# Real three dimensional PCA!

By Razib Khan | February 15, 2011 10:14 pm

Sort of. Check out what Doug McDonald’s done with javascript. The HGDP populations in three dimensions.

CATEGORIZED UNDER: Genomics
• http://mengbomin.wordpress.com/ Meng Bomin

That’s pretty damn sweet.

• Bob

Now, that’s quite sweet!

However, this doesn’t look as I’d expect. I’d thought there would be a 2-d projection there that was familiar. There’s no view (that I can see) that shows, for example, the separation between african and non-african populations that I’m used to seeing in the 2-d plots.

Am I not understanding what I’m looking at? This just used the first 3 eigenvectors rather than the first 2?

Cheers,
–Bob

• http://sep.stanford.edu/sep/jon/ Jon Claerbout

A nice start, but needs a few basic user controls like speed and screen size. Also, fundamentally, rotations should be possible in two directions (latitude and longitude).

• http://mengbomin.wordpress.com/ Meng Bomin

@Bob:

However, this doesn’t look as I’d expect. I’d thought there would be a 2-d projection there that was familiar. There’s no view (that I can see) that shows, for example, the separation between african and non-african populations that I’m used to seeing in the 2-d plots.

Well, if you notice, most of the Eurasian populations are relatively equidistant from the Sub-Saharan African cluster, forming something akin to a planar triangle. The clusters that deviate from that are the Amerindians, who are farther away from the Africans than the Eurasians as well as the Melanesians and Papuans, who are marginally closer in this plot, though I suspect they’d show themselves to be greater outliers if we could plot, see, and/or envisage more than three dimensions.

Obviously there’s a lot more than that. It’s generally a feast for the eyes.

Am I not understanding what I’m looking at? This just used the first 3 eigenvectors rather than the first 2?

Assume so.

• http://abugblog.blogspot.com Blackbird

I do like this so much more than Structure plots, I agree with Meng Bomin, a feast for the eyes. The isolated position of the Melanesians and Papuans is surprising, though; the closest to them are the Nganasans from Siberia!

• onur

The isolated position of the Melanesians and Papuans is surprising, though

It isn’t so surprising given their very different physical appearance from their neighbors except Australian Aboriginals (who aren’t covered in this 3-d plot).

• Insightful

Blackbird: The isolated position of the Melanesians and Papuans is surprising, though

Onur: It isn’t so surprising given their very different physical appearance from their neighbors…

Onur, what is surpising–given their ‘relatively’ similar physical appearances is that the two most divergent human populations on earth are the San Bushman (from southern Africa) and the Papuans from Melanesia. This was in fact stated by Dienekes:

Blackbird said that Papuans are closest to the Nganasans from Siberia. How can this be??..

• onur

Onur, what is surpising–given their ‘relatively’ similar physical appearances is that the two most divergent human populations on earth are the San Bushman (from southern Africa) and the Papuans from Melanesia.

Insightful, that isn’t surprising either. I would expect Australoids (the race of New Guineans, Australian Aboriginals and Melanesians) and San to be genetically close to each other only if they lived geographically close to each other (say, in the same continent). But they are geographically very remote from each other (I am only taking into account the distance by land of course), so it isn’t so surprising them to be genetically so divergent despite their relatively similar looks. What they have geographically in common is living close to the Equator to this day uninterruptedly for hundreds of thousands of years (ancestors of Australoids arrived their current territories tens of thousands of years ago from Africa passing through close-to-the-Equator parts of Eurasia, so their line has always lived close to the Equator), and this explains a lot of similarities in their appearances.

• onur

Blackbird said that Papuans are closest to the Nganasans from Siberia. How can this be??..

I think Blackbird is wrong. It isn’t clear from that 3-d plot which of the covered populations Papuans and Melanesians are closest to.

• onur

But they are geographically very remote from each other

And this has been so for tens of thousands of years.

• Insightful

But how do you explain the fact that Yorubans (west African bantu-related people) are genetically closer to the Chinese than to San Bushman even though Yorubans are geographically closer to the bushman??..

Dienekes pointed this out:

• Perahu

But how do you explain the fact that Yorubans (west African bantu-related people) are genetically closer to the Chinese than to San Bushman even though Yorubans are geographically closer to the bushman??..

African populations diverged earlier (~100 kya) than Out of Africa populations. (~60 kya). San Bushmen were relatively isolated from West African populations for a long time.

• onur

But how do you explain the fact that Yorubans (west African bantu-related people) are genetically closer to the Chinese than to San Bushman even though Yorubans are geographically closer to the bushman??..

