Peaks and Pitfalls: How to avoid artefacts in demographic analyses

Changing default parameters can prevent false peaks in population size estimations.

In the final chapter of my PhD thesis, I explored the impact of hybridization on the evolutionary history of the True Geese (eventually published in BMC Evolutionary Biology). One of the genomic analyses involved a pairwise sequentially Markovian coalescent (PSMC) approach which uses estimates of coalescent times to infer changes in effective population size (Ne) over time. This technique – developed by Heng Li and Richard Durbin – is nicely explained by David Reich in his book Who We Are and How We Got Here:

A 2011 paper by Heng Li and Richard Durbin showed that the idea that a single person’s genome contains information about a multitude of ancestors was not just a theoretical possibility, but a reality. To decipher the deep history of a population from a single person’s DNA, Li and Durbin leveraged the fact that any single person actually carries not one but two genomes: one from his or her father and one from his or her mother. Thus it is possible to count the number of mutations separating the genome a person receives from his or her mother and the genome the person receives from his or her father to determine when they shared a common ancestor at each location. By examining the range of dates when these ancestors lived—plotting the ages of one hundred thousand Adams and Eves—Li and Durbin established the size of the ancestral population at different times. In a small population, there is a substantial chance that two randomly chosen genome sequences derive from the same parent genome sequence, because the individuals who carry them share a parent. However, in a large population the chance is far lower. Thus, the times in the past when the population size was low can be identified based on the periods in the past when a disproportionate fraction of lineages have evidence of sharing common ancestors.

The PSMC results from my study revealed that some goose species showed a marked increase in effective population size during the Last Glacial Maximum (about 110,000 to 12,000 years ago). We interpreted this pattern as a consequence of “population subdivision and occasional gene flow, leading to higher levels of heterozygosity and consequently higher estimates of Ne.” However, a recent paper in the journal Current Biology suggests that these population peaks might represent artefacts due to issues with default parameters.

Two examples from my PSMC analyses of several goose genomes that revealed dramatic peaks in recent population sizes. From: Ottenburghs et al. (2017).

Sophisticated Simulations

Leon Hilgers and his colleagues were analyzing several turtle species when they “detected a curious pattern of dramatic peaks followed by even more extreme population collapses.” This pattern occurred across distantly related turtle species from around the globe. Moreover, a literature search revealed similar patterns in countless other taxa, such as horses (where it was linked to expanding and contracting grasslands) and the Cape Buffalo (where it was explained by colonization of new areas). What is going on here?

First, the researchers examined whether the extreme peaks might reflect genuine increases in effective population size. However, when they simulated genomes based on the expected population history, no such peaks were detected. Similarly, simulations incorporating population fragmentation failed to produce extreme peaks (except in cases with high levels of post-fragmentation gene flow). Taken together, these simulation results suggest that the observed peaks are unlikely to represent real biological phenomena.

Effective population size (Ne) estimates for three turtle species with extreme Ne peaks before population collapses (indicated with asterisks). From: Hilgers et al. (2025).

Splitting Time Intervals

Perhaps the extreme peaks are due to a technical artifact related to PSMC parameter settings? Indeed, when the researchers adjusted some parameter settings, the peaks disappeared from the PSMC plots. To understand this finding, we need to dive into the details of the p-parameter which divides time into a series of intervals, after which PSMC estimates an effective population size for each interval.

The default setting for this parameter looks like this: “-p 4 + 25 x 2 + 4 + 6”. This series of numbers can be broken down into a set-up of time intervals. You start with four small intervals for recent times, followed by 25 groups of two intervals for the mid-range time. Next, there are four small intervals for somewhat ancient times. And the final six intervals are merged together in the very distant past.

The artefact could be avoided by splitting the first time window into two windows (so ‘‘-p 2 + 2 + 25 x 2 + 4 + 6’’ instead of the default ‘‘-p 4 + 25 x 2 + 4 + 6’’). This adjustment, nicely demonstrated in additional analyses of several primate species, prevents the default setting from fixing the first four intervals into a single large window. In the default model, PSMC infers one Ne for this broad window and cannot capture population changes within it. When population declines occur during this period, the model thus may overcompensate by inflating Ne estimates in the preceding window. Splitting the first window into two can circumvent this issue.

