Sometimes results precede an appropriate interpretative framework. I am beginning to think that that was the case with the explosion in analysis of Y and mtDNA phylogenies around the year 2000. This was a research program which took the direct male (Y) and female (mtDNA) lineages of humans, inferred a phylogenetic tree, and drew historical and demographic inferences. Part of the reason that this research program flourished is that these lineages are easy to model as a tree, because there isn’t recombination and sex on these lineages. The genealogy of these markers is actually a straightforward tree. And, in the case of mtDNA it is copious, so extraction was and is relatively easy.

This period not only resulted in an explosion of research, but also several seminal popularizations, in particular Spencer Wells’ Journey of Man, Bryan Sykes’ Seven Daughters of Eve, and Steve Olson’s Mapping Human History. The way this method worked involved taking the distributions and diversity of particular Y and mtDNA lineages present in the world today, and inferring back to the past routes of human expansions assuming that the tree-like model of genetic fission reflected a serial founder process of human migration. On the largest scale this work did confirm and extend previous research which implied a dominant signal “Out of Africa,” as well as broad geographic racial clusters. But on a finer scale the method likely led us astray, in large part because it seems that the model of human expansion and diversification was just too simple. Empirically, the rise of whole-genome analysis via SNP-chips in the middle years of the 2000s and today ancient DNA has revolutionized and clarified our understanding of the human past. Many of the inferences made from Y and mtDNA turn out to be wrong, even if the original results are broadly robust.

And that is a key point, because the original Y and mtDNA results are correct, and there has been an enormous body of research already in this area, they can be leveraged in understanding the human past when slotted into the framework sculpted on the edges by autosomal SNP data and ancient DNA. A serial founder model where the tree-like phylogeny was recapitulated spatially and temporally always had “anomalies.” In light of new results those anomalies may actually make much more sense. In particular, the common lineages which span Sub-Saharan Africa and western Eurasia (e.g., R1b above) are less perplexing in light of a model of periodic Eurasian back migration after the initial Out of Africa event. Admixture between long distinct lineages resolves issues relating to haplogroups with discordant geographic distributions. As long a the results are correct and methodologically sound, more data is better.

The term “eugenics” has very negative connotations today. Nevertheless, in some ways society is moving in a direction which results in “eugenical” outcomes, insofar as allele frequencies and genotypes are skewed from what would otherwise be the case if natural processes operated without human volition.* Probably the most obvious case in modern medicine is the high rate of abortion of fetuses which are inferred to carry the genetic profile of an individual with Down Syndrome. In the future this sort of instance will be more general, as high quality prenatal genome sequencing along with progress toward understanding of the basis of inheritance of Mendelian diseases will avail parents of many choices. This will naturally result in a lot of discussion and debate about ethics and values.

But there is a less high-tech and ethically fraught form of eugenics, which is nevertheless culturally controversial. One of the overlooked aspects of the 2009 paper Reconstructing Indian population history is that it found that many Indian populations had an excess of homozygosity, likely due to long standing endogamous practices encouraged by the caste system. This, despite customs which enforce exogamy for Hindus across much of India, in particular the North. Within the abstract the authors suggest then that “there will be an excess of recessive diseases in India.” Recently I spoke to a young woman of Jat background whose parents are very traditional. I told her the issue relating to homozygosity, and communicated that to gain the benefits of masking genetic load one need not go genetically and culturally very far. An individual of the same ethnicity and religion would be sufficient, so long as they were not of the same caste (jati).**

This is not only a South Asian issue, as evident in a recent paper in The European Journal of Human Genetics, Genotyping of geographically diverse Druze trios reveals substructure and a recent bottleneck. Following up on earlier work it confirms some structure within the Lebanese Arab population. This should not be surprising, as the distribution of ethno-religious groups within Lebanon is not geographically arbitrary. Whether they live in Beruit today, the Maronite Christians for example often have a ancestral background from around Mount Lebanon. Additionally, the Muslim groups within Lebanon have been subject to a proportion of admixture with foreign populations over the past ~1,000 years. Nevertheless, it does seem that overall the Lebanese of all sects derive from a common ancestral group, and exhibit more affinities with each other on the whole than with non-Lebanese populations (the Druze have been subject to a stronger bottleneck than the other sects, explaining why they are distinct genetically).

