Copyright 0 1991 by the Genetics Society of America

Molecular Evidence for Multiple Origins of Hybridogenetic Clones (: )

Joseph M. Quattro,* John C. Aviset and Robert C. Vrijenhoek* *Center for Theoretical and Applied Genetics, Cook College, Rutgers University, New Brunswick, New Jersey 08903-0231, and ‘Department $Genetics, University of Georgia, Athens, Georgia 30602 Manuscript received September 6, 1990 Accepted for publication October 1 1, 1990

ABSTRACT Hybrid matings between the sexual species and produced a series of diploid all-female lineages of P. monacha-lucida that inhabit the Rio Fuerte of northwestern Mexico. Restriction site analyses of mitochondrial DNA (mtDNA) clearly revealed that P. monacha was the maternal ancestor of these hybrids. The high level of mtDNA diversity in P. monacha was mirrored by similarly high levels in P. monacha-lucida; thus hybridizations giving rise to unisexual lineages have occurred many times. However, mtDNA variability among P. monacha-lucida lineages revealed a geographical component. Apparently the opportunity for the establishment of unisexual lineages varies amongtributaries of the Rio Fuerte. We hypothesize thata dynamic complex of sexual and clonal appear to participate in a feedback process that maintains genetic diversity in both the sexual and asexual components.

ENETIC studies have revealed that clonally re- versity among the hemiclonal monacha genomes of P. G producing, all-female “species” of vertebrates monacha-lucida fromthe Rio Fuerte (VRIJENHOEK, are the products of hybridization between congeneric ANGUSand SCHULTZ1977). For brevity, we refer to sexual species (SCHULTZ 1969;SITES et al. 1990; VRI- distinct hemiclones defined by electrophoresis as “E- JENHOEK et al. 1989).A new unisexual (all-female) types” and identify them by the Roman numerals lineage can be established if genetic differences be- following the biotype designation (e.g., ML/VIII). All tween the hybridizing entities are sufficient to disrupt the allozymic diversity distinguishing Rio Fuerte hem- recombinant processes during meiosis, without se- iclones also exists in P. monacha from this river. How- verely limiting viability, fecundity, and other charac- ever, endemic allozymes, found onlyin discrete P. teristics affecting fitness of the hybrids (MORITZ et al. monacha populations, often mark co-occurring hemi- 1989; VRIJENHOEK1989). With such a narrowwindow clones, suggesting that these ML lineages arose during of opportunity, it is not surprising thatthe origins and independent hybridization events in partially isolated establishment of new unisexual vertebrates have been tributaries. ecologically, geographically, and phylogenetically con- Subsequent tissue grafting studies of Rio Fuerte P. strained (DENSMOREet al. 1989; DENSMORE,WRIGHT monacha-lucida identified 18 histocompatibility clones and BROWN1989; MORITZ, WRIGHTand BROWN (H-types) hidden within the eight E-types (ANGUSand 1989; VRIJENHOEK1989; VYASet al. 1990). SCHULTZ 1979). What proportion of this additional An exception might be unisexual fishes of the genus variation was due to multiple hybridization events vs. Poeciliopsis which have arisen independently in several postformationalmutations within E-type lineages river systems of Sonora and Sinaloa, Mexico. Hybrid- could not be determined. However,evidence for post- ization between P. monacha and several sexual species formational divergence was found in expanded elec- in this genus has produced allodiploid and allotriploid trophoretic surveys that included more loci and uni- unisexual biotypes (SCHULTZ1977). In this study, we sexual fish from additional river systems (SPINELLA focus on the allodiploid biotype P. monacha-lucida and VRIJENHOEK1982; VRIJENHOEK1984). Further- (designated ML), which reproduces by hybridogene- more, genetic studies of several ML strains revealed sis, a hemiclonal mechanism that transmits only the that hemiclonal M genomes carried recessive lethals monacha (M) chromosome set to developing ova and subvitals that probably arose post-formationally (SCHULTZ1969). The lucida (L)chromosomes are (LESLIEand VRIJENHOEK1978, 1980). Apparently, excluded during a premeioticcell division, precluding some hybridogenetic strainshave existed long enough synapsis and crossing over (CIMINO1972). Sperm from to accumulate a substantial load of mutations, but the P. lucida males is required for fertilization and resto- data could not be used to infer relative evolutionary ration of diploidy. ages of hemiclonal lineages. Protein electrophoresis uncovered considerable di- Restriction site analysis of mitochondrial DNA has

C;r~~ct~

FIGURE 1.-The Rios Fuerte, Mayo, and Sinaloa of northern Mexico and associated tri- butaries. Sampling localities (arrows) are indi- cated by site abbreviations: CA, the Arroyo Aguajita de El Cajon;JA, the Arroyo deJaguari and its feeder streams (Arroyos de 10s Platanos and Nachapulon); SP, Rio San Pedro; RG, Ran- cho Guamichil; and GO, Arroyo Seco ca. Co- ronado. Refer to text for more detailed site descriptions.