Analyses of several primate species illustrate how the artefact disappears by splitting the first window (default setting = 4, red line) into two windows (setting = 2+2, yellow line). From: Hilgers et al. (2025).

A Call for Caution

This study underscores the need for careful evaluation of complex analytical methods. New genomic tools are often rapidly applied to preferred study systems (such as the True Geese, in my case), but their outputs warrant cautious interpretation. Dramatic increases in effective population size might reflect real environmental or climatic events, or they can be artefacts of the method. Researchers should therefore invest time to thoroughly understand a method’s assumptions and limitations before applying it. Not an easy choice under the time pressure of today’s “publish-or-perish” environment.

References

Hilgers, L., Liu, S., Jensen, A., Brown, T., Cousins, T., Schweiger, R., Guschanski, K. & Hiller, M. (2025). Avoidable false PSMC population size peaks occur across numerous studies. Current Biology, 35(4), 927-930.

Ottenburghs, J., Megens, H. J., Kraus, R. H., Van Hooft, P., van Wieren, S. E., Crooijmans, R. P., Ydenberg, R. C., Groenen, M. A. M. & Prins, H. H. T. (2017). A history of hybrids? Genomic patterns of introgression in the True Geese. BMC Evolutionary Biology, 17(1), 201.

Featured image: PSMC plot from Ottenburghs et al. (2017) BMC Evolutionary Biology

Do plumage patterns predict hybridization between subspecies of the Variable Seedeater?

Comparing contact zones leads to a surprising conclusion.

Let’s start with a thought experiment. Imagine three avian subspecies: one entirely black, and two others with a pied black-and-white plumage pattern. Which of these subspecies do you think would be most likely to hybridize? You might predict that the two pied subspecies will interbreed, while any combination involving the black subspecies will not. That is a perfectly logical prediction if you assume that the birds will recognize potential mates based on shared plumage patterns. A nice example of how plumage can function as a premating isolation barrier.

This very set-up (one black subspecies and two pied ones) is not merely hypothetical. It occurs in the Variable Seedeater (Sporophila corvina), a small passerine bird that breeds from southern Mexico through Central America to the northwestern regions of South America. A recent study published in the journal Molecular Ecology put this prediction to the test, examining three independent contact zones where different pairs of subspecies meet and potentially interbreed.

Three Contact Zones

Time to put some names on our hypothetical subspecies. The three Variable Seedeater taxa occur in close proximity across Costa Rica and Panama. The nominate subspecies (S. c. corvina) is almost entirely black and inhabits the Caribbean slope, ranging from northeastern Costa Rica through central Panama. Along the Pacific slope, from central Costa Rica to eastern Panama, you can find the two pied subspecies (S. c. hoffmanni and S. c. hicksii) which are both characterized by white collars, bellies, and rumps. They differ mainly in the extent of a white patch on the throat.

These subspecies meet in several regions where interbreeding may occur. A well-documented hybrid swarm exists between S. c. corvina and S. c. hicksii in central Panama, while the contact zone between S. c. hoffmanni and S. c. hicksii appears to be a more gradual intergradation along the Pacific slope in the Panamanian Veraguas Province. Finally, recent observations from the Central Valley in Costa Rica point to a third potential contact zone, where populations of S. c. hoffmanni and S. c. corvina come into secondary contact.

Distribution of the thee subspecies across Costa Rica and Panama. Different pairs of subspecies meet in three independent hybrid zones. From: Ocampo et al. (2023).

Hybrid Triangles

Diego Ocampo and his colleagues collected genetic samples from each of the three contact zones. Using a comprehensive suite of molecular techniques, they quantified the extent of hybridization among the subspecies. One particularly insightful method – which I like to call “hybrid triangles” – combines measures of heterozygosity and hybrid index to visualize an individual’s genetic make-up. In this triangular space, the two parental subspecies occupy the lower corners, while first-generation hybrids cluster toward the top. The sides and interior of the triangle represent various types of backcrosses, illustrating the continuum of genetic mixing between “pure” and admixed individuals.

