A few months ago I asked this community if modern genome science had tested an “Adam and Eve” hypothesis that the human lineage has passed through short, sharp bottleneck of two at some point in its history. While this question may sound bizarre to some, it is one that is often asked by those with a background in Abrahamic faiths. My post has therefore been taken up and discussed extensively on the Skeptical Zone and Biologos Forum over the past few months, as well as by various blogs.
The claim that genomic methods have been used to test and reject an "Adam and Eve" hypothesis was central to the recent book Adam and the Genome. My post, which critiqued the arguments made in that book, has received a broad level of explicit or tacit agreement in subsequent online discussions. More adequate ways of testing the hypothesis have been suggested, and preliminary results have been obtained.
Here I will share some of the lessons I have learned from these discussions and from further reading. These are somewhat tentative, and not all are based on published peer reviewed literature. In a short blog I cannot do not do full justice to all the contributions that have been made by various scientists within the online fora, so as far as possible I will try to provide direct links to the contributions of others.
Here are the lessons I have learned so far:
1) The “Adam and Eve” hypothesis was advocated by geneticists in the 1970s and 1980s
In his 1975 book Adam and the Ape, R. J. Berry (Professor of Genetics at University College London 1974-2000) wrote:
“...recent work on the distribution of biochemical variants can most easily be interpreted to mean that man has been through a ‘bottleneck’ in numbers during his recent evolutionary past. The reasons for claiming this are … based on the frequencies of variants of the blood protein haemoglobin. These demand that during the past million years, the human species comprised a single pair for one generation, or an effective population of 10,000 for half a million years, or something in between [here he cites J. M. Smith “The causes of polymorphism” in Variation in Mammalian Populations, 26th Symposium of the Zoological Society of London, Academic Press, 1970, edited by R. J. Berry and H. N. Southern]. This assumes that the variants considered are selectively neutral, which is only partially true. Nevertheless it is strong circumstantial evidence that man has passed through an ‘Adam and Eve’ situation in comparatively recent times.” (p. 42)
In PNAS in 1980, Wesley Brown suggested on the basis of mitochondrial DNA variation that for humans “the amount of sequence heterogeneity observed, 0.18%, could have been generated from a single mating pair that existed 180-360 x 10^3 years ago, suggesting the possibility that present-day humans evolved from a small mitochondrially monomorphic population that existed at that time.” (p.3609)
Allan Wilson and colleagues suggested in a 1985 Biological Journal of the Linnean Society review that “the human species” could have passed though a short sharp bottleneck because it “has an anomalously low level of mtDNA variability, in spite of an apparently normal level of nuclear variability.” (p.393).
These and other studies led to a fairly extensive literature about “mitochondrial Eve”, later joined by “Y-chromosome Adam”. For example, Dorit et al. calculated in Science in 1995 that the two could have co-occured around 270,000 years ago.
2) Studies by Ayala et al. in the 1990s rejected the hypothesis
In a 1994 PNAS paper, Francisco Ayala and colleagues appeared to disprove the “Adam and Eve” hypothesis. Examining the huge allelic diversity of the human major histocompatibility complex (MHC) gene DRB1 they calculated “that human ancestral populations could never have been smaller than two or three thousand individuals at any time over the last several million years.” They explicitly tested and rejected “the possibility of occasional population bottlenecks”.
In 1996, Ayala and Escalante followed this up with a publication on the MHC gene DQB1, looking at trans-species variation. They found that “Many human alleles are more closely related to pongid and cercopithecoid alleles than to other human alleles.” They argued that “The persistence of 12 DQB1 allelic lineages over the last 6 Myr of hominid evolution implies of necessity that no fewer than six individuals could have existed at any one time throughout that time.” Using coalescence theory, they estimated that “these polymorphisms require human populations of the order of N= 100,000 individuals for the last several million years.”
(3) Increasing knowledge of MHC evolution may have undermined the case made by Ayala et al.