become a useful tool forstudying the origins and dogens from the Rio Fuerte. By integrating informa- evolutionary ages of unisexual vertebrates (MORITZ, tion from allozymes, tissue grafting, and mtDNA we DOWLINCand BROWN1987). Because it is maternally address the following questions: (1) is P. monacha the inherited, mtDNA contains in its sequence the evolu- maternal progenitor to all of P. monacha-lucida lin- tionary history of an all-female lineage. Additionally, eages; (2) are certain P. monacha populations more the molecule’s rapid rate of sequence evolution facil- prone to producingunisexual lineages; and (3) is there itates microevolutionary studies below the species evidence for mutationally divergent, potentially long- level. A preliminary analysis of mtDNA restriction lived unisexual lineages. Finally, we comment on the fragment polymorphism inPoeciliopsis (AVISEand recent decline of P. monacha populations in three river VRIJENHOEK1987), revealed three points relevant to systems encompassing its complete range, and hope- the present study. First, P. monacha and P. lucida have fully raise concern regardingconservation of the com- widely divergent mtDNAs, allowing unequivocal iden- mon ancestor to all unisexual lineages in the genus Poeciliopsis. tification of P. monacha as the maternal progenitor of two naturally occurring P. monacha-lucida strains MATERIALS ANDMETHODS (ML/VII and ML/VIII). Second, these natural hybri- dogens, plus five laboratory strains, clearly demon- Poeciliopsis monacha was originally described as having a strated that mtDNA is maternally inherited. Finally, relictual distribution in headwater tributaries of the Rios Fuerte, Mayo, and Sinaloa of northwestern Mexico (MILLER this study revealed no evidence that the naturalstrains 1960; VRIJENHOEK1979b). Prior to 1978, P. monacha were had diverged mutationally from one another or from found at a single locality in the Rio Sinaloa (Arroyo Seco, the coexisting P. monacha population. ca. Rancho Coronado, Sinaloa). It might be extinct in this In the presentstudy, we conducteda survey of river, since all subsequent attempts to capture this species were unsuccessful. No laboratory strains from this popula- mtDNA diversity in P. monacha, the common ancestor tion were available for study. Prior to 1989, P. monacha of all unisexual forms of Poeciliopsis, and compared were found at two sites in the Guamlichil tributary of Rio its variation to that found in P. monacha-lucida hybri- Mayo (ca. El Tabelo and ca. Rancho Guamlichil, Sonora). mtDNA Variation in Poeciliopsis 393

TABLE 1 according to the manufacturers' specifications with a battery of 17 restriction endonucleases having four (MsPI), five Sources of P. monacha and P. monacha-lucida and their (AuaI,AuaII, HincII),and six (BamHI, BclI, BglI, BglII, electrophoretic andmtDNA haplotypes BstEII, EcoRI, HindIII, NdeI, PstI, PuuII, SpeI, StuI, Xbal) base recognition sequences. DNA fragments were end-la- Strain beled with ["S]dNTP(s) and separated in 0.8-1.8% agarose Locality and biotype Wild Laboratory E-type" Mt-typeb gels. After electrophoresis, gels were dried under vacuum onto a filter paper backing and exposed to X-ray film at Rio Fuerte room temperature for 24-72 hr. A. San Pedro Each restriction digest profile for a specific endonuclease P. monacha 1 2 was given an arbitrary letter code. Composite scores, rep- P. monacha 1 8 resenting a letter code for each of 17 gel-fragment profiles, P. monacha 1 11 wereassigned to each individual. Sinceall gel-fragment P. monacha 1 14 patterns could be related by specific site gains or losses, a P. monacha-lucida 2 I 11 matrix of the minimum number of site changes necessary P. monacha-lucida 1 I1 14 to interrelate composite haplotypes was constructed. A min- P. monacha-lucida 1 IV 7 imum length network linking observed haplotypes was con- B. Guiricoba structed by inputting the mutation matrix into the minimum P. monacha 6 7 spanning tree algorithm of the NTSYS statistical package P. monacha-lucida 1 I 11 (ROHLF,KISHPAUCH and KIRK 1974). The resulting network P. monacha-lucida 1 I 12 was compared to the consensus of the most parsimonious P. monacha-lucida 1 I1 13 trees obtained with the CONSENSE and MIX computer 4 I1 14 P. monacha-lucida programs (FELSENSTEIN1990). We evaluated the fit of the P. monacha-lucida 1 Ill 9 resulting network to the input matrix by the