Hybrid Sterility, Haldane's Rule and Speciation in Heliconius cydno and H. melpomene

Corresponding author: James Mallet, Department of Biology, University College London, 4 Stephenson Way, London NW1 2HE, United Kingdom., j.mallet{at}ucl.ac.uk (E-mail)

Communicating editor: J. B. WALSH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Most genetic studies of Haldane's rule, in which hybrid sterility or inviability affects the heterogametic sex preferentially, have focused on Drosophila. It therefore remains unclear to what extent the conclusions of that work apply more generally, particularly in female-heterogametic taxa such as birds and Lepidoptera. Here we present a genetic analysis of Haldane's rule in Heliconius butterflies. Female F₁ hybrids between Heliconius melpomene and H. cydno are completely sterile, while males have normal to mildly reduced fertility. In backcrosses of male F₁ hybrids, female offspring range from completely sterile to fully fertile. Linkage analysis using the Z-linked triose-phosphate isomerase locus demonstrates a "large X" (Z) effect on sterility. Expression of female sterility varies among crosses in this and a previous study of Heliconius. Sterility may result from the production of normal but infertile eggs, production of small infertile eggs, or from a complete failure to develop ovarioles, which suggests multiple routes to the evolution of hybrid sterility in these Heliconius species. These results conform to the expectations of the "dominance" rather than "faster male" theories of Haldane's rule and suggest that relatively few loci are responsible. The two species are broadly sympatric and hybridize in the wild, so that female hybrid sterility forms one of several strong but incomplete barriers to gene flow in nature. The effect of female sterility is comparable to that of selection against non-mimetic hybrids, while mate choice forms a much stronger barrier to gene transfer.

HALDANE'S rule has proved an enduring generalization. It states that when one sex is absent, rare, or sterile in the F₁ offspring of a cross between two races or species, that sex is most commonly the heterogametic sex (HALDANE 1922 ). It holds for >95% of 324 crosses from six classes and several insect orders (LAURIE 1997 ) and seems to be a standard feature of the evolution of complete sterility and inviability (COYNE and ORR 1989A , COYNE and ORR 1997 ). However, a universal explanation has proved elusive.

Recently a degree of consensus has been reached that Haldane's rule is largely explained by the "dominance theory," with a contribution in Drosophila from "faster male evolution," and possibly "faster X evolution" (WU and DAVIS 1993 ; TURELLI and ORR 1995 , TURELLI and ORR 2000 ; LAURIE 1997 ; ORR 1997 ; TURELLI 1998 ). All of these theories rely on the existence of complementary genes (BATESON 1909 ; DOBZHANSKY 1936 ; MULLER 1940 ) that can produce hybrid inviability or sterility between two populations without either having passed through an adaptive valley. Divergent alleles become fixed at different loci between populations and cause incompatibility only when brought together in novel hybrid genotypes. The dominance theory stems directly from this epistatic model. If members of complementary sets of genes are found on the sex chromosomes, and act as recessive alleles in their effect on hybrid sterility, the heterogametic sex will suffer incompatibilities before the homogametic sex. This will produce Haldane's rule and may be sufficient to explain it (TURELLI and ORR 1995 ). However, several observations from studies of Diptera have inspired other theories. Within Drosophila, cases of Haldane's rule for sterility vastly outnumber those for inviability (WU and DAVIS 1993 ). Attached-X hybrid females with homozygous X chromosomes fail to show sterility in crosses where male hybrids are sterile (COYNE 1985 ), but do show inviability in cases where normal male hybrids are inviable (ORR 1993 ). Sterility factors therefore seem to accumulate more rapidly in males and are sex specific, unlike inviability loci. Male sterility is also found in Aedes mosquito hybrids, which lack a hemizygous X so that the dominance theory cannot act (PRESGRAVES and ORR 1998 ). These observations may be explained by "faster male" evolution, that is, the faster evolution of male sterility alleles, due either to greater sensitivity of spermatogenesis to disruption or to a rapid divergence of male reproductive characters driven by sexual selection (WU and DAVIS 1993 ; WU et al. 1996 ). Of course, this theory cannot contribute to sterility in female-heterogametic taxa. The incidence of Haldane's rule may also be promoted by "faster X" evolution, where hemizygosity enhances selection of favorable recessive alleles on the X chromosome (HALDANE 1924 ; CHARLESWORTH et al. 1987 ). For these alleles to produce sex-limited effects in hybrids and contribute to Haldane's rule, faster X explanations must act in conjunction with either the dominance theory or a chance correlation across taxa between the sex that is heterogametic and the sex-specific expression of incompatibilities (ORR 1997 ). Two detailed introgression studies provide little support for a greater density of sterility factors on the X chromosome than on autosomes, ruling out a major contribution from faster X theory (HOLLOCHER and WU 1996 ; TRUE et al. 1996 ). They do, however, provide evidence for both dominance and faster male evolution, with many recessive autosomal sterility factors effective only when made homozygous and a greater density of male than female sterility factors. In sum, the dominance theory seems to provide a fundamental explanation of Haldane's rule across all taxa for sterility and inviability, while faster male evolution contributes to the disproportionate representation of hybrid male sterility in Drosophila.

