I recently had a chat with José Cerca about his recent piece in Molecular Ecology about how we define parallel and convergent evolution. Questions are in bold.

What got you interested in this topic?

I was always fascinated by how the environments drive evolutionary change, and parallel/convergent evolution are really fascinating as they allow us noticing natural selection and seeing what just keeps evolving across the tree of life.

Great! This paper obviously stems from getting tangled on the various definitions of parallel v. convergent evolution. Was there a project in particular that got you thinking along these lines?

Yeah! So this goes back to another MolEcol paper, which I co-first-authored with Will Sowersby ( https://onlinelibrary.wiley.com/doi/abs/10.1111/mec.16139). Will and his team were focusing on the midas’ cichlid fishes, where these thick lips (Amphilophus labiatus) were present in two different lakes. Will is an ecologist, and his data clearly showed that the A. labiatus had significantly different lips and diet than other species in the lake, A. citrinelus. When I did the genomics of the system … it was quite a mess to untangle. The PCA showed some geographical segregation of PC1 and a weak’ish species segregation on PC2 – so both A. citrinelus and A. labiatus were segregating on PC2 let’s say, ‘in parallel’. This was a mess and we had hybridization across the board, and a lot of phylogenetic mess. So when it was time to write the paper, we entered this rabbit hole: is it parallel? Is it not? In the end we said it wasn’t. A collaborator, David Marques, wrote to me personally saying that, for him, that was clearly parallel evolution … and this made me want to read up the literature and find a good answer (spoilers – I ended up agreeing with David!)

That’s pretty relatable! That’s an anxiety that we all share, right – that some one we look up to will point out a flaw in our work after publication. Even though it is nerve wracking, it is clearly constructive – it certainly has been in this case!

Yeah! Regarding flaws … One thing I liked about the ‘eLife experiment’ is seeing papers as ‘not writen in stone’. Papers can have mistakes, as long they’re not ‘unscientific’.… But ultimately, it’s always nerve wracking.

The paper largely deals with semantics, which terms to use and when. Do you think semantics are particularly important?

Yes. Totally. I think semantics are important; two reasons: first, as we diversity science. As a non-native speaker I regularly have to be constantly checking nuances on terms. Second, I think many times we could avoid unproductive discussions if we agreed our definitions are just not the same. Semantics is this thing that allows us to see whether we disagree in the mechanisms/interpretations, or just in the way we define things. I learnt this in a debating competition: sometimes just nailing down the definition of a term gets us really much further. One word I think a lot about semantics is respect: does ‘I respect that!’ mean you tolerate something, or that you look up to that? I am not really certain and I am sure if you ask different people from different social backgrounds/countries/first-languages, you’ll get different answers.

I agree with you – even if sometimes arguments around semantics can be quite draining! The argument around parallel/convergence probably gets repeated fairly often.

I really hope the paper makes a strong argument to stopping these discussions and that we can all get behind repeated evolution. As I went through different cases I kept thinking: it doesn’t matter whether we call this parallel/convergent evolution, as long as we recognize it repeatedly happens. An alternative – not explored in the paper, and which is messier – is encouraging people to state their definition of parallel/convergent on their paper introductions. But again, this will keep the mess.

I kept getting back to the cichlid work of Claudius Kratochwil. He has these elegant (natural) experiments on transposable element (TE) evolution in cichlids and I kept thinking … it’s kinda cool things just re-evolve and how his work shows the genomic basis of all these things, but perhaps we do not even care for the initial phenotype. For instance, the evolution of melanic stripes in fishes – does it really matter if before selection there were circles, or no melanin at all? Perhaps that may be cool for someone interested in phenotypes, but if your question is on molecular mechanisms – not so much; and so, you can just call it repeated evolution of stripes and move on.

Yeah, I think that would get messy! In the paper, you talk about “gene reuse” as opposed to parallel/convergent evolution. So that’s a kind of catch all. Do you think it matters that population genetic studies may not be looking at phenotypes at all?

I just really think that things get really messy at the genetic level. I kept going through this examples that are on the paper, say – if a phenotype re-evolves from different paralogs, that are 99.9% similar. Is it parallel? Is it convergent? It’s just a rabbit hole. Where do you draw the line? In the end, and mostly due to the extremely helpful review of Maddie James and editorial work of Loren Rieseberg, I just went and got gene reuse, and went for: let’s just circumscribe parallel/convergent evolution to the phenotype. I really think it became a more robust framework because of that – but I also expect that to ruffle the feathers of molecularly oriented people 🙂

I got the gene reuse umbrella from a really cool paper from some colleagues of yours in BC (https://royalsocietypublishing.org/doi/full/10.1098/rspb.2012.2146) [TB is based at UBC in Vancouver]

Right, but what if you looked at some signal of selection (e.g. sweep signals or something like that) and find patterns in homologous regions. At the phenotype perhaps there have been anti-collinear trajectories, to use your term. Does that matter?

