rationalskepticism.org

Originally posted by Calilasseia in response to the above:

And while tracking that scientific paper down, I found another one. This one freely downloadable into the bargain.

Speciation By Hybridisation In Heliconius Butterflies by Jesús Mavárez, Camilo A. Salazar, Eldredge Bermingham, Christian Salcedo, Chris D. Jiggins and Mauricio Linares, Nature, 441: 868-871 (15th June 2006) [Full paper downloadable from here]

Mavárez et al, 2006 wrote:Speciation is generally regarded to result from the splitting of a single lineage. An alternative is hybrid speciation, considered to be extremely rare, in which two distinct lineages contribute genes to a daughter species. Here we show that a hybrid trait in an animal species can directly cause reproductive isolation. The butterfly species Heliconius heurippa is known to have an intermediate morphology and a hybrid genome1, and we have recreated its intermediate wing colour and pattern through laboratory crosses between H. melpomene, H. cydno and their F1 hybrids. We then used mate preference experiments to show that the phenotype of H. heurippa reproductively isolates it from both parental species. There is strong assortative mating between all three species, and in H. heurippa the wing pattern and colour elements derived from H. melpomene and H. cydno are both critical for mate recognition by males.

The authors continue with:

Mavárez et al, 2006 wrote:Homoploid hybrid speciation—hybridization without change in chromosome number—is considered very rare2–4. This has been explained by the theoretical prediction that reproductive isolation between hybrids and their parents is difficult to achieve3,5,6. However, if a hybrid phenotype directly causes reproductive isolation from parental taxa, this difficulty can be overcome. Such a role for a hybrid phenotype has been convincingly demonstrated only in Helianthus sunflowers7. In animals, the evidence for homoploid hybrid speciation is less convincing. Putative hybrid species are known with mixed genomes8–11, but in these examples shared genetic variation could also be a result of introgression subsequent to a bifurcating speciation event.

Heliconius cydno and H. melpomene are two closely related species that overlap extensively in lower Mesoamerica and the Andes12. Speciation in these butterflies has not involved any change in chromosome number13 but is instead associated with shifts in colour patterns that generate both assortative mating and postzygotic isolation due to predator-mediated selection14–17. Heliconius cydno is black with white and yellow marks, whereas H. melpomene is black with red, yellow and orange marks. Both species exhibit strong positive assortative mating based on their wing colour patterns and also differ in habitat use18 and host plant preference19, but interspecific hybrids do occur at low frequency in the wild15. Heliconius heurippa has an intermediate wing pattern, which has led to the suggestion that this is a hybrid species1,20. Its hindwing is indistinguishable from that of sympatric H. m. melpomene, whereas the yellow band on its forewing is similar to that of parapatric H. cydno cordula. Ecologically, H. heurippa is most similar to H. cydno, which it replaces geographically in the eastern Andes of Colombia. Here we first establish that H. heurippa is currently genetically isolated from its putative parents and provide evidence that its genome is of hybrid origin. A Bayesian assignment analysis using 12 microsatellite loci scored in populations from Panama, Colombia and Venezuela divides H. cydno (n = 175), H. melpomene (n = 167) and H. heurippa (n = 46) individuals into three distinct clusters (Fig. 1). Hence, H. heurippa is genetically more differentiated than any geographic race sampled of either species. Moreover, analyses of polymorphism at two nuclear genes (Invected and Distal-less) show no allele sharing between H. cydno and H. melpomene, whereas the H. heurippa genome appears as an admixture, sharing allelic variation with both putative parental species (Supplementary Fig. 2, and C.S., C.D.J. and M.L., unpublished observations).

So, the authors begin by noting that the wing pattern of Heliconius heurippa is intermediate between that of local races of Heliconius melpomene and Heliconius cydno, and ask the question whether or not this is because Heliconius heurippa is a hybrid between individuals from those two races of Heliconius melpomene and Heliconius cydno. Suspicions that this might be the case were reinforced, when a genetic analysis demonstrated that certain genes present in Heliconius heurippa were admixtures of those found in Heliconius melpomene and Heliconius cydno, whilst the genes in question show NO such admixture in the other two species.

Moving on ...

Mavárez et al, 2006 wrote:To test the hypothesis of a hybrid origin for the H. heurippa colour pattern, we performed inter-specific crosses between H. cydno cordula and H. m. melpomene to reconstruct the steps of introgressive hybridization that could have given rise to H. heurippa. The colour pattern differences between H. m. melpomene and H. cydno cordula are determined largely by three co-dominant loci controlling the red and yellow bands on the forewing and the brown pincer-shaped mark on the ventral hindwing (see Fig. 2a)21,22. Most H. cydno × H. melpomene F1 hybrids seem intermediate to both parents (Fig. 2a), with both a yellow (cydno) and a red (melpomene) band in the median section of the forewing, whereas the ventral side of the hindwing shows a reduced brown mark intermediate between the parental species.