Birds and Lepidoptera may play an important role in distinguishing the contributions of the three theories. Their females are heterogametic yet both groups still display Haldane's rule. Therefore, heterogamety rather than sex is critical. Faster male theories predict the reverse of Haldane's rule, and models suggest that the dominance theory can produce Haldane effects in females even when opposed by faster evolution of male hybrid sterility (TURELLI and ORR 2000 ). In Lepidoptera the Z (X) chromosome represents a relatively small proportion of the genome (1/20–1/30), which could slow the origin of Haldane's rule effects (TURELLI and BEGUN 1997 ). However, for unexplained reasons other genetic differences between species show a bias toward Z-linkage (PROWELL 1998 ). Detailed genetic analyses on rates of accumulation of male relative to female sterility factors and on Z relative to autosomal effects have yet to be performed within the Lepidoptera (HOLLOCHER and WU 1996 ; TRUE et al. 1996 ) and better genetic studies of the taxon are required.

Here we study the genetic basis of sterility between Heliconius cydno and H. melpomene (Lepidoptera: Nymphalidae). These sister species are sympatric across much of Central and Andean South America below 1500 m (BROWN 1979 ; BROWER 1996 ). Both are unpalatable and warningly colored, and their speciation has been accompanied by a shift in Müllerian mimicry. H. melpomene is black with red and yellow markings and mimics H. erato, while H. cydno is black with yellow or white markings and usually mimics H. sapho or H. eleuchia (LINARES 1997 ; JIGGINS et al. 2001B ). This shift in mimetic allegiance appears to have driven speciation, resulting both in selection against hybrids due to their intermediate non-mimetic color pattern and in assortative mating through pleiotropic effects on mate recognition (JIGGINS et al. 2001B ). Color pattern divergence probably followed a change in habitat use, bringing the two into contact with different suites of potential comimics. H. melpomene is found in second growth and H. cydno in forest understorey, but there is considerable overlap and the two species commonly fly together (BENSON 1978 ; SMILEY 1978B ; WAAGE et al. 1981 ; MALLET and GILBERT 1995 ). They also differ in their degree of specialization among Passiflora host plants (SMILEY 1978A ). Despite these differences the species hybridize in the wild, albeit rarely: F₁ and backcross hybrids are known from across their range, although they probably form <0.1% of natural populations (MALLET et al. 1998B ).

A previous study has shown female hybrid sterility between H. melpomene races from French Guiana and those from Panama and Colombia (JIGGINS et al. 2001A ). Here we investigate hybrid incompatibility between the genetically more divergent sister taxa, H. cydno and H. melpomene, and review what is known of the genetics of hybrid sterility in all these crosses.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Collection localities are given in Table 1. Crosses involving H. melpomene rosina and H. cydno chioneus from Panama and H. melpomene melpomene from French Guiana were performed in Gamboa, Republic of Panama, between August 1998 and March 2000. Crosses involving Colombian butterflies were made in La Vega, 50 km northwest of Santafé de Bogotá, Colombia. H. melpomene vulcanus and H. cydno were collected in the west of the Cauca valley, in the Dagua Pass. In this region three races of H. cydno form a hybrid zone among H. cydno zelinde, H. cydno cydnides, and H. cydno weymeri (LINARES 1997 ), but with no detectable reduction in hybrid fertility or viability. Stocks of H. melpomene melpomene and H. cydno cordula were collected from the foothills of the eastern slopes of the Andes.

After mating, females were kept individually in 1 x 1 x 2-m outdoor insectaries and supplied daily with pollen sources (Lantana and Psiguria) and artificial nectar (10% sugar solution). Passiflora vines were provided for oviposition, mainly Passiflora menispermifolia and P. edulis. Eggs were collected daily and kept individually in small plastic pots with moist cotton wool to maintain humidity. To avoid cannibalism caterpillars were reared individually until the third instar and then in groups of two to eight on new growth of P. biflora in Panama and P. edulis in Colombia. After pupation they were transferred to baskets until eclosion. The number of eggs laid, hatch rates, and eclosing butterflies were recorded. Females were dissected to confirm that they had received a spermatophore at mating and to assess the level of ovary development. F₁ males and females and female offspring of backcrosses were tested for fertility.

Crosses were attempted in all possible combinations, but very strong asymmetrical mate choice prevented one direction of interspecific cross (H. melpomene female x H. cydno male) from Panama populations, and this direction of cross was obtained only once each using Colombia and French Guiana females (Table 2). Crosses between the two species in Colombia were largely restricted to stocks from within a geographic region. Results from interracial crosses within H. melpomene are described separately (JIGGINS et al. 2001A ).

