The Blog

2021-

2021-07-12 g
MAMAS, DON'T LET YOUR BABIES GROW UP TO BE HARMED OR KILLED BY HOMO ERECTUS HYDRIDS

Editor's Note:
A realistic interpretation of this study calls into question the "out of Africa" hypothesis for modern humans at any time after the Eemian Interglacial. A dispassionate student of human evolution could infer that the "ghost archaic" introgression noted below in Sub-Saharan Africans was contributed by a population of Homo erectus surviving into the last 60,000 to 30,000 years (though the interbreeding could have been occurring further into the past). The unique physical characteristics of modern Africans could certainly have come from Homo erectus. Modern West Africans found to have up, "to 19% of their genetic ancestry from an archaic population that diverged before the split of Neanderthals and modern humans," would account for their physical similarities with recent reconstructions of Homo erectus.

Further study is needed to determine how modern West Africans with the greatest percentage of that "ghost archaic" introgression differ (physically, cognitively and socially, for example) from those with the least percentage of that introgression. Further comparisons of those with 19% "ghost archaic" (Homo erectus) introgression with Europeans and East Asians would also be instructive; though I fear such studies, if carried out, would be suppressed.

Further corroboration for these musings comes from the absence of Neanderthal genes in Sub-Saharan Africans and the absence of the "ghost archaic" genes in non-African populations.

While ancient hominin fossils are well known from the Rift Valley of East Africa, the birthplace of the genus, Homo, could well have been elsewhere. I would look wherever prosimians occur for signs of ancient hominin and human fossils.

See also: A world map of Neanderthal and Denisovan ancestry in modern humans

Recovering signals of ghost archaic (Homo erectus] introgression in African populations

Abstract

While introgression from Neanderthals and Denisovans has been documented in modern humans outside Africa, the contribution of archaic hominins to the genetic variation of present-day Africans remains poorly understood. We provide complementary lines of evidence for archaic introgression into four West African populations. Our analyses of site frequency spectra indicate that these populations derive 2 to 19% of their genetic ancestry from an archaic population that diverged before the split of Neanderthals and modern humans. Using a method that can identify segments of archaic ancestry without the need for reference archaic genomes, we built genome-wide maps of archaic ancestry in the Yoruba and the Mende populations. Analyses of these maps reveal segments of archaic ancestry at high frequency in these populations that represent potential targets of adaptive introgression. Our results reveal the substantial contribution of archaic ancestry in shaping the gene pool of present-day West African populations.

INTRODUCTION

dmixture has been a dominant force in shaping patterns of genetic variation in human populations (1). Comparisons of genome sequences from archaic hominins to those from present-day humans have documented multiple interbreeding events, including gene flow from Neanderthals into the ancestors of all non-Africans (2), from Denisovans into Oceanians (3) and eastern non-Africans (4, 5), as well as from early modern humans into the Neanderthals (6). However, the sparse fossil record and the difficulty in obtaining ancient DNA have made it challenging to dissect the contribution of archaic hominins to genetic diversity within Africa. While several studies have revealed contributions from deep lineages to the ancestry of present-day Africans (7–12), the nature of these contributions remains poorly understood.

RESULTS

We leveraged whole-genome sequence data from present-day West African populations and archaic hominins to compute statistics that are sensitive to introgression in the history of these populations. Specifically, we tabulated the distribution of the frequencies of derived alleles (where a derived allele is determined relative to an inferred human ancestor) in the analyzed African populations at single-nucleotide polymorphisms (SNPs) for which a randomly sampled allele from an archaic individual was observed to also be derived. Theory predicts that this conditional site frequency spectrum (CSFS) is expected to be uniformly distributed when alleles are neutrally evolving under a demographic model in which the ancestor of modern and archaic humans, assumed to be at mutation-drift equilibrium, split with no subsequent gene flow between the two groups (13, 14). This expectation is robust to assumptions about changes in population sizes in the history of modern human or archaic populations. Further, we show that this expectation holds even when there is population structure or gene flow in the history of the archaic population (see Materials and Methods).
[...]
Given that none of the current demographic models are able to fit the observed CSFS, we explored models where present-day West Africans trace part of their ancestry to (A) a population that split from their ancestors after the split between Neanderthals and modern humans, (B) a population that split from the ancestor of Neanderthals after the split between Neanderthals and modern humans, or (C) a population that diverged from the ancestors of modern humans and Neanderthals before the ancestors of Neanderthals and modern humans split from each other (fig. S2 and section S3). Each of these models of admixture (which we refer to as models A, B, and C, respectively) can yield a U-shaped CSFS. The increase in the counts of low derived allele frequency SNPs is largely due to the introduction of the derived allele from the introgressing population at sites that are fixed for the ancestral allele. The increase in the counts of the high-frequency SNPs is largely due to the introduction of the ancestral alleles at sites that are fixed for the derived allele.