Dienekes pointed this out:

Dienekes is wrong. Yorubans are genetically much closer to San than they are to Chinese.

Here are the FST figures:

Yoruban-San: 0,0976
Yoruban-Chinese: 0,1913
San-Chinese: 0,2480

• Neurobiogeek

Sorry for the naive question, but as someone who has worked in genomics for years I’m puzzled as to why a 3-D PCA plot is such a challenge. Is it that the data set is too large for applications like Spotfire or Partek to handle? I commonly generate and view interactive 3-D PCA plots to help analyze large microarray data sets. Partek can generate such a plot from 100 samples with >20,000 variables per sample almost instantly.

• http://blogs.discovermagazine.com/gnxp Razib Khan

#14, looking for something to display on web. does it output gifs?

• Insightful

Dienekes is wrong. Yorubans are genetically much closer to San than they are to Chinese.

Here are the FST figures:

Yoruban-San: 0,0976
Yoruban-Chinese: 0,1913
San-Chinese: 0,2480

Onur, Dienekes is referring to Table S6.2 of “Genetic history of an archaic hominin group from Denisova Cave in Siberia”, which shows that the divergence time of Yoruba/Chinese is smaller than that of Yoruba/San. Here is the link to Table S6.2

http://www.nature.com/nature/journal/v468/n7327/extref/nature09710-s1.pdf

Dienekes also said that the Fst you give are probably derived using some sort of microarray (I’ve given Fst estimates for Sub-Saharans vs. Paleoafricans myself here: http://dienekes.blogspot.com/2010/12/human-genetic-variation-first.html), and most SNPs in these chips are ascertained on Hapmap or other such data collections. Paleoafricans are polymorphic in sites where Eurasians (and African farmers) are monomorphic, so there is an ascertainment bias.

Also that the genetic divergence dates of the Denisova paper are the “state of the art” as they are based on full genome sequences (see also Figure 1 of the Denisova paper showing San as outgroup to other modern humans).

• Insightful

By the way, when I stated in post# 16 above that I’ve given Fst estimates for Sub-Saharans vs. Paleoafricans myself I was speaking as Dienekes in first person. He was telling me this. The second link to what he was referring to in post# 16 above may not work (the first link does work) so you can click on the second link here and it should work:

http://dienekes.blogspot.com/2010/12/human-genetic-variation-first.html

• Neurobiogeek

#15, Yes it does, as well as jpeg, png, and svg. I’m not selling Partek. I’m sure that Spotfire does the same. The image files are a static snapshot of the 3-D plot. You can interact with the plot within the application to rotate, zoom in/out, as well as to use color, shape, and size to denote factor variables.

• onur

Estimated genetic divergence dates (as their name implies, they are just estimates) don’t indicate genetic distances, Fst does; and the Fst figures, ADMIXTURE and MDS plots and dendrogram Dienekes provided in his above blog thread clearly show that Yorubans and San are genetically relatively close to each other and away from Chinese.

• Justin Giancola

onur- Just the person I was looking for! Maybe you can help me out! Am I wrong in thinking you are a computer programmer? And Turkish!? My computer I think has a virus that I can’t seem to shake as it has jumped from Chrome to Opera, in that it makes my google searches display in Turkish! or as if looked up from Google Turkey!?

• onur

Did you scan your computer for viruses?

• Justin Giancola

Yes but it keeps turning up clean. I run AVG Anti-Virus (the free one) and I have Spysweeper and Malwarebytes, keep coming up ok. And I do have the current Ad-Aware, though it won’t seem to run. And I have other issues like my internet straight up not working sometimes, showing I have no network when my internet and cable are the same source and the cable works so how is that possible? I feel like they all might be related as this Turkish thing correlates in time with the other issues and has been persistent despite repairing other flaws. Maybe have a recommendation for a program or something I’m not utilizing?

• onur

When did it start? When was the last time that your computer was OK? What programs/files did you install/download/use and which sites did you visit around the time the problem started?

• Insightful

Onur, this is from the Journal of Nature in Feb 2010:

We performed principal component analysis (PCA) using the EIGENSTRAT software on 174,272 autosome-wide SNPs common across the data sets (generated using 1M or 610K Illumina, or Affymetrix SNP6.0 arrays). Data on 10 Bushmen and 20 Xhosa were projected with 20 Yoruba and 20 Europeans from available (HapMap and Coriell) data, and 5 Bushmen (SAN) from the Human Genome Diversity Panel (HGDP) data. Population-wide PCA defines the Bushmen as distinct from Niger-Congo populations as from Europeans (Fig. 4a).

http://genetics.med.harvard.edu/reich/Reich_Lab/Publications_files/2010_Nature_Priya_Nick_Bushman_Paper.pdf

• onur

Insightful, PCA doesn’t mean genetic distance, Fst is a much better indicator of genetic distance.