Statistical analysis:
Counts of egg hatch rates were analyzed using a likelihood test based on a ß-binomial model (BETA-BINO program by Ziheng Yang; see JIGGINS et al. 2001A ). This method takes account of genetic or environmental variation in hatch rate between replicate broods within each type of cross, when the real interest is in variation between cross types. Rather than assuming a simple binomial distribution of hatch rates, the binomial parameter is allowed to vary across replicate broods within each cross type according to a ß distribution, which can accommodate skewed or even bimodal distributions. Crosses can then be compared using estimates of the ß-distribution parameters of mean, variance, and their standard errors. The program provides five alternative models for a data set containing replicate broods of several classes: (1) a classical binomial parameter for each class, which assumes a zero brood-to-brood variance; (2) a single ß mean and variance for the entire data set; (3) different means for each class but a single variance; (4) a single mean but different variances; and (5) a different mean and variance for each class. Further comparisons can be made by fitting models after combining parts of the data set. For example, to compare the fertility of different male genotypes while controlling for female genotype, models are fitted before and after combining brood classes of the same maternal genotype but different paternal genotype. These models are then compared using likelihood (L) ratio tests, for which the test statistic asymptotically follows a ² distribution (EDWARDS 1972 ). Colombian broods were analyzed separately as hatch rates of control broods were lower than those in Panama, presumably for environmental reasons. For sex ratio, where a simple binomial was the best-fitting model, support limits were obtained at the parameter values that led to a decrease in log_e likelihood of two units, approximately equivalent to 95% confidence intervals (EDWARDS 1972 ).

Linkage analysis:
Intron 4 of the sex-linked triose-phosphate isomerase (Tpi) gene was amplified using primers situated in the surrounding exons. Evidence for sex linkage, primer sequences, and PCR conditions is described by JIGGINS et al. 2001A . This intron contains a 39-bp insertion that is common (but not fixed) in the Panama H. cydno population, hereafter the "cydno insertion," but absent in H. melpomene. This length variation was used to follow segregation of Tpi in Panama backcross broods: alleles were separated on 6% acrylamide gels run for 4 hr at 125 V and stained using ethidium bromide. All broods proved informative with regard to the segregation of alleles in female offspring, having F₁ fathers heterozygous for the insertion.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Crosses between the sister species in Panama:
Female hybrids between H. cydno and H. melpomene were completely sterile (Table 2). In the sympatric Panamanian cross 25 F₁ females were tested, producing 209 apparently normal eggs, not one of which hatched. Dissections confirmed that these females had received a spermatophore at mating and had developed normal ovaries (Table 3). Fifteen of the females failed to lay eggs despite surviving >15 days, by which time fertile control broods had invariably begun laying. Male hybrids were fully fertile, and there was no significant difference in hatch rate between their offspring and those of control broods .

Sex ratio across all broods was best described by a simple binomial model with an average proportion of females of 0.516 (support limits 0.491, 0.542), across 1553 adults from 48 broods. There was no evidence for differences in the sex ratio of adults emerging from nine Panama and French Guiana control, F₁, and Panama backcross brood classes , suggesting an absence of Haldane's rule on female inviability.

When fertile F₁ males were backcrossed to females of the two parental species, female offspring were recovered showing the full range of fertility but with pronounced bimodality between complete sterility and apparently normal fertility (Table 4A and Table 4B, and Fig 1). In the backcross to H. cydno, complete sterility was usually manifested as a failure to lay eggs, while from the backcross to H. melpomene, sterile females typically laid eggs that did not hatch.

View larger version (37K): In this window In a new window Download PPT slide   Figure 1. Hatch rates of eggs laid by female offspring of the (A) backcross to H. melpomene MP x (CP x MP) and from (B) all backcrosses to H. cydno C x(C x M). Hatched bar, genotype at the triose-phosphate isomerase locus, with superscripts for the alleles of H. melpomene and H. cydno; ?, for individuals that were not analyzed.

There was strong linkage between sterility and Z-linked Tpi in female offspring from the backcross to H. melpomene (Table 4A). All six females that were completely sterile had the cydno Tpi insertion, while all five females that showed at least some fertility lacked the insertion and therefore had that region of the H. melpomene Z chromosome . In contrast, in the backcross to H. cydno there was no association between Tpi genotype and sterility (Table 4B, ).

Crosses between H. melpomene from French Guiana and H. cydno from Panama:
Interspecific hybrid females of this cross were also sterile (Table 2). None of the 10 female offspring of H. cydno female x H. melpomene male laid any eggs. Dissections showed that the failure to lay eggs in these females was due to a complete failure to develop ovaries (Table 3). Only two females could be produced from the reciprocal cross and, although both were sterile, one actually laid eggs. There was also a reduction in F₁ male fertility (Table 2). Broods fathered by F₁ males had hatch rates of 0.615 ± 0.041 with a H. cydno female and 0.611 ± 0.124 with H. melpomene, significantly lower than the rates of 0.859 ± 0.039 and 0.901 ± 0.027 for control broods of H. cydno and Guiana H. melpomene, respectively . This is hybrid male sterility, rather than incompatibility when crossing parents of different genotypes, since there was no evidence of a reduction in hatch rate among eggs fertilized by a heterospecific male when F₁ and control broods were compared .

