"Understanding the Genetics of Intelligence : Can Height Help? Can Corn Oil?", Johnson 2010; apropos of the sudden flurry of interest in BGI & Hsu...
"Although the subject is controversial, identifying the specific genes that contribute to general cognitive ability (GCA) has seemed to have good prospects, at least among psychological traits. GCA is reliably and validly measured and strongly heritable, and it shows genetically mediated physiological associations and developmental stability. To date, however, results have been disappointing. Human height shows these measurement characteristics even more strongly than GCA, yet data have indicated that no individual gene has more than trivial effects and this is also true for corn oil. The potential for environmental trigger of genetic expression, long recognized in evolutionary and developmental genetics, as applied to these seemingly disparate traits, can help us to understand the apparent contradiction between the heritability of intelligence and other psychological traits and the difficulty of identifying specific genetic effects.
Three recent genome-wide association studies of height, including some 63,000 people, however, reported a total of 54 specific genetic variants involved (Visscher, 2008). Together, these variants might have accounted for about 5% of the variance in height. Two of these variants were reported in all three studies, and eight were reported in two. A subsequent study showed replication accounting for 1% of the variance (Sammalisto et al., 2008). Do we need more and larger studies to uncover all the genes involved, or have we learned enough to conclude that there are few, if any, specific genes with any consistently meaningful effects on height?"
[loaded question there; the GWAS clearly shows there are no genes with large effects, but "consistently meaningful effects" is a weird way of describing "small".]
"Height shares some other characteristics with GCA that make this a reasonable conjecture. Like GCA but unlike some other continuous traits such as skin color, height is a developmental trait, normally monotonically increasing throughout childhood and then stabilizing in adulthood until old age. This means that there are opportunities throughout development for environmental perturbations that could either interrupt or accentuate expression of their contributing genes. Height and GCA are also correlated, about .2, and the correlation is primarily genetically mediated (Silventoinen, Posthuma, van Beisterveldt, Bartels, & Boomsma, 2006), so some of the same genes might even be involved. Both height and GCA are affected by known mutations considered abnormal, but these are rare and do not account for much ‘‘normal’’ population variance. Both show considerable national mean differences, and the national means vary with national income and national income disparity (Lynn, 2008; Steckel, 2008), suggesting either the presence of national differences in relevant gene frequency or important environmental variables, or both.
There are some other puzzles involving height that are potentially relevant to GCA. Though its secular trend has generally been increasing, there have been many exceptions. For example, northern Europeans in the 11th century averaged 3 inches taller than their descendents in the 18th, which is generally attributed to increased urbanization and industrialization during this period, which initially resulted in poorer nutrition and living conditions due to crowding, absence of light, and so on. Over the last 140 years, due to stagnation of increases in U.S. height but continuing increase in the Netherlands, average U.S. height has gone from being 3 inches taller than the Dutch to 3 inches shorter (Komlos & Lauderdale, 2007). Average Canadian height, however, continues to increase, although not as fast as Dutch (Cranfield & Inwood, 2007). South Koreans average about 6 inches taller than North Koreans (e.g., Schwekendiek, 2009), a difference almost impossible to attribute to genetics. Taken together, all of these puzzles suggest large environmental effects on this 80% heritable trait. That is, depending on environmental circumstances, the height of any particular individual could range considerably, but within population groups, most observed differences are strongly genetically influenced because relevant environmental circumstances are similar."
[Just as Khan reminds us: when the environment improves, the remaining variability becomes due more to genetics.]
"If we select for breeding only those plants or livestock (or whatever) that are at least, say, 1 standard deviation above the mean on some trait, the standarized difference between the mean in the original generation and the mean in the offspring generation is heritability, or the proportion of variance in the original generation that can be attributed to genetic influences. Of course we do not selectively breed humans, but psychologists interested in the presence of genetic influences on behavior realized that they could make use of the underlying quantitative genetic concepts. In selective breeding, the offspring generation presumably is more genetically homogeneous than the original, and thus shows less variance in the trait. Geneticists had thought that, if they kept on doing this, over time all the genetic variance would be eliminated as genes associated with lower levels of the trait were selected out, leaving only fixed genes for higher levels of the trait. As in all areas of science, however, it makes sense to test even the ideas that seem most obvious. Thus, in one of the longest-running experiments ever, since 1896 geneticists at the University of Illinois have been looking at corn’s response to selection for oil content (Hill, 2005). Like GCA and height in humans, corn oil production is a continuous, polygenic trait, and corn survives well with a broad range of oil production levels.
