F 1 hybrid ("Fx" X "C") fe­ X "F" matings, in which 1:3 and 1:5 males were mated to the F 1 hybrid male ratios favored the males. In others, such ("C" X "F"), the two types of male as "F" X "C" and ("F" X "C") X "C," gametes, one marked by the S-gene and one the ratio is apparently reversed, with two by the s-allele, combined with the female or three times as many females as males. gamete containing the W-chromosome to Differential mortality is probably negligi­ produce half spot-fin and half clear-fin ble, since broods sometimes contained as daughters. If expression of the spot-fin many as 10 young. character were controlled by multiple fac­ One female of Hybrid Mating I ("C" X tors involving even as few as two loci, four "F"), that produced 39 female and 16 male types of gametes would be produced by offspring (all of which had the dorsal-fin the F 1 hybrid male: AB, Ab, aB, and abo spot), was remated to a male of her own The first three of these combined with the species after the "F" sperm was exhausted. W-chromosome would result in spot-fin In the new combination she produced 47 fe­ females, whereas only the latter would pro­ males and 56 males (all of which lacked duce clear-fin individuals. Since the ratio the dorsal-fin spot), thus restoring the 1: 1 of spot-fin to clear-fin progeny in all three sex-ratio. In view of this result and the crosses was 1: 1 and not 3: 1, it is safe to evidence against a lethal mechanism, it conclude that multiple factors are not in­ seems likely that the sex-determining fac­ volved. tors of "C" and "F" have a difference in The phenotypes of only one of the 14 potential. This matter is further pursued crosses (tables 3 and 4) do not conform in the discussion of sex-determination in to the expected ratio. When the F 1 off­ both the unisexual and bisexual forms. spring of "C" X "F" were inbred in Hy­ DISCUSSION brid Mating XIV, a 3: 1 ratio was expected. All but six of the 31 offspring, however, Sufficient biological and experimental had either a weak or a well-developed spot. evidence is now available to permit an eval­ The phenotypic ratio in this cross evidently uation of the Poeciliopsis problem in terms was altered by the deleterious effects of of what is known of one-sided sex-inherit­ hybridization. Brood size data and em- ance in other animals. Since the literature REPRODUCTIVE MECHANISM OF POECILIOPSIS 317

treating this subject is too extensive for a formosa virgins produced no young; but complete review, only a few examples of when placed with males of M. latipinna or the various reproductive processes are pre­ M. sphenops (species with which M. sented. formosa is sympatric) large broods resulted. It was concluded that Mollienesia formosa Parthenogenesis and Gynogenesis reproduces by a process of gynogenesis, Parthenogenesis is a common phenome­ wherein sperm is essential to initiate de­ non in the animal kingdom; one form or velopment of the embryo but does not ac­ another has been found in nearly all of the tually take part in formation of the zygote. major phyla. Beatty (1957), White (1954), A similar situation has since been re­ and Suomalainen (1950) have discussed at ported (Lieder, 1955) wherein certain pop­ length the various types of parthenogenesis ulations of goldfish (Carassius auratus) and have indicated that nowhere among the specified as the silver goldfish (C. a. vertebrates has a species or strain been gibelio) produced only female progeny. found that normally reproduces without Here again stimulus to the ova was pro­ fertilization. Although numerous examples vided by other species, in this case a gold­ may be cited in which cell division was fish and a carp. initiated in fish, amphibian, and mammal­ The evidence that parthenogenetic or ian eggs by physical or chemical stimula­ gynogenetic reproduction occurs in fishes, tion, most of the artificially produced in­ and that three of the four species for which dividuals failed as early embryos. Only it has been postulated belong to the same the Beltsville Small White turkey has pro­ family as Poeciliopsis, suggests that a sim­ duced viable progeny from unfertilized and ilar mode of reproduction might be func­ apparently unstimulated eggs (Olsen and tional in this genus. In order for either Marsden, 1954 and Kosin, Sato, and process to provide a suitable explanation Nagra, 1958). All of these parthenoge­ for the production of only female offspring netic birds were males, some of which it is essential that no paternal characters reached maturity and yielded small amounts appear in any of the progeny and that the of semen containing motile spermatozoa sperm