Crosses between Colombian sister species:
In both the Eastern Andean foothills and Cauca valley crosses, female F₁ hybrids of H. cydno female x H. melpomene male were completely sterile, either failing to lay eggs or laying eggs that never hatched (Table 2). Hybrid males were fully fertile: hatch rates of their offspring were higher than those of control broods, although not significantly so . Some partially fertile F₁ females were produced in the reciprocal cross of an Eastern Andes H. melpomene female x H. cydno male from the Cauca valley (Table 2). Females were tested from a single brood, with three laying few eggs that never hatched, and the remaining four producing eggs with hatch rates of 20%.

Fertile F₁ males backcrossed to H. cydno produce female offspring with the full range of fertility (Table 4C and Table 4D, Fig 1B). The ratio of completely sterile to fertile females in backcrosses does not differ significantly between the eastern foothills, Cauca valley, or Panama backcrosses to H. cydno .

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Female F₁ hybrids between H. cydno and H. melpomene are completely sterile in five of the crosses described here, and their fertility is dramatically reduced in the sixth. In contrast, males have normal fertility in all but one of the crosses, with fertility reduced only in hybrids between H. cydno from Panama and H. melpomene from French Guiana. The lack of variation in sex ratio across control and F₁ broods suggests that there are no sex differences in viability. Counting as three examples the Heliconius studies (the interracial melpomene cross and reciprocal cydno x melpomene crosses) brings the total to 17 cases of sex-limited sterility in Lepidoptera, all but one of which conform to Haldane's rule, compared to 71 lepidopteran examples of unisexual hybrid inviability, of which only three do not conform (DAVIES et al. 1997; LAURIE 1997 ; JIGGINS et al. 2001A ). This is the reverse of the pattern seen in Drosophila, where sterility predominates (WU and DAVIS 1993 ; LAURIE 1997 ), although it may be that in Lepidoptera too few species pairs have been studied carefully enough to detect hybrid sterility since inviability is more easily demonstrated. The dominance theory can explain why both taxa conform to Haldane's rule for sterility and inviability regardless of which sex is heterogametic (TURELLI and ORR 1995 ), while faster male evolution may explain the strong excess of sterility in Drosophila (WU and DAVIS 1993 ; TURELLI 1998 ).

The association between sterility and Tpi genotype shown here provides evidence of a large Z (X) effect on sterility, as is common in Drosophila (COYNE and ORR 1989B ) and similar to that shown in interracial crosses in H. melpomene (JIGGINS et al. 2001A ). The locus forms a terminal marker in linkage maps of the Z chromosome in H. erato and Helicoverpa armigera (N. FLANAGAN, personal communication; D. HECKEL, personal communication) so that linkage suggests factor(s) in that region able to produce complete sterility. The effect of introgression of this region of the Z chromosome is asymmetric. Introgression of the H. cydno region into a predominantly H. melpomene genotype produces complete sterility, while there is no detectable linkage between sterility and the Tpi locus of H. melpomene when introgressed into a predominantly H. cydno genetic background. Such asymmetric effects are predicted by Muller's model of complementary gene evolution (MULLER 1942 ). Sterility in the backcross to H. cydno must be produced by different factors, which must be less tightly linked to Tpi if they are on the H. melpomene Z chromosome. However, dominant-dominant and dominant-recessive interactions between autosomal loci cannot be ruled out for this direction of cross.

Here we find sterility in females with the H. cydno Tpi allele on a Panama H. melpomene autosomal background. Interestingly, in hybrids between geographic races of H. melpomene from Panama and French Guiana, sterility is associated with the same Z-linked Tpi marker: in this case it is the French Guiana Tpi that is associated with sterility on a Panama autosomal background while the reverse cross produces no sterility and therefore no association with Tpi (JIGGINS et al. 2001A ). According to recent molecular evidence, the two H. melpomene races are most closely related, with H. cydno as an outgroup (BELTRAN et al. 2002 ; V. BULL, unpublished results) in spite of previous evidence for paraphyly of H. melpomene relative to H. cydno (BROWER 1996 ). Two possible explanations exist for the observed pattern of sterility in the three taxa. First, sterility is due to a Z-linked region involved in female fertility and ovary development, which has diverged rapidly between populations. Independent substitutions in the Tpi regions of both H. cydno and Guianan H. melpomene must then have occurred to cause sterility on a Panamanian H. melpomene autosomal background. Second, and more parsimoniously, sterility in both cases is due to the same derived Z-linked and autosomal alleles substituted within H. melpomene in Panama. In both crosses, sterility is found in females with an autosomal background predominantly derived from Panamanian H. melpomene, but lacking the Z-linked Tpi region of that population. Sterility arises when an individual lacks the epistatic sex-linked genes of Panama H. melpomene needed to complement the autosomes of that population, providing good evidence for complementary genes (MULLER 1940 , p. 203). Sterility caused by the absence of this Z-linked region would therefore make the H. cydno and French Guiana H. melpomene regions behave as recessive "loss-of-function" alleles, as predicted under the dominance theory. However, loci causing sterility are likely to be sex specific, as in Drosophila. Ideally, therefore, we would need fertility data from the equivalent of attached-Z males to test between sex specificity (faster female evolution in this case) and the dominance theory as the principal cause of Haldane's rule in this example. However, no such strains exist and it is not even certain that the resulting W genotype would be phenotypically male under lepidopteran sex determination.