...Beginning at the same average 5%, oil production decreased steadily and reached effectively 0 at about the 84th generation, at which point these plants were no longer viable. After 50 years, some of the corn that had been selected for high oil content was selected instead for low oil content, and vice versa. This produced the green and blue lines in Figure 3, respectively. After another 5 years, selection was reversed again for some of the corn on which selection had been reversed at 50 years. This produced the red line in Figure 3.
Geneticists had expected that, as each selected line became more genetically homogeneous from generation to generation, its ability to respond to selection in the opposite direction would decrease because the genes involved in opposite levels of corn oil would no longer be present in that selected line. That was not, however, what happened. In fact, the responses to selection after 50 and 55 generations were as great as the original responses (the slopes of the lines beginning at 50 generations in Fig. 3). The only ways for this to occur are for genes to contribute to oil production against some genetic backgrounds but not others, or for newly arising genes (in the form of mutations) to become involved.
...Given their very different selection histories, it is quite likely that this same level of oil content is produced by very different combinations of genes in the two lines. About 50 identified genetic loci appear to account for about 50% of the genetic variation in corn oil content (Laurie et al., 2004), but the others remain unknown.
Despite the much greater level of control possible in evaluating the genetics of corn oil and the huge commercial payoff that could be realized by identifying the genes involved, we have done only marginally better in understanding the genetics of corn oil content than we have in understanding the genetics of human height. There is no reason to think that what goes on with the genetics of corn oil is necessarily relevant to human height, let alone GCA or any other psychological trait, but it is very reasonable to think that it might be."
[50% of variation is still a lot and may be worth embryo-selecting for.
"Although the subject is controversial, identifying the specific genes that contribute to general cognitive ability (GCA) has seemed to have good prospects, at least among psychological traits. GCA is reliably and validly measured and strongly heritable, and it shows genetically mediated physiological associations and developmental stability. To date, however, results have been disappointing. Human height shows these measurement characteristics even more strongly than GCA, yet data have indicated that no individual gene has more than trivial effects and this is also true for corn oil. The potential for environmental trigger of genetic expression, long recognized in evolutionary and developmental genetics, as applied to these seemingly disparate traits, can help us to understand the apparent contradiction between the heritability of intelligence and other psychological traits and the difficulty of identifying specific genetic effects.
Three recent genome-wide association studies of height, including some 63,000 people, however, reported a total of 54 specific genetic variants involved (Visscher, 2008). Together, these variants might have accounted for about 5% of the variance in height. Two of these variants were reported in all three studies, and eight were reported in two. A subsequent study showed replication accounting for 1% of the variance (Sammalisto et al., 2008). Do we need more and larger studies to uncover all the genes involved, or have we learned enough to conclude that there are few, if any, specific genes with any consistently meaningful effects on height?"
[loaded question there; the GWAS clearly shows there are no genes with large effects, but "consistently meaningful effects" is a weird way of describing "small".]
"Height shares some other characteristics with GCA that make this a reasonable conjecture. Like GCA but unlike some other continuous traits such as skin color, height is a developmental trait, normally monotonically increasing throughout childhood and then stabilizing in adulthood until old age. This means that there are opportunities throughout development for environmental perturbations that could either interrupt or accentuate expression of their contributing genes. Height and GCA are also correlated, about .2, and the correlation is primarily genetically mediated (Silventoinen, Posthuma, van Beisterveldt, Bartels, & Boomsma, 2006), so some of the same genes might even be involved. Both height and GCA are affected by known mutations considered abnormal, but these are rare and do not account for much ‘‘normal’’ population variance. Both show considerable national mean differences, and the national means vary with national income and national income disparity (Lynn, 2008; Steckel, 2008), suggesting either the presence of national differences in relevant gene frequency or important environmental variables, or both.