So, the authors produced some experimental crosses, and noticed that those experimental crosses produced individuals possessing wing pattern intermediate between those of the parents. However, they didn't just produce single-generation crosses, instead, they tested the effects that would arise from multiple crossings across several generations, and the results were extremely illuminating to put it mildly! But I'm jumping the gun here a little ... let's see what the authors have to reveal to us, shall we?

Mavárez et al, 2006 wrote:Female F1 hybrids resulting from crosses between H. melpomene and H. cydno are sterile in accordance with Haldane’s rule1,23, and thus only male F1 hybrids backcrossed to either H. cydno cordula or H. m. melpomene females resulted in offspring. Backcrosses to H. melpomene produced offspring very similar to pure H. m. melpomene, and further backcross generations never produced individuals with forewing phenotypes similar to H. heurippa (Fig. 2a). However, after only two generations a phenotype virtually identical to H. heurippa (Supplementary Fig. 3) was produced by backcrossing an F1 male to an H. cydno cordula female and then mating selected offspring of this cross (Fig. 2b). In offspring of crosses between these H. heurippa-like individuals the pattern breeds true, showing that they are homozygous for the red forewing band (BB) and the absence of brown hindwing marks (brbr) characteristic of H. melpomene, and similarly homozygous for the yellow forewing band (NNNN) derived from H. cydno. The pattern of these H. heurippa-like individuals also breeds true when crossed to wild H. heurippa (Fig. 2b), implying that pattern genes segregating in our crosses are homologous with those in wild H. heurippa.

Oh, now look at that for a spectacular set of results!

First of all, the authors crossed Heliconius melpomene with Heliconius cydno to produce F1 hybrids, then back-crossed the fertile males with females of each species. Back-crossing with Heliconius melpomene resulted in melpomene wing patterns reappearing, but back-crossing the F1 hybrids with Heliconius cydno to produce the F2 generation, then mating selected offspring of the F2 generation, produced individuals that were virtually identical to Heliconius heurippa!

But it gets even better. When the laboratory produced Heliconius heurippa analogues were mated to wild type Heliconius heurippa, they produced fertile offspring and the wing patterns bred true!.

These crossing experiments, as a consequence, constitute compellingly strong evidence that Heliconius heurippa resulted from a similar process occurring among hybrid butterflies in the wild. Not only did the authors reproduce the likely crossing sequence that produced Heliconius heurippa in the wild, thus providing a repeatable test of the relevant speciation mechanism, but the laboratory crosses were interfertile with the wild type Heliconius heurippa, further strengthening the hypothesis advanced by the authors.

Moving on ...

Mavárez et al, 2006 wrote:Furthermore, in a wild population of sympatric H. m. melpomene and H. cydno cordula in San Cristóbal, Venezuela, we observed natural hybrids at an unusually high frequency (8%), including some individuals very similar to our laboratory-produced H. heurippa-like butterflies (Fig. 2b). Microsatellite data show that these individuals have genotypes indistinguishable from that of H. cydno and must therefore be at least fifth-generation backcrosses (Supplementary Fig. 4). This shows that multiple generations of backcrossing can occur in the wild and that female hybrid sterility is not a complete barrier to introgressive hybridization. The fact that the H. heurippa pattern can be generated by laboratory crosses between H. melpomene and H. cydno, and is also observed in wild hybrids between the two species, establishes a probable natural route for the hybrid origin of H. heurippa.

Well, at this point, one is tempted to say, QED. The authors could hardly have asked for better, could they? Not only did their laboratory crosses reproduce virtually identical Heliconius heurippa analogues, that were furthermore interfertile with wild Heliconius heurippa, but they observed hybrids in the wild that included individuals matching both the wild type Heliconius heurippa and the authors' laboratory analogues!

Not satisfied with this, however, the authors then turned their attention to the next part of the speciation process, and performed some experiments to determine if an isolating mechanism was in place, which would reinforce speciation. Let's take a look at those experiments, shall we?

Mavárez et al, 2006 wrote:The next step in species formation is reproductive isolation. We therefore tested the degree to which H. heurippa is isolated from H. melpomene and H. cydno by assortative mating. No-choice mating experiments showed a reduced probability of mating in all interspecific comparisons, with H. heurippa females particularly unlikely to mate with either H. cydno or H. melpomene (Table 1). When a male of each species was presented with a single female, H. heurippa males were tenfold more likely to court their own females than the other species (Supplementary Fig. 5). In mating experiments with choice, there was similarly strong assortative mating, although occasional matings between H. cydno and H. heurippa were observed (Table 2). Isolation due to assortative mating, on average more than 90% between H. heurippa and H. melpomene and more than 75% between H. heurippa and H. cydno, is therefore considerably greater than that caused by hybrid sterility (about 25% isolation between H. heurippa and H. melpomene, and zero between H. heurippa and H. cydno)1 or predator selection against hybrids (about 50%)24. Therefore, strong assortative mating, in combination with geographic isolation from H. cydno and postzygotic isolation from H. melpomene has contributed to the speciation of H. heurippa.