Mantel matrix P. monacha-lucida 1 IV 9 comparison test (SMOUSE,LONG and SOKAL1986). The P. monacha-lucida 1 xv 10 significance of the resulting matrix correlation coefficient C.Jaguari was evaluated by comparison to calculated test statisticsafter P. monacha 3 3 1 1000 random permutations of the input matrix. Nucleon P. monacha 1 3 diversity, h, a measure of mtDNA haplotype variability, was 7 P. monacha 4 calculated according to the formula of NEI and TAJIMA P. monacha-lucida 2 1 VI I 1 (1981). P. monacha-lucida 2 1 VI11 1 Rio Mayo P. monacha 1 5 RESULTS P. monacha 2 6 Refer to Table 3. Ten restriction endonucleases produced polymor- * Refer to Table 2. phic fragment profiles, revealing 13 mtDNA haplo- types among the 44 specimens.Divergence among Recent attempts to capture this species at these sites have these 13 haplotypes couldbe attributed to changes at been unsuccessful. Fortunately three independent isofemale 16 restriction sites (Table 2). The minimum length lineages from Rancho Guamkhil wereavailable for the present analysis. Headwater tributaries of the Rio Fuerte network interrelating these haplotypesis illustrated in still maintain P monacha populations. We examined 18 wild- Figure 2. The total number of required mutational caught females and three isofemale lines from three head- steps along branches of the network connecting hap- water tributaries in this river (Figure 1 and Table 1): the lotypes (320) is slightly largerthan the minimum Arroyo Aguajita de El Caj6n which connects with the Ar- numberinferred from the composite haplotypes royo de Guiricoba (ca. El Cajon, Sonora); the Arroyo de Jaguari and its feeder streams (Arroyos de 10s Platanos and (3 16), indicating only slight distortion in the network Nachapulon; ca. Agua Caliente, Sonora); and the Rio San (normalized Mantel Z statistic = 0.99, p(random Z 2 Pedro (approximately 22 km upstream from the San Pedro, observed Z) = 0.001). The distortion is attributable Sonora). mtDNA patterns in the sexual species were com- to two instances of homoplasy: parallel losses of the pared with those of three isofemale strains and 17 wild AuaI" site in haplotypes 12 and 14, and parallel gains females of P. monacha-lucida from the same localities. Morphological phenotypes were used to distinguish P. of the BstEIF sitein haplotypes 1 and 6 (Table 2, monacha from unisexualbiotypes in the field (SCHULTZ Figure 2). 1969). Wild-caught specimens werereturned to the culture ConsiderablemtDNA diversity was found in P. facility at Rutgers University, where protein electrophoresis monacha (Table 1). Seven haplotypes were observed was used to distinguish among diploid and triploid biotypes. among 18wild females and threeisofemale lines from Electrophoretic techniques, diagnostic genotypes, and gene dosage patterns for 25 loci were described in earlier publi- the Rio Fuerte (nucleon diversity: h = 0.72 f 0.06), cations (VRIJENHOEK 1975; VRIJENHOEK,ANGUS and and two unique haplotypes were observed among the SCHULTZ1977, 1978). For the present study, we added a three Rio Mayo lines.The eight Rio Fuerte haplotypes 26th locus, Pep-lgg (leucyl-glycyl-glycine peptidase EC were not evenly distributed among the threelocalities. 3.4.1 1) (for methods see MULVEYand VRIJENHOEK1981). The mostcommon variant (haplotype 7) was fixed Mitochondrial DNA was isolated from fresh tissues fol- lowing the methods of LANSMANet al. (1 983). Because of (100%) in theGuiricoba sample of P. monacha; it low yields of mtDNA, we used only fresh female specimens occurred at lower frequency in the Jaguari sample 225 mm standard length. Purified mtDNAs were digested (36%), and was absent from the San Pedro sample. 394 J. M. Quattro, J. C. Avise and R. C. Vrijenjhoek TABLE 2 Restriction site polymorphism in Poeciliopsis mtDNA, expressed as binary characters (0, absent, 1, present)

Restriction site Aval I BsfEll Hind111 Aval - EroRI Hincll Mspl Pstl XbaI Spcl Hxplotvpe a bcdefg h I jkl n n P '1 1 1 1 100011 1 0 111 1 0 0 0 2 1 100000 1 0 111 1 0 0 0 3 1 100000 0 0 111 1 0 0 0 5 1 100000 0 1 111 1 0 0 0 6 1 100001 0 1 111 0 1 0 0 7 1 100000 1 0 111 1 0 0 1 8 1 100000 1 0 011 1 0 0 1 9 1 110000 1 0 110 1 0 1 0 10 1000000 1 0 111 1 0 0 1 11 1000000 1 0 101 1 0 0 1 12 0 000000 1 0 101 1 0 0 1 13 1 101100 1 0 111 1 0 0 0 14 0 101100 1 0 111 1 0 0 0 A nlinimunl length network based on these data is shown in Figure 2.

