Natural hybridization between species provides an opportunity to study the mechanisms that maintain independent lineages and may help us understand the process of speciation. The New Zealand tree wētā species Hemideina thoracica produces F₁ hybrids where it lives in sympatry with two closely related species: Hemideina crassidens and Hemideina trewicki. This study looked at the viability and fertility of F₁ hybrid wētā between H. thoracica and H. crassidens that were collected from the wild and kept in captivity. The hybrids appeared to have normal viability from the late juvenile stage, with all male wētā maturing at a late instar. Male F₁ hybrids displayed normal mating behavior and one male produced offspring in captivity. In contrast to Haldane’s rule, female F₁ hybrids appeared to be infertile; they refused to mate and did not produce eggs. No evidence of Wolbachia infection was identified in any of the three North Island Hemideina species.

Haldane’s rule, Hybridization, introgression, sexual exclusion, Wolbachia

Natural hybridization between species provides an opportunity to study the mechanisms that maintain independent lineages and may help us understand the process of speciation (Butlin 1987, Barton and Gale 1993). Hybrid animals often have lower fitness than individuals of their parent species (Harrison 1993 and references therein) as a result of incompatible genetic combinations, such as Dobzhansky-Muller incompatibilities (Dobzhansky 1937, Muller 1942, Bolnick and Near 2005), mismatched chromosomes (Shaw and Wilkinson 1980, Searle 1993), incompatible cellular organelles (e.g. mitochondria: Ellison and Burton 2008) and symbiotic bacteria (e.g. intracellular Wolbachia infections: Bordenstein et al. 2001; gut bacteria: Brucker and Bordenstein 2013). However, low hybrid fitness also arises through natural and sexual selection when an intermediate phenotype is a disadvantage (Svedin et al. 2008).

If hybrids are viable they might nevertheless have limited fertility. If fertility of F₁ hybrids is very low, fertility levels usually improve in subsequent generations of backcross hybrids (Mallet et al. 1998, Mallet 2005, Descimon and Mallet 2009), and even limited fertility provides the possibility for introgression that can have important consequences for species interactions. Adaptive alleles that arise in one population/species can be transferred to another through hybridization, known as adaptive introgression, as seen in wing coloration in Heliconius butterflies (Pardo-Diaz et al. 2012). In some cases, when hybridization reduces fertility a unimodal hybrid (or tension) zone forms. Unimodal hybrid zones are geographically constrained; with most individuals in the zone having mixed ancestry, and the width of the hybrid zone depending on hybrid disadvantage and dispersal of the species (Jiggins and Mallet 2000). If hybridization is more limited, a bimodal hybrid zone may result, where parental forms overlap and predominate with a few individuals of mixed ancestry amongst them. Bimodal hybrid zones are typically associated with assortative mating so, in tree weta, reproductive character displacement is a likely outcome that increases assortative mating (Dieckmann and Doebeli 1999, Jiggins and Mallet 2000). Where two species compete for the same resources, hybridization may prevent one from successfully out-competing the other. Alternatively sexual exclusion, where one species (often the males of that species) out-competes the other for mates, can limit fitness through reproductive interference (Gröning and Hochkirch 2008), as observed between the tetrigids: Tetrix ceperoi and Tetrix subulata (Hochkirch et al. 2007).

