The genetic correlation between educational attainment, intracranial volume and IQ is due to recent polygenic selection on general cognitive ability. Evidence for opposite directional selection on stature.
This is the first paper to report evidence of a cross-population genetic correlation between educational attainment and IQ. Principal component analysis (PCA) was used to identify the underlying genetic structure resulting from recent directional selection. Converging evidence from two different genetic databases (1000 Genomes and ALFRED, comprising 14 populations and 8 racial groups, respectively ) and two independent sets of genes (14 in total)found by GWAS to have a significant association with IQ and Educational Attainment, reveals a high correlation (around 0.9) between these two social science constructs across populations and ethnic groups. This correlation in turn provided convergent validity for the genetic measures identified with PCA. Two alleles associated with higher intracranial volume were positively correlated to the two PCs.
A separate analysis carried out on an independent set of alleles related to human stature suggests that this trait was subject to different directional selective pressures and provided evidence for the discriminant validity of alleles supposedly related to IQ and educational attainment.One-way ANOVA showed that the average frequencies of height increasing alleles for three racial groups (East Asian, Sub-Saharan African and European) differed significantly and that Africans had significantly higher frequencies than East Asians.
A significant negative genetic correlation (r around 0.9) between height and intelligence was found at a cross-population level, contrasting the generally positive correlation found within populations. Evolutionary explanations in terms of Allen’s rule and the theory of cold winters are attempted. The GWAS hits for height were from studies carried out on different ethnic groups, and the converging results suggests that the same genes have similar effects across racial groups.
This paper provides evidence that Principal Component Analysis and ANOVA can be used to detect deviations from random drift and detect directional selection on polygenic traits in recent human evolutionary history.
Back to [Archive] Post-review discussions
This is a fascinating paper and certainly deserves publication. Two points:
You describe selection for human intelligence as "recent" throughout your paper. How recent is "recent"? This point is important because the word "recent" has different meanings for different people, particularly in this context. For many anthropologists, selection for human intelligence took place in the Palaeolithic. For others, it occurred especially when modern humans entered sub-Arctic and Arctic environments some 30,000 years ago. For others, like Henry Harpending and Greg Cochran, it took off 10,000 years ago with the advent of agriculture. Finally, Gregory Clark sees another phase of selection beginning 1,000 years ago, when stable, State-pacified societies began to arise in Western Europe, thereby giving a reproductive advantage to the nascent middle class.
There is probably some truth in all of these theoretical models. You seem to favour the second one, but what about the others? Will it be possible to estimate the time frame when these different derived alleles rose to their current frequencies in non-African populations?
I would have elaborated on your statement: "colder climates pose two evolutionarily novel problems that would have required high intelligence to solve: finding food and keeping warm" (Note: replace "poses" with "pose"). You might note here that this is not simply a hypothesis. It has been shown that technological complexity inversely correlates with mean temperature among hunter-gatherers. A good reference is Hoffecker (2002, p. 10):
"The technology of recent hunter-gatherers is also influenced by temperature and diet. Both the diversity of tool types and the complexity of individual tools and weapons […] increase as effective temperature and the percentage of plant foods in the diet decline […]. This apparently reflects the need for greater foraging efficiency in habitats where resources are available for limited periods of time. Recent hunter-gatherers in cold environments also tend to make increased use of storage technologies and untended facilities (e.g., traps and snares). The former represent another adaptive response to seasonal variations in resource availability, while the latter reflect an efficient approach (i.e., reduced mobility) to collecting unpredictable and widely dispersed resources […]. Finally, modern hunter-gatherers in northern environments produce relatively complex technology for heat conservation and cold protection (e.g., tailored fur clothing)."
Another good reference is Zvelebil (2009, p. 170):
"In such [non-tropical] areas, one or two seasonally abundant resources may be relied on to produce the critical storable surplus for the lean seasons. This would require short periods of intensive harvest and precise scheduling during those times of the year when these resources were available. In such ‘time-stressed environments’, time was at a premium and hunter-gatherer societies responded by developing time-saving devices: by budgeting their time and by preparing in advance more sophisticated, but also more complicated tools designed for the specific tasks involved. The development of capture facilities, such as pits, traps, weirs, and nets can be also seen as time-saving devices. Another technological requirement for effective exploitation of seasonal resources consists of storage. […] These technological developments, combined with the development of the microlithic industry, could be called, with some justification, the original industrial revolution."