A search for the parameters for models A and B that produce the best fit to the CSFS results in a trifurcation, i.e., models in which the introgressing population splits off from the modern human population at the same time as the modern human–Neanderthal. Models A and B fail to fit the observed CSFS even at their most likely parameter estimates (KS P = 3.3 × 10−15 and P = 5.6 × 10−6, respectively; section S3) because of insufficient genetic drift in the African population since the split from the introgressing population (section S4.2). In addition, we show in appendix B that the spectrum for model A is expected to be symmetric, which is not observed in the data (Fig. 1). Model C, on the other hand, is consistent with the data (KS P = 0.09), suggesting that part of the ancestry of present-day West Africans must derive from a population that diverged before the split time of Neanderthals and modern humans. In addition to the goodness-of-fit tests, we examined the likelihood of the best-fit parameters for each of the models and found that model C provides a significantly better fit than other models (model C having a higher composite log likelihood than the next best model Δℒℒ = ℒℒNextbestmodel − ℒℒC = −6806 when we condition on the Vindija Neanderthal genome and Δℒℒ = −6240 when we condition on the Denisovan genome; table S4 and Materials and Methods). Our analyses provide support for a contribution to the genetic ancestry of present-day West African populations from an archaic ghost population whose divergence from the ancestors of modern humans predates the split of Neanderthals and modern humans.

We applied approximate Bayesian computation (ABC) to the CSFS to refine the parameters of our most likely demographic model (model C) (section S5). Given the large number of parameters in this demographic model, we fixed parameters that had previously been estimated (15) and jointly estimated the split time of the introgressing archaic population from the ancestors of Neanderthals and modern humans, the time of introgression, the fraction of ancestry contributed by the introgressing population, and its effective population size. We determined the posterior mean for the split time to be 625,000 years before the present (B.P.) [95% highest posterior density interval (HPD): 360,000 to 975,000], the admixture time to be 43,000 years B.P. (95% HPD: 6000 to 124,000), and the admixture fraction to be 0.11 (95% HPD: 0.045 to 0.19). Analyses of three other West African populations (ESN, GWD, and MSL) yielded concordant estimates for these parameters (Fig. 2 and table S7). Combining our results across the West African populations, we estimate that the archaic population split from the ancestor of Neanderthals and modern humans 360 thousand years (ka) to 1.02 million years (Ma) B.P. and subsequently introgressed into the ancestors of present-day Africans 0 to 124 ka B.P. contributing 2 to 19% of their ancestry. We caution that the true underlying demographic model is likely to be more complex. To explore aspects of this complexity, we examined the possibility that the archaic population diverged at the same time as the split time of modern humans and Neanderthals and found that this model can also produce a U-shaped CSFS with a likelihood that is relatively high, although lower than that of our best-fit model (Δℒℒ = −2713 for the Neanderthal CSFS and Δℒℒ = −2597 for the Denisovan CSFS, KS P ≤ 2.9 × 10−6). Our estimates of a large effective population size in the introgressing lineage (posterior mean of 25,000; 95% HPD: 23,000 to 27,000) could indicate additional structure. We find that the Ne of the introgressing lineage in YRI and MSL is larger than that in the other African populations, possibly due to a differential contribution from a basal West African branch (20).