I think the terms collinear/confluent are not mutually exclusive with convergent/parallel evolution – they just explore different dimensions: the first on how phenotypes are evolving and the latter about observed/end points.
If you find signal of selections, and experimentally validate them (we don’t do this enough in evolutionary biology, given how convenient genomics became), and find patterns in homologous genes, you can just say – hey I have parallel evolution (if you show ancestral phenotypes to be the same) with gene reuse at homologous genes. Does it make sense?

Yes, I think so. Ok, last question: What was the hardest part about this project?

There were two particular aspects. First, the semantic, because I noticed sometimes I’d send the manuscript to people and they’d be put off by statements which to me were very reasonable. When I asked them about what they meant and we sat and chatted – it ended up being about semantics (of simple things – such as similarity). Second, it was putting the framework together. If you see the acknowledgements (big shoutout to everyone there), people kept coming up with holes and with exceptions which sometimes made me have to revamp the whole thing.

Cerca, J. (2023). Understanding natural selection and similarity: Convergent, parallel and repeated evolution. Molecular Ecology. https://doi.org/10.1111/mec.17132

In a recent paper in Molecular Ecology, Shang et al. used population genomic data to characterize patterns of genetic diversity and divergence across the genomes of 8 different species of poplar. With an extensive population genomic dataset, they characterize patterns of introgression and signals of linked selection across the various species’ genomes. They found striking correlations in the landscapes of genetic variation across the species’ genomes, suggesting conserved patterns of linked selection across the divergence gradient of their sampled species.

We sent some questions to Huiying Shang and Thibault Leroy, the corresponding authors of this work, to get more detail on this study.

What led to your interest in this topic / what was the motivation for this study? 

We are fascinated by speciation, the process that generates biodiversity. An ongoing challenge in the discipline is  identifying so-called speciation genes, i.e., molecular variation that generates reproductive barriers to gene flow. One general strategy to identify these genomic regions involves scanning across pairs of genomes and looking for specific signals in variation of nucleotide diversity and differentiation across the genome. However, similar signals amongst the genomic “landscapes” can be generated by other evolutionary forces, in such a way that the landscapes are known to be highly heterogeneous and therefore complex to interpret. Relatively little remains known about the relative contribution of the different evolutionary forces and how these landscapes evolve with time. Fascinated by some pioneering works in monkeyflowers (Stankowski, et al. 2019 PLoS Biology 17) and in avian species (Irwin, et al. 2018 Molecular Ecology 27), Christian Lexer planned to use a series of Populus species pairs along a divergence gradient as models, with pairs spread from an early to a late stage of speciation.  In addition, the genus Populus has a wide geographic distribution, a small diploid genome,  and experiences extensive gene flow between some species pairs. We therefore decided to resequence 200 whole-genomes to investigate the relative contribution of different evolutionary processes to these landscapes.

What difficulties did you run into along the way? 

Our general hypothesis was the main driving force promoting genetic differentiation changes during the course of speciation. Reflected in the genome, different drivers lead to differentiated genomic footprints regarding the local variation in nucleotide diversity (π), relative (F_ST) and absolute sequence divergence (D_XY). By investigating the genomic landscapes of a series of species pairs along a divergence gradient, we thought we can  identify the main evolutionary force that contributes to the genomic landscapes of differentiation based on theoretical expectations under different evolutionary scenarios. But unfortunately, it is not as simple, in fact, linked selection played a prime role in shaping the genomic landscapes across all species pairs. Christian Lexer was the main initiator of many of these ideas for the project, the sudden loss of our colleague drove us all to see this project through in honour of his memory. We think he would have been fascinated with the scientific journey this project ultimately provided and the new opportunities opening up in speciation genomics

What is the biggest or most surprising innovation highlighted in this study? 

We found significantly positive correlations between D_XY and π in all species pairs, though the correlation coefficients became weaker as divergence increased. This was not expected, as in the early stage of speciation, we observed extensive gene flow between species and expected our results to be in agreement with a ‘divergence with gene flow’ scenario (including negative correlations between D_XY and π). Even at early stages of divergence,  our results were more consistent with a prime role of linked selection, especially background selection, in shaping the genomic landscapes.

Moving forward, what are the next steps in this area of research?

Our approach is interesting but remains limited by its empirical nature, because our interpretation is mostly based on verbal models. Methodological developments allowing new methods to compare the different scenarios – virtually allowing to quantify the associated parameters – could be extremely helpful in order to aid future interpretations.  In the future, the development of such methods, trained with extensive demo-genetic simulations, could greatly contribute to a better understanding of the effects of the divergence process on genome-wide molecular patterns.

Describe the significance of this research for the general scientific community in one sentence.

Speciation is a complex continuous process involving multiple evolutionary factors contributing to genomic differentiation, but the main force involved may vary at different stages.

Describe the significance of this research for your scientific community in one sentence.

We found that the genomic landscapes of diversity and divergence are mostly shaped by linked selection, along with gene flow and standing genetic variation, and that this result holds true independently of the stage across the divergence gradient.