FIGURE2."An unrooted minimum length net- work summarizing relationships among observed mtDNA haplotypes in sexual Poeciliopsis monacha g" 0 (open circles) and unisexual P. monacha-lucida (stip- Ill pledcircles). For P. monacha, circles are drawn in Ill proportion to the observed number of individualswith the respective haplotype. P. monacha-lucida circles are proportional to the numberof different electromorph -- b clones observed with the respective haplotype. Slashes represent the number of mutations necessary to inter- relate haplotypes, and letter codes refer to specific restriction sites in Table 2. The boxes and dashed -- e -- k lines are alternative pathways in the network, and are explained in the text. Haplotype 4, which occurs only in triploid unisexual Poeciliopsis, is not shown, and will be described elsewhere (J. M. QUATTRO,unpub- @--@ lished results). """ ? I I I I "QI I a+? *a a -- ?+a I I I I 612 "_ ? ""2 @-@ Haplotype 1 characterized 55% of the Jaguari sample, related to Rio Fuerte P. monacha haplotypes (Figure but was notfound elsewhere. Remarkably, four 2). Rio Mayo P. monacha haplotypes are distinguished unique haplotypes were found in a sample of four P. by a unique HincII' fragmentprofile. monacha females from the Rio San Pedro (haplotypes Addition of the Pep-Zgg locus to the earlier suite of 2, 8, 1 1and 14). 25 gene loci (VRIJENHOEK1984; VRIJENHOEK,ANGUS Sequence divergence between P. monacha and P. and SCHULTZ1978) identified a new E-type (ML/XV) lucida mtDNA is extensive (p = 11.5%) (AVISEand from the Arroyo de Guiricoba. Composite allozyme VRIJENHOEK1987). Fragment profiles for all restric- genotypes of the seven E-types observed in this study tion enzymes examined in this and the previous study are listed in Table 3. All the allozymes discriminating diagnostically differentiate between the two species. among E-types are encoded by common alleles in P. Thus we could unequivocally assign maternal paren- monacha (6VRIJENHOEK 1979b). The seven E-types tage to the unisexual samples examined in this study. exhibited eight distinct mtDNA haplotypes, or Mt- All P. monacha-lucida females contained monacha type types (Table 1). Some E-types could be decomposed mtDNAs that were identical with or mostclosely into distinct Mt-types, for example: ML/I was marked mtDNA Variation in Poeciliopsis 395 TABLE 3 (R. J. SCHULTZ,personal communication). As found Allozyme haplotypes (E-types) of the monacha genome in P. previously with a small sample of hybridogenetic lin- monacha-lucida eages (AVISEand VRIJENHOEK1987), paternalleakage does not appear to occur in P.monacha-lucida. We Diagnostic loci found no evidence for heteroplasmy of monacha and E-type Aat-3 Est-5 Ldh-1 Ck-A" Pep-lgg Pgd lucida mtDNAs in any of the unisexual specimens. 1 bf b a a a Compared with other unisexual vertebrates, P. mon- 11 b C b a a a acha-lucida is exceptional, because its mtDNA diver- Ill a C b a a a sity essentially matches that of its maternal ancestor, IV b d b a b a P. monacha.In contrast, parthenogeneticlizards of the VI1 b e a b a a VI11 b d a a a C genera Cnemidophorus and Heteronotia arefar less XV bf b a b a diverse than their closest sexual relatives (DENSMORE et al. 1989; DENSMORE,WRIGHT and BROWN1989; a Previouslycalled Mp-3 because it encodes a soluble muscle protein (VRIJENHOEK,ANGUS and SCHULTZ1978). MORITZ, WRIGHTand BROWN1989). In general, par- thenogenetic lizards appear to have arisen as a result by Mt-types 1 1and 12; ML/II by Mt-types 13 and 14; of a limited number of hybridization events. Similarly, and ML/IV by Mt-types 7 and 9.Conversely, some E- the gynogenetic fish Menidia clarkhubbsi exhibits low types carried identical mtDNA haplotypes (e.g., ML/ mtDNA diversity relative to its sexual relative P. pen- VI1 and ML/VIII carriedMt-type 1; ML/III andML/ insulae (ECHELLEet al. 1989), butmultiple events were IV carried Mt-type 9). To facilitate identification of involved in producingdifferent clonal lineages. these hemiclones, we append the Mt-type to the E- MtDNA haplotypes in a limited sample of the gyno- type designations (e.g., ML/IV.7 US. ML/IV.9). genetic fish Phoxinus eos-neogaeus also indicate multi- The numbers of sexual and hybridogenetic fish in ple hybrid origins (GODDARD, DAWLEYand DOWLING the sample were similar (24 vs. 20, respectively). Four 1989). Whether multiple origins are characteristic of of the mtDNA variants found in P. monacha-lucida the hybridogenetic waterfrog, Rana esculenta, is un- also were observed in the sample of P. monacha (hap- clear, because the patterns of mating and mode of lotypes 1, 7, 1 1 and 14), and four others (9, 10, 12 hemiclonal reproduction are more complex than in and 13) were not. Mt-type 9 differed by three unique Poeciliopsis (SPOLSKYand UZZELL 1986). mutations from the nearest P. monacha