References
Hoffecker, J.F. (2002). Desolate Landscapes. Ice-Age Settlement in Eastern Europe. New Brunswick: Rutgers University Press.
Zvelebil, M. (2009). Hunters in Transition: Mesolithic Societies of Temperate Eurasia and Their Transition to Farming, Cambridge University Press.
You describe selection for human intelligence as "recent" throughout your paper. How recent is "recent"? This point is important because the word "recent" has different meanings for different people, particularly in this context. For many anthropologists, selection for human intelligence took place in the Palaeolithic. For others, it occurred especially when modern humans entered sub-Arctic and Arctic environments some 30,000 years ago. For others, like Henry Harpending and Greg Cochran, it took off 10,000 years ago with the advent of agriculture. Finally, Gregory Clark sees another phase of selection beginning 1,000 years ago, when stable, State-pacified societies began to arise in Western Europe, thereby giving a reproductive advantage to the nascent middle class.
There is probably some truth in all of these theoretical models. You seem to favour the second one, but what about the others? Will it be possible to estimate the time frame when these different derived alleles rose to their current frequencies in non-African populations?
I would have elaborated on your statement: "colder climates pose two evolutionarily novel problems that would have required high intelligence to solve: finding food and keeping warm" (Note: replace "poses" with "pose"). You might note here that this is not simply a hypothesis. It has been shown that technological complexity inversely correlates with mean temperature among hunter-gatherers. A good reference is Hoffecker (2002, p. 10):
"The technology of recent hunter-gatherers is also influenced by temperature and diet. Both the diversity of tool types and the complexity of individual tools and weapons […] increase as effective temperature and the percentage of plant foods in the diet decline […]. This apparently reflects the need for greater foraging efficiency in habitats where resources are available for limited periods of time. Recent hunter-gatherers in cold environments also tend to make increased use of storage technologies and untended facilities (e.g., traps and snares). The former represent another adaptive response to seasonal variations in resource availability, while the latter reflect an efficient approach (i.e., reduced mobility) to collecting unpredictable and widely dispersed resources […]. Finally, modern hunter-gatherers in northern environments produce relatively complex technology for heat conservation and cold protection (e.g., tailored fur clothing)."
Another good reference is Zvelebil (2009, p. 170):
"In such [non-tropical] areas, one or two seasonally abundant resources may be relied on to produce the critical storable surplus for the lean seasons. This would require short periods of intensive harvest and precise scheduling during those times of the year when these resources were available. In such ‘time-stressed environments’, time was at a premium and hunter-gatherer societies responded by developing time-saving devices: by budgeting their time and by preparing in advance more sophisticated, but also more complicated tools designed for the specific tasks involved. The development of capture facilities, such as pits, traps, weirs, and nets can be also seen as time-saving devices. Another technological requirement for effective exploitation of seasonal resources consists of storage. […] These technological developments, combined with the development of the microlithic industry, could be called, with some justification, the original industrial revolution."
References
Hoffecker, J.F. (2002). Desolate Landscapes. Ice-Age Settlement in Eastern Europe. New Brunswick: Rutgers University Press.
Zvelebil, M. (2009). Hunters in Transition: Mesolithic Societies of Temperate Eurasia and Their Transition to Farming, Cambridge University Press.
For completion about the colder climate theory, you might mention the exchange between these:
Templer, Donald I., and Hiroko Arikawa. "Temperature, skin color, per capita income, and IQ: An international perspective." Intelligence 34.2 (2006): 121-139.
Jensen, Arthur R. "Comments on correlations of IQ with skin color and geographic–demographic variables." Intelligence 34.2 (2006): 128-131.
Hunt, Earl, and Robert J. Sternberg. "Sorry, wrong numbers: An analysis of a study of a correlation between skin color and IQ." Intelligence 34.2 (2006): 131-137.
Templer, Donald I., and Hiroko Arikawa. "The Jensen and the Hunt and Sternberg comments: From penetrating to absurd." Intelligence 34.2 (2006): 137-139.
Specifically, Hunt and Sternberg wrote that:
Perhaps Peter Frost knows whether there are any anthropological studies of technology and parasite prevalence among HG's.