The genetic analyses revealed some intriguing and unexpected patterns. The researchers detected extensive hybridization, primarily involving later-generation backcrosses, in the contact zone between the black S. c. corvina and the pied S. c. hicksii. Similarly, substantial genetic mixing occurred between the two pied subspecies, S. c. hoffmanni and S. c. hicksii. In contrast, there was no clear evidence of hybridization between S. c. corvina (black) and S. c. hoffmanni (pied). Not exactly the pattern we predicted from our initial thought experiment.

Hybrid triangles – combining heterozygosity and hybrid index – show the genetic composition across the three contact zones. Parental individuals are located in the lower corners, whereas first-generation hybrids can be found at the top. Backcrosses are scattered in the interior of the triangle. From: Ocampo et al. (2023).

Premating Isolation

The researchers reported “two different outcomes in two contact zones with similar patterns of plumage divergence (i.e., black vs. pied plumage), suggesting that factors other than just plumage color may also shape the different patterns of hybridization found.” If not plumage, then what other behavioral mechanisms could explain the differing dynamics across the contact zones? In their discussion, the authors propose that divergence in song or breeding phenology might serve as premating barriers.

Overall, our results suggest that divergence in plumage colour is important in reducing gene flow between these populations, but is not sufficient to stablish complete reproductive isolation. Other factors, such as reproductive timing or divergence in song, might explain why hybridization is reduced in one of the contact zones but not in the others, a pattern of isolation that is likely to change with time.

However, a follow-up study on song structure and playback responses revealed that song alone is not a strong isolating mechanism. This leaves differences in the timing of breeding as a promising alternative explanation, although this hypothesis awaits formal testing.

Overall, this study highlights the difficulty of predicting the outcome of secondary contact. To hybridize or not to hybridize? That is the question.

References

Ocampo, D., Winker, K., Miller, M. J., Sandoval, L., & Uy, J. A. C. (2023). Replicate contact zones suggest a limited role of plumage in reproductive isolation among subspecies of the variable seedeater (Sporophila corvina). Molecular Ecology, 32(13), 3586-3604.

Ocampo, D., Sandoval, L., & Uy, J. A. C. (2025). Weak premating reproductive isolation despite divergence in secondary sexual traits in the variable seedeater. Animal Behaviour, 221, 123072.

Featured image: Variable Seedeater (Sporophila corvina) © Mdf | Wikimedia Commons

Are polar bears “re-writing their DNA” in response to climate change?

More work is needed to confirm these speculations.

We are currently living in an attention economy where capturing someone’s attention often matters more than producing high-quality work. This incentive has given rise to shallow TikTok videos, annoying Instagram influencers, and clickbait headlines on news websites. Unfortunately, I fear that science journalism is also becoming a casualty of this battle for online attention.

This week, I came across several news articles claiming that “polar bears could be rewriting their DNA to adapt to warmer climates” (BBC) or that “a statistically significant link has been found between rising temperatures and changing DNA” (The Guardian). Such bold claims immediately set off my alarm bells. When headlines report dramatic findings, the most reliable approach is to go back to the source: read the original scientific paper and evaluate the evidence for yourself.

Transposable Elements

The strong claims in these headlines are based on a recent paper published in the journal Mobile DNA. In this study, the researchers analyzed transcriptome data from polar bear subpopulations in two regions of Greenland: the cooler North-East (NEG) and the warmer South-East (SEG). Their analysis focused on the activity of transposable elements in both populations.

These so-called “jumping genes” are DNA sequences that can move around the genome via copy-and-paste or cut-and-paste mechanisms. Because their movement is largely random, some transposable elements may insert themselves near protein-coding genes, where they can potentially alter gene expression. For this reason, transposable elements are often considered a possible source of rapid evolutionary adaptation. I have previously covered some avian examples, such as crows and warblers.

Based on their analyses, the authors argue that transposable elements have contributed to the adaptation of the southeastern polar bear population to warming temperatures. However, as Carl Sagan famously said, “Extraordinary claims require extraordinary evidence.” Personally, I do not find the evidence presented in this study particularly convincing.

Temperature Effect?

The researchers report more than 1,500 transposable elements that are differentially expressed between the two polar bear populations. When temperature was added as a covariate in their analyses, they concluded that “samples clustered based on temperature.” However, upon examining the associated PCA figure, I do not observe clearly separated clusters. More importantly, this claim is not supported by any formal statistical testing. For instance, a PERMANOVA analysis could have assessed whether the purported clusters differ significantly from one another.