haplotype (2), The present data clearly corroborate earlier allo- but it was found in two distinct E-types sampled from zyme studies implicating multiple hybridization events the Guiricoba locality, ML/III.9 and ML/IV.9. The E-typesof these hemiclones differ with respect to as the primary source of hemiclonal diversity in P. common P. monacha allozymes at three polymorphic monacha-lucida (VRIJENHOEK1984; VRIJENHOEK,AN- loci (Table 3), presumably as a result of independent GUS and SCHULTZ1978). However, useof mtDNA origins. Thus, mtDNA haplotype 9 must have existed restriction site analyses alone might underestimate the in P. monacha prior to the origin of these hemiclonal opportunity for multiple origins in unisexual verte- lineages. Mt-types 10, 12, and 13of ML/XV.10, ML/ brates. For example, a single hybridization of P. mon- I. 12 andML/II. 13) each differedby a single mutation acha X P. lucida would give rise to sibling P. monacha- from the nearestP. monacha haplotypes in the network lucida strainsthat differ in nuclear genotypes, but (Figure 2). These orphan haplotypes probably exist, share identical mtDNAs (AVISE and VRIJENHOEK or existed, in P.monacha; however, we cannot rule 1987; WETHERINGTONet al. 1989). Similarly, inde- out the possibility that ML/I.12 and ML/II.13 rep- pendent hybridizations involving separate P. monacha resent one-step mutations within clonal lineages (rep- females from the same maternal lineage also give rise resented as rectangles in Figure 2). to such pairs. Despite these conservative features of mtDNA, we were able to discriminate between clones marked by DISCUSSION identical allozyme genotypes. For example, MLIIV.7 Given the large amount of sequencedivergence from the Rio San Pedro and ML/IV.9 from the Guir- between P. monacha and P.lucida mtDNAs (AVISE icoba tributary carry distinct mtDNA haplotypes. It and VRIJENHOEK1987), we could unequivocally assign was previously shown that ML/IV E-types from these maternal parentage to each unisexual specimen. P. tributaries had distinct H-types (ANGUS andSCHULTZ monacha is the maternal progenitor of all the P. mon- 1979), butwe could not ascertain whetherthe H-types acha-lucida hybridogens examined in this study. This arose independently. In this case, independent hybrid finding confirms earlier laboratory studies, wherein events are likely to capture identical E-types, because matings of P. monacha females x P.lucida males ML/IV contains the most frequent P. monacha alleles sometimes produce viable hybridogenetic lineages at all polymorphic loci (compare Table 3 with VRIJEN- (SCHULTZ1973b), but the reciprocal matings do not HOEK 197913). Mutational divergence of the mtDNAs 396 J. M. Quattro,J. C. Avise and R. C. Vrijenjhoek

in ML/IV.7 and ML/IV.9 is less likely because these 1987). P. monacha females from the Rio San Pedro haplotypes are separated by at least four mutations. were more likely to yield viable and fertile hybrido- Unfortunately, strains used in the earlier tissue graft- genetic lineages than P. monacha from any other re- ing studies were not maintained, preventing a com- gion within the Rio Fuerte. prehensive evaluation of relationships among E-types, Surveys of allozyme diversity in hybridogens from H-types, and Mt-types. other rivers also revealed endemic hemiclones marked Tissue grafting studies revealed that some P. mon- by local P. monacha allozymes. The Rio Sinaloa, just acha-lucida hemiclones are widely distributed in the south of the Rio Fuerte, has several endemic hemi- Rio Fuerte (ANGUS and SCHULTZ1979). We found clones of P. monacha-lucida (VRIJENHOEK1984). Sim- hemiclone ML/I.l 1 in both the San Pedro and Guir- ilarly, the Rio Mayo, just north of the Rio Fuerte, has icoba tributaries, more than 150 km apart. Similarly, endemic hemiclones of P. monacha-occidentalis (VRI- hemiclone ML/lI.14 was captured at bothsites. Given JENHOEK, ANGUSand SCHULTZ1977). Clearly, multi- the high diversity of allozyme and mtDNA genotypes ple independent originsof hybridogenetic strains have in P. monacha, it is unlikely that two hemiclones would occurred within all major drainage systems that con- freeze identical nuclear and mitochondrial genotypes tain P. monacha. duringindependent origins. Furthermore, mtDNA Although multiple independent origins explain haplotypes 1 1and 14were not observed in P. monacha most of the mtDNA diversity in P. monacha-lucida, from Guiricoba locality. Although we cannot exclude mutations might also have contributed. Some P. mon- the possibility that these haplotypes were once more acha-lucida individuals have identical allozyme geno- broadly distributed in P. monacha, it is more parsi- types but harbor mtDNAvariants differing by a single monious to propose that ML/I. 11and ML/II.14 each mutation. We previously arguedthat independent