I enjoyed your paper!

We have empirical datasets that correspond to your theory. For example, I was one author on a paper showing "islands of divergence" between subspecies of Heliconius butterflies, where the loci that showed up as Fst peaks were known colour pattern loci.

See: Nadeau, N.J., Whibley, A.C., Jones, R.T., Davey, J.W., Dasmahapatra, K.K., Baxter, S.W., Quail, M.A., ffrench-Constant, R.H., Blaxter, M., Mallet, J., & Jiggins, C. 2012. Genomic islands of divergence in hybridizing Heliconius butterflies identified by large-scale targeted sequencing. Philosophical Transactions of the Royal Society B: Biological Sciences 367:343–353. https://doi.org/10.1098/rstb.2011.0198

Needless to say, Cruickshank & Hahn (2014) found fault with this, as well as with all other examples of islands of speciation, for sometimes spurious reasons. We had used the wrong measure of Fst, hence the regions of the genome that should have shown Fst ~ 0 were in fact shown as Fst >> 0. So actually the Fst peaks are much more real that appeared to be the case in the paper. We know this because this is a parapatric hybrid zone between mimicry colour pattern "races" that we designate as subspecies, and the two regions of the genome that show Fst peaks are in regions are colour pattern switch loci that have strong effects on survival, due to mimicry. So the pattern is real, as evidenced by the narrow hybrid zone (~10 km wide) between very divergent colour pattern forms with mobile butterflies (which move up to several km between hatching and reproducing). It's worth reading Cruickshank & Hahn's critique of Nadeau et al. to see if it makes sense. I don't think it does. When Fst ~ 0 in most of the genome, it's hard to argue that this isn't due to gene flow.

Meanwhile, in a later study from our group, Martin et al. (2013) again showed that pattern (this time with a better measure of Fst), for the aglaope/amaryllis hybrid zone studied by Nadeau et al. (their Fig. 4, top panel). However, they showed that in spite of abundant gene flow between sympatric Heliconius melpomene and H. cydno (actual per generation m < 0.001, leading to up to around 40% genomic replacement of the H. melpomene genome by H. cydno genomic information in Panama), and in spite of strong colour pattern differences, the colour pattern loci were not clearly visible against the generally highly diverged genetic background (Fig 4, panel 7 down from top). Martin, S.H., Dasmahapatra, K.K., Nadeau, N.J., Salazar, C., Walters, J.R., Simpson, F., Blaxter, M., Manica, A., Mallet, J., & Jiggins, C.D. 2013. Genome-wide evidence for speciation with gene flow in Heliconius butterflies. Genome Research 23:1817–1828. https://doi.org/10.1101/gr.159426.113 .

So you could infer from this, that Heliconius cydno and Heliconius melpomene had speciated, and produced whole-genome reproductive isolation, whereas in parapatric contact, Heliconius melpomene aglaope and H. melpomene amaryllis, had not. So what is surprising about that?

However, at the time of publication of Cruickshank & Hahn 2014, we were already sitting on another dataset, from deep in the Amazon. We've never really published this properly on its own, but it does form part of a recent paper on the hybrid origin of Heliconius elevatus. See: Rosser, N., Seixas, F., Queste, L.M., Cama, B., Mori-Pezo, R., Kryvokhyzha, D., Nelson, M., Waite-Hudson, R., Costa, M., Elias, M., Eleres de Figueiredo, C.M., Freitas, A.V.L., Joron, M., Kozak, K., Lamas, G., Pereira Martins, A.R., McMillan, W.O., Ready, J., Rueda-Muñoz, N., Salazar, C., Salazar, P., Schulz, S., Shirai, L.T., Silva-Brandão, K.L., Mallet, J., & Dasmahapatra, K.K. 2024. Hybrid speciation driven by multilocus introgression of ecological traits. Nature (London) 628:811–817. https://doi.org/10.1038/s41586-024-07263-w . See Fig.3. Heliconius elevatus and its majority parent species, H. pardalinus, co-occur across most of the southern Amazon basin, a vast region, and appear to have homogenized around 98% of their genomes (Fst ~ 0), leading to a spiky pattern of Fst "islands of speciation."

Cruickshank & Hahn were arguing even against Hahn's own Turner et al. 2005 paper which first promoted the idea of islands of speciation, on the basis of now hard-to-interpret data they obtained form Anopheles gambiae M and S forms; today recognized as An. gambiae sensu stricto, and An. coluzzi. (Turner, T.L., Hahn, M.W., & Nuzhdin, S.V. 2005. Genomic islands of speciation in Anopheles gambiae. PLoS Biology 3:e285). Since then, Turner and Hahn have been consistently engaged in trying to downplay this original argument, and the Cruickshank and Hahn paper was perhaps, they thought, the final "nail in the coffin" of the idea.

But why? I believe that they, Turner & Hahn in particulr, could not conceive of coexisting species that could sustain an amount of gene flow that would homogenize large portions of the genome, and yet leave species-determining traits intact; it made no sense in view of what they regarded as "good," reproductively isolated, species.

I think the problem seems to be that perhaps the "islands of speciation" of Turner et al. 2005 are highly transitory, on the basis of our empirical data. There is probably only a brief period where gene flow is sufficient, but also divergent polygenic selection is sufficient, to maintain Fst islands of speciation, as in Anopheles and Heliconius elevatus/pardalinus. Soon after, the species either collapse, or become "congealed" (genomic divergence everywhere), like Heliconius cydno and H. melpomene in Panama.

On the other hand, it seems to me incredibly important in the understanding of speciation that this intermediate state exists along the speciation continuum. If speciation mainly occurs by ecological divergent selection, then models like yours are relevant to speciation theory. Yes, you can have a few loci evolving, that provide local adaptation, and then you can have smaller effect loci evolving that hitch-hike along with those, and then you can have genome-wide LD among islands that helps small effect loci evolving elsewhere in the genome, long before you get total genome-wide "congealing", as Nosil et al. have put it. Heliconius elevatus and H. pardalinus are clearly separate species with different colour patterns (involved in mimicry), behaviour (including mating behaviour, and flight performance), host plant selection, pheromones, etc., etc. The F1 hybrids are very rare in the wild (although do occur, maybe 1 in 1000 matings). And yet 98% of their genome is more or less shared with Fst ~ 0, as a result of this trickle of gene flow.

As a long term warrior on behalf of the importance of hybridization and introgression, I am not fond of the terminology of "reproductive isolation", "barriers" and "barrier loci" that you use in your paper. There's a lot of gene flow between closely related psecies, and there is a continuum. Is speciation really a result of erecting barriers? That seems very mid-20th Century (think Hitler, Stalin, etc.)! Speciation is not about building walls around populations, it's about exploiting novel ecological opportunities, which may or may not result in epistatic problems with maintaining both adaptive suites in the same population. Classic "barriers" between species are by-products of speciation, not a result of the requirement of speciation to produce barriers.

Anyway, I hope you find this commentary interesting!