I have a speculative hypothesis. While intelligence can combat parasites (there is a global strong negative r (IQ x parasite prevalence), there is a threshold effect. Slight increases in intelligence from a low level (compared to present day humans) have no effect on combating parasites. It is first once one gets to the level where one can develop advanced medicine and proper sanitation that intelligence starts to be a factor.
Contrarily, on the colder climate theory, there seems to be no such threshold effect. Even slight increases in intelligence helps. I think of it in the fitness landscape. If moving in some direction does not go uphill, then natural selection will not move in that direction (except by drift), even if begins to go uphill later on in that direction. Evolution has no foresight.
Templer, Donald I., and Hiroko Arikawa. "Temperature, skin color, per capita income, and IQ: An international perspective." Intelligence 34.2 (2006): 121-139.
Jensen, Arthur R. "Comments on correlations of IQ with skin color and geographic–demographic variables." Intelligence 34.2 (2006): 128-131.
Hunt, Earl, and Robert J. Sternberg. "Sorry, wrong numbers: An analysis of a study of a correlation between skin color and IQ." Intelligence 34.2 (2006): 131-137.
Templer, Donald I., and Hiroko Arikawa. "The Jensen and the Hunt and Sternberg comments: From penetrating to absurd." Intelligence 34.2 (2006): 137-139.
Specifically, Hunt and Sternberg wrote that:
One might equally argue that the greater challenge of living in equatorial regions, such as fending off myriad parasitic diseases, should render equatorial people more intelligent.
Perhaps Peter Frost knows whether there are any anthropological studies of technology and parasite prevalence among HG's.
I have a speculative hypothesis. While intelligence can combat parasites (there is a global strong negative r (IQ x parasite prevalence), there is a threshold effect. Slight increases in intelligence from a low level (compared to present day humans) have no effect on combating parasites. It is first once one gets to the level where one can develop advanced medicine and proper sanitation that intelligence starts to be a factor.
Contrarily, on the colder climate theory, there seems to be no such threshold effect. Even slight increases in intelligence helps. I think of it in the fitness landscape. If moving in some direction does not go uphill, then natural selection will not move in that direction (except by drift), even if begins to go uphill later on in that direction. Evolution has no foresight.
Thanks for the useful references. I've incorporated them in the manuscript.
I understand that in anthropology "recent" means an event that occurred after the out of Africa migration, that is after about 60Kya. However, there is not full accord on the usage of this term. Additionally, "very recent" usually refers to the Holocene.
I think that recent means any event that contributes to the genetic divergence between human populations, as apparent from the use of this term in several papers, for example the Pickrell et al. (2009) being an example.
When this actually occurred is difficult to ascertain, although from my data it looks like there was continuos selection until very recent times, as suggested by the divergence between Native Americans and East Asians (thus after migration across Bering Strait, about 15kya). It is more difficult to assess whether there was selection before the Holocene. However, such relatively big frequency differences across many alleles probably require a longer time scale than 10ky. Most of the intelligence boosting alleles were present in the pre-out of Africa population, thus some selection must have taken place in “non recent” times as well. They seem not to be mutations that arose after human populations split 50Kya, otherwise they’d be at near zero frequencies among sub-Saharan Africans. At least, this is true for most of them. The table below analyzes the top 3 hits from Rietveld et al. The second (rs11584700 G)has very low among Africans, and its presence there could be the result of recent admixture with non-African populations or a more recent, independent mutation event. There are two alleles that are present only in Europe (and not in Asia or Africa) and probably originated after the split between Europeans and Asians , and thus can be considered European-specific mutations(rs2054125 T, rs13188378G ) . The same cannot be said about the 3 IQ increasing alleles reported in this paper or the other 7 educational attainment alleles, which are present at substantial frequencies among Africans (4-40%). Thus, probably the majority of the intelligence boosting mutations arose in non-recent times (before the out of Africa migration) but it is possible that some of them have a much more recent origin. The fact that the GWAS was carried out on Europeans prevents us from knowing which mutations are Asian or African-specific. If rs11584700 G emerged after the out-of-Africa migration, then it must have appeared before the split between European and Asian populations, as suggested by its frequencies among East Asians (31%).
Overall, it seems like there was continued selective pressure for increased intelligence since before the great human migration till very recent times (after the farming revolution).
I understand that in anthropology "recent" means an event that occurred after the out of Africa migration, that is after about 60Kya. However, there is not full accord on the usage of this term. Additionally, "very recent" usually refers to the Holocene.
