Another paper on Introgressions (April 2026)

Continuing with the wide variety of introgression / admixture papers published over the past few years, today I add a new preprint (in Biorxiv, and therefore not peer-reviewed) published a few days ago, on April 12, 2026: Inferring hominin history with recurrent gene flow from single unphased genomes and a two-locus statistic. Nicholas W Collier, Simon Gravel, Aaron P Ragsdale. bioRxiv 2026.04.11.717825; doi: https://doi.org/10.64898/2026.04.11.717825

Through the use of a very particular statistical model (described at the beginning of the paper, and well over my statistical abilities to understand), and genetic analysis of the autosomal DNA, the authors suggest a population structure and admixture, and population sizes for modern humans, Neanderthals, Denisovans, and super-archaics that mix to and fro over the past million years. The paper assumes "a fixed mutation rate of 1.3 × 10−8 per bp per generation and a generation time of 29 years" (I have previously posted about mutation rate, its variability, and generation times and the combined effect of them on calculating timelines.

The paper reports the following events and dates:

• Neandertal-Denisovan Common ancestor or ND lived from 779 to 726 kya and lasted for ~50.000 years.
• Ancestral Neanderthals or AN that around 123 kya split into Altai people in Siberia - who later became extinct, and the Western Neandrthals or WN
• Anatomically Modern Humans or AMH introgressed into AN 250 kya ago, and 110 kya into the western Neanderthal (WN) group, which later evolved into the Croatian Vindija and Chagyrskaya (Siberia) lineages.
• Denisovans received gene flow from a ghost lineage, a "Superarchaic" S that may be Homo erectus, it had split from our ancestors 2 million years ago.
• They reckon that the Ust’Ishim people from East Central Siberia dated to around 45 ky were the first humans in Eurasia to split from the other branches after the Out of Africa Event.

The arrows in the chart show the introgression: "broken one-headed arrows denote instantaneous gene flow events; solid double-headed arrows denote continuous gene flow." The percentages, and population sizes (N_e) are also represented:

Figure 6: Early hominin history in Eurasia with recurrent gene flow. From Nicholas W Collier, Simon Gravel, Aaron P Ragsdale, 2026.

The timeline is the following:

TND→AMH (ky) AMH–ND split time 798 CI: 748 – 827
TAN→Den (ky) AN–Denisova split time 688 CI: 639 – 734
TWN→Alt (ky) WN–Altai split time 123 CI: 117 – 137
TCha→Vin (ky) Chagyrskaya–Vindija split time 60.5 CI: 57.3 – 67.7
TYor→OOA (ky) Yoruba–OOA split time 56.9 CI: 53.6 – 60
TOOA→BE (ky) OOA–BE split time 54.7 CI: 49.7 – 57.6
TLos→Stu (ky) Loschbour→Stuttgart admixture time 29.4 CI: 13.2 – 35.9

The authors conclude that "Using these advances, we inferred a demographic model that broadly explained observed H2 patterns and integrated major supported features in hominin evolution, including recurrent interbreeding between Neanderthals and AMH, introgression from a distantly-related, unsampled lineage to Denisovans, and population structure in western Eurasian AMH."

There is no Denisovan to AMH admixture in this model, it seems to only focus on Northern and Western Eurasians, and does not consider Eastern, Southern or Southeastern Asians and Oceanians.

Effective Populations

I found the Effective population sizes to be of interest (the N_e). As you can see, the Ancient basal root at the top of the image (A) has a large population from which the Superarchaics (S), the modern humans (AMH) split from and conserve a large population size, the ancestor of Neanderthals and Denisovans (ND) has a tiny population and remain that size, so do the N, Denisovans, and the original Out of Africa migration group (bottleneck). The Yoruba people retain a large population.

The paper says that the original Ancestral population had an N_e of 16500 individuals (CI: 15900 – 17300) and then it says "We fixed the effective size of the Superarchaic lineage (S) to 20,000" So the superarchaics splitting from the Ancestral line into Eurasia didn't suffer a bottleneck? Why?

Yet the other groups splitting from the Ancestral group did! The authors explain this large Superarchaic population size as follows: "We justified fixing the population size of S with the observation that changing the effective size of a ghost lineage which makes a small ancestry contribution to a sampled lineage has a negligible effect on E[H2]." So, their model and formulation allows these unrealistic assumptions.

The upper part of the image further down, shows how the Superarchaic introgression into Denisovans affects the population sizes and dates, their model calculates an outcome with a minor impact on effective populations or the timelines.

Interestingly, they note that small effective population sizes may be an artifact, because they can be "plausibly explained by geographic population structure. With spatial structure, recent ancestors are expected to live in closer proximity, and to therefore have a higher probability of sharing parents, than ancient ancestors. Strong structure therefore causes recent coalescence rates to be larger than ancient ones,a pattern which is interpreted as a small recent effective size in a panmictic model."

The paper also notes that "using a lower mutation rate inflated effective size and time parameters, while a higher rate diminished them." They show tables with the effects of different mutation rates as can be seen in the lower part of the image below. The image shows the effects of Superarchaic introgression into Denisovans and the effect of different mutation rates on Ages and N_e of the hominin clades. The original can be seen in tables S8 and S9 in the Supplementary Information of this paper:

The impact of a slower mutation rate can be seen in the older split between ND and our lineage and an older ND split into Denisovans and Neanderthals, but does not affect on more recent events. The impact of mutation rates on the effective population size (N_e) is also variable, some populations have a bigger effective population (A, AMH, Altai, Vindija, Chagyrskaya, Denisovan) while others smaller (ND, Yoruba).

I have already posted about mutation rates and how it impacts on N_e and heterozygosity, and mentioned the same effect reported by the authors of this paper: For a given heterozygosity, lower mutation rates increase effective population size. I also posted about the effect of mutation rates on dating splits, lower rates lead to deeper (older) split dates.

However, this paper does not explain why the same mutation rate affects N_e in opposite ways (there is a complex explanation about their model in the Appendix that mentions some effects on the Effective Population size). It does admit that the model, like all models is a simplification of reality: "... we made many approximations to simplify our models. We treated populations as discrete entities, with random mating, piecewise-constant sizes, and instantaneous divergence. Some of these assumptions allow us to model the evolution of HR statistics, while others are useful for formally testing tractable demographic models. Of course, the true evolutionary history includes unmodeled populations, continuously fluctuating population sizes, population structure induced by the spatial distribution of individuals, and variable migration rates... We also made a number of simplifying biological assumptions. We assumed that the genome-wide average germline mutation rate was constant across all lineages throughout the modeled period... We also assumed a constant generation time for all lineages throughout the studied period."

Interesting work.

Introgression Index

Visit my index post, with all the introgression posts in one single place.

---

Patagonian Monsters - Cryptozoology, Myths & legends in Patagonia Copyright 2009-2026 by Austin Whittall ©