Inversions, not introgression, behind supergenes in Atlantic cod

In Atlantic cod, migratory and stationary ecotypes are separated by supergenes – tightly linked sets of genes that together encode a complex phenotype.
Inversions, not introgression, behind supergenes in Atlantic cod

In the coastal waters around the rugged Lofoten islands in Northern Norway, two types of Atlantic cod co-occur seasonally: A stationary ecotype that lives there year-round, and a migratory ecotype that spends most of the year in the Arctic Barents Sea but comes down to the Lofoten islands during the spawning season. It is this massive migration by millions and millions of fish, travelling on the order of a thousand kilometers, that is driving one of the planet's most productive fisheries, the Norwegian cod fishery. But what is driving the migration?

Atlantic cod (WaterFrame / Alamy Stock Photo)

It turns out that the different lifestyles of stationary and migratory cod ecotypes are associated with so-called supergenes; large regions on the genome with tightly linked genes that interact and are thus jointly able to encode complex phenotypes – such as migration across vast distances. The presence of supergenes in Atlantic cod had been suggested by earlier studies based on limited numbers of genetic markers, but otherwise, little was known about them when a team of researchers at the University of Oslo began to apply high-throughput genome sequencing to research the biology of Atlantic cod. Together with colleagues, Kjetill S. Jakobsen, Sissel Jentoft, and Bastiaan Star had secured funding from the Norwegian Research Council to sequence hundreds of Atlantic cod genomes to investigate the genetic basis of phenotypic variation in this species. The funding also covered a postdoc position to which I successfully applied, so that I became part of the cod research group at the University of Oslo and got my hands on the wealth of sequencing data that they had produced.

The data puzzled me at first. The supergenes were clearly visible as regions of strong differentiation between the ecotypes. But whether these regions were the result of massive inversions suppressing recombination, or of introgression from another species, was not clear to me at this stage. My first attempt at distinguishing between these two possibilities was rather unconventional: I polled the participants of the 2015 Congress of the European Society for Evolutionary Biology (or at least those that I managed to attract to my poster). The poll ended in favor of inversions, which received ten votes, while two participants voted for introgression and one voted for a combination of the two.

My poster at the 2015 Congress of the European Society for Evolutionary Biology

The second attempt was more laborious, but also more conclusive: To test for the presence of inversions, our team produced a long-read based genome assembly for a stationary Atlantic cod that could then be compared to the already existing assembly for a migratory Atlantic cod. And to rule out introgression, I expanded the dataset by including previously published genome sequences of outgroup species. The comparison of the two genome assemblies clearly supported the presence of inversions, while the larger dataset revealed that introgression from other species had nothing to do with the origin of supergenes in Atlantic cod. Thus, those ESEB Congress participants that had voted "Clearly inversions" were right, and I could simply have trusted their collective expertise instead of running more analyses.

However, as we report in our paper in Nature Ecology and Evolution, the larger dataset also had other benefits, as it now also allowed me to date the origin of the supergenes based on age estimates for outgroup divergences. And it turned out that the supergenes were surprisingly old, with ages up to almost 2 million years. Such old supergenes are not predicted by theory, as, on the one hand, the suppression of recombination in supergene regions is expected to lead to an accumulation of deleterious mutations, and on the other hand, processes like gene conversions and double crossover are expected to erode the differentiation between supergene haplotypes over time. But while we found evidence for both gene conversion and double crossover, we found no signs of any erosion of the differentation, and also no evidence for any accumulation of deleterious mutations. This suggests that the level of gene conversion and double crossover may be just at the sweet spot  in Atlantic cod: high enough to prevent the accumation of deleterious mutations, but low enough to allow the persistence of differentiation.