had single origins and then spread to othersites in the origins could produce hemiclones with identical allo- Rio Fuerte. zyme genotypes but distinct mtDNAs, yetin some Conversely, other hemiclones appearendemic to cases post-formational mutation of mtDNA within a certain tributaries of the Rio Fuerte. For example, hemiclonal lineage also seems likely. For example, the hemiclones ML/VII. 1 and ML/VIlI. 1 bear allozymes mtDNAs of E-type ML/I.l 1 and ML/I.12 differ by a and mtDNA haplotypes that areunique to P. monacha single mutation. ML/I. 12 could have arisen directly fromthe Jaguari tributary and its feeder streams. from ML/I. 11, orit might have arisen independently With diallelic polymorphisms at five loci in P. monacha from a P. monacha lineage harboring mtDNA haplo- from this tributary, we have the capacity to identify type 12 that was not included in our sample (Figure 32 E-types at this site (VRIJENHOEK,ANGUS and 2). Similarly possibilities exist for the hemiclonal pair SCHULTZ 1978), butonly ML/VII and ML/VIII have ML.II.13 and ML/II.14. Both the ML/I and ML/II been found in surveys involving several thousand in- E-types are composed of common alleles segregating dividuals over a 15-yr period (SCHENCKand VRIJEN- in Rio Fuerte P. monacha; thus, independent origins, HOEK 1986). The present addition of three mtDNA giving rise to identical E-types are likely. haplotypes tothe suite of charactersfound in P. With the limited sample available for this study, and monacha at this site raises thenumber ofpossible the shrinking number of P. monacha populations re- hemiclonal combinations to 96, butstill we found only maining in nature, we cannot easily determine how two. Similarly, tissue grafting studies identified no important mutation within hemiclones has been as a additional H-type variation beyond the major differ- source of mtDNA diversity for these fish. Despite the ences defining ML/VII andML/VIII (ANGUS and indirect evidence for postformational mutation in SCHULTZ 1979).Hybridization experiments involving hemiclonal mtDNAs and in enzyme-determining gene P. monacha females from the Jaguari tributary were loci (SPINELLAand VRIJENHOEK1982; VRIJENHOEK less likely to produce viable hybridogenetic lineages 1984), genetic differencesdo not gobeyond the broad than females from the San Pedro or Guiricoba tribu- variation segregating in extant populations of P. mon- taries (WETHERINGTON,KOTORA and VRIJENHOEK acha. Thus, theorigins of hemiclonal Poeciliopsis must 1987). Perhaps opportunities forthe origin and estab- have occurred within the time scaleof matriarchal lishment of new hybrid lineages are limited at this site. lineage diversification in P. monacha. Without excep- Hybrid origins appear to be more likely in the Rio tion, molecular genetic studies of unisexual verte- San Pedro than in other tributaries of the Rio Fuerte. brates have concluded that the unisexual populations Comparatively more mtDNA haplotypes found in P. are recently evolved (DENSMOREet al. 1989; DENS- monacha-lucida were associated with P. monacha hap- MORE, WRIGHTand BROWN 1989; ECHELLEet al. lotypes sampled from the San Pedro than either the 1989; GODDARD, DAWLEYDOWLING and 1989; MOR- Guiricoba or Jaguari populations. This conclusion is ITZ et al. 1989; MORITZ, WRIGHT and BROWN 1989). supported by thelaboratory hybridization experi- We find it curious that P. monacha, a sexual species ments(WETHERINGTON, KOTORA and VRIJENHOEK characterized by relatively small and highly subdi- mtDNA Variation in Poeciliopsis 397 vided relictual populations, contains so much genetic and JOSE CAMPOY-FAVELA. Also,we thank DEANA. HENDRICKSON for help in the field and BILL NELSONfor assistance in the labora- diversity. It was previously hypothesized that the sex- tory. Specimens used in this study were collected under permits ual species might recapture variation through rare #412.2.1.3.0folio4815issuedjointlybytheDepartmentodePesca, hybridizations with the widely dispersed and more by Secretaria de Desarrollo Urbanoy Ecologia, and by the Depar- abundant hybridogens (VRIJENHOEK1979b). Hemi- tamento de Exteriores, Mexico, D.F. ANGELICA NARVAEZ, Office clonal variation within P. monacha-lucida populations of the U.S. Science Attache, Mexico,was instrumental in obtaining these permits. Research was supported by grants BSR88-05360 can enter the sexual gene pool if hybridogens mate U.C.A.) andBSR88-05361 (R.C.V.) from the National Science with P. monacha males (e.g., M’L X MM). The result- Foundation, the Theodore Roosevelt Memorial Fund, American ing M’M progeny, if viable and fertile,reproduce Museum of Natural HistoryU.M.Q.), and by the Leatham-Steinetz- sexually and have normal meiosis (LESLIEand VRIJEN- Stauber Graduate Research Fund, Rutgers UniversityU.M.Q). HOEK 1980; SCHULTZ 1973a; VRIJENHOEKand