Furthermore, the authors focus exclusively on temperature as an explanatory variable. However, it is highly likely that the environments of the two populations differ along many additional dimensions. Why not include other factors such as food availability or snow cover to evaluate which environmental variables best explain the observed differences in the expression of transposable elements? By considering temperature alone, the analysis presents an overly narrow (and potentially biased) interpretation of the data.

Principal Component Analysis of SEG and NEG polar bear populations including temperature and location. From: Godden et al. (2025).

Positive Selection?

Irrespective of the issues surrounding temperature, I am also unconvinced by the supposed adaptive role of the transposable elements. While the researchers report several differentially expressed genes, they do not test whether these genes have been subject to positive selection. Given the short evolutionary timescale – these populations diverged roughly 200 years ago – one would expect particularly strong and detectable signatures of selection if these genetic changes were truly adaptive.

Additionally, the study does not directly estimate the timing of transposable element activity. The authors state that “we observed a peak of younger, diverged LINEs in the SEG population of polar bears compared to the reference TEs present in the genome,” but they do not assign a concrete timeframe to this observation. If this apparent burst of transposable element activity occurred thousands or even millions of years ago, the claim that it reflects adaptation to recent climate change would be severely weakened.

Finally, the researchers rely on Gene Ontology (GO) term enrichment analysis to pinpoint the putative functions of the differentially expressed genes. I have never been a fan of this approach. You will always end up with a list of possible processes for which you can develop an adaptive story. For instance, the researchers link terms related to “oxidative stress and lipid peroxidation” to polar bear diet and pregnancy, while genes associated with “gated channel and chloride channel activity” are interpreted in the context of thermoregulation. These explanations might be plausible, but they fall short of providing hard evidence for adaptation. At best, such genes should be viewed as candidates for future functional studies, not as definitive proof of adaptive genetic change.

Genetic divergence of several transposable elements compared to a reference copy allowed the researchers to estimate the relative timing of their activity. The lower divergence of some LINEs in the SEG population suggest a recent burst in activity. From: Godden et al. (2025).

Jumping to Conclusions

Taken together, this study offers an interesting first glimpse into the diversity and activity of transposable elements in polar bear genomes. However, the claim that these “jumping genes” are actively helping polar bears adapt to warming temperatures is not supported by the analyses. It would indeed be exciting if future studies were to substantiate these speculations, but until then, we should be very cautious in jumping to conclusions.

One aspect that particularly concerns me is the casual way the scientists misrepresent evolution in their communication with the media. In The Guardian, for example, the lead author claims that “polar bears in the warmest part of Greenland are using ‘jumping genes’ to rapidly rewrite their own DNA,” which misleadingly suggests that the bears are actively modifying their genomes in response to their environment. That is not how evolution works. Such phrasing reinforces a deeply ingrained misunderstanding of evolutionary theory, implying foresight or intent where none exists. We need to do better in explaining how evolution really works.

References

Godden, A.M., Rix, B.T. & Immler, S. (2025) Diverging transposon activity among polar bear sub-populations inhabiting different climate zones. Mobile DNA 16:47.

Featured image: Polar bear (Ursus maritimus) © Arturo de Frias Marques | Wikimedia Commons

Water off a manakin’s back: How sexual and natural selection shape plumage coloration in manakins

Plumage coloration mechanisms shed light on introgression patterns.

The main title of my PhD thesis, Crossing Species Boundaries, referred to the exchange of genetic material between different goose species. With the advent of genomic data, it has become clear that this process – technically known as introgression – is widespread across the Tree of Life. In some cases, introgression even extends beyond a single species boundary: some genes can move through multiple species, with one species acting as a genetic bridge between others (see this blog post on Darwin’s finches). A recent study in the journal Science Advances traced the journey of the genetic variant of the BCO2 gene as it passed through three different manakin species (genus Manacus).

The most likely scenario, therefore, is one in which the BCO2 allele for pigmented collars arose once in M. aurantiacus, was subsequently transferred through hybridization to vitellinus, spread throughout its range, and then partially spread into a third species, M. candei, in western Panama.