While we have chosen to represent the genetic contribution of the African ghost population as a single discrete interbreeding event, a more realistic model could include low levels of gene flow in a structured population over an extended period of time. Previously proposed models of ancestral structure in Africa do not fit the CSFS [KS P < 2 × 10−16 for the model described in (21) and KS P < 2 × 10−16 for the model proposed in (14); fig. S18], although we observe that the model of ancestral structure proposed by Yang et al. does produce a slight U-shape. We explored additional models of population structure in Africa (22) in which a lineage split from the ancestor of the modern humans with split times ranging from 100 to 550 ka B.P. and continued to exchange genes with the modern human population until the present with migration rates ranging from 2.5 × 10−5 to 2 × 10−2 migrants per generation. While these models of continuous gene flow produce a U-shaped CSFS for low migration rates and deep splits, they do not provide an adequate fit to the empirical CSFS over the range of parameters considered (KS P ≤ 2.3 × 10−5; section S6 and figs. S14 and S15). We used our ABC framework to explore a more detailed model of continuous migration in which we varied split time, migration rate, and effective population size of the introgressing lineage. Simulations under the best fitting model produce a CSFS that does not adequately fit the data (KS P = 1.83 × 10−6). A possible reason why the continuous migration models that we have explored do not fit the data is that these models can be considered as extensions of model A with multiple admixture events. We have shown that these models can only produce symmetric CSFS, unlike the CSFS that we observe in the data (appendix B). Thus, deep population structure within Africa alone cannot not explain the data (section S6).

Given the uncertainty in our estimates of the time of introgression, we wondered whether jointly analyzing the CSFS from both the CEU (Utah residents with Northern and Western European ancestry) and YRI genomes could provide additional resolution. Under model C, we simulated introgression before and after the split between African and non-African populations and observed qualitative differences between the two models in the high-frequency–derived allele bins of the CSFS in African and non-African populations (fig. S40). Using ABC to jointly fit the high-frequency–derived allele bins of the CSFS in CEU and YRI (defined as greater than 50% frequency), we find that the lower limit on the 95% credible interval of the introgression time is older than the simulated split between CEU and YRI (2800 versus 2155 generations B.P.), indicating that at least part of the archaic lineages seen in the YRI are also shared with the CEU (section S9.2).

We then attempted to understand the fine-scale distribution of archaic ghost ancestry along the genomes of present-day Africans. We used a recently developed statistical method (ArchIE) that combines multiple population genetic statistics to identify segments of diverged ancestry in 50 YRI and 50 MSL genomes without the need for an archaic reference genome (section S7) (23). Briefly, the method uses summary statistics computed from present-day genome sequences as input to a logistic regression model to estimate the probability that a haploid segment of an individual genome (defined as a contiguous region of length 50 kilobases) is archaic. While the parameters of the model are estimated by simulating data under a model that closely matches the demographic history relating Neanderthals and non-Africans, we found that ArchIE has 68% power to detect archaic segments at a false discovery rate of about 7% under our best-fit demographic model, confirming that its inferences are robust and sensitive to archaic introgression in Africa.

On average, ≃6.6 and ≃7.0% of the genome sequences in YRI and MSL were labeled as putatively archaic in ancestry. We sought to test whether the putatively archaic segments identified in YRI and MSL traced their primary ancestry to other African populations (8–10) or to known archaic hominins such as the Neanderthals or Denisovans. We computed the divergence of these segments to a genome sequence from each of six populations: southern African KhoeSan, Jul‘hoan; two Central African pygmy populations (Biaka and Mbuti); and two archaic hominin populations (Neanderthal and Denisovan). We expect segments introgressed from any of these populations to be less diverged relative to nonarchaic segments. On the contrary, the putatively archaic segments are more diverged, consistent with their source not being any of these populations (Fig. 3C and section S7.1). Merging the putatively archaic segments across individual genomes, we obtained a total of 482 and 502 Mb of archaic genome sequence in the YRI and MSL, respectively. We estimated the distribution of the time to the most recent common ancestor (TMRCA) between segments labeled archaic and those labeled nonarchaic using the pairwise mode of multiple sequentially Markovian coalescent (MSMC) (Fig. 3B and section S7.2) (24) and observed that the TMRCA is larger for the putatively archaic class of segments. Specifically, we find that the median nonarchaic segment coalescent time is 0.865 Ma ago for both populations, while the median archaic segment coalescent time is 1.51 Ma ago for YRI and 1.15 Ma ago for MSL (1.69- and 1.23-fold increases in age for YRI and MSL, respectively).