SCHULTZ1974). Thus,P. monacha-lucida might serve LITERATURE CITED as a reservoirof genetic variation, connectingisolated ANGUS,R. A., and R. J.SCHULTZ, 1979 Clonaldiversity in the populations of P. monacha, and buffering the loss of unisexual fish Poeciliopsis monacha-lucida: a tissue graftanalysis. nuclear and organeller variation. Evolution 33: 27-40. Circumstantially, we found mtDNA haplotypes in AVISE,J. C., and R. C. VRIJENHOEK,1987 Mode of inheritance P. monacha-lucida that mightbe intermediates be- and variation of mitochondrial DNA in hybridogenetic fishes tween sexual haplotypes. For example, ML/XV. 10 of the genusPoeciliopsis. Mol. Biol. Evol. 4: 514-525. CIMINO,M. C., 1972 Eggproduction, polyploidization and evo- mightconnect haplotypes 7 and11 in P. monacha lution in adiploid all-female fish ofthe genus Poeciliopsis. (Figure 2). Alternatively, haplotype 10 might have Evolution 26 294-306. been frozen from a P. monacha lineage that we failed DENSMORE,L. D., IIl., J. W. WRIGHT and W. M. BROWN, to sample, or has since disappeared.Although we 1989 Mitochondrial-DNA analyses and the origin and relative cannotdemonstrate the feedback process with the age of parthenogenetic lizards (genus Cnemidophorus). 11. C. neomexicanus and theC. tesselatus complex. Evolution 43: 943- present data, its feasibility has been shown clearly in 957. the laboratory (LESLIEand VRIJENHOEK1980), and DENSMORE,L. D., HI., C. C. MORITZ,J. W. WRIGHT and W. M. we suspect it also occurs in nature. Ultimately, we BROWN,1989 Mitochondrial-DNAanalyses and the origin must view monacha genomes as dynamic assemblages and relative age of parthenogenetic lizards (genusCnemidopho- of genes that exist in at least two states: as recombining rus). IV. Ninesexlineatus-group unisexuals. Evolution 43: 969- 983. units in a sexual species and as potentially recombining ECHELLE,A. A., T. E. DOWLING,C. C. MORITZand W. M. BROWN, units frozen in the hemiclonal genomes of the hybri- 1989 Mitochondrial-DNA diversity and the origin of Men- the dogens. idia clarkhubbsi complex of unisexual fishes (Atherinidae). Ev- It is unfortunatethat P. monacha, the common olution 43: 984-993. sexual ancestor to all the known unisexual-hybrid FELSENSTEIN,J., 1990 PHYLIP ManualVersion 3.3. University Herbarium, University of California, Berkeley, Calif. biotypes of Poeciliopsis, is presently in danger of ex- GODDARD,K. A., R. M. DAwLEYand T. E. DOWLING,1989 Origin tinction. It might already be extinct in the Rios Mayo and genetic relationships of diploid, triploid, and diploid-tri- and Sinaloa, and it is seriously threatened in the San ploidmosaic biotypes in the Phoxinus eos-neogaeus unisexual Pedro andGuiricoba tributaries, wherenative habitats complex, pp. 268-280 in Evolutionand Ecology of Unisexual have beenaltered with theintroduction of exotic Vertebrates, edited by R. DAWLEY andJ. BOGART.New York State Museum, Albany, N.Y. fishes, primarily Tilapia spp. If P. monacha disappears LANSMAN,R. A., R. 0. SHADE,J. F. SHAPIRAand J. C.AVISE, from a river, the opportunity for de novo origins of 1983 The use of restriction endonucleases to measure mito- hybridogenetic lineages also disappears. Previous chondrial DNA sequence relatedness in natural populations. studies revealed that theecological health of unisexual 111. Techniques and potential applications. J. Mol. Evol. 80 populations depends on the opportunity for de novo 1969-1971. LESLIE,J. F., and R. C. VRIJENHOEK,1978 Genetic dissection of hybrid origins and interclonal selection (VRIJENHOEK clonally inherited genomes of Poeciliopsis. I. Linkage analysis 1979a; WETHERINGTON,KOTORA and VRIJENHOEK and preliminary assessment of deleterious geneloads. Genetics 1987; WETHERINGTON,SCHENCK and VRIJENHOEK 90: 801-811. 1989; WETHERINGTONet al. 1989). Ironically, just as LESLIE,J. F., and R. C.VRIJENHOEK, 1980 Consideration of variability in hemiclonal populations depends on var- Muller’s ratchet mechanism through studies of genetic linkage and genomic compatibilities in clonally reproducing Poeciliop- iation frozen from P. monacha, variability in P. mona- sis. Evolution 34: 1105-1 115. cha might also depend on feedback from the genetic MILLER,R. R., 1960 Four new species of viviparous fishes, genus reservoir in the unisexual populations. We should seek Poeciliopsis from northwestern Mexico. Occas. Pap. Mus.Zool. means to preserve all components of this dynamic Univ. Mich. 433: 1-9. genetic complex. MORITZ, C.C., T. E. DOWLINGand W. M. BROWN, 1987 Evolution of mitochondrial DNA: relevance for popu- lation biology and systematics. Annu. Rev. Ecol. Syst. 18: 269- We gratefully acknowledge the staff of the Centro Ecologicode 292. Sonora, Hermosillo, Mexico, for their collaboration and hospitality, MORITZ, C. C., J.W. WRIGHT andW. M. BROWN, 1989 particularly LOUDESJU~REZ-ROMERO, ALEJANDRO VARELA ROMERO Mitochondrial-DNA analyses and the origin and relativeof age 398 J. M. Quattro, J. C. Avise and R. C. Vrijenjhoek