In any case, the major finding in these results is that there are elvated levels of homozygosity in individuals who are Lebanese who are the product of marriages between “unrelated” parents. Additionally, individuals of Muslim background who are the product of first cousin marriages show evidence of being descended from common ancestors recently across multiple paths, suggesting a more ubiquitous practice of cousin marriage within this group. This mean that “first cousins” in the Muslim community often exhibit a relatedness greater than that of idealized first cousins because they are part of an extended inbred lineage.

In the figure to the right you can see the distribution of runs of homozygosity for different Lebanese sects comparing individuals who are the product of unrelated parents and those who are first cousin offspring (against a reference set of Europeans). You note that in the 0.5 to 1 Mb range of homozygosity tract length Europeans are highly enriched in comparison to Arab Lebanese. This is almost certainly due to the bottlenecks that Europeans have been subject to over the past ~50,000 years. Most Middle Eastern populations a priori should have a higher long term effective population size assuming a serial founder effect, as well as the ecological context of the Ice Ages. But at 1-2 Mb you see that long term inbreeding shows up in all the Lebanese groups. Notice that the further you go up in tract length the more unequal the ratio between the first cousin offspring (FCO) and those who are the products of “unrelated individuals” (URO). This is what you expect. But even at greater than 16 Mb you see that Lebanese who are unrelated still have many more of these segments than Europeans. This is a major tell that cryptic relatedness is a bigger issue within the Lebanese population than among Europeans.

The recommendation here then would be simple: Lebanese Arabs should marry individuals who are culturally aligned with them (e.g., Greek Orthodox Lebanese marrying Greek Orthodox Palestinians, Maronites with individuals from other Arab Christian groups in alignment with Rome, etc.), but have a different genetic history. This shouldn’t be controversial, but it can be (e.g., warning about the dangers of cousin marriage in terms of birth defects has resulted in accusations of Islamopobhia in Britain because of the power of identity politics in a multicultural society). Yet to some extent I believe that first cousin marriages will decline throughout the Middle East because of the demographic transition. If you don’t have many first cousins, then the potential for first cousin marriages declines greatly. But as I noted above in many Indian Hindu groups exogamy is normative, but there is still an issue with elevated homozygosity, so even without first cousin marriages there could be some gains in utility on the margin from outbreeding further than just near relatives.

Finally, aside from the straightforward genetic issues, there may be social benefits to the breakdown of clan structures which are the driving force behind population wide elevated homozygosity. Keeping it “all in the family” may not be conducive to broader national trust and cohesion. Readers probably know enough about the modern Middle East so that I don’t need to elaborate on this issue angle….

* In a narrow sense eugenics should actually result in allele frequency changes, but in modern practice this is not always the case. E.g., the screening for individuals with Tay-Sachs carrier alleles changes the genotype frequency in the population so it is not in Hardy-Weinberg Equilibrium.

** Another thing that might be feasible to be to attempt to infer potential enrichment of homozygosity by looking at genotypes of pairs of potential mates within the same caste.

Nice review, Advances and limits of using population genetics to understand local adaptation. In particular the focus here is on the insights one can derive from new genomic methods (e.g., think of SFS analysis). But they on a cautious, perhaps even down, note:

Many of the important questions in local adaptation being pursued with population genetics approaches begin – rather than end – with identifying loci responsible for variation. It is therefore important to realize that a full accounting of local adaptation at the molecular level goes beyond having high-quality data to analyze and statistical methods to identify causative genes. The crux of the challenge is that most ecologically important traits responsible for local adaptation are quantitative, and identifying all of the genes responsible for variation in quantitative traits is likely not possible. Even the cumulative explanatory power of individual loci identified in human genotype–phenotype association studies, which often involve tens of thousands of individuals, is generally only a small percentage of the phenotypic variation….