• Insightful

Onur, I don’t have the time right now, but I suggest you do a little research of the latest Fst figures based on full genome sequences for san (which have only come into existence in 2010) in comparison to Yoruba and East Asian and you will find that Yoruba are genetically closer to Chinese than to San bushman…

• http://blogs.discovermagazine.com/gnxp Razib Khan

i don’t want to get into the genetic distance argument, though i probably lean to onur’s side (PCA isn’t even alway scaled to easily translate into distance), but, let’s be careful about using physical similarity to adduce genetic relatedness. there is obvious a lot of independent/convergence in trait values across different human linages. we see it in cranial capacity on the deep time scale, and we see it in skin color within the last 20-30,000 years.

• Justin Giancola

It started right in the beginning of December, right in time for finals! The only thing I downloaded out of the ordinary was a some bit torrent, probably an album or movie. Sites I’m always all over the place, but most I have good confidence in.

• zach

I’m not a geneticist, so this may be a naive quote or quandary (from the late Melvin Moss): “This is very beautiful. It is neat, it is modern technology, and it is fast. I am just wondering very seriously about the biological validity of what we are doing with this machine.” That is, the 3D plot looks very cool, but what exactly does it mean, and why?

• onur

Justin, I am currently searching for an effective anti-virus program for the specific virus (yes, it is a virus!) that infected your computer. I don’t want to recommend you a program which I am not so sure if it will work or not.

• onur

Onur, I don’t have the time right now, but I suggest you do a little research of the latest Fst figures based on full genome sequences for san (which have only come into existence in 2010) in comparison to Yoruba and East Asian and you will find that Yoruba are genetically closer to Chinese than to San bushman…

Did you see their genetic distance figures (Fst, ASD or IBS), and where?

• onur

Justin, are you the person with the nickname “Missingtrees” in the Google Chrome help forum?

• Justin Giancola

no I am not…but righteous help onur!

• onur

It seems he/she is in trouble with the same virus:

• Insightful

Did you see their genetic distance figures (Fst, ASD or IBS), and where?

The quickest way to get these genetic distance figures is from Dienekes. Onur, when you get a chance ask Dienekes for the latest of these figures using the full genome sequence. Be sure to be specific to get what you want.

• onur

Dienekes reads this blog, so I don’t see a need to contact him; if he has something to say, he can say without my contact. Insightful, what you claim is in contradiction with all of the genetic research covering Yorubans, San and Chinese I have seen so far, so until I see your evidence I see no reason to believe you.

• http://dienekes.blogspot.com Dienekes

Yoruba are genetically closer to Han Chinese based on FGS than they are to San Bushmen. That’s clear from the recent Denisova paper.

Table S6.2
Genetic divergence
San-Yoruba = 0.0913 x human-chimpanzee
Yoruba-Han = 0.0811 x human-chimpanzee

As I explain in my post, a possible reason for the low Fst distances between San and yoruba using microarrarays is the fact that San samples were not used in the ascertainment process. That is, a location where San (or other Palaeoafricans, who are the most diverse populations on the planet) are polymorphic (have e.g., A and G alleles) and the resst of mankind are monomorphic (have e.g., only the G allele) will not be discovered as polymorphisms using a panel that does not include San Bushmen, or not enough of them to discover low-frequency polymorphisms of that population.

Given that the HapMap has 0 San and the HGDP 5 unrelated ones, I very much doubt that most of the genetic variability in that population has been captured in commecial microarrays behind the quoted Fst numbers.

• KRANG

But dienekes acording to that figure, han is closer to yoruba then french is to yoruba, but on FST, europeans are closer to africans then east asians are?

Arent FST bether in calculating genetic distances?

• onur

Still, I cannot say that I am convinced until seeing a measure that gives more direct and reliable information about genetic distances like Fst. We need more genetic tests than that to change the current paradigm of intra-modern human genetic relationships.

• http://www.kinshipstudies.org German Dziebel

“‘But how do you explain the fact that Yorubans (west African bantu-related people) are genetically closer to the Chinese than to San Bushman even though Yorubans are geographically closer to the bushman??..’