We examined the frequencies of archaic segments to investigate whether natural selection could have shaped the distribution of archaic alleles (fig. S40). We found 33 loci with an archaic segment frequency of ≥50% in the YRI (a cutoff chosen to be larger than the 99.9th percentile of introgressed archaic allele frequencies based on a neutral simulation of archaic introgression with parameters related to the time of introgression and admixture fraction chosen conservatively to maximize the drift since introgression; section S7.3 and fig. S40) and 37 loci in the MSL. Some of these genes are at high frequency across both the YRI and MSL, including NF1, a tumor suppressor gene (83% in YRI, 85% in MSL), MTFR2, a gene involved with mitochondrial aerobic respiration in the testis (67% in YRI, 78% in MSL), HSD17B2, a gene involved with hormone regulation (74% in YRI, 68% in MSL), KCNIP4, which is a gene involved with potassium channels (73% in YRI, 69% in MSL), and TRPS1, a gene associated with trichorhinophalangeal syndrome (71% in YRI, 75% in MSL; Table 1). Three of these genes have been found in previous scans for positive selection in the YRI: NF1 (25, 26), KCNIP4 (27), and TRPS1 (28). On the other hand, we do not find elevated frequencies at MUC7, a gene previously found to harbor signatures of archaic introgression (29).

DISCUSSION

Our analyses document introgression in four present-day West African populations from an archaic population that likely diverged before the split of modern humans and the ancestors of Neanderthals and Denisovans. A number of previous studies have found evidence for deeply diverged lineages contributing genetic ancestry to the Pygmy (8, 9) and Yoruba (7, 30) populations. Analyses of ancient African genomes have revealed that stone-age hunter-gatherers from South Africa diverged from other modern-day populations >260,000 years (31) B.P. and that present-day West African populations trace part of their ancestry to a basal lineage that diverged before the split of the southern African San (20) (although an alternative model consistent with their data includes a complex pattern of isolation by distance between western, eastern, and southern African populations). Placing our results within the context of the complex patterns of deep divergences in the African populations will require the analysis of a diverse set of African populations that include the southern African San populations, as well as the inclusion of ancient African genomes that lack signals of recent admixture that are present in the present-day San populations (32).

One interpretation of the recent time of introgression that we document is that archaic forms persisted in Africa until fairly recently (33). Alternately, the archaic population could have introgressed earlier into a modern human population, which then subsequently interbred with the ancestors of the populations that we have analyzed here. The models that we have explored here are not mutually exclusive, and it is plausible that the history of African populations includes genetic contributions from multiple divergent populations, as evidenced by the large effective population size associated with the introgressing archaic population. Relatively, recent fossils with archaic features (or combinations of archaic and modern human features) have been found in the fossil record in Africa and the Middle East. While anatomically modern humans appear in the fossil record around 200,000 years ago, fossils with a combination of archaic and modern features can be found across sub-Saharan Africa and the Middle East until as recently as 35,000 years ago (34). Examples of these fossils include a cranium from Iwo Eleru (33) and human remains from Ishango (35) that have been interpreted as being consistent with deep structure and representing a complex history of interaction between modern and archaic hominins in Africa.

The signals of introgression in the West African populations that we have analyzed raise questions regarding the identity of the archaic hominin and its interactions with the modern human populations in Africa. Analysis of the CSFS in the Luhya from Webuye, Kenya (LWK) also reveals signals of archaic introgression, although our interpretation is complicated by recent admixture in the LWK that involves populations related to western Africans and eastern African hunter-gatherers (section S8) (20). Non-African populations (Han Chinese in Beijing and Utah residents with northern and western European ancestry) also show analogous patterns in the CSFS, suggesting that a component of archaic ancestry was shared before the split of African and non-African populations. A detailed understanding of archaic introgression and its role in adapting to diverse environmental conditions will require analysis of genomes from extant and ancient genomes across the geographic range of Africa. (read more)

______________________

Permission is hereby granted to any and all to copy and paste any entry on this page and convey it electronically along with its URL,
______________________