But I want to note that they cite a paper from 2012, and the work in capturing the fraction of phenotypic variation from genomic variation for height has gotten much better since then. So perhaps in that way that is a reason for optimism. Though, as they note, humans are a best case scenario since the sample sizes are enormous. Just a nice reminder of the limitations of the ‘post-genomic’ era.

A very fertile area of evolutionary science is the understanding and modeling of human culture. But it’s hard, which is why my aspiration is to be an evolutionary geneticist in a more classical sense. Not only is it hard, but people don’t appreciate it, because they think they understand “culture.” Robert Boyd and Peter Richerson’s The Origin and Evolution of Cultures is an excellent introduction (with Not by Genes Alone for a more popular audience), but Alex Mesoudi’s Cultural Evolution: How Darwinian Theory Can Explain Human Culture and Synthesize the Social Sciences is a more recent survey which I think is very worthwhile. Complementary to the modelers, who draw from classical genetic frameworks (going back to Cultural Transmission and Evolution by Cavalli-Sforza and Feldman), are the cognitive anthropologists, represented by Dan Sperber and his colleagues (Explaining Culture: A Naturalistic Approach). Another way to look at this is that it is cultural anthropology without the Post-Modern Voodoo. In other words, it strives for disciplinary relevance through a fidelity to reality, rather than political and social impact.

This is all to highly recommend you read an excellent post on cultural evolution, Why Cultural Evolution Is Real (And What It Is). I don’t link to blogs much because it is usually more useful to read the paper, or the book, but in this case I think the writer really reduced the essentials of the field down to a manageable extent. Hopefully it will encourage readers to pursue their own avenues of inquiry.

It is common for strong results from population genetics to be confused when it is translated for public consumption. The best example is that of “mtDNA Eve.” Despite the big warning label that mtDNA Eve is one of many female ancestors, the public has gained the impression that she is the female ancestor. A similar problem is cropping up with the Khoisan paper which reports that they went through a relatively mild bottleneck in comparison to other modern human populations. There’s a reason I titled the post The Least Bottlenecked Humans of All. It’s a defensible reduction of the results. In contrast many popular treatments are translating the results into the conclusion that “the Khoisan had the largest population of all human groups at some point in the past.” The reason I avoided this formulation is that plainly stated I doubt that at any time the Khoisan as we understand them, a genetically-culturally coherent group in southern Africa, had the largest population of all. Humans of various sorts have been common across Afro-Eurasia for over a million years. Is it plausible that ancestors of the Khoisan had the largest populations of all? Anne Gibbon’s somewhat cautiously stated piece in Science, Dwindling African tribe may have been most populous group on planet, relays the sentiment which I share:

Other researchers agree that it’s likely that the Khoisan descend from a large population. But because sampling of African genomes is still so spotty, not everyone is yet convinced that the Khoisan “was the largest population on Earth at some point,” says evolutionary geneticist Pontus Skoglund of Harvard University. “Many African populations are not included for comparison,” he says, so it is possible that some of the diversity seen in the Khoisan was inherited from recent interbreeding that cannot yet be detected.

Either way, the study makes it clear that even though the Khoisan are genetically diverse by today’s standards, even they carry just a fraction of our ancestors’ genetic legacy over the past 120,000 years. “It is quite staggering how much extraordinary genetic variation and ethnic diversity was present but is now lost,” Skoglund says. The Khoisan, retaining more than the rest of us, offer a rare window to look back in time at some of that diversity.

The biggest gap in the current study is that many extinct lineages were not included. Obviously they couldn’t be included, because they’re extinct, though at some point in the future ancient DNA or (more likely in the African context) reconstruction of ancient genomes from extant populations which have absorbed them, might allow for a better understanding of Pleistocene human population sizes. Population genomics is powerful, but it has limits. We need to be cautious about assuming that what we can illuminate with current methods is all that can be conceived in our natural philosophy.