African populations diverged earlier (~100 kya) than Out of Africa populations. (~60 kya). San Bushmen were relatively isolated from West African populations for a long time.”

Or San Bushmen simply mutated faster and diverged from the rest of the humanity, including Yoruba, more radically, as they colonized South Africa and got isolated from everybody. From linguistics we know that they speak click languages that no other human group has. They must have acquired them after they had branched off from the rest. Also, if San Bushmen have existed as a separate population and an isolate, then they must have developed language and culture independently from everybody else, since archaeologically there’s not much evidence for behavioral modernity prior to 40-50K. This is unlikely, however, which means San Bushmen are derived from the rest, not the other way around.

• gcochran

“Also, if San Bushmen have existed as a separate population and an isolate, then they must have developed language and culture independently from everybody else, since archaeologically there’s not much evidence for behavioral modernity prior to 40-50K”

I’ve been waiting for someone , even a nut like Dziebel, to notice that it now looks as if the splits between human populations far predate any archaeological evidence of of ‘ behavioral modernity’, that misbegotten concept.

• http://dienekes.blogspot.com Dienekes

But dienekes acording to that figure, han is closer to yoruba then french is to yoruba

Not really, as the difference is 0.0812 vs 0.0811 and the authors state that the standard errors of the estimates are in the 0.0005-0.0007 range, hence you can’t say that one distance is great than the other, and the common-sense conclusion is that Han and French are about equi-distant from Yoruba.

Arent FST bether in calculating genetic distances?

When we have a large number of FGS examples from the relevant populations, we can calculate genome-wide Fst among them. Unfortunately the 1000 genomes project does not include Palaeoafricans, a surprising omission, so I don’t know when that day will come.

To reiterate: microarray chips are good at providing genome-wide dense coverage of polymorphism: they provide calls on a few hundred thousand variable sites. The problem is that you don’t know whether a site is variable or not until you test a particular population, and Palaeoafricans such as San or Pygmies, being the most variable people of all are likely to harbor variation at sites where the rest of mankind does not.

• KRANG

Dinekes most fst values have caucasoids closer to africans, but what are the FST values from 1000 genome project tell us?
By the way dienekes, only 3-4 million snps vary between human populations, FST based on those snps should give correct numbers on distances, correct me if am wrong.

One more question dienekes, are san and papuan really the most distant from yoruba?

• http://www.kinshipstudies.org German Dziebel

“I’ve been waiting for someone , even a nut like Dziebel, to notice that it now looks as if the splits between human populations far predate any archaeological evidence of of ‘ behavioral modernity’, that misbegotten concept.”

Since gcochran again decided to knuckle-walk into this discussion some rather primitive interpretations, I have to remind him that San Bushmen are culturally and linguistically modern humans. And they are biologically modern, too. If they’d separated from the rest of the humanity before modern human behavior emerged (around 200-100K according to various molecular watches) they 1) would’ve been replaced by behaviorally modern humans around 40-50K (with or without interbreeding) and we wouldn’t have known they existed; 2) if they had survived their cultural and linguistic pattern would’ve been totally aberrant. Neither of these two scenarios are borne out by facts. I suspect that genetic phylogenies and the molecular watch are simply off. San Bushmen are relatively recent arrivals (likely from East Africa) who acquired some special features as a result of isolation. And they may have mutated faster. If they are polymorphic (A and G, for example) in the loci where other humans are monomorphic (G only) it means their A mutation emerged on top of the common human G mutation and hasn’t replaced the G mutation yet across the population. Again, this suggests that they derived from the rest relatively recently.

As for behavioral modernity being a misbegotten concept, it’s uncontroversial that human beings display a pattern of adaptation qualitatively distinct from that of earlier hominids. We outlived them and we spread all over the globe. Our cognitive, communicational, technological, adaptive, etc. skills are unparalleled. In the archaeological record, anything suggestive of similar skills starts popping up consistently and unambiguously only 40K years ago or so. Gcochran, you’ve got some catching up to do!

• Sandgroper

“In the archaeological record, anything suggestive of similar skills starts popping up consistently and unambiguously only 40K years ago or so.”

But there are some much earlier examples.

• http://www.kinshipstudies.org German Dziebel

“But there are some much earlier examples.”