Three different sterility phenotypes appear in female hybrids involving H. melpomene (Table 3):

1. In the cross described here between female Panamanian H. melpomene and male H. cydno, F₁ females lay normal-sized eggs that never hatch. In this cross alone it is possible that "sterility" is actually produced by zygote inviability or failure of fertilization despite normal spermatophore transfer.

2. The sterile eggs of hybrids are much smaller in interracial crosses between female H. melpomene from Panama and male H. melpomene from French Guiana (JIGGINS et al. 2001A ). (In the reciprocal cross, egg size and fertility are normal.)

3. The third sterility phenotype is more extreme: hybrid females between French Guiana H. melpomene and Panama H. cydno fail to develop ovaries at all.

These distinct sterility phenotypes back up evidence from Tpi linkage to suggest that at least some different loci are involved in each case, although a parsimonious explanation of the results involving Panamanian H. melpomene (see previous paragraph) implies some genes in common.

At least four features of these crosses suggest that relatively few loci each of major effect are important in producing sterility:

1. The asymmetry seen in crosses between French Guiana and Panama or Colombia H. melpomene implies that few loci are involved, so that, by chance, incompatibilities have arisen in one but not both reciprocal crosses (MULLER 1942 ).

It has been asserted on the basis of an unpublished analysis that such asymmetry might arise even if large numbers of loci produce sterility (TURELLI and ORR 1995 , p. 395). However, many other examples of heterogeneity between the crosses studied here add to an impression that a few different major-effect loci, rather than many loci of individually small phenotypic effect, are important in each cross. For instance,

2. there are strong effects of loci linked to Tpi in some crosses but not others, and

3. the sterility phenotypes are very different between the different crosses. Neither would be expected if sterility were a polygenic effect of steadily accumulating genetic differences.

Finally,

4. the distribution of sterility in the backcrosses is strongly bimodal, with fertile females recovered after only a single generation of backcrossing.

If many loci combine to produce sterility, one might expect to see a range of intermediate fertility levels. Alternatively, if major-effect sterility factors were scattered throughout the genome, most offspring should have low fertility because in the backcrosses there is the opportunity for many additional interactions between dominant and recessive autosomal factors. Tpi-linked sterility on the H. cydno Z chromosome is therefore unusual in comparison with the rest of the genome. While some of these heterogeneities might be explained by multiple genes and threshold effects, the combination of effects seems most easily explained if several (rather than many) genomic regions are important, at least some of which have major effects on sterility.

While female sterility is complete in F₁ hybrids between a H. cydno female and H. melpomene male, it remains uncertain if sterility is normally present in hybrids from the reciprocal cross since few were produced due to strong behavioral isolation. There is sterility in both directions of cross between H. cydno from Panama and H. melpomene from French Guiana. In the Colombian crosses there is variation among females from the one brood of H. melpomene female by H. cydno male, with some showing complete sterility and others having low fertility with 20% of eggs hatching. This variation within F₁ females is difficult to explain, but both parents come from regions where interspecific hybridization occurs, H. melpomene melpomene with H. heurippa in the Eastern Andean foothills and H. cydno with H. melpomene vulcanus in the Cauca valley and Pacific slopes (LINARES 1989 ; MALLET et al. 1998B ). Populations may therefore carry introgressed polymorphisms for loci causing interspecific sterility, which then segregate in our F₁ cross. On the basis of the linkage between sterility and Tpi, we predict that female offspring from the unobtainable H. melpomene female x H. cydno male Panama cross should also be sterile. These would have a H. cydno Z chromosome and heterozygous H. melpomene autosomes. This is similar to the interaction that causes sterility in the backcross to H. melpomene shown by Tpi linkage, but which consists of the H. cydno Z together with on average 50% homozygous and 50% heterozygous H. melpomene autosomes. These females were completely sterile.