The BCO2 gene influences collar coloration in manakins, producing a spectrum that ranges from white to yellow to orange across species. Previous studies have demonstrated that females prefer males with yellow collars, and that these yellow-collared males display higher levels of aggression. Since collar coloration plays a role in both female mate choice and male–male competition, yellow coloration (and the genes underlying it) represents a prime target of sexual selection.

Feather Microstructures

Taken together, these findings help explain how the BCO2 variant spread across three manakin species under the influence of sexual selection. However, the researchers also uncovered an intriguing twist: yellow collars may come at a fitness cost. To explore the potential role of natural selection, it is necessary to first understand the different mechanisms that shape plumage coloration in these birds.

Additional analyses revealed that a single carotenoid pigment (lutein) is responsible for the yellow color in males, with pigment concentration closely correlating with collar coloration. Variation in other pigments also played a role: the presence or absence of melanin accounted for additional patterns, such as the olive-green belly observed in some manakin species, which itself is a target of sexual selection. Finally, features of feather microstructure further modified plumage appearance, adding another layer of complexity to coloration.

Microscopic analyses of different manakin species (and their hybrids) revealed feather microstructures that influence plumage coloration. From: Lim et al. (2024).

Water Repellency

The combination of lutein pigmentation and feather microstructure may influence water repellency, an essential trait for species inhabiting humid tropical forests. To test this, the researchers applied a model from textile science to estimate how well manakin feathers shed water. Their results showed that “M. vitellinus and M. aurantiacus, the two species with carotenoid-pigmented collar feathers and accompanying microstructural modifications, had the poorest ability to shed water when compared to the species with white collars, while hybrids showed intermediate water shedding ability.”

This mechanism may help explain the introgression pattern of collar width across the hybrid zone from M. vitellinus into M. candei. In M. vitellinus, the pigmented collar is considerably narrower than the broad white collar of M. candei. The researchers speculate that this narrower hindneck collar could mitigate the reduced water repellency associated with lutein-pigmented feathers, thereby limiting potential fitness costs. In this way, natural selection may favor reduced collar width, driving the introgression of the narrow vitellinus-like collar into candei. An interesting idea that remains to be tested.

Differences in collar width between M. vitellinus, M. candei, and their hybrids. From: Lim et al. (2024).

Tricky Trade-off

These findings highlight how plumage coloration in manakins emerges from the dynamic interplay of sexual and natural selection. On the one hand, sexual selection favors bright yellow collars that enhance male attractiveness and competitiveness, driving the spread of associated genetic variants across species. On the other hand, natural selection imposes constraints through carotenoid-based pigmentation and feather microstructure that reduce water repellency in humid environments. Ultimately, the balance between sexual and natural selection gives rise to the striking plumage patterns of these manakins. And understanding the underlying mechanisms makes you appreciate these beautiful birds even more.

References

Bennett, K. F., Lim, H. C., & Braun, M. J. (2021). Sexual selection and introgression in avian hybrid zones: Spotlight on Manacus. Integrative and Comparative Biology, 61(4), 1291-1309.

Lim, H. C., Bennett, K. F., Justyn, N. M., Powers, M. J., Long, K. M., Kingston, S. E., et al. (2024). Sequential introgression of a carotenoid processing gene underlies sexual ornament diversity in a genus of manakins. Science Advances, 10(47), eadn8339.

McDonald, D. B., Clay, R. P., Brumfield, R. T., & Braun, M. J. (2001). Sexual selection on plumage and behavior in an avian hybrid zone: experimental tests of male‐male interactions. Evolution, 55(7), 1443-1451.

Parsons, T. J., Olson, S. L., & Braun, M. J. (1993). Unidirectional spread of secondary sexual plumage traits across an avian hybrid zone. Science, 260(5114), 1643-1646.

Stein, A. C., & Uy, J. A. C. (2006). Unidirectional introgression of a sexually selected trait across an avian hybrid zone: a role for female choice?. Evolution, 60(7), 1476-1485.

Featured image: Golden-collared Manakin (Manacus vitellinus) © Francesco Veronesi | Wikimedia Commons

Genetic time travel with three enigmatic bird species from New Zealand

What can ancient DNA tell us about the evolutionary history of these birds?