serve no purpose beyond that of a (Poole, 1959). stimulating agent. Although intraspecific Among the teleost fishes a few special cases are known. These involved virgin marker genes, suitable for testing such a births from two females of Lebistes reticu­ hypothesis are not known for "C" and "F," latus and from one female of Xiphophorus interspecific matings involving the trans­ helleri (Stolk, 1958). The ovaries of each mission of the dorsal-fin spot have pro­ female were found to be infested with the vided the necessary character. In Hybrid phycomycete Ichthyophorus hoferi Plehn­ Mating II (table 3) it was demonstrated Mulsow, which apparently provided a toxic that a clear-fin "Cx" female mated to a substance responsible for the artificial stim­ spot-fin "F" male resulted in 23 female ulation of the ova. More than 64 offspring progeny, all of which had the spot. Cross­ were produced by this pathological parthe­ ing a spot-fin "Fx" female in Hybrid Mat­ nogenetic process; these, without excep­ ing III with a clear-fin male gave 37 off­ tion, were females. spring, all with the pigment condition of The famous Mollienesia formosa, men­ the male. The consistent expression of pa­ tioned early in this paper, was studied for ternal characters by unisexual progeny in 12 years by Hubbs and Hubbs (1932, these and other hybrid experiments (table 1946) and Hubbs (1955) during which 3) is conclusive evidence that partheno­ time 21 generations of over 8,000 offspring genesis or gynogenesis cannot explain the were produced, all of which were pheno­ all-femaleness of Poeciliopsis strains "Cx" copies of the female parent. Isolated M. and "Fx ." 318 R. JACK SCHULTZ

Lethal Mechanism or postnatal stage, at either of which it would be readily detectable. The percent­ Abnormal sex-ratios resulting from dif­ ages of deaths among offspring from the ferential viability have been reported nu­ four strains of females were as follows: merous times. Most of these, however, do "Cx," 7.70/0; "C," 14.20/0; "F 15.00/0; not result in complete annihilation of one x," and "F," 35.40/0. From "Cx" females, en­ sex. The hybrid cross of Drosophila mela­ tire broods of 20 and 34 offspring were nogaster X D. similans is exceptional and raised to adulthood without a single loss. more closely resembles the situation in Poe­ Since so few deaths occur among the prog­ ciliopsis, in that only females are present eny of the unisexual strains, a lethal mech­ among the mature offspring. Sturtevant anism is unsubstantiated for the period be­ ( 1920) found that most of the dead and tween birth and sexual differentiation. dying larvae of the cross are males. A It is still possible, however, for the ex­ similar mechanism is known in Drosophila pression of a lethal gene to be limited to prosaltens (Cavalcanti, Falcao, and Castro, the prenatal stage. One would expect under 1957). In this species, as in the two spe­ such circumstances to find brood sizes of cies of Poeciliopsis, both unisexual and bi­ unisexual females reduced to about half sexual strains of females are known. Al­ those of the bisexual strains, but no such though normal D. prosaltens mothers, like reduction was found. The brood size of those of Poeciliopsis strains "C" and "F," each unisexual strain was about equal to produced no daughters of the unisexual that of its bisexual counterpart (figs. 2 and types, the unisexual flies differed from 3). "C" females in the smaller size ranges "Cx" and "Fx" by occasionally giving rise produced slightly more young than "Cx" to the bisexual form. It was believed by females, but larger "Cx" females had the above authors that the unisexual type broods that were twice the size of those females carry a cytoplasmic factor 0 of any other species of Poeciliopsis. Fur­ (omikron), which is reproduced only in thermore, dissected ovaries of "Cx" fe­ those individuals homozygous or heterozy­ males contained rarely more than one de­ gous for the "sex-ratio" gene, Sr' The generating embryo, even when as many as omikron plasmagene, which is transmitted 30 healthy ones were developing. The hy­ exclusively through the eggs of "sex-ratio" pothesis that unisexuality is a result of a females, kills only the XY embryos. lethal system acting during either the pre­ It is not inconceivable that a similar natal or postnatal period must, therefore, lethal system is responsible in Poeciliopsis be rejected. for the lack of male offspring from strains "Cx" and "Fx." Transmission of the char­ Sex-Inversion acter of the dorsal-fin spot is not contradic­ Sex-inversions have been found in a vari­ tory to such a mechanism. In Hybrid Mat­ ety of