Closely related species of Heliconius typically have several incomplete barriers to gene flow that can include assortative mating, hybrid sterility, selection against hybrids due to increased predation on their non-mimetic color patterns, and the divergent ecology of the parental species (MALLET et al. 1998B ). Speciation is probably most closely associated with the origin of assortative mating and ecological divergence in Heliconius and other taxa (MALLET et al. 1998B ; JIGGINS and MALLET 2000 ; KIRKPATRICK and RAVIGNE 2002 ). Hybrid sterility is neither necessary nor sufficient for speciation. For example, H. erato and H. himera show no evidence for a reduction in viability or fertility of interspecific hybrids, and yet the two remain distinct in hybrid zones characterized by assortative mating and probably strong ecological selection against hybrids (JIGGINS et al. 1997 ; MCMILLAN et al. 1997 ; MALLET et al. 1998A ). In contrast, unidirectional female hybrid sterility has been found between the French Guiana and Panama races of H. melpomene (JIGGINS et al. 2001A ), although these are traditionally considered part of the same species and are thought to be connected via continuous populations (BROWN 1979 ; MALLET 1986 ). Hybrid sterility may therefore arise as a consequence of reproductive or geographic isolation more than as a direct cause of speciation. If spatial isolation or assortative mating limit gene flow, even weak divergent selection in the two populations may lead to the accumulation of epistatic incompatibilities in hybrids. Sterility can of course promote further divergence by reducing gene flow still further and providing a selection pressure for increased assortative mating through reinforcement. In H. cydno and H. melpomene sterility currently forms one of three strong barriers reducing the fitness of F₁ and later generation hybrids. Genomic incompatibility resulting in female sterility produces strong selection against F₁ hybrids, of 50% when averaged across both sexes, and backcrosses produce females with an average reduction in fertility of 80%, so that the total barrier due to F₁ and backcross sterility is 0.7 (i.e., 0.5 + 0.5 x 0.5 x 0.8). Ecological incompatibility causing selection by predators against hybrid color patterns is of a similar strength, 50% if similar to that estimated from a H. erato hybrid zone (MALLET and BARTON 1989 ). Nearly half of the backcross offspring should escape this increased predation on hybrids, because their color patterns are similar to those of the parentals due to modifier loci and linkage of several major color pattern loci (R. E. NAISBIT, unpublished results), so that the overall barrier due to mimicry is 64%. Mate choice between the two species is by far the strongest barrier to gene transfer, since the species hybridize extremely rarely in nature, only with great difficulty in no-choice tests, and never in laboratory choice tests (NAISBIT et al. 2001 ). This is accompanied by disruptive sexual selection against hybrids causing reduced mating success, forming an overall mating barrier of >99.9% (MALLET et al. 1998B ; NAISBIT et al. 2001 ).

A consequence of these multiple incomplete barriers to gene flow is that certain hybrid genotypes will suffer disproportionately while others escape, creating a selectively permeable species boundary. As in hybrid zones, different regions of the genome will cross the species boundary and introgress with varying ease, with some halted by selective barriers and others relatively unrestricted (BARTON 1979 , BARTON 1983 ; RIESEBERG et al. 1999 ). Since backcrossing occurs with both parental species (NAISBIT et al. 2001 ), introgression should be possible in both directions (MALLET et al. 1998B ; GILBERT 2003 ). Selection will oppose the transfer of some genes, such as color pattern genes or sterility factors. Strongly selected loci will also affect associated regions, for example, the Tpi locus or mitochondrial DNA, the introgression of which is prevented by female sterility (SPERLING 1993 ). However, universally advantageous alleles should be barely affected if loosely linked. Even the strong Haldane's rule sterility found here between a pair of sympatric Heliconius species may not result in complete evolutionary independence in nature. In Heliconius, hybrid sterility clearly contributes to speciation, but it seems neither entirely necessary for speciation nor very effective at preventing gene flow after speciation has occurred.

	FOOTNOTES

¹ Present address: Laboratoire d'Ecologie Animale et Entomologie, Institut de Zoologie, Université de Neuchâtel, CH-2007 Neuchâtel, Switzerland.

	ACKNOWLEDGMENTS

We are very grateful to the Smithsonian Tropical Research Institute where the Panamanian crosses were carried out, and to ANAM (Autoridad Nacional del Ambiente) for permission to work in Panama. Fernando Arango kindly let two of us perform crosses at his farm at La Vega village in Colombia. For saving caterpillars from starvation and humans from boredom we thank Richard Woods, Camilla Paresce, Luis Alberto Taylor R., Catalina Estrada, Margarita Beltrán, Lianne O'Donovan, Rebecca Coe, Vanessa Bull, Cris Thompson, Yaneth Sánchez, and Carolina Guaneme. Diana Pérez deserves special thanks for providing partial support to Camilo Salazar during the development of this study. Thanks also go to Ziheng Yang for the use of his BETABINO program to analyze hatch rate data. The project was funded by a Natural Environment Research Council grant and a studentship from the Biotechnology and Biological Sciences Research Council. Funding for Colombian crosses was partially provided by Fondo Colombiano de Investigaciones Cientificas y Proyectos Especiales Francisco Jose de Caldas COLCIENCIAS, grant CO-1204-05-510-93, and Fundación Para la Promoción de la Investigación y la Tecnología del Banco de la República, grant 734.

Manuscript received July 2, 2001; Accepted for publication May 8, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BARTON, N. H., 1979  Gene flow past a cline. Heredity 43:333-339.