Tree wētā (Orthoptera: Anostostomatidae: Hemideina) are a genus of seven nocturnal arboreal insects, with high morphological and ecological similarity (Field and Bigelow 2001, Dewhurst 2012, Bulgarella et al. 2014). They are hypothesized to have speciated in allopatry during the Pliocene or earlier (Trewick and Morgan-Richards 2005), and they now have broadly parapatric distributions (Trewick and Morgan-Richards 1995, Bulgarella et al. 2014). Hybridisation in the wild between the New Zealand tree wētā Hemideina thoracica (White) and both Hemideina crassidens (Blanchard) and Hemideina trewicki Morgan-Richards has recently been confirmed with genetic data, but so far only F₁ hybrids have been confirmed (Mckean 2014, Mckean et al. 2016). These three Hemideina species each have distinct karyotypes (i.e. different numbers of chromosomes with some differing in size and shape; Morgan-Richards 1995, 1997, 2000, Mckean et al. 2015). Karyotype differences are generally seen as presenting barriers to gene flow by disrupting meiosis and rendering F₁ hybrids infertile. However, some tree wētā species naturally comprise multiple chromosome races that are capable of interbreeding in the wild (Morgan-Richards 1997, 2000, Morgan-Richards et al. 2000, Morgan-Richards and Wallis 2003). The apparent tolerance of chromosome rearrangements displayed in this orthopteran lineage might influence fertility of interspecies hybrids. Karyotype, mtDNA haplotypes, and alleles at four nuclear DNA loci were found to differentiate parent populations of H. thoracica and H. trewicki in a large area of sympatry in Hawke’s Bay. These markers (except mtDNA, which is maternally inherited) were heterozygous in individuals who were phenotypically intermediate in abdominal coloration (orange rather than yellow or brown), abdominal bands (faint rather than striking or non-existent), abdominal stripe (a series of spots rather than a stripe or the absence of a stripe) and the number of spines on the prolateral hind tibia (typically between the three spines seen in H. thoracica and the four in H. crassidens/H. trewicki, with a half-sized medial spine on each leg being common, or three spines on one leg and four on the other). A similar situation was seen in the Manawatu area of sympatry between H. thoracica and H. crassidens, where karyotype, mtDNA and three nuclear DNA markers were found to differentiate the two species (with some introgression detected relative to allopatric populations). All individuals, which had an intermediate phenotype (the same phenotype as for H. thoracica × H. trewicki hybrids), were heterozygous for these markers (Mckean et al. 2016). Whether hybridization occurs between H. crassidens and H. trewicki is currently unknown due to the morphological similarities of these two species, and unknown distribution boundaries due in part to clearance of native forest where the two are hypothesized to have historically met (Trewick and Morgan-Richards 1995). A lack of gene flow suggests that H. thoracica × H. trewicki hybrids, which are found at a frequency of 1% of wētā in sympatry, are infertile, but genetic and morphological data suggest a low, but potentially significant, level of introgression between H. thoracica and H. crassidens, where hybrid frequency is ~3 in every 100 wētā (Mckean et al. 2016).

Introgression is the signal of past hybridization, and an ability to successfully hybridize might be of fundamental importance to the future of a species, while climates and environments continue to change (Grant and Grant 1993, Allendorf et al. 2001, Becker et al. 2013, Taylor et al. 2015, Sivyer et al. 2018). There is evidence that H. crassidens formerly occupied much of central North Island that is now the range of H. thoracica (Bulgarella et al. 2014). Isolated populations of H. crassidens remain in regions of high elevation, which suggests they are adapted to colder environments. With global warming, H. crassidens might continue to be displaced, but this depends on the ecological and sexual interaction between species. Adult tree wētā often form harems in tree cavities during the summer and autumn (Wehi et al. 2013), and mixed species harems in areas of overlap suggest that species recognition is not complete (Trewick and Morgan-Richards 1995, Wehi et al. 2017). Of F₁ hybrids collected in the wild, the majority had an H. thoracica father and H. crassidens mother (Mckean et al. 2016), which suggests sexual exclusion by H. thoracica males; as in Hemideina spp. the females do not appear to actively choose their mates. Although females will sometimes resist mating, resistance times are similar whether mating occurs or not (Field and Jarman 2001). This gives rise to the possibility that hybridization may be an important factor in the coexistence/exclusion of these species.

Haldane (1922) observed that where one sex is absent, rare, or infertile in F₁ hybrids, it is usually the heterogametic sex. In tree wētā this is the male, as tree wētā, like most Orthoptera, have an XO sex determination system where females have two copies of the sex chromosome and males one (White 1940, Morgan-Richards 1997, Morgan-Richards and Wallis 2003). Based on this, if there is a difference between the sexes, we would expect male F₁ hybrids to have lower viability and/or fertility than female F₁ hybrids.