There’s a new open access paper on human Y chromosome phylogenies, The Y-chromosome tree bursts into leaf: 13,000 high-confidence SNPs covering the majority of known clades, which is just as interesting because of a reference than the paper itself. The paper cited is “Batini, C, P Hallast, D Zadik, et al. submitted. Large-scale recent expansion of European patrilineages shown by population resequencing. Nature Comms.” As I’ve mentioned earlier it was clear from several posters at ASHG that whole genome sequencing of larger sample sizes is making much more clear how recent and rapid the expansion of several of the common lineages across Eurasia was. In particular I’m thinking of R1a and R1b. We know this already from earlier work, but these latest results just confirm it. The truth is visually obviously. Notice in the figure above how R1b fans out in an instant from a single node. This reflects recent rapid population expansion.

But when The Journey of Man and The Seven Daughters of Eve were being written in the early 2000s it was assumed that Y and mtDNA phylogenies could tell us about human prehistory in totality. In hindsight that was probably asking for too much. But because they are sex specific these two lineages can inform us a great deal about social structures. In particular, rapid expansion of Y chromosomal lineages in the recent past may indicate the rise of patriarchy, as powerful males began to see polygyny as a viable strategy due to their monopolization of the resources of whole societies.

If I had to predict, I believe that female lineages tend to be characterized by isolation by distance dynamics. In contrast, male lineages are pruned periodically by rapid expansions and admixtures by groups which are able to marginalize or exterminate competitors over vast swaths of territory. The Genghis Khan effect is just the most prominent of this sort of phenomenon. It is certain that in prehistory this occurred as well, as new technologies such as agricultural triggered social revolutions, and the rise of specialization and stratification among humans.

Making my way through Steve Pinker’s The Sense of Style: The Thinking Person’s Guide to Writing in the 21st Century, one thing I really appreciate is his lucid deconstruction and refutation of the tyranny of The Elements of Style, also known as “Strunk & White”. It’s not that I haven’t run into this line of thinking before from linguists. But it’s interesting to note that Pinker points out that Strunk & White violate the very maxims they promote in a sentence asserting the importance of particular grammatical constructions (e.g., using passive voice to exhort active voice). From what I can gather part of the issue is that the authors themselves were not linguistic scholars, so lacked the expertise to pass judgments on issues which they were only vaguely aware of. Nevertheless, they’ve attained a certain oracular status among many Americans. I know this because over the years of my blogging many readers have exhorted me to read Strunk & White to remedy my deficiencies as a prose stylist. At some point in the mid-2000s I tried, but found it all rather tedious. This makes sense, as I always detested the grammar section of English class. My teachers were sometimes perplexed, because I obviously read and wrote with a high aptitude.

Read More

Spent much of yesterday at BAPG XI. Props to the Coop lab for putting it on. And speaking of which, want to give a special shout out to Gideon Bradburd, who presented his “SpaceMix” method. The preprint will be out soon, but I think the best way to understand it is what TreeMix is to standard phylogenetic trees, SpaceMix is to standard PCA. SpaceMix attempts to fit pulse admixture events against a landscape defined by geographic isolation by distance dynamics.

Also, big shout out to Dmitri Petrov for starting the BAPG meetings over four years ago (along with sponsors, such as Ancestry.com and CPB at Davis). From what I’m to understand the various regional meetings took their inspiration from BAPG, so it’s kind of a big deal. Though no doubt BAPG benefits from the high density of research universities from UCSC to UCD south to north, with Stanford and UCB anchoring in the middle, all in a small commutable area.

Note: In case readers aren’t clear, “open thread” means you can talk about/ask about anything (within reason).