BARTON, N. H., 1983  Multilocus clines. Evolution 37:454-471.

BATESON, W., 1909 Heredity and variation in modern lights, pp. 85–101 in Darwin and Modern Science, edited by A. C. SEWARD. Cambridge University Press, Cambridge, UK.

BELTRÁN, M., C. D. JIGGINS, V. BULL, M. LINARES, and J. MALLET et al., 2002  Phylogenetic discordance at the species boundary: comparative gene genealogies among Heliconius butterflies. Mol. Biol. Evol. in press.

BENSON, W. W., 1978  Resource partitioning in passion vine butterflies. Evolution 32:493-518.

BROWER, A. V. Z., 1996  Parallel race formation and the evolution of mimicry in Heliconius butterflies: a phylogenetic hypothesis from mitochondrial DNA sequences. Evolution 50:195-221.

BROWN, K. S., 1979 Ecologia geográfica e evolução nas florestas neotropicais. Universidade Estadual de Campinas, São Paulo, Brasil.

CHARLESWORTH, B., J. A. COYNE, and N. H. BARTON, 1987  The relative rates of evolution of sex chromosomes and autosomes. Am. Nat. 130:113-146.

COYNE, J. A., 1985  The genetic basis of Haldane's rule. Nature 314:736-738.[Medline]

COYNE, J. A. and H. A. ORR, 1989a  Patterns of speciation in Drosophila. Evolution 43:362-381.

COYNE, J. A., and H. A. ORR, 1989b Two rules of speciation, pp. 180–207 in Speciation and Its Consequences, edited by D. OTTE and J. A. ENDLER. Sinauer Associates, Sunderland, MA.

COYNE, J. A. and H. A. ORR, 1997  "Patterns of speciation in Drosophila" revisited. Evolution 51:295-303.

DAVIES, N., A. AIELLO, J. MALLET, A. POMIANKOWSKI, and R. E. SILBERGLIED, 1997  Speciation in two neotropical butterflies: extending Haldane's rule. Proc. R. Soc. Lond. Ser. B 264:845-851.

DOBZHANSKY, T., 1936  Studies on hybrid sterility. II. Localization of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21:113-135.[Free Full Text]

EDWARDS, A. W. F., 1972 Likelihood. Cambridge University Press, Cambridge, UK.

GILBERT, L. E., 2003  Adaptive novelty through introgression in Heliconius wing patterns: evidence for shared genetic "tool box" from synthetic hybrid zones and a theory of diversification, in Ecology and Evolution Taking Flight: Butterflies as Model Systems, edited by C. L. BOGGS, W. B. WATT and P. R. EHRLICH. University of Chicago Press, Chicago in press.

HALDANE, J. B. S., 1922  Sex ratio and unisexual sterility in hybrid animals. J. Genet. 12:101-109.

HALDANE, J. B. S., 1924  A mathematical theory of natural selection and artificial selection. Part 1. Cambr. Phil. Soc. Trans. 23:19-41.

HOLLOCHER, H. and C.-I WU, 1996  The genetics of reproductive isolation in the Drosophila simulans clade: X vs. autosomal effects and male vs. female effects. Genetics 143:1243-1255.[Abstract]

JIGGINS, C. D. and J. MALLET, 2000  Bimodal hybrid zones and speciation. Trends Ecol. Evol. 15:250-255.[Medline]

JIGGINS, C. D., W. O. MCMILLAN, P. KING, and J. MALLET, 1997  The maintenance of species differences across a Heliconius hybrid zone. Heredity 79:469-505.

JIGGINS, C. D., M. LINARES, R. E. NAISBIT, C. SALAZAR, and Z. H. YANG et al., 2001a  Sex-linked hybrid sterility in a butterfly. Evolution 55:1631-1638.[Medline]

JIGGINS, C. D., R. E. NAISBIT, R. L. COE, and J. MALLET, 2001b  Reproductive isolation caused by colour pattern mimicry. Nature 411:302-305.[Medline]

KIRKPATRICK, M. and V. RAVIGNÉ, 2002  Speciation by natural and sexual selection: models and experiments. Am. Nat. 159:S22-S35.

LAURIE, C. C., 1997  The weaker sex is heterogametic: 75 years of Haldane's rule. Genetics 147:937-951.[Medline]

LINARES, M., 1989 Adaptive microevolution through hybridization and biotic destruction in the Neotropics. Ph.D. Thesis, University of Texas, Austin, TX.

LINARES, M., 1997 Origin of neotropical mimetic biodiversity from a three-way hybrid zone of Heliconius cydno butterflies, pp. 93–108 in Tropical Biodiversity and Systematics: Proceedings of the International Symposium on Biodiversity and Systematics in Tropical Ecosystems, edited by H. ULRICH. Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn, Germany.

MALLET, J., 1986  Hybrid zones of Heliconius butterflies in Panama and the stability and movement of warning colour clines. Heredity 56:191-202.