animals, but the incidence is strong­ ing III (fig. 11), in which "Fx" was mated est in those with sex-determining mecha­ to "C," the ysys genotype of the males nisms still in a primitive state (White, was not represented among the F 1 progeny; 1954) . Cyprinodont fishes fall into this only clear-fin, W-ys, females were pro­ category. Their sex-chromosomes are so duced. In the same cross but with normal little differentiated as to be morphologically females, the ysys genotype was expressed indistinguishable from autosomes. The sex by the F 1 males as a dorsal-fin spot of in­ of animals of this type is often unstable, termediate intensity. and subject to alteration by both internal In spite of the attractiveness of a lethal­ and external factors. mechanism explanation of unisexuality in Among the cyprinodont fishes examples Poeciliopsis, biological evidence offers no of sex-inversion are known. For example, support. A lethal gene in poeciliids might in the guppy, Lebistes reticulatus, and the produce its effect during either a prenatal medaka, Oryzias latipes, the normal XX REPRODUCTIVE MECHANISM OF POECILIOPSIS 319

( Cjl )- XY (!i) relationship is occasionally In Hybrid Mating III (fig. 2), a spot­ upset by an imbalance of minor sex-de­ fin "Fx" female was mated to a clear-fin termining genes on the autosomes (Winge "C" male. Since the gene for spot-fin is and Ditlevsen, 1947; Aida, 1936). Mat­ located on only one "Fx" chromosome, two ings of these reversed XY females and XX types of gametes should normally occur. males to normal individuals resulted in prog­ One of these, which is marked by a chromo­ eny with extremely abnormal sex-ratios. some lacking the spot-gene locus, when In the swordtail, Xiphophorus helleri, un­ combined with either of the clear-fin male usual sex-ratios have been attributed to gametes results in clear-fin females. The sex-determining systems relying on hor­ other female gamete, marked by the spot­ monal control (Essenberg, 1923 and Friez, fin gene, should in combination with the 1933) . clear-fin male gametes result in male off­ Kosswig (1954) presented a generalized spring in which the spot-fin is expressed theory of sex-determination for fishes in with reduced intensity. If these spot-fin which he stressed both environmental and males were genetically produced but re­ genetic factors. versed to females, half of the offspring of The evidence that sex-inversions do oc­ the "Fx" X "C" cross would have a spot cur in cyprinodonts suggests that one such on the dorsal fin. It can, therefore, be system might be functioning in Poeciliopsis concluded that autosomal or cytoplasmic strains "Cx" and "F-." External, internal, factors or hormonal or physical influences genetic, and cytoplasmic influences have all do not reverse the sex of "Cx" or "Fx" been considered. males to females. In rearing viviparous fishes, standardi­ zation of the external environment is im­ Incompatibility Mechanisms practical; consequently, offspring from In a number of plants, for example Zea, each of the four strains of females were Datura, and Oenothera, incompatibility reared under all available aquarium con­ systems that have arisen between the style ditions. Regardless of the light, tempera­ and pollen grains cause differential pollen­ ture, time of year, space, food, or any tube growth. Since one pollen tube may other physical variable in the laboratory, grow slower than the other, incompletely, females of the bisexual strains produced or not at all, only one type of gamete may both male and female progeny and females reach the ovary. of the unisexual strains produced only fe­ In viviparous fishes a similar relationship male young. might exist between the gonaduct and the If sex-reversal is to be considered a char­ spermatozoa, whereby viability or motility acteristic of "Cx" and "Fx," it is more of the sperm is adversely affected. If likely to be operative through an internal the male gametes are of two types, one process. Histological evidence, presented carrying the female-determining factors above, is strongly contradictory to the con­ and the other male determiners, failure of cept of an inversion brought about by the latter to reach the ova could result in endocrine imbalance. Sectioned gonads only female progeny. from "Cx" young killed at the time of Hybrid matings between "C-Cx " and sexual differentiation revealed no evidence "F-Fx ," however, demonstrate the unlike­ of male origin. This, however, does not ex­ lihood of such a system operating within clude some autosomal or cytoplasmic con­ the unisexual strains (table 3). Gametes trol that may gain prominence over the of the "C" X "F" F 1 male were marked sex-chromosomes prior to the formation of by either the clear-fin gene, s, or the spot­ a distinct gonad. It is in the transmission fin gene, S. When, in Hybrid Mating IV, data of the dorsal-fin spot that evidence a clear-fin "Cx" female was mated to this contradicts sex-inversion in any form. male, both phenotypes were represented 320 R. JACK SCHULTZ