On a related note, a new paper attempted to find evidence for classic selective sweeps in our lineage comparison to the Neandertal and Denisovan populations, something feasible because of the existence of high quality ancient genomes. Let me quote from the abstract:

… We tested sites at which humans are fixed for the derived (i.e., nonchimpanzee allele) whereas the Neanderthal and Denisovan genomes are homozygous for the ancestral allele. We observe only weak differences in statistics indicative of selection between functional categories. When we compare patterns of scaled diversity or use our ABC approach, we fail to find a significant difference in signals of classic selective sweeps between regions surrounding nonsynonymous and synonymous changes, but we detect a slight enrichment for reduced scaled diversity around splice site changes. We also present a list of candidate sites that show high probability of having undergone a classic sweep in the modern human lineage since the split from Neanderthals and Denisovans.

As the heirs of the snottily termed “anatomically modern” humans it is hard for us to not think that we came and conquered due to some glittering genetic mutation. The lack of pervasive classic sweeps since the divergence of “us from them” confirms that evolution didn’t go into magical overdrive in our lineage. We are not “more derived” on the Great Chain of Being with chimpanzees as a reference, but simply different. The laundry list of candidate sites will be interesting, but must always be interpreted cautiously. These sorts of results dovetail nicely with the reality that the behavioral chasm between “archaic” and anatomically modern humans until about ~50,000 years ago wasn’t that large at all. Though changes in the fossil and archaeological record can seem punctuated and is discontinuous, I wouldn’t bet against the possibility that genomics points to a dynamic where change between the lineages was driven by gradual evolution via selection on standing variation.

Finally, there’s a recent paper on the nasal anatomy of Neandertals which made some media headlines recently. The most sensational conclusion drawn from nasal architecture is that Neandertals were a separate species from us. As I don’t know much about anatomy, and I don’t care much about species status, this isn’t something I’ll explore beyond just reporting that. But, another aspect is the fact that it seems the morphology of Neandertals has often been compared to modern humans, and judged by modern human standards. It seems entirely likely that Neandertals had different adaptations and an alternative morphological trajectory from our own lineage. Though I don’t think they were as different as some scholars would like to think, we need to be cautious about turning them into a version of ourselves, or at least our Out of Africa selves. The hominin lineage contains multitudes, and we are but one.

The Khoisan are not the oldest people on the face of this earth, they simply have been the lest impinged by population crashes over the past ~200,000 years. This is not a shocking assertion, but it is supported with greater robustness by a new paper in Nature Communications, Khoisan hunter-gatherers have been the largest population throughout most of modern-human demographic history. Unfortunately the standard tropes that come to the fore whenever the Khoisan are the objects of genetic scrutiny have emerged fully formed. But the worst problem is that one of the authors seems to be contributing to the misleading perceptions. Here are the press release

“Khoisan hunter/gatherers in Southern Africa have always perceived themselves as the oldest people,” said Prof Schuster, an NTU scientist at the Singapore Centre on Environmental Life Sciences Engineering (SCELSE) and a former Penn State University professor.

“Our study proves that they truly belong to one of mankind’s most ancient lineages, and these high quality genome sequences obtained from the tribesmen will help us better understand human population history, especially the understudied branch of mankind such as the Khoisan.

Though I think I know what Dr. Schuster is saying, I also believe that it is important remember that both you and I are also descendants of one of mankind’s most “ancient lineages.” We’re all equally descended from ancient lineages, because we’re all descended from ancient proto-humans. The Khoisan are not fossils preserved from a bygone age, they’re a modern people who descend from those same archaic Africans that we descend from. The phylogenetic distinction of the Khoisan is that their ancestors seem to have diverged from other human beings rather early on. In fact, it is defensible to suggest that they departed from the family tree of other humans at the first fork in the road, with the ancestors of Nilotic and Bantu Africans separating from the ancestors of Eurasians rather later on (also, it seems that the lineage that led up to the Khoisan after their divergence from the proto-non-Africans/agriculturalist Africans diversified and are represented by the Hadza and Pygmies, in addition to the Khoisan).