Shang, H., Field, D. L., Paun, O., Rendón-Anaya, M., Hess, J., Vogl, C., Liu, J., Ingvarsson, P. K., Lexer, C., & Leroy, T. (2023). Drivers of genomic landscapes of differentiation across a Populus divergence gradient. Molecular Ecology, https://doi.org/10.1111/mec.17034

In a recent paper in Molecular Ecology, Portinha et al. used population genomic data to analyse the speciation history of two closely related species of wood ants, Formica polyctena and F. aquilonia. Using a demographic modelling approach, the authors reconstruct the history of divergence for multiple heterospecific pairs of populations. In all cases, the authors found that there was evidence for divergence with gene flow. However, for a sympatric population pair sampled in Finland there was evidence for substantially elevated gene flow between the species. Their findings imply that population genomic analysis of speciation history may be geographically variable for particular species.

We sent some questions to Beatriz Portinha and Pierre Nouhaud, the corresponding authors of this work, to get more detail on this study.

What led to your interest in this topic / what was the motivation for this study? 

Knowledge on the demographic and speciation histories is essential for understanding
contemporary genomic patterns in natural populations, which is why we wanted to
reconstruct it for the emerging Formica model system. Our study species, Formica polyctena
and F. aquilonia, are known to hybridize naturally in Southern Finland, where their hybrids
have been studied for over 10 years (Kulmuni et al., 2010; Martin-Roy et al. 2021). We
wanted to test whether a similar divergence history was consistently inferred across the
European ranges of both species, or whether the Finnish populations would stand apart,
possibly because of gene flow mediated by hybrid populations in the area.

What difficulties did you run into along the way? 

Formica polyctena and F. aquilonia had a limited genomic toolbox when we started the
project, and we initially relied on a distant and non-contiguous reference assembly.
Meanwhile, our group assembled a high quality reference genome (Nouhaud et al., 2022),
which improved the quality of our inferences.

The demographic modelling software we used, fastsimcoal2, can simulate a large panel of
evolutionary scenarios. When planning this study, we wanted to design models that
considered alternative scenarios for the divergence of the species which would be as
biologically meaningful as possible, while keeping the number of models low enough that the
project 1) would not be a huge computational burden and 2) would be executable in the
available time frame (Beatriz’s MSc. project, funded by Erasmus+ and Societas pro Fauna et
Flora Fennica). This was an especially important aspect as we used four distinct population
pairs to reconstruct the history of the two species, so each model had to be run, at least, four
different times.

What is the biggest or most surprising innovation highlighted in this study? 

We found that there was already bidirectional gene flow occurring in Finland before the
hybridization events that led to the present-day hybrid populations. This was not suspected
before, as there is no evidence in the literature, and it suggests that F. polyctena in Finland
may be admixed, which is supported by the fact that we have not found non-admixed F.
polyctena individuals in Finland.

Moving forward, what are the next steps in this area of research?

The divergence history we inferred between F. polyctena and F. aquilonia can be used to
run simulations about the evolution of the hybrid populations, which is what we did in a
subsequent work (Nouhaud et al. 2022). In the longer run, it would also be important to
extend this work by reconstructing the divergence history of the whole F. rufa species group,
which encompasses 5 species (including F. aquilonia and F. polyctena) and where gene flow
is prevalent (Seifert, 2021).

Describe the significance of this research for the general scientific community in one sentence.

Genomes from individuals sampled thousands of kilometers apart tell the same ancient
history, while their most recent history may be different.

Describe the significance of this research for your scientific community in one sentence.

The divergence history between two species can be reliably and consistently inferred from a
small number of individuals sampled across the species’ ranges.

Portinha, B., Avril, A., Bernasconi, C., Helanterä, H., Monaghan, J., Seifert, B., Sousa, V. C., Kulmuni, J., & Nouhaud, P. (2022). Whole-genome analysis of multiple wood ant population pairs supports similar speciation histories, but different degrees of gene flow, across their European ranges. Molecular Ecology, 31, 3416– 3431.

Kulmuni, J., Seifert, B. & Pamilo, P. (2010). Segregation distortion causes large-scale
differences between male and female genomes in hybrid ants. Proceedings on the National
Academy of Sciences, 107(16), 7371-7376.

Martin-Roy, R., Nygård, E., Nouhaud, P. & Kulmuni, J. (2021). Differences in thermal
tolerance between parental species could fuel thermal adaptation in hybrid wood ants.
American Naturalist, 198(2), 278-294.

Nouhaud, P., Beresford, J. & Kulmuni, J. (2022). Assembly of a hybrid Formica aquilonia× F.
polyctena ant genome from a haploid male. Journal of Heredity, esac019, 1-7.

Nouhaud, P., Martin, S. H., Portinha, B., Sousa, V. C. & Kulmuni, J. (2022). Rapid and
repeatable genome evolution across three hybrid ant populations. bioRxiv.

Seifert, B. (2021). A taxonomic revision of the Palaearctic members of the Formica rufa
group (Hymenoptera: Formicidae) – the famous mound-building red wood ants.
Myrmecological News, 31, 133-179.