among the female progeny in a 1: 1 ratio, mechanisms that might be responsible for thus demonstrating that each gamete was the unisexual strains of Poeciliopsis. equally effective in fertilizing ova. CONCLUSIONS Polyploidy Results obtained from studying the re­ Although polyploidy has been a signifi­ productive biology of Poeciliopsis strains cant factor in the evolution of plants, it "Cx" and "Fx" reduce considerably the has played a minor role in the animal number of possible explanations for their kingdom. The fact that most animals with unisexuality. Inheritance of the dorsal-fin reduplicated sets of chromosomes or in­ spot contributed most to the final conclu­ creases of one or more elements are either sion. In hybrid crosses between the bisex­ parthenogenetic or hermaphroditic strongly ual types, "C" and "F," the spot-fin con­ suggests that bisexual reproduction acts as dition is determined by a pair of alleles a barrier to polyploidy (Muller, 1925). such that in "F," SS resulted in the fin­ The claim of Svardson (1945) and Kupka spot; in "C," ss resulted in no spot; and in ( 1948) that polyploidy has occurred in the F 1 hybrid progeny, Ss showed incom­ salmonid fishes is based on conflicting evi­ plete dominance with the spot character dence and is not widely accepted. Among present in dilute form. When hybrid mat­ the vertebrates a single triploid newt, Tri­ ings involved either of the unisexual strains, turus vulgaris, is the only known naturally the pattern of inheritance differed. Invari­ occurring example (Book, 1940). Experi­ ably the fin condition of the male was mentally produced triploid and tetraploid transmitted unaltered to the F 1 progeny. Amphibia are not uncommon, but they in­ If a hybrid male (Ss) was the father, two variably are sterile; consequently, it is phenotypes were represented, strong spot quite unlikely that polyploidy has played and no spot in a 1: 1 ratio, with the in­ a part in the evolution of Amphibia (White, termediate condition absent. 1954) . It was concluded from a series of hybrid In fishes, although an exclusively female matings using all four strains with "C" and F 1 generation might be produced by an "F" males that the unisexual strains differ increased dosage of female determiners from the males and females of the bisexual through polyploidy, the problem of sterility strains by at least one chromosome. This remains. Were this not so, there would still nonhomologous chromosome, which for be no reason to believe that the second convenience was labelled "W," has no ef­ generation would also be all-female since fect on the dorsal-fin spot. In the unisex­ segregation should result in some gametes ual progeny expression of fin pigment de­ with the normal haploid chromosome num­ pends on the paternal chromosome, "Y," ber. Inasmuch as the unisexual strains of which carries the gene-S for fin-spot or its Poeciliopsis have failed to give rise to a allele-s for no spot. Either of these genes single male in five generations, it seems un­ combined with the "w" chromosome is al­ likely that polyploidy or aneuploidy is in­ lowed full expression, but when paired with volved. Furthermore, the cytological evi­ each other in the bisexual strains the op­ dence is completely negative. The haploid posing effect results in dilution. chromosome number of "C" and "F" When an "F," (W-YS) female is crossed males, which serve as mates to both the with a "C" (Y'Y") male, clear-fin, W-ys, unisexual and the bisexual strains, is in­ hybrid females are produced, but the ysys variably 24. Embryos obtained from "Cx" weak-spot individuals are not represented and "Fx " females have the diploid number among the progeny as either males or fe­ 48, which is the same as that of embryos males. The same phenomenon occurs in the from the bisexual strain, "c." Polyploidy "Cx" (W-YS) X "F" (ysYS) cross. Only can, therefore, be excluded from the list of WrY", strong-spot, females are obtained REPRODUCTIVE MECHANISM OF POECILIOPSIS 321