Given their apparent tolerance of karyotype variation, the high degree of infertility in wētā might have another source. Wolbachia is an endosymbiotic intracellular bacteria that infects a large proportion of the arthropod and nematode phyla (Werren et al. 2008). In arthropods, Wolbachia is estimated to infect about 65% of species (Hilgenboecker et al. 2008) including many grasshoppers and crickets (Werren and Windsor 2000, Mandel et al. 2001, Bella et al. 2010). Wolbachia is known to manipulate the reproductive biology of many of its hosts to its own advantage (Werren et al. 2008). Some of the currently known host-reproductive manipulations include male killing, induction of parthenogenesis, feminization of genetic males, forced production of haploid individuals in haplodiploid systems, and cytoplasmic incompatibility (Werren 1997). Wolbachia is hypothesized to have a role in arthropod speciation via induction of cytoplasmic incompatibility (Werren 1998 and references therein). Wolbachia infections appear to be responsible for maintaining hybrid zones between the well-studied grasshopper subspecies Chorthippus parallelus parallelus and Chorthippus parallelus erythropus via two different forms of cytoplasmic incompatibility (Bella et al. 2010). Although recently detected in New Zealand insects (Bridgeman et al. 2018), it is not known whether the tree wētā lineage (with many hybrid zones; Morgan-Richards and Wallis 2003) contains this intracellular parasite.

Here, we describe the viability and fertility of hybrids between Hemideina thoracica and H. crassidens, using F₁ hybrids collected in the wild and held in captivity. We sought evidence of Wolbachia infections to assess whether this common intracellular parasite has potential to limit reproductive compatibility among these wētā species.

The size of H. thoracica × H. crassidens hybrids fell within the normal range expected for the parent species (with males at the larger end), and many hybrids were found as adults in the wild, therefore we have no positive evidence of hybrid inviability or abnormal development. There could be some inviability early in development, during the pre-hatching or early instar phases, but it appears that at least by the time F₁ hybrids have reached the larger instars (5^th to 7^th), they are as successful as a typical wētā of either parent species. Female tree wētā all mature at the tenth instar but males can mature at the eighth, ninth or tenth instar (Spencer 1995, Kelly and Adams 2010), resulting in a wide size range of adult males. All F₁ hybrid males in this study matured at the tenth instar, which may be important in understanding their reproductive success (if any). Male tree wētā compete for females via competition for tree cavities (resources) that females use as refuges during the day (Spencer 1995, Field 2001). Tenth instar males have much larger mandibles than eighth instar adult males, which are used during male-male competition, but which also limit their mobility, leading to the hypothesis that smaller males actively search for and mate with females which are away from their tree cavities foraging at night (Field 2001, Kelly 2004, Kelly 2006a, Kelly and Adams 2010). If male F₁ hybrids all mature at a later instar, it is unclear whether this is an advantage or disadvantage for reproductive success. As the ratios of other male wētā maturing at different instars may play a part in determining success in controlling harems, more research into reproductive success among males in this location is needed to determine the outcome of these 10^th instar hybrid males.

Our observations of mating were limited to experimental pairs (rather than harems, which are common in the wild; Wehi et al. 2013) and this might have influenced the behaviors exhibited and observed. Female wētā were not given a choice of mate, which could have also influenced mating behavior. However, mating behavior appeared to be normal for our limited sample of F₁ hybrid males when paired with adult H. thoracica and H. crassidens females. One of the hybrid males had been found in the wild with a harem consisting of two adult H. crassidens females. This male produced offspring in captivity, hence it is likely that this male, along with at least some others, are behaving in the wild in a manner typical of males from the parent species. One significant limitation of this study is the lack of control mating crosses and mating behavior comparisons for parental species from the same populations, as previous mating studies were conducted with Hemideina crassidens from southern populations (Field and Jarman 2001, Kelly 2006b, c). There have been few studies of H. thoracica mating behavior (Wehi et al. 2013) and there is a general lack of understanding of mating outcomes in this lineage (Field and Jarman 2001). Unfortunately, the triggers for embryo development and hatching are unknown for tree wētā, making laboratory crosses difficult and prone to failure, as evidenced by the control wētā embryos failing to hatch. Therefore no inferences can be drawn about the success of these crosses relative to parental crosses, but given that laboratory crosses are sometimes successful when virgin females are mated to single males, it does not appear that sperm competition has to take place in these species to induce fertility in females (Morgan-Richards 2000, Stringer 2001, present study), and can probably be ruled out as an explanation for lack of fertility.