Yes, agree. I don’t have a good explanation for bows and arrows and for pressure flaking technologies in the Middle Stone Age in Africa. They are aberrant, sporadic and create problems for my theory. But we need to have DNA from sites like Blombos Cave to determine whether it matches San Bushmen-specific mutations. Without ancient DNA, the two populations – one ancient and another modern – remain apples and oranges. Phenotypically, Bushmen-looking skulls are some 7,000 years old. The 36,000-year-old Hofmeyr skull from South Africa clusters with Paleolithic Eurasians. To imagine that Bushmen haven’t split off from any other population for 200K years, and this is what genetics tells us, is difficult for me.

• Justin Giancola

Oh onur I was reminded of another aspect of the virus! Almost every time you chose the first link of a google search it steals it after it starts moving to the page and redirects you to some bullshit page, not specifically an ad, but probably a fake site. And it makes it very difficult for you to go back to the search query. VERY annoying.

• onur

Justin, I am sorry to say this but you may need to spare some money and buy a serious antivirus program. But I suggest you first download and try all free antivirus programs available. If they don’t work and you don’t want to spend your money on an antivirus program or have doubts about their usefulness, just format your hard drive and reinstall your OS.

• Insightful

German Dziebel: The 36,000-year-old Hofmeyr skull from South Africa clusters with Paleolithic Eurasians.

That is very interesting. 36,000 years ago, Eurasians would have been roughly 20,000 years out of Africa or at least as long as Amerindians have been in the Americas. And of course in the paleolithic all humans would have been about 50% closer to Y Chromosome Adam (the most recent common male ancestor of all of us today). So across continents there must have been more similarity not only in cultures but in phenotypes or to the proto-phenotype…

• http://www.kinshipstudies.org German Dziebel

“So across continents there must have been more similarity not only in cultures but in phenotypes or to the proto-phenotype…”

So it appears… Hofmeyr, Zhoukoudian, Paleoindian skulls all seem to cluster together. I suppose that what we would expect to see if the founding migration happened around 40-50K. Unless small sample size hides important regional differences that existed prior to 10K. Then, regional differentiation resulted in the formation of Australoid, Caucasoid, Khoisanid, Black African, Mongoloid phenotypes in the past 10K years. Some of the distinctive features of those recent regional phenotypes may be much older than 10K (both the Bushman canine and shovel-shaped incisors frequent in, respectively, San Bushmen and Asians-Amerindians have close matches in Neanderthals), but the unique combination of Mongoloid, etc. features looks recent.

However, the Hofmeyr skull is a very inconvenient case for the recent out of Africa migration theory because one would expect deep South Africa to bring to the fore skulls with the pronounced retention of ancient African morphology (AMH and archaic) and the continuation of this morphology, however modified, further downstream to modern African skulls (San Bushmen would be an ideal candidate for such retentions). But this is not the case. Instead, we see non-African morphology in South Africa at 36,000 followed by the formation of African-specific Khoisanid and Black African phenotypes.

“Eurasians would have been roughly 20,000 years out of Africa”

More likely, the other way around….

• occamseraser

See JJ Shea’s recent article in Current Anthropology on the demise of the rorshach concept of “behavioral modernity” in archeology. R.I.P.

“Paleolithic archaeologists conceptualize the uniqueness of Homo sapiens in terms of “behavioral modernity,” a quality often conflated with behavioral variability. The former is qualitative, essentialist, and a historical artifact of the European origins of Paleolithic research. The latter is a quantitative, statistically variable property of all human behavior, not just that of Ice Age Europeans. As an analytical construct, behavioral modernity is deeply flawed at all epistemological levels. This paper outlines the shortcomings of behavioral modernity and instead proposes a research agenda focused on the strategic sources of human behavioral variability. Using data from later Middle Pleistocene archaeological sites in East Africa, this paper tests and falsifies the core assumption of the behavioral modernity concept—the belief that there were significant differences in behavioral variability between the oldest H. sapiens and populations younger than 50 kya. It concludes that behavioral modernity and allied concepts have no further value to human origins research. Research focused on the strategic underpinnings of human behavioral variability will move Paleolithic archaeology closer to a more
productive integration with other behavioral sciences.”