So what is the figure at the top of this post telling us? It’s showing you the effective population in the past by using the variation within the genome of individuals. In any given generation only a fraction of the population reproduces, and among the fraction which reproduces there is variation in output (e.g., some have one offspring live to adulthood, others have four). This increases random genetic drift by reducing population size and increasing variance of the sampling outcome. Drift tends to remove variation as some alleles fix and others go extinct due to the fluctuations that it produces. What happens when populations go through a bottleneck is that a lot of the variation that was present in a population can be squeezed out of it, and it takes a long time for that variation to replenish itself through mutation alone. This is why the long term effective population of the Khoisan is so much larger than the Han Chinese, despite there being four orders of magnitude more Han Chinese in the world today than Khoisan. As you can see in the figure the Khoisan were not subject to nearly as strong a bottleneck in the past. The small long term effective population size of non-Africans, reflected in their relative genetic homogeneity in comparison to Africans, is due to the bottlenecks that occurred as the ancestors of non-Africans were migrating across the world. About ~30,000 years ago the effective population of the Khoisan was ~10,000, while that of the ancestors of Eurasians was ~1,000. Both groups were going through a bottleneck, but the proto-Eurasians were experiencing a much stronger one. The ancestors of agriculturalist Africans, the vast majority of Sub-Saharan Africans today, were also experiencing a bottleneck, but a milder one.

One major caveat with the time scale above is that it is sensitive to the mutational rate. The figure above uses ~2.5 X 10^-8 value, while some estimates are close to ~1.25 X 10^-8. Though the authors brush this problem off, arguing that paleoclimate models can fit both parameter values in terms of explanations for the demographic patterns, from what I know the paleonathropology more neatly fits the dates from the higher mutation rate. With the lower value the divergence between Khoisan and non-Khoisan is almost contemporaneous with the first fossil evidence of anatomically modern H. sapiens in Africa. But it is true that in either scenario non-Africans are subject to a strong bottleneck, long evident from the genetic data, and that the Khoisan have relatively large long term effective populations.

To put my cards on the table I do not accept the paleoclimatic explanation for why the Khoisan in particular had a large population while the ancestors of agriculturalist Africans and non-Africans did not. First, in Science Pontus Skoglund observes that “Many African populations are not included for comparison….” Yes. He also points out that the Khoisan could have gotten their diversity from interbreeding, something that has support in the literature. That being said, that’s not my biggest problem with the explanatory model outlined in the paper. In the blog post yesterday I reviewed results which imply that the Igbo of Nigeria have some admixture from a hunter-gatherer group genetically closer to the Khoisan than to to the Pygmies. This is peculiar because geographically the Igbo are far closer to the Pygmy populations. So what’s going on here?

Human populations move. Paleoclimate models are often predicated on the idea that populations were geographically fixed for eons. But there’s a fair amount of circumstantial evidence that Khoisan-like people were extant as far north as Ethiopia in the recent past. The major element of the ancestry of the Bantu speaking majority in South Africa was resident in Cameroon ~3,000 years ago. A hybrid Eurasian-Sub-Saharan African populations in the highlands of Ethiopia only emerged in the past 3,000 years. And so on. We don’t know that the ancestors of agriculturalist Africans were resident in the northern fringe of Sub-Saharan African ~100,000 years ago. We don’t if the ancestors of the modern Khoisan have such deep time history in southern Africa.

In light of evidence of Eurasian back-migration into Sub-Saharan Africa in deep antiquity it is quite reasonable to suggest that the intermediate population dynamics of the agricultural Sub-Saharan populations in this paper is simply due to the fact that they are a hybrid between long resident hunter-gatherer populations with some affinity with Khoisan and Pygmies, and descendants of the original Out of Africa migration. The blogger Dienekes Pontikos has been proposing a flavor of this model for a long time. Initially I thought it was a cranky fixation of his which had little basis in reality, but over the years I’ve come around to the position that it’s a plausible theory. At least as plausible as the myriad attempts to reconcile genetic patterns with paleoclimate models.

Citation: Khoisan hunter-gatherers have been the largest population throughout most of modern-human demographic history.