There are some other puzzles involving height that are potentially relevant to GCA. Though its secular trend has generally been increasing, there have been many exceptions. For example, northern Europeans in the 11th century averaged 3 inches taller than their descendents in the 18th, which is generally attributed to increased urbanization and industrialization during this period, which initially resulted in poorer nutrition and living conditions due to crowding, absence of light, and so on. Over the last 140 years, due to stagnation of increases in U.S. height but continuing increase in the Netherlands, average U.S. height has gone from being 3 inches taller than the Dutch to 3 inches shorter (Komlos & Lauderdale, 2007). Average Canadian height, however, continues to increase, although not as fast as Dutch (Cranfield & Inwood, 2007). South Koreans average about 6 inches taller than North Koreans (e.g., Schwekendiek, 2009), a difference almost impossible to attribute to genetics. Taken together, all of these puzzles suggest large environmental effects on this 80% heritable trait. That is, depending on environmental circumstances, the height of any particular individual could range considerably, but within population groups, most observed differences are strongly genetically influenced because relevant environmental circumstances are similar."
[Just as Khan reminds us: when the environment improves, the remaining variability becomes due more to genetics.]
"If we select for breeding only those plants or livestock (or whatever) that are at least, say, 1 standard deviation above the mean on some trait, the standarized difference between the mean in the original generation and the mean in the offspring generation is heritability, or the proportion of variance in the original generation that can be attributed to genetic influences. Of course we do not selectively breed humans, but psychologists interested in the presence of genetic influences on behavior realized that they could make use of the underlying quantitative genetic concepts. In selective breeding, the offspring generation presumably is more genetically homogeneous than the original, and thus shows less variance in the trait. Geneticists had thought that, if they kept on doing this, over time all the genetic variance would be eliminated as genes associated with lower levels of the trait were selected out, leaving only fixed genes for higher levels of the trait. As in all areas of science, however, it makes sense to test even the ideas that seem most obvious. Thus, in one of the longest-running experiments ever, since 1896 geneticists at the University of Illinois have been looking at corn’s response to selection for oil content (Hill, 2005). Like GCA and height in humans, corn oil production is a continuous, polygenic trait, and corn survives well with a broad range of oil production levels.
...Beginning at the same average 5%, oil production decreased steadily and reached effectively 0 at about the 84th generation, at which point these plants were no longer viable. After 50 years, some of the corn that had been selected for high oil content was selected instead for low oil content, and vice versa. This produced the green and blue lines in Figure 3, respectively. After another 5 years, selection was reversed again for some of the corn on which selection had been reversed at 50 years. This produced the red line in Figure 3.
Geneticists had expected that, as each selected line became more genetically homogeneous from generation to generation, its ability to respond to selection in the opposite direction would decrease because the genes involved in opposite levels of corn oil would no longer be present in that selected line. That was not, however, what happened. In fact, the responses to selection after 50 and 55 generations were as great as the original responses (the slopes of the lines beginning at 50 generations in Fig. 3). The only ways for this to occur are for genes to contribute to oil production against some genetic backgrounds but not others, or for newly arising genes (in the form of mutations) to become involved.
...Given their very different selection histories, it is quite likely that this same level of oil content is produced by very different combinations of genes in the two lines. About 50 identified genetic loci appear to account for about 50% of the genetic variation in corn oil content (Laurie et al., 2004), but the others remain unknown.
Despite the much greater level of control possible in evaluating the genetics of corn oil and the huge commercial payoff that could be realized by identifying the genes involved, we have done only marginally better in understanding the genetics of corn oil content than we have in understanding the genetics of human height. There is no reason to think that what goes on with the genetics of corn oil is necessarily relevant to human height, let alone GCA or any other psychological trait, but it is very reasonable to think that it might be."
[50% of variation is still a lot and may be worth embryo-selecting for.