MALLET, J. and N. H. BARTON, 1989  Strong natural selection in a warning-color hybrid zone. Evolution 43:421-431.

MALLET, J. and L. E. GILBERT, 1995  Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol. J. Linn. Soc. 55:159-180.

MALLET, J., W. O. MCMILLAN, and C. D. JIGGINS, 1998a  Estimating the mating behavior of a pair of hybridizing Heliconius species in the wild. Evolution 52:503-510.

MALLET, J., W. O. MCMILLAN and C. D. JIGGINS, 1998b Mimicry and warning color at the boundary between races and species, pp. 390–403 in Endless Forms: Species and Speciation, edited by D. J. HOWARD and S. H. BERLOCHER. Oxford University Press, New York.

MCMILLAN, W. O., C. D. JIGGINS, and J. MALLET, 1997  What initiates speciation in passion-vine butterflies? Proc. Natl. Acad. Sci. USA 94:8628-8633.[Abstract/Free Full Text]

MULLER, H. J., 1940 Bearings of the ‘Drosophila’ work on systematics, pp. 185–268 in The New Systematics, edited by J. S. HUXLEY. Oxford University Press, London/New York.

MULLER, H. J., 1942  Isolating mechanisms, evolution and temperature. Biological Symp. 6:71-125.

NAISBIT, R. E., C. D. JIGGINS, and J. MALLET, 2001  Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene.. Proc. R. Soc. Lond. Ser. B 268:1849-1854.[Medline]

ORR, H. A., 1993  Haldane's rule has multiple genetic causes. Nature 361:532-533.[Medline]

ORR, H. A., 1997  Haldane's rule. Annu. Rev. Ecol. Syst. 28:195-218.

PRESGRAVES, D. C. and H. A. ORR, 1998  Haldane's rule in taxa lacking a hemizygous X. Science 282:952-954.[Abstract/Free Full Text]

PROWELL, D. P., 1998 Sex linkage and speciation in Lepidoptera, pp. 309–319 in Endless Forms: Species and Speciation, edited by D. J. HOWARD and S. H. BERLOCHER. Oxford University Press, New York.

RIESEBERG, L. H., J. WHITTON, and K. GARDNER, 1999  Hybrid zones and the genetic architecture of a barrier to gene flow between two sunflower species. Genetics 152:713-727.[Abstract/Free Full Text]

SMILEY, J., 1978a  Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora.. Science 201:745-747.[Abstract/Free Full Text]

SMILEY, J. T., 1978b The host plant ecology of Heliconius butterflies in northeastern Costa Rica. Ph.D. Thesis, University of Texas, Austin, TX.

SPERLING, F. A. H., 1993  Mitochondrial DNA variation and Haldane's rule in the Papilio glaucus and P. troilus species groups. Heredity 71:227-233.

TRUE, J. R., B. S. WEIR, and C. C. LAURIE, 1996  A genome-wide survey of hybrid incompatibility factors by the introgression of marked segments of Drosophila mauritiana chromosomes into Drosophila simulans.. Genetics 142:819-837.[Abstract]

TURELLI, M., 1998  The causes of Haldane's rule. Science 282:889-891.[Free Full Text]

TURELLI, M. and D. J. BEGUN, 1997  Haldane's rule and X-chromosome size in Drosophila.. Genetics 147:1799-1815.[Abstract]

TURELLI, M. and H. A. ORR, 1995  The dominance theory of Haldane's rule. Genetics 140:389-402.[Abstract]

TURELLI, M. and H. A. ORR, 2000  Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154:1663-1679.[Abstract/Free Full Text]

WAAGE, J. K., J. T. SMILEY, and L. E. GILBERT, 1981  The Passiflora problem in Hawaii: prospects and problems of controlling the forest weed P. mollissima (Passifloraceae) with Heliconiine butterflies. Entomophaga 26:275-284.

WU, C.-I and A. W. DAVIS, 1993  Evolution of postmating reproductive isolation: the composite nature of Haldane's rule and its genetic bases. Am. Nat. 142:187-212.

WU, C.-I, N. A. JOHNSON, and M. F. PALOPOLI, 1996  Haldane's rule and its legacy: Why are there so many sterile males? Trends Ecol. Evol. 11:281-284.

This article has been cited by other articles:

				Y. Tao, S. Chen, D. L. Hartl, and C. C. Laurie Genetic Dissection of Hybrid Incompatibilities Between Drosophila simulans and D. mauritiana. I. Differential Accumulation of Hybrid Male Sterility Effects on the X and Autosomes Genetics, August 1, 2003; 164(4): 1383 - 1397. [Abstract] [Full Text] [PDF]

				M. Beltran, C. D. Jiggins, V. Bull, M. Linares, J. Mallet, W. O. McMillan, and E. Bermingham Phylogenetic Discordance at the Species Boundary: Comparative Gene Genealogies Among Rapidly Radiating Heliconius Butterflies Mol. Biol. Evol., December 1, 2002; 19(12): 2176 - 2190. [Abstract] [Full Text] [PDF]