• http://www.kinshipstudies.org German Dziebel

Occamseraser, I obviously have a lot to say regarding Shea’s paper. In a nutshell, I think it’s an attempt to substitute the Africanist bias for the European bias. Africa has a bunch of sapient-looking fossils that are very old by extra-African standards (many of them re-dated in the 1980s from much more recent dates) but it doesn’t have European-style revolutionary toolkits. Let’s debunk European Paleolithic standards to help out-of-Africa prevail among archaeologists. The truth is the whole debate is completely misplaced into an isolationist archaeology/paleoanthropology debate about “behavioral modernity” that neither of them have the right data to discuss. It’s really an exercise from modern human cultural and linguistic variation down into the Pleistocene, not the other way around. It’s beyond doubt that modern humans are behaviorally unique. If archaeologists can’t circumscribe it on the basis of garbage pits and debitage, they shouldn’t even try. And Shea admits it himself, “There is more to human uniqueness than symbol use, language, and complex cognition. Homo sapiens differs from our near primate relatives in a variety of interesting ways. If we are looking for root causes of evolutionary change, the first place to look is among variation in reproductive strategies (Darwin 1859). For example, H. sapiens’ reproductive strategies involving pooled energy sources effectively balance slow child development with rapid birth rates (Kramer and Ellison 2010). These qualities almost certainly underwrote our species’ population growth, rapid geographic dispersals, construction of broad ecological niches, development of complex technology, and reliance on extensive social networks. While language, symbolism, and complex cognition might have played a role in the institution of such reproductive strategies, they are far more likely epiphenomena developed in service of those strategies. It will be difficult for us as Paleolithic archaeologists to investigate variation in human reproductive strategies with the kinds of data we are accustomed to discussing.”

Shea correctly notes, “New Guinea and Australia were demonstrably occupied by H. sapiens from 40 to 45 kya onward, and yet few of the traits used to define modern human behavior appear in this region consistently until mid-Holocene times (Habgood and Franklin 2008; O’Connell and Allen 2007; Stern 2009). It is difficult to reconcile this evidence with the notion of behavioral modernity as a significant and universal human quality. There is no evidence for any Pleistocene “human revolution” in Sahul comparable with that observed in western Eurasia.” But then he doesn’t mention the New World, which has notoriously elusive modern human record earlier than 15,000. It’s precisely in the New World and Sahul that we have problems detecting modern human behavioral signatures in the archaeological record AND there’s extreme diversity of languages, cultures and reproductive strategies that are fully compatible with the great antiquity of modern human behavior.

Shea’s notion of “behavioral variability” should encompass both archaeological and ethnological/linguistic patterns of variability across the whole globe. “The more you know the less you need” dictum that he keeps reiterating is a good paradigm to explain the archaeological record in the Americas: low population size, reliance on soft technologies and scavenging rather than hunting (Monte Verde is a clear example of such a decisively non-Upper Paleolithic and Late Stone Age adaptation) results in the proliferation of languages, kinship systems and folkloric traditions but not in archaeologically-paleontologically visible signatures of human behavioral and physical presence.

• Justin Giancola

thanks onur. If I did have :0 !BUY! one what do you use or recommend? point me towards a good free one? I seem to be using stuff that comes highly recommended already.

• onur

Justin, since you are trying to find a way to get rid of a specific virus, you need an antivirus program specifically dealing with that virus. And since I don’t know the name or type of that virus or which antivirus programs are effective against it, I cannot recommend any specific antivirus program for your problem. That is why I refrained from specific antivirus program recommendations in the first place.

But maybe Dienekes can help you, as reportedly he is too a computer programmer.

• Sandgroper

“New Guinea and Australia were demonstrably occupied by H. sapiens from 40 to 45 kya onward, and yet few of the traits used to define modern human behavior appear in this region consistently until mid-Holocene times…”

They had to get there in the first place, no mean technological feat.

• Sandgroper

Plus there is rock art clearly dated to much earlier than Lascaux. Pretty sophisticated stuff too, clearly recognisable animals and such (and interestingly, some depictions which appear to be of animals which became extinct in the Pleistocene, like Thylacoleo carnifex).

Maybe that’s Shea’s point – he says “few of the traits”, not “none”.

• http://www.kinshipstudies.org German Dziebel

“he says “few of the traits”, not “none”.”

Yes, he refers to the fact that the modern human behavioral package (points, beads, ochre, rock art, burials, etc.) seems to be assembled slowly in Sahul, over 30,000 years. But some individual components thereof are visible from 40,000K on. Habgood and Franklin, The revolution that didn’t arrive: A review of Pleistocene Sahul, have a nice Fig. 9 to illustrate this process.

• Justin Giancola

Dang…this is one hell of virus! Leave to the Turks to take-over your shit and hold on to it with a death grip. maybe I’ll pick up Turkish through all the google searches…maybe that was their plan all along!

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE

Sign up to get the latest science news delivered weekly right to your inbox!

### Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

### About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at http://www.razib.com