Nests, Neurons, and the Evolution of Behavior. How and Why Do Brain Cells Die?Biology 

Nests, Neurons, and the Evolution of Behavior

The Genius of Genome-Wide Association

Only recently has the field of behavioral genetics starting giving us the tools to examine some of this variance, and its phenotypic expression, in much higher resolution. One such method, genome-wide complex trait analysis (GCTA), estimates the chance of genetic similarity across hundreds of thousands of variants for each pair of individuals and relates this genetic similarity to phenotypic similarity for a particular trait. This analysis uses genetic variants called single-nucleotide polymorphisms (SNPs) that differ between individuals at a specific nucleotide. GCTA has consistently revealed a strong genetic component to cognitive ability and to socioeconomic status that corroborates twin studies, while making fewer assumptions.

Another powerful and promising tool in modern behavior genetics is genome-wide association (GWA), which is used to identify specific genetic correlates of observable traits. This method works by overlaying the genomic data from a large number of participants, which covers a wide range of common genetic variants, and looking for associations with phenotypic data from the participants. For example, if one version of a defined string of DNA shows up in participants with schizophrenia at a rate far above the population average, this string of DNA can be thought of as “associated” with schizophrenia.

The first successful GWA study, which found two genetic variants associated with macular degeneration, was performed in 2005. Since then, over 200 diseases and traits have been studied with this technique, and nearly 4,000 SNPs have been isolated. The GWA design has been used to track down SNPs that influence traits and susceptibilities such as longevity, schizophrenia, Alzheimer’s, various cancers, and more. The beauty of the genome-