So, the females of the new species, Heliconius heurippa, exhibited strong preference for other male Heliconius heurippa, with probabilities of out-crossing being 0.073 with Heliconius melpomene males and 0.022 with Heliconius cydno males. Male Heliconius heurippa again exhibited strong preference for female Heliconius heurippa, with probabilities of outcrossing being 0.1 with Heliconius melponeme and 0.44 with Heliconius cydno females. The table in the paper also demonstrates that the parent species also show strong assortative mating, though exhibit enough tendency to hybridise with each other to produce the offspring needed to generate Heliconius heurippa in the first place (hybridisation rate approximately 8%).

However, apart from mating experiments, the authors conducted some other experiments too. Let's take a look at these shall we?

Mavárez et al, 2006 wrote:We next investigated the role of colour pattern in mate choice. Experiments with dissected wings showed that both elements of the forewing colour pattern of H. heurippa were necessary for the stimulation of courtship (Fig. 3). H. heurippa males were less than half as likely to approach and court the H. m. melpomene or the H. cydno cordula pattern than their own (Fig. 3).When either the red or yellow bands were experimentally removed from the H. heurippa pattern, this led to a similar reduction in its attractiveness, demonstrating that both hybrid elements are necessary for mate recognition by male H. heurippa (Fig. 3).

So in this experiment, the authors demonstrated that visual cues are important to Heliconius heurippa, and that experimental manipulation of the wing pattern to mask certain features reduces their attractiveness as visual stimuli to mating.

Mavárez et al, 2006 wrote:Similar results were obtained when these experiments were replicated with printed-paper models (Fig. 3), showing that the colour pattern itself was the cue rather than pheromones associated with the dissected wings. Additional experiments showed that males of both H. m. melpomene and H. cydno cordula showed a greatly reduced probability of approaching and courting the H. heurippa pattern than their own (Supplementary Figs 6 and 7). Given the incomplete postzygotic reproductive isolation between all three species1, this pattern-based assortative mating must have a continuing role in generating reproductive isolation between H. heurippa and its relatives.

Nice. The above experiments established that visual stimuli reproduce the same pattern of assortative mating behaviour even in the absence of pheromones, demonstrating that visual cues are the primary means of stimulating courtship behaviour in these butterflies, and that those visual cues exert strong effects upon mate preference, leading to the assortative mating patterns seen above.

Mavárez et al, 2006 wrote:Novel patterns in Heliconius probably become established through a combination of genetic drift and subsequent fixation of the novel pattern driven by frequency-dependent selection25. Such an event could have established the hybrid H. heurippa pattern as a geographic isolate of H. cydno. Subsequently, the pattern was sufficiently distinct from both H. melpomene and H. cydno that mate-finding behaviour also diverged in parapatry, generating assortative mating between all three species (Supplementary Fig. 8). This two-stage process indicates a possible route by which the theoretical difficulty of a rapid establishment of reproductive isolation between the hybrid and the parental taxa could have been overcome5,6. Furthermore, because we are proposing divergence in mate behaviour in a geographically isolated population, reinforcement or some other form of sympatric divergence is not required for speciation to occur.

Our study provides the first experimental demonstration of a hybrid trait generating reproductive isolation between animal species, and the first example of a hybrid trait causing pre-mating isolation through assortative mating. None of the theoretical treatments of homoploid hybrid speciation have considered the effects of
assortative mating5,6. If variation for mate preference were incorporated, the theoretical conditions favouring hybrid speciation might not be as stringent as has been supposed. Finally, two other species, H. pachinus20 and H. timareta26, have also been proposed as having H. cydno/H. melpomene hybrid patterns, indicating that this process might have occurred more than once. However, whether these cases represent a particularity of Heliconius or a common natural process that has been undetected in other animal groups studied less intensively remains a matter of further study. Suggestively, other proposed cases of homoploid hybrid speciation in animals occur in well-studied groups such as African cichlids8–10 and Rhagoletis flies11.

So, the authors were able to reproduce a wild speciation event in the laboratory, produce laboratory analogues of the new species that were interfertile with wild type members of that species, and demonstrate the existence of assortative mating preferences producing a reproductive isolation barrier between the new species and the parents once the new species existed. Furthermore, this mechanism of speciation has been erected as a probable model in other well-studied groups of organisms, including those particular favourites of mine among the vertebrates, African Cichlid fishes.

I'll now see if I can obtain this latest paper and cover that in detail, though I suspect sadly it's still behind a paywall (sigh).

Thanks for the tip off, CJ, otherwise I wouldn't have found this paper either!