I love a good time travel story, such as Back to the Future, Interstellar or the Terminator movies. Recently, I picked up Julian May’s Saga of the Pliocene Exile, a book series in which a group of 22nd-century outcasts escape their present by stepping through a time-portal into Earth’s Pliocene epoch (between 5 and 2.5 million years ago). Of course, real time travel remains beyond our reach, and reviving extinct species is still more hype than reality (despite dubious claims from the biotech company Colossal Biosciences, see this blog post). However, we’re not entirely without options: by analyzing genetic data, we can reconstruct events in the distant past and still experience some form of scientific time travel.

In this blog post, we will journey to New Zealand to trace the evolutionary stories of three remarkable birds: two extinct species – the Eastern Moa (Emeus crassus) and the Moho (Porphyrio mantelli) – and one survivor – the South Island Takahē (Porphyrio hochstetteri) – which is now clinging on as a relict population. What insights can genetic data reveal about their deep evolutionary past?

The Story of the Eastern Moa

Let’s begin with the Eastern Moa, a giant flightless bird that disappeared around the year 1400. Alexander Verry and his colleagues extracted DNA from museum specimens and sequenced the mitochondrial genomes of 46 individuals to reconstruct its evolutionary history. Their findings suggest that Eastern Moa endured the climate shifts of the Pleistocene within a single refugium, from which it later expanded after the Last Glacial Maximum.

Several lines of evidence support this scenario. First, the mitochondrial DNA revealed no deeply divergent lineages, indicating survival in one refugium rather than several. Second, samples from the proposed refugial area contained the highest levels of genetic diversity, which is a hallmark of long-term persistence. As populations spread into new regions, diversity typically declines, only to rise again as their numbers grow. Consistent with this pattern, the researchers observed a clear increase in genetic diversity when comparing Pleistocene and Holocene samples.

A map of the South Island of New Zealand showing samples and sites inside (purple) and outside (green) of the putative refugial area occupied by eastern moa during the Last Glacial Maximum (purple shading). The accompanying network illustrates the relationships between the mitochondrial haplotypes. From: Verry et al. (2022).

The Story of the Moho and Takahē

After our stop in the Pleistocene, we jump further back in time. About four million years ago, a bird arrived in New Zealand which would later give rise to two flightless species: the Moho and the South Island Takahē. Fast forward to much more recent times, where both species faced dramatic declines following human arrival: first with Polynesian settlement between 1250 and 1300, and later with European colonization in the 18th century. One species went extinct, while the other managed to persist.

In contrast to the sustained human population pressure in the North Island that likely led to the extinction of moho, the reduced human pressure in the southern South Island apparently allowed takahē to persist as a small isolated population in the remote, mountainous Fiordland region. 

The human impact is clearly visible in the DNA of the South Island Takahē. A temporal comparison of mtDNA showed a marked loss of genetic diversity, coinciding with the human arrival about 750 years ago (and an additional decline across the Pleistocene-Holocene transition).

Temporal haplotype networks of mitochondrial DNA across three discrete time periods (based on subfossil bones, museum specimens/skins and contemporary samples) show a clear decline in genetic diversity for the South Island Takahē. Each circle corresponds to a distinct haplotype with the size proportional to its frequency. From: Verry et al. (2024).

A Molecular Time Machine

These studies show how ancient genomes allow scientists to peer into past populations, retrace their evolutionary journeys, and even uncover patterns that are invisible in the fossil record. Who needs a time machine when you have DNA?

References

Verry, A. J., Mitchell, K. J., & Rawlence, N. J. (2022). Genetic evidence for post-glacial expansion from a southern refugium in the eastern moa (Emeus crassus). Biology Letters, 18(5), 20220013.

Verry, A. J., Mas‐Carrió, E., Gibb, G. C., Dutoit, L., Robertson, B. C., Waters, J. M., & Rawlence, N. J. (2024). Ancient mitochondrial genomes unveil the origins and evolutionary history of New Zealand’s enigmatic takahē and moho. Molecular Ecology, 33(3), e17227.

Featured image: South Island Takahē (Porphyrio hochstetteri) © Bernard Spragg | Wikimedia Commons