and the ysy. genotype is again lost. In and it is suggested above that the sex­ these and other crosses it was noted that, determining potential differs among the regardless of whether the gametes provided various species of Poeciliopsis. In the group by the male carried the S-gene or its s­ with which we are here concerned sex-de­ allele, transmission occurred, yet in the termining factors in "F" males and females unisexual females, only the W-type gamete are stronger than those of "C." Although was transmitted. If the fate of the paternal these sex-determiners are balanced in in­ chromosome can be determined, unisexual­ traspecific matings, the balance is not ity in "Cx" and "F-" will be explained. maintained in hybrid crosses. Consequently, The possibility that all YY males pro­ when "C" females are mated to "F" males duced by unisexual females die prior to the stronger male-producing genes of "F" maturity was rejected on the basis of the combine with the weaker female-producing low prenatal and postnatal mortality; hence genes of "C" to result in a sex-ratio of loss of the V-chromosome takes place be­ 1 <;' : 3 ~ ~ (table 3). Excessive mortality fore fertilization. An almost complete lack was not a feature of this experiment. In of degenerating ova within "C," and "Fx" the reciprocal cross ("F" x "C") stronger ovaries, excluded the possibility that the female-determining factors combined with V-bearing gametes died. It must, therefore, the weaker male determiners, and the ratio be concluded that only the W-type egg is was shifted in the opposite direction. Mat­ produced and that the V-chromosome is uration division within the F 1 hybrids re­ lost during oogenesis. In each generation, sults in a mixing of the strong and weak then, only the W-bearing ovum combines elements, promoting an intermediate con­ with the V-bearing sperm to produce a WY dition. The gametes of the "C" X "F" hy­ female. brid males, therefore, are not as strongly Although transmission data from the hy­ male-determining as those of "F," yet their brid crosses with the unisexual females in­ potential exceeds that of "C." When such dicates a WY-chromosome system, this a male is backcrossed to a "C" female, designation is unjustifiable in terms of the male offspring still slightly outnumber the bisexual hybridization data, in which the females, but not significantly so, since dif­ marked chromosomes segregated randomly ferential mortality tends to equalize the and behaved as autosomes. In Hybrid ratio. In the "C" X "F" hybrid female, Mating XII, when an F 1 hybrid ("C" X the same mixing of sex-determiners takes "F") female was backcrossed to the clear­ place, to result in ova that tend toward fin "C" male, all of the female progeny maleness. Combining these gametes with should have been W·y· and, hence, clear­ strong male-determining gametes of "F" fin, whereas all of the males, YSY·, should results in nearly all male progeny: only 4 have had the dilute spot (table 4). Since out of 27 were females (table 4). both spot-fin and clear-fin males and spot­ Hybrid crosses involving members of the fin and clear-fin females were produced in genus Poeciliopsis, not considered in this a 1: 1:1:1 ratio, it is questionable whether study, suggest that a similar situation exists the marked chromosomes are by themselves for other closely related species. In some responsible for sex-determination. Conse­ the difference in potential of sex factors is quently, there is no reason to consider the greater than in "C" X "F," whereas in chromosomes with the spot-fin locus as sex­ others it is less defined. chromosomes. In fact there is evidence that This relationship of strong and weak sex­ sex-determination in "C" and "F" is in determiners bears some resemblance to the part, if not completely, dependent on genes genetic mechanism in the gypsy moth, dispersed among the autosomes. It is Lymantria dispar. In this species the male­ shown above that progeny from certain hy­ and female-determining factors are in brid crosses have unbalanced sex-ratios, equilibrium in pure-bred races, but in in- 322 R. JACK SCHULTZ

terracial crosses involving "weak" females responsible for the modified characters and "strong" males the F 1 generation is noted above but also have female-deter­ completely male in appearance (Gold­ mining factors too strong to be counter­ schmidt, 1931, 1932, 1934). Half of these, balanced by the male determiners on the however, proved to be intersexes. In Ly­ corresponding chromosomes. In hybrid mantria, females are heterogametic, so that combinations between closely related spe­ the X- (or Z-) chromosome is male-deter­ cies of Poeciliopsis, such as "C" and "F," mining and variable in potency, but the the high degree of genic compatibility per­ female-determining factors evidently are mits normal pairing and segregation of