A small share of the genome held the clearest signal
Researchers studying six populations of H. erato found a pattern consistent with reproductive isolation beginning to develop while gene flow continues. Genetic differentiation, changes in accessible chromatin and structural variation appeared across the genome, leading the authors to describe a possible multilocus architecture rather than a barrier centered on one DNA site. The study does not establish that any of these changes causes the populations to stop mating or producing viable offspring.
The team built its comparison around a pangenome aligned across the six populations: demophoon, hydara, notabilis, etylus, favorinus and chestertonii. It examined adjacent population pairs in hybrid zones alongside non-hybrid-zone comparisons, then combined genome-wide Fst scans, ATAC-seq profiles and forward simulations. Here, Fst is a measure of genetic differentiation between populations, while ATAC-seq was used to compare regions of chromatin that were accessible in the samples.
The Fst scan used nonoverlapping 1-kilobase windows, grouped adjacent windows into blocks of 10 and required two separate 0.99-quantile criteria for a candidate signal.
It identified 91 elevated-Fst regions covering 2.943 megabases, or 0.427% of the pangenome, in demophoon-hydara. In etylus-notabilis, it identified 76 regions covering 2.916 megabases, or 0.423%. Both sets also had elevated Dxy, a measure of sequence difference between populations, relative to background. The authors describe that combination as consistent with reduced gene flow.
The regions contained 54 genes in the first comparison and 88 in the second, including the wing-pattern genes WntA, optix and cortex. That distribution is central to the authors' argument that early reproductive isolation can have a multilocus basis. The gene annotations remain candidates, however, rather than direct functional evidence.
The regulatory evidence was broader and less tidy
For the main ATAC-seq analysis, a peak had to be present in all three samples from a group and meet a reciprocal minimum overlap of 50%. The study also used a two-out-of-three peak list for supplementary analyses. The strict main rule favored reproducibility, but the authors note that it can miss variable regulatory peaks.
Under the conservative definition, unique peaks made up just 1.48% of all peaks, or 439. Most, 68.82%, were polymorphic, meaning they varied across populations. The result does not support a picture in which most regulatory regions are exclusive to one population.
In the two hybrid zones, the overall Fst distributions showed no significant shift associated with differentiated ATAC-seq peaks. The KS test, which compares the shapes of two distributions, gave a D statistic of 0.06 with an adjusted P value of 0.2 for demophoon-hydara, and a D statistic of 0.07 with an adjusted P value of 1 for notabilis-etylus. The study therefore found no consistent genome-wide association between chromatin-accessibility differences and elevated genetic differentiation in those zones.
Some non-hybrid-zone comparisons produced a different result. The shift was significant for demophoon-favorinus, where D was 0.11 and P was below 0.01, and for demophoon-etylus, where D was 0.13 and P was below 0.01. Demophoon-chestertonii showed no significant shift, with D 0.01 and P 0.46. The uneven pattern is compatible with the idea that regulatory divergence becomes easier to detect when gene flow is reduced, but it does not show that reduced gene flow caused the chromatin differences.
The genomic signals also overlapped less than a reader might expect. Only five of 385 differentiated ATAC-seq peaks, about 1%, overlapped candidate Fst regions in demophoon-hydara, while none of 380 overlapped in the other hybrid zone. The authors point to the different genomic scales measured by the two methods as one possible reason.
A second look reduced the apparent exclusivity
A further check asked whether apparent population-specific regulatory sequences held across multiple individuals, rather than arising from a single reference comparison. Of the original 439 candidate peaks, only 40, or 9.1%, met the study's criteria for population-specific nonhomology. The reassessment classified 75% as shared, 23% as variable, fewer than 1% as uninformative and 0.5% as exclusive. The retained 40 peaks averaged 51.2% overlap with transposable elements.
That reassessment is a warning against treating every apparent population-specific peak as a fixed difference. Most candidates were shared or variable once more individuals were considered, and the exact number of nonhomologous regulatory elements remained approximate because of sampling and reference bias. The result supports the authors' broader claim that single-reference analyses can overestimate fixed population-specific variation.
What the signals can and cannot tell us
Forward simulations supplied another qualification. Fst-based detection of barrier loci was most informative at intermediate divergence. High gene flow reduced the contrast with background differentiation, while advanced divergence did the same. A weak or absent signal, including the nonsignificant chestertonii comparison, cannot by itself show that regulatory barrier loci are missing.
Taken together, the study supports a picture of incipient reproductive isolation built from many genetic, regulatory and structural differences, with detection depending partly on gene flow, sampling and the scale of the method. It does not identify a gene, peak or structural variant that causes mating failure, infertility or poor hybrid viability. Direct reproductive measurements and functional tests would be needed to connect these molecular candidates to actual barriers.
The article is peer-reviewed and open access. Its authors state that the data and code needed to evaluate and reproduce the work are available in the paper, supplementary materials and cited repositories, with custom code on Zenodo and GitHub and sequencing data in NCBI BioProjects.
Paper data and sources
Original title: Multilocus basis of incipient reproductive isolation in hybridizing populations is revealed by pangenomic and epigenetic divergence.
Authors: Angelo A Ruggieri, Francesco Cicconardi, Nicolò Bellin et al.
Journal/Repository: Science advances
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1126/sciadv.adz6665
Original paper