I think that recent means any event that contributes to the genetic divergence between human populations, as apparent from the use of this term in several papers, for example the Pickrell et al. (2009) being an example.
When this actually occurred is difficult to ascertain, although from my data it looks like there was continuos selection until very recent times, as suggested by the divergence between Native Americans and East Asians (thus after migration across Bering Strait, about 15kya). It is more difficult to assess whether there was selection before the Holocene. However, such relatively big frequency differences across many alleles probably require a longer time scale than 10ky. Most of the intelligence boosting alleles were present in the pre-out of Africa population, thus some selection must have taken place in “non recent” times as well. They seem not to be mutations that arose after human populations split 50Kya, otherwise they’d be at near zero frequencies among sub-Saharan Africans. At least, this is true for most of them. The table below analyzes the top 3 hits from Rietveld et al. The second (rs11584700 G)has very low among Africans, and its presence there could be the result of recent admixture with non-African populations or a more recent, independent mutation event. There are two alleles that are present only in Europe (and not in Asia or Africa) and probably originated after the split between Europeans and Asians , and thus can be considered European-specific mutations(rs2054125 T, rs13188378G ) . The same cannot be said about the 3 IQ increasing alleles reported in this paper or the other 7 educational attainment alleles, which are present at substantial frequencies among Africans (4-40%). Thus, probably the majority of the intelligence boosting mutations arose in non-recent times (before the out of Africa migration) but it is possible that some of them have a much more recent origin. The fact that the GWAS was carried out on Europeans prevents us from knowing which mutations are Asian or African-specific. If rs11584700 G emerged after the out-of-Africa migration, then it must have appeared before the split between European and Asian populations, as suggested by its frequencies among East Asians (31%).
Overall, it seems like there was continued selective pressure for increased intelligence since before the great human migration till very recent times (after the farming revolution).
Duxide,
We should distinguish between two things: 1) the time of origin of the derived alleles; and 2) the increase in their incidence to present levels. Since homo sapiens is a young species, selection has tended to act on existing genetic variability (as opposed to waiting for new mutations to arise). What interests me is the trajectory of these derived alleles. Did they increase in incidence over time at roughly the same rate? Or did they exist at low levels for a long time before undergoing a rapid increase much later?
I know you can't answer these questions with the current data, but would it be possible to chart this kind of trajectory by looking at adjacent stretches of DNA? This seems to be the method most commonly used. Of course, retrieval of ancient DNA would be an even better method.
Chuck,
I'm skeptical about the role of parasite load. Parasite load would have decreased dramatically once early modern humans had left the tropics and entered seasonal temperate environments where the ground is frozen for part of the year. Thus, when we look at the correlation between mean temperature and technological complexity in hunter-gatherers, we should see a sharp increase in complexity between tropical and temperate environments. But this isn't what we see. There is just a steady increase with decreasing mean temperature
We should distinguish between two things: 1) the time of origin of the derived alleles; and 2) the increase in their incidence to present levels. Since homo sapiens is a young species, selection has tended to act on existing genetic variability (as opposed to waiting for new mutations to arise). What interests me is the trajectory of these derived alleles. Did they increase in incidence over time at roughly the same rate? Or did they exist at low levels for a long time before undergoing a rapid increase much later?
I know you can't answer these questions with the current data, but would it be possible to chart this kind of trajectory by looking at adjacent stretches of DNA? This seems to be the method most commonly used. Of course, retrieval of ancient DNA would be an even better method.
Chuck,
I'm skeptical about the role of parasite load. Parasite load would have decreased dramatically once early modern humans had left the tropics and entered seasonal temperate environments where the ground is frozen for part of the year. Thus, when we look at the correlation between mean temperature and technological complexity in hunter-gatherers, we should see a sharp increase in complexity between tropical and temperate environments. But this isn't what we see. There is just a steady increase with decreasing mean temperature
Duxide,
We should distinguish between two things: 1) the time of origin of the derived alleles; and 2) the increase in their incidence to present levels. Since homo sapiens is a young species, selection has tended to act on existing genetic variability (as opposed to waiting for new mutations to arise). What interests me is the trajectory of these derived alleles. Did they increase in incidence over time at roughly the same rate? Or did they exist at low levels for a long time before undergoing a rapid increase much later?