cytoplasmic. chromosomes during meiosis. The "new" Although interracial crosses of Lymantria chromosomes contained by -c,: and "Fx," dispar and interspecific crosses of Poeciliop­ however, are too dissimilar to pair with the sis are comparable on the basis of varying male chromosomes. Consequently, paternal degrees of sex-determining potential, here characters expressed by each generation fail the similarity ends. In Lymantria the ratio to be incorporated into the F 1 ova and the of males to females is not really shifted but two unisexual strains remain perpetual hy­ intersexes take the place of what normally brids. would constitute one of the sexes. Further­ In vertebrate animals selective matura­ more, in F 1 hybrids of Poeciliopsis the sex­ tion has not been previously verified either determining factors are redistributed in genetically or cytologically, but details of such a way that equilibrium is eventually the process have been worked out for a restored, as would be true with polygenic number of insects in the families Mar­ inheritance. garoidae, Lacanoidae, Sciaridae, Cecidomy­ If in the bisexual strains, sex is deter­ idae, and Pediculidae (White, 1954). The mined totally or in part by a number of means by which chromatin is eliminated genes dispersed among the autosomes, the varies. In general, oogenesis is normal but hypothesis that femaleness in "C,' and in the spermatogonial divisions certain "F-" is attributable to a single chromosome chromosomes or the entire genome of one referred to as "W" is open to question. It sex or the other is discarded in a bud of is a more reasonable assumption that sex cytoplasm resembling a polar body. Ex­ here is also controlled by a number of pulsion of certain chromosomes may even genes. Since chromosomes carrying these occur during early cleavage stages. genes do not undergo random assortment With the information now at hand the during oogenesis, but behave as a unit, in­ two examples of unisexuality in the family terpretation of the data is the same as Poeciliidae can be compared. Basically though a single nonhomologous sex-chro­ the two mechanisms are different. In M ol­ mosome were involved. Genes carried by lienesia formosa, fertilization does not in­ this unit are responsible for such differ­ clude gametic fusion; hence the phenotype ences between the bisexual and unisexual of the offspring is totally dependent on the strains as: dentition, ova size, size of young genotype of the mother; the all-female prog­ at birth, maximum size of adults, age and eny of Poeciliopsis, however, express char­ size at maturity, etc. acters of both parents. In spite of this It is probable that in the recent past a basic difference there are ways in which species of Poeciliopsis, not closely related the two systems are alike. In M. formosa to "C" or "F," through hybridization the egg nucleus contains only maternal altered the genomes of the strains now chromatin; the same is true in mature ova known as "Cx" and "F;." Some of the of Poeciliopsis strains "Cx" and "F-" since foreign chromosomes were retained by the the paternal chromatin (for the most part unisexual strains in subsequent backcrosses. at least, if not in total) is lost during These chromosomes not only carry genes meiosis. In individuals with either type of REPRODUCTIVE MECHANISM OF POECILIOPSIS 323

mechanism, the unisexual condition pre­ vided by the Department of Fisheries and vents mutations from becoming part of a the Museum of Zoology. Grants from the common gene pool. Any change is restricted Horace H. Rackham School of Graduate to the clone in which it originates. The Studies made available the live stocks used obligatory heterozygous nature of the uni­ in this study and part of the field expenses sexual Poeciliopsis, however, would seem for the 1959 expedition to Mexico. A Na­ to hold certain advantages over the gyno­ tional Science Foundation grant (NSF genetic type of M ollienesia in that each G-4858) to Dr. Miller provided the author generation of "C," or "Fx" utilizes the with employment as research assistant and genetic variability originating within the later as research associate in the biosys­ bisexual strains. Since these paternal con­ tematic studies of Poeciliopsis, part of tributions are not incorporated in the "Cx" which form the basis of this report. and "Fx" genetic systems but are merely "on loan," the unisexual strains are, in SUMMARY effect, as isolated from their bisexual 1. The viviparous poeciliid fishes of the counterparts as is M ollienesia formosa genus Poeciliopsis comprise approximately from its males of M. latipinna and M. 16 species that