In contrast to the males, the female F₁ hybrids did not show typical mating behavior, but this may be irrelevant to fertility if they cannot produce eggs. The lack of egg production in all five F₁ female hybrids is probably biologically important, despite the small sample, because it contrasts with that observed in adult H. crassidens females kept in the same conditions (Table 2). A lack of eggs was also never observed in more than 50 mature parent females of both species that were dissected (N.E.M. personal observation, 2012–2013). The absence of eggs suggests that F₁ females may typically be infertile, whereas at least some F₁ male hybrids are fertile, as was evident from the offspring produced in captivity.

Male F₁ hybrids being partially fertile while females are infertile contrasts with the usual variation between the sexes in reduced fertility (Haldane’s rule) and may be of interest for future research. Haldane’s rule applies across many animal taxa, including others with a XO sex determination system (Haldane 1922, and one analysis found that it applied in 99% of 223 cases of sex-specific hybrid sterility and 90% of 115 cases of sex-specific hybrid inviability (Laurie 1997). Infertility as opposed to inviability appears to be the most normal sex-skewed outcome, as heterogametic infertility is known to outnumber heterogametic inviability about 10:1 in Drosophila and mammals (Wu and Davis 1993). The mechanism behind Haldane’s rule is still unclear although two main hypotheses are X chromosome to autosome imbalance, and incompatibilities between the sex chromosomes. It is also possible that there are multiple causes underlying this phenomenon, but with so few sex determination systems for comparison a conclusive inference is elusive (Coyne 1985, Wu and Davis 1993, Turelli 1998). There is also evidence that the genetic basis of inviability in heterogametic hybrids differs from the genetic basis for infertility (Coyne 1985). Exceptions such as this may eventually shed light on why this rule applies so well to the majority of species. There are some contradictions to Haldane’s rule in other XO systems, such as the field crickets Teleogryllus oceanicus and Teleogryllus commodus (Moran et al. 2017), although whether contradictions to Haldane’s rule are more common in XO systems is unknown.

One question remaining unanswered in the present study is where the barriers to reproduction are. As bimodal hybrid zones are typically associated with pre-mating rather than post-mating barriers (Jiggins and Mallet 2000 and references therein), the situation here is somewhat unusual. As there does not appear to be assortative mating between these species pairs (Field and Jarman 2001, Morgan-Richards et al. 2001, Wehi et al. 2017), it suggests that barriers are more likely to be the result of genetic constraints. It is not known at what stage the production of F₁ hybrids is limited, but as intermediate forms are far less common than expected if the species were freely interbreeding (Mckean et al. 2016) some reproductive constraint must operate. Barriers are hypothesised to be at the post-mating pre-zygotic stage or early in development, and it is possible that the wētā use unknown behavioral mechanisms to limit interbreeding. A bimodal hybrid zone in two species of chrysomelid beetles (Chyrsochus cobaltinus and C. auratus) also involves stronger post-zygotic barriers than pre-zygotic barriers (Peterson et al. 2005), so the association of assortative mating and bimodal hybrid zones has exceptions. A later study of these same beetles also showed a significant sex-bias in the production of offspring (most had mtDNA haplotypes and hence mothers from one species), despite mating occurring in both directions in the wild, and offspring in both sex-pairings being produced in equal numbers and with equal viability in laboratory crosses at the first instar (Monsen et al. 2007). The proposed explanation was asymmetric post-mating pre-zygotic barriers, or possible asymmetric inviability later in development. As H. thoracica and H. crassidens appear to exhibit both a bimodal hybrid zone in the apparent absence of pre-mating barriers, and also a sex-biased production of F₁ offspring, there may be some similarities in the mechanisms causing reproductive isolation in these disparate species pairs. Examples such as these may give insights into how bimodal hybrid zones are typically formed and maintained.