The papers by Ayala et al. appeared to banish the “Adam and Eve” hypothesis to scientific oblivion, but subsequent progress in knowledge of MHC evolution suggests that it may not be the last word after all. This has been pointed out by Joshua Swamidass, who is Assistant Professor in the Laboratory and Genomic Medicine Division at Washington University in St Louis, on his website “Peaceful Science”. Among other points, he shows that that the Ayala papers were significantly undermined by a study in Nature Genetics in 1998 by Bergstrom et al. which argued from intron sequence diversity that DRB1 alleles in humans have diversified more recently than suggested by Ayala et al.. In addition, Swamidass notes that a 2016 BMC Evolutionary Biology paper by Vahdati and Wagner suggests that convergent evolution to be common at human MHC loci, and this does not appear to have been accounted for by Ayala and colleagues. These, and several other relevant points and papers can be found here, by which Swamidass seems to make a powerful case that Ayala’s work should not be seen as the last word on this issue.
(4) The “Adam and Eve” hypothesis has not been rigorously tested with genome-wide human diversity data
This is something that became very apparent in the discussion following my previous blog post. Chapter Three of Adam and the Genome, by Dennis Venema of Trinity Western University, claimed that multiple lines of genomic evidence show that the human lineage has never dropped below a population size of several thousands since divergence from chimpanzees. Venema responded to my post with two blog posts (here and here) and over 160 forum posts, making many attempts to substantiate his claims. These contributions, together with others by population geneticists (e.g. here, here and here), have confirmed that the extreme human bottleneck hypothesis has not been rigorously tested in the modern genomic literature, unless we have all missed something.
(5) A test could be made using genome-wide allele frequency spectra (AFS)
In a significant contribution to the discussion, Steve Schaffner of the Broad Institute suggested that unfolded SNP allele frequency spectra (AFS) from African populations in the 1000 genomes project could be used to test for a recent bottleneck. He posted about this on the Biologos forum here and in subsequent comments. Using a script that he has made available on Github, he simulated the AFS expected under various simple recent-bottleneck scenarios with simple subsequent demographics. He showed that these did not fit well with the real data. A bottleneck had to be at least 500,000 years ago before anything approaching a good fit was simulated.
This approach seems interesting, but several papers have been published in the past few years that question the value of AFS for demographic inference and underline that a wide range of complex demographic, phylogenetic, selective and mutational processes, together with sampling strategies, can influence them. I listed some of these papers here and Schaffner has responded here.
(6) A test can be made from genome-wide estimates of the time to the most recent four lineages (TMR4L)
Building on Joe Felsenstein’s comment “it would be even better to find more polymorphisms involving multiple haplotypes and put all that information together”, Joshua Swamidass has done a preliminary analysis in which he sought to examine genome-wide patterns of multiple haplotypes and put them all together to examine whether or not a bottleneck of two is plausible in the human lineage in the last few million years. He did this by estimating the time to the most recent four lineages for each recombination block in the autosomal genome (unlike mitochondrial genomes and Y chromosomes, four gene lineages can pass through a bottleneck of two within autosomes). More detail can be found here. In brief, this analysis suggested that a two-person bottleneck is plausible over approximately 500,000 years ago. Again, this is a preliminary analysis, and could be developed further with more accurate phylogeny building, and more sophisticated statistical approaches. I think that Swamidass intends to work this up for a publication.
(7) Introgression from Neanderthals and Denisovans makes everything more complicated
Evidence for introgression from Neanderthal and Denisovan populations into modern human genomes provides a major complication for the "Adam and Eve" hypotheses. If Neanderthals and Denisovans are defined as being within the “human” lineage, then a two-person bottleneck could only have occurred before the divergence of modern humans, Neanderthals and Denisovans. If, on the other hand, we define only modern humans as truly “human”, and posit a bottleneck within the human lineage, with later admixture of DNA variation from “non-humans”, then a more recent bottleneck is possible, and the AFS and TMR4L approaches will be made very difficult to apply, unless admixed alleles can be filtered out.
To my mind, the question has now moved on from “Is an 'Adam and Eve' bottleneck inconsistent with human genetic diversity data?” to “At what timescale could an 'Adam and Eve' bottleneck be consistent with human genetic diversity data?” It would be good to see the AFS and TMR4L approaches followed up with published analyses, and I would be very glad to know of other approaches that could be used to test or date the hypothesis.
Poster image: The Expulsion of Adam and Eve from Paradise, Benjamin West, 1791. NGA Images.