Genomic islands of differentiation in seedeaters are mainly the outcome of selective sweeps

New statistical methods point to several soft sweeps that acted on standing genetic variation.

When you compare the genomes of two related species, you will observe a heterogenous distribution of genetic differentiation. Some genomic regions will be very similar, while other are drastically different. In recent years, evolutionary biologists have tried to unravel the evolutionary processes underlying these differentiated genomic regions – also known as “islands of differentiation” (I have covered a few of these studies on birds, including wood-warblers, white-eyes and hummingbirds). Two main explanations are currently under debate. One model suggests that these genomic islands contain loci that contribute to reproductive isolation. When two species interbreed, these barrier loci are expected to be immune to introgression. Hence, they will diverge while the remainder of the genome is homogenized by introgression. Alternatively, local peaks in genetic differentiation might be the result of species-specific selective sweeps. To discriminate between these two explanations, the majority of studies resorted to population genetic summary statistics (e.g., Fst, Dxy, etc.). A recent study in the journal PNAS took a different approach and applied some newly developed statistical methods to this conundrum.

Ancestral Recombination Graph

In 2017, Leonardo Campagna and his colleagues compared the genomes of several Capuchino seedeaters (genus Sporophila). Their analyses uncovered 25 genomic islands of differentiation, containing genes involved in plumage pigmentation. It remained to be determined whether these genomic islands arose because of they contribute to reproductive isolation or because they were the target of species-specific selection. In the new study (led by Hussein Hejase), the researchers revisited these genomic islands with novel statistical tools.

The first method is the ancestral recombination graph (ARG), which describes both the genealogical relationships as well as the changes in those relationships along the genome due to historical recombination events. This approach was recently used to detect introgression between archaic humans, Neanderthals and Denisovans. With regard to the debate of barrier loci vs. selective sweeps, the ARG-approach can be applied to test a particular prediction involving the TMRCA. This abbreviation stands for “time to most recent common ancestor” and concerns the timepoint where two genetic samples find their common ancestor (or in jargon, when they coalesce). Recent selective sweeps are expected to reduce the TMRCA, because the genetic variants that survived the selection process can probably trace their common ancestor back to that event. Based on this reasoning, the researchers developed a statistical test to detect these species-specific selective sweeps. They found that 23 of the 25 genomic islands showed signs of recent selective sweeps in at least one seedeater species.

Example of a selective sweep involving the gene SLC45A2. The selection event results in a reduction in the TMRCA which is visible in the gene tree as a group of samples with short branches (topright figure). The topleft figure shows the situation for a neutral genomic region. From: Hejase et al. (2020) PNAS.

Machine Learning

Next, the researchers turned to machine learning. They trained a machine learning algorithm with simulated data to discriminate between selective sweeps and neutral evolution. Using this approach, they “identified large numbers of apparent species-specific sweeps, many of which coincided with Fst peaks or otherwise occurred nearby genes involved in the regulation of melanogenesis.” One important caveat is that this method is sensitive to biases in the choice of parameters for simulations. The authors have tried to cope with this potential bias by simulating various evolutionary scenarios and validating the outcomes with independent methods. Indeed, the observation that both the ancestral recombination graph and the machine learning analyses point to a preponderance of selective sweeps is certainly a good sign. All in all, it seems likely that most genomic islands of differentiation can be explained by recent, species-specific selective sweeps. However, this conclusion does not rule out the involvement of barrier loci. The authors put it nicely in the discussion.

Thus, both models likely contributed to differentiation in the regulatory sequence of this gene, but at different times and in different species. Notably, the distinction between the two paradigmatic models may not be absolute, since loci that experienced early barriers to gene flow could later undergo selective sweeps, and loci that underwent species-specific sweeps could lead to reduced hybrid fitness resulting in barriers to gene flow.


Hejase, H. A., Salman-Minkov, A., Campagna, L., Hubisz, M. J., Lovette, I. J., Gronau, I., & Siepel, A. (2020). Genomic islands of differentiation in a rapid avian radiation have been driven by recent selective sweeps. Proceedings of the National Academy of Sciences117(48), 30554-30565.

Featured image: Tawny-bellied Seedeater (Sporophila hypoxantha) © Hector Bottai | Wikimedia Commons

2 thoughts on “Genomic islands of differentiation in seedeaters are mainly the outcome of selective sweeps

  1. […] Interestingly, the genetic variants in these three genomic regions were also found in other seedeater species. The combination in the Ibera Seedeater, however, was unique for this species. The researchers noted that “this result implies that the S. iberaensis phenotype likely arose through the reshuffling of standing genetic variation that already existed within the other southern capuchinos, providing a mechanism for rapid speciation without the long period required for relevant mutations to arise de novo.” An intriguing conclusion that nicely aligns with previous work on this radiation (see for example this blog post). […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s