Female infertility would prevent mtDNA passing the species boundary (introgressing), and this may explain why no evidence of mtDNA introgression has been seen in previous studies (Bulgarella et al. 2014, Mckean et al. 2016), despite evidence of a low level nuclear DNA and possible phenotypic introgression (Mckean et al. 2016). The low number of F₁ hybrids seen in the wild suggests that these wētā are forming a bimodal hybrid zone (Mckean et al. 2016), and with reduction in fertility of at least a 50% (due to female infertility), production of hybrids is probably costly. The most likely outcome in this scenario would appear to be reproductive character displacement or niche divergence limiting hybridization and loss of reproductive potential, as loss of reproductive compatibility allowing merging of the species would be unlikely with such a significant reduction in fertility (Dieckmann and Doebeli 1999, Jiggins and Mallet 2000). However, if one species uses sexual exclusion to monopolize mates of the other, it could enable for the range expansion of this species. Given that most F₁ hybrids (8/9) were shown to have a H. thoracica father, this may in part explain how H. thoracica has been able to displace H. crassidens from much of its former range as climate has warmed since the last glacial maximum (Bulgarella et al. 2014). Introgression of adaptive alleles (e.g. cold tolerance) could enable continued expansion of H. thoracica, otherwise the hybrid zone is likely to settle where environmental selection and mate competition are at equilibrium. Further work to determine if male F₁ hybrids are more likely to successfully reproduce with H. thoracica or H. crassidens females would be valuable, because a bias at this point could influence introgressive asymmetry.

Neither of the two methods employed here provided evidence of Wolbachia infection in Hemideina. The primer pairs that amplify DNA from the common Wolbachia supergroups that infect arthropods (Simoes et al. 2011), and the mapping software used with NGS were sensitive enough to detect infection by another bacterial parasite, so it is highly likely that these wētā did not contain Wolbachia although other New Zealand Orthoptera do (Bridgeman et al. 2018). A Chlamydia-like infection was detected in one of the wētā in this study however, and as this bacteria also functions as an intracellular parasite (Wyrick 2000), it may be of interest. Wētā make good candidates for sexually transmitted diseases, as they generally have some level of promiscuity and have overlapping adult generations (Knell and Webberley 2004).

Our sample of hybrid individuals was small, due to the low frequency of hybrids in the wild (Mckean et al. 2016), however, the results are biologically significant (i.e. F₁ males having at least some fertility, the infertility of female F₁ hybrids, thus being an exception to Haldane’s Rule), and raise questions about future interaction and survival of these species. In summary, both male and female F₁ hybrids are capable of reaching maturity, and although Wolbachia is not involved in limiting hybridization, there is at least a 50% (probably higher) reduction in F₁ hybrid fertility due to female infertility, which might have a strong limiting effect on introgression in the wild. There appears to be a contrast between complete failure by female F₁ hybrids to produce eggs and partial fertility of some male hybrids, suggesting this system provides another exception to Haldane’s rule that in interspecific hybrids the heterogametic sex (in this case males; XO) will have lower fertility than the homogametic sex (female Hemideina XX).

Tree wētā are an interesting group for evolutionary studies, in part because they appear to have a high tolerance for chromosome rearrangement that leads to many intraspecific hybrid zones. Much remains unknown about wētā biology, particularly with regard to species coexistence and production of hybrids where these wētā meet in sympatry lends an extra layer of complexity to the situation. Given that these species meet in different zones of sympatry across the country (and in different species combinations), there is the possibility that different mechanisms have, or will, evolve in different areas, which could be another promising area for further study.

We thank Mariana Bulgarella, Emily Koot, Shaun Neilson, Anne Kim and Priscilla Wehi for help collecting hybrid wētā. This work was supported by the Massey University Research Fund (M. Morgan-Richards; MURF2013: Can wētā coexist?). The paper was much improved by constructive comments from Mike Ritchie and two anonymous reviewers.