chiefly inhabit the Pacific sphenops. It is possible that the selective slope of Mexico but occur as far south as segregation of Poeciliopsis represents a Colombia. Two allopatric populations, here transition stage to gynogenesis and thence designated "C" and "F," each have, in to parthenogenesis, but it is equally pos­ addition to normal males and females, an sible that these two unisexual mechanisms aberrant, unisexual strain. These strains, are merely leftovers from the hybridization "Cx" and "Fx ," in five laboratory genera­ of animals with inefficient isolating mech­ tions, gave birth only to female progeny. anisms. In attempting to determine the mechanism If the latter situation obtains, it is dif­ by which all-female broods are produced, ficult to imagine that such a dead-end the reproductive biology and genetics of phenomenon has persisted or could long both "C-Cx" and "F-Fx" were investigated. persist in nature. It, therefore, might still 2. In Poeciliopsis fertilization is internal. be possible to reconstruct the unisexual The embryos develop within the ovarian strains. Investigations currently being con­ follicles by a process of superfetation, ducted toward this end have been promis­ whereby two or more stages of embryos ing; and although a unisexual type has not occur in the ovary simultaneously. Births been produced, male progeny were obtained occur at intervals of 7 to 14 days, with as when "Cx" was mated to males of a dif­ many as 34 offspring in a single brood; ferent species group. It is believed that however, broods of fewer than 12 are more with continued experimentation in this common. area the "Cx" and "Fx" strains will ulti­ 3. Examination of ovaries and brood mately be synthesized in the laboratory. data collected over the four-year period of this investigation eliminated the possibility ACKNOWLEDGMENTS that the lack of male offspring from "ex" The author wishes to express his deepest and "Fx" females is caused by a lethal gratitude to Dr. Robert R. Miller for sug­ system operative on either the gametes, em­ gesting the problem and for giving freely bryos, or juveniles. Genetic data and ex­ of his time and helpful suggestions through­ amination of sectioned gonads provided out this study. The splendid cooperation evidence against sex-inversion. Inheritance of Dr. and Mrs. Carl L. Hubbs in supervis­ of paternal characters by the unisexual ing the initial handling of live shipments type progeny invalidates parthenogenesis from Mexico made the project feasible. or gynogenesis, and polyploidy was rejected Aquaria and other equipment were pro- on the basis of chromosome counts. 324 R. JACK SCHULTZ

4. In hybrid crosses between "C" and Poeciliopsis. The reproductive mechanisms "F" there is a high degree of genetic com­ of these two exclusively female-producing patibility. Meiotic divisions of the F 1 prog­ groups, however, are different. In M. eny are apparently normal, with random formosa sperm provided by males of two segregation of chromosomes. Reciprocal other species serves only to stimulate the differences in the sex-ratio suggest, how­ ova into development, there being no ever, that the potency of the sex-determin­ gametic fusion, whereas in Poeciliopsis ing factors of one species exceeds that of fertilization is normal and paternal char­ the other. There is no genetic or cytological acters are expressed by the offspring. evidence that the bisexual strains have sex chromosomes. It appears that, to a large LITERATURE CITED extent if not completely, sex-determination AIDA, T. 1936. Sex reversal in Aplocheilus la­ depends on autosomal genes. tipes and a new explanation of sex differentia­ 5. Inheritance of a marker gene (dorsal­ tion. Genetics, 21: 136-153. fin spot) in both the bisexual and the uni­ BOOK, J. A. 1940. Triploidy in Triton taeniatus Laur. Heridatas, 26: 107-114. sexual strains bears significantly on the CAVALCANTI, A. G. L., D. N. FALCAO, AND L. E. mechanism involved in the production of CASTRO. 1957. "Sex-ratio" in Drosophila pro­ all-female broods. When crosses involved saltens-a character due to interaction between the bisexual-type female, transmission data nuclear genes and cytoplasmic factors. Amer. Nat., 90: 327-329. indicated autosomal inheritance with in­ ESSENBERG, J. M. 1923. Sex-differentiation in complete dominance. When the unisexual the viviparous teleost Xiphophorus helleri strains were used in the same crosses, how­ Heeke!. Bio!. Bull., 45: 46-97. ever, the character appeared to be sex­ FRIEsz, E. 1933. Untersuchungen tiber die Gesch­ lechtsumkehr bei Xiphophorus helleri Heeke!. linked with the female heterogametic, WY, Roux Arch. Entw. Mech. Organ., 129: 255­ and the fin-spot (S) dominant to its allele, 355. no-spot (s). Expression of the character GOLDSCHMIDT, R. 1931. Analysis of intersex­ in each "all-female" generation was, there­ uality in the gypsy moth. Quart. Rev. Bio!., fore, totally dependent upon the S locus of 6: 125-142. --. 