parthenogenetic lizards (genus Cnemidophorus). 111. C. uelox lution, edited by C. L. MARKERT.Academic Press, New York. and C. exsanguis. Evolution 43: 958-968. ~RIJENHOEK,R. C., 1979a Factors affecting clonal diversity and MORITZ, C., W.M. BROWN,L. D. DENSMORE,J. W. WRIGHT, D. coexistence. Am. Zool. 19: 787-797. VYAS, S. DONNELLAN,M. ADAMS and P. BAVERSTOCK, ~RIJENHOEK,R. C., 1979b Genetics of a sexually reproducing fish 1989 Genetic diversity and the dynamics of hybrid pdrthen- in a highly fluctuating environment. Am. Nat. 113: 17-29. ogenesis in Cnemidophorus (Teiidae) and Heteronotia (Gekoni- VRIJENHOEK,R. C.,1984 The evolutionof clonal diversity in dae), pp. 87-1 12 in Euolution and Ecology of Unisexual Verte- Poeciliopsis, pp. 399-429 in Euolutionary Genetics of Fishes, ed- brates, edited by R. DAWLEY andJ. BOGART.New York State ited by B. J. TURNER.Plenum Press, New York. Museum, Albany, N.Y. VRIJENHOEK,R. C., 1989Genetic and ecological constraints on MUINEY, M., and R. C. VRIJENHOEK,1981 Characterization of the origins andestablishment of unisexual vertebrates, pp. 24- Biomphalariaglabrata strains by starch-gel electrophoresis. 31 in Evolution and Ecology of Unisexual Vertebrates, edited by Biochem. Genet. 19 1169-1 179. R. DAWLEY andJ.BOGART. New York State Museum, Albany, NEI, M., and F. TAJIMA, 1981 DNA polymorphism detectable by N.Y. restriction endonucleases. Genetics 97: 145-163. VRIJENHOEK,R. C., R. A. ANGUSand R. J. SCHULTZ,1977 KOHLF, F.J., J. KISHPAUCHand D. KIRK, 1974 Numerical Tax- Variation and heterozygosity in sexually vs. clonally reproduc- onomy System of Multivariate Statistical Programs. ing populations of Poeciliopsis. Evolution 31: 767-78 1. SCHENCK,R. A,, and R. C. VRIJENHOEK,1986 Spatial andtem- VRIJENHOEK,R. C., R. A. ANGUSand R. J. SCHULTZ, 1978 poral factors affectingcoexistence among sexual and clonal Variation and clonal structure in a unisexual fish. Am. Nat. forms of Poeciliopsis. Evolution 40: 1060-1070. 112: 41-55. SCHULTZ,R. J., 1969 Hybridization, unisexuality and polyploidy VRIJENHOEK,R. C.,and R. J. SCHULTZ,1974 Evolution of a in the teleost Poeciliopsis (Poeciliidae) and other vertebrates. trihybrid unisexual fish (Poeciliopsis, Poeciliidae). Evolution 28: AI]). Nat. 103: 605-6 19. 205-3 19. VRIJENHOEK,R. C., R. M. DAWLEY,C. J. COLEand J. P. BOGART, SCHULTL,R. J., 1973a Origin and synthesisof a unisexual fish, 1989 A list of knownunisexual vertebrates,pp. 19-23 in pp. 207-2 1 1 in Genetic and Mutagenesis ofFish, edited by J. H. Evolutionand Ecology of UnisexualVertebrates, edited by SCHOEDER. Springer-Verlag,New York. R, DAWLEY andJ. BOGART.New York State Museum,Albany, SCHULTZ,R. J., 1973b Unisexual fish: laboratory synthesis of a N.Y. “species.” Science 179: 180- 181. VYAS,D. K., C. MORITZ,D. M. PECCININI-SEALE,J. W. WRIGHT SCHULTZ,R. J., 1977 Evolution and ecology of unisexual fishes. and W. M. BROWN,1990 The evolutionary history of par- Evol. Biol. 10277-33 1. thenogenetic Cnemidophoruslemniscatus (Sauria: Teiidae). 11. SITES, J.W., JR., D. M. PECCININI-SEALE,C. MORITZ,J. W. WRIGHT Maternal origin and age inferred from mitochondrial DNA and W. M. BROWN,1990 The evolutionary history of par- analyses. Evolution 44: 922-932. thenogenetic Cnemidophoruslemniscatus (Sauria: Teiidae). I. WETHERINGTON,J. D., K.E. KOTORA and R. C. VRIJENHOEK, Evidence for a hybrid origin. Evolution 44: 906-921. 1987 A test of the spontaneous heterosis hypothesis for uni- SMOUSE,P. E., J. C. LONG and R. R. SOKAL,1986 Multiple sexual vertebrates. Evolution 41: 721-731. regression and correlation extensions of the Mantel Test of WETHERINGTON,J. D., R. A. SCHENCKand R. C. VRIJENHOEK, matrix correspondence. Syst. Zool. 35: 627-632. 1989 Origins and ecological success of unisexual Poeciliopsis: SPINELLA,D. G., and R. C. VRIJENHOEK,1982 Genetic dissection the Frozen Niche Variation model, pp. 259-276 in The Ecology of clonally inherited genomes of Poeciliopsis. 11. Investigation and Euolution of Poeciliid Fishes, edited by G. A. MEFFE and F. of a silent carboxylesterase allele. Genetics 100: 279-286. F. SNELSON,JR. Prentice-Hall, Englewood Cliffs, N.J. SPOISKY,C., and T. UZZELL,1986 Evolutionaryhistory ofthe WETHERINGTON,J. D., S. C. WEEKS,K. E. KOTORA and R. C. hybridogenetic hybrid frog Rana esculenta as deduced from VRIJENHOEK,1989 Genotypic and environmental compo- mtDNA analyses. Mol. Biol. Evol. 3: 44-56. nents of variation in growth and reproduction of fish hemi- VRIJENHOEK,R. C.,1975 Gene dosage in diploid and triploid clones (Poecilzopsis: Poeciliidae ). Evolution 43: 635-645. unisexual fishes, pp. 463-475 in Isozymes IV. Genetics and Euo- Communicating editor:J. R. Pow~1.1.