I know you can't answer these questions with the current data, but would it be possible to chart this kind of trajectory by looking at adjacent stretches of DNA? This seems to be the method most commonly used. Of course, retrieval of ancient DNA would be an even better method.
Could we compare with analogous genes on the Chimpanzee genomes (Bonobo, Common)? Or do we need genomes from fossils of previous hominids?
I know you can't answer these questions with the current data, but would it be possible to chart this kind of trajectory by looking at adjacent stretches of DNA? This seems to be the method most commonly used. Of course, retrieval of ancient DNA would be an even better method.
Could we compare with analogous genes on the Chimpanzee genomes (Bonobo, Common)? Or do we need genomes from fossils of previous hominids?
As you imply,answering these questions goes beyond the scope of the present paper. I do not see how retrieval of ancient DNA could be of help, unless we had frequencies among ancient populations, but it's doubtful we can get DNA samples from more than a couple of individuals for each ancient population. The derived alleles by definition are human specific mutations but of course, unless these mutations are present only in one race (i.e. Europe) and not in the others, as seems to be the case for some of the educational attainment boosting alleles, we cannot date the mutation event. Most of them appeared in Africa, but it's difficult to assess their frequencies before the great human diaspora. Unfortunately the current methods are based on the idea that strong selective sweeps act on a few genes and thus have very little power of detecting subtle changes at many loci.
The genetic correlation between educational attainment, intracranial volume and IQ is due to recent polygenic selection on general cognitive ability. Evidence for opposite directional selection on stature.
"This component can be interpreted as indicating the strength of natural selection (Piffer, 2014)."
Some IQ/AQ alleles are probably pleiotropic and subject to mixed selection. Some are probably functionally less important than others. If the polygenetic selection model for IQ/AQ is correct, I would guess, without much thought on the matter, that the alleles that load more on your natural selection factor would have larger IQ/AQ effects. Does this make sense? If so, is it testable?
The paper seems fine and should be published.
Some IQ/AQ alleles are probably pleiotropic and subject to mixed selection. Some are probably functionally less important than others. If the polygenetic selection model for IQ/AQ is correct, I would guess, without much thought on the matter, that the alleles that load more on your natural selection factor would have larger IQ/AQ effects. Does this make sense? If so, is it testable?
Yes, this is testable in theory. However, it's difficult to know which alleles have an effect on IQ, let alone knowing how much stronger this effect is compared to other alleles. A better measure is the significance value, which is an indicator of signal to noise ratio. That is, alleles with higher GWAS significance tend to load more strongly on my PC. I've got preliminary evidence that this is the case but I need a larger N to achieve significance. I used this method of correlated vectors in my previous paper, with the height increasing alleles, where I showed that the SNPs with lower p values load more strongly on the PC.
Simple Statistical Tools to Detect Signals of Recent Polygenic Selection.
Davide Piffer, IBC 2014, vol. 6, article no. 1, pp. 1-6 | doi: 10.4051/ibc.2014.6.1.0001
You can download the paper here: http://www.ibc7.org/article/journal_v.php?sid=317&page=1
This is an excellent paper and I recommend it for publication.
As 3 reviewers have agreed to publish this, I attach the last version of the document and the data set.
I have made some changes in formatting and presentation. Can Piffer verify this as acceptable?
Ideally, the abstract should be smaller. Right now it's about 1/13th of the length of the article itself. In general, abstract for shorter papers should be less than 150 words. The one in this paper is 305.
Ideally, the abstract should be smaller. Right now it's about 1/13th of the length of the article itself. In general, abstract for shorter papers should be less than 150 words. The one in this paper is 305.
Yes your changes are ok. I've corrected some typos. I have attached the updated version.
Here's a third version.
Corrected the spacing between tables at the end, deleting tons of empty lines.
Removed some grey background from some text (not sure why it was there).
Some other small minor corrections.
Corrected the spacing between tables at the end, deleting tons of empty lines.
Removed some grey background from some text (not sure why it was there).
Some other small minor corrections.
Here is the fourth revision. Replaced commas with dots in table 13. Should be fine now.
Here is the fourth revision. Replaced commas with dots in table 13. Should be fine now.
Here's the PDF output. If the author can confirm it is right, then I will publish.
Yes I confirm it's right.
© Copyright 2024 OpenPsych - All Rights Reserved