1932. Untersuchungen zur Genetik der the paternal chromosome. All evidence in­ geogrophischen Variation. III. Abschliessendes dicated that the male chromosome is not tiber die Geschlechtsrassen von Lymantria dis­ transmitted through the F 1 ova but is lost par. Roux Arch. Entw. Mech. Organ., 126: during oogenesis. 277-324. 6. It is postulated that the nonhomolo­ 1934. Lymantria. Bibliogr. Genet. II: 1-186. gous or "W" chromosome of "C," and "F-" GORDON, M. 1952. Sex determination in Xi­ is in reality not a sex-chromosome, but phophorus (Platypoecilus) maculatus. III. Dif­ rather one of several chromosomes intro­ ferentiation of gonads in platyfish from broods duced into the genomes of the unisex­ having a sex ratio of three females to one ual females through hybridization. These male. Zoologies, 37: 91-100. chromosomes, which are inherited as a unit, --. 1957. Physiological genetics of fishes. In Physiology of Fishes. Vo!. 2. Acad. Press Inc., have female-determining factors too strong New York. pp. 431-501. to be counterbalanced by the paternal fac­ HUBBS, C. L. 1955. Hybridization between fish tors provided by "C" or "F" males. They species in nature. Syst. Zoo!., 4: 1-20. furthermore are genetically too dissimilar -- AND L. C. HUBBS. 1932. Apparent par­ thenogenesis in nature, in a form of fish of to pair with the paternal chromosomes dur­ hybrid origin. Science, 76: 628-630. ing oogenesis. Asynapsis probably results -- AND --. 1946. Breeding experiments with in the expulsion of the paternal chromo­ the invariable female, strictly matroclinous somes in a polar body. fish, M ollienisia formosa. Rec. Gen. Soc. Am., 7. The only other naturally occurring No. 14: 48. KOSSWIG, C. 1954. Zum Geschlechtsbestimungs­ unisexual vertebrate is Mollienesia formosa, problem bei den Zahnkarpfen. Rev. Fac. Sci. a cyprinodont fish in the same family as Univ. Istanbul, 19: 187-190. REPRODUCTIVE MECHANISM OF POECILlOPSIS 325

KOSIN, 1. L., 1. SATO, AND H. NAGRA. 1958. Stud­ OZTAN, N. 1954. Cytological investigation of ies of parthenogenesis in the domestic turkey the sexual differentiation in the Anatolian and chicken. Eleventh World's Poultry Con­ cyprinodontids. Rev. Fac. Sci. Univ. Istanbul, gress, Mexico City, Mexico. 19: 245-280. KRUMHOLZ, L. A. 1948. Reproduction in the POOLE, H. K. 1959. The mitotic chromosomes western mosquitofish, Gambusia affinis affinis of parthenogenetic and normal turkeys. Jour. (Baird and Girard), and its use in mosquito Hered., 50: 151-154. STOLK, A. 1958. Pathological parthenogenesis control. Ecol. Monogr., 18: 1-43. in viviparous toothcarps. Nature, 181: 1,660. KUPKA, E. 1948. Chromosomale Verschiedenheiten STURTEVANT, A. H. 1920. Genetic studies on bei schweitzerischen Coregonen (Felchen). Rev. Drosophila simulans. 1. Introduction. Hybrids Suisse Zool., 55: 285-293. with Drosophila melanogasier, Genetics, 5: LIEDER, U. 1955. Mannschenmangel und natiir­ 488-500. liche Parthenogenese bei der Silberkarausche SUOMALAINEN, E. 1950. Parthenogenesis in ani­ Carassius auraius gibelio. Naturwissenschaften, mals. Advanc. Genet., 3: 193-253. 42: 590. SVARDSON, G. 1943. Studien tiber den Zusarnmen­ MAKINO, S. 1951. An atlas of the chromosome hang zwischen Geschlechtsreife und Wachstum numbers in animals. Iowa State College Press, bei Lebistes. Meddel. Rept. Swedish State pp. 1-290. Inst. Freshw. Fish. Res. Drottningholm, 21: MILLER, R. R. 1960. Four new species of vivip­ 1-48. arous fishes, genus Poeciliopsis, from north­ 1945. Chromosome studies in Salmonidae. western Mexico. Occ. Papers Mus. Zool. Univ. Ibid., 23: 1-151. Mich., No. 619: 1-11. TURNER, C. L. 1937. Reproductive cycles and -- AND R. J. SCHULTZ. 1959. All-female superfetation in poeciliid fishes. Biol. Bull., 72: 145-164. strains of the teleost fishes of the genus WHITE, M. J. D. 1954. Animal cytology and Poeciliopsis. Science, 130: 1,656-1,657. evolution. Cambridge Univ. Press. 454 pp. MULLER, H. J. 1947. The production of muta­ WICKBOM, T. 1943. Cytological studies in the tions. Jour. Hered., 28: 259-270. family Cyprinodontidae. Hereditas, 29: 1-24. OLSEN, M. W., AND S. J. MARSDEN. 1954. Nat­ WINGE, 0., AND E. DITLEVSEN. 1947. Color in­ ural parthenogenesis in turkey eggs. Science, heritance and sex determination in Lebistes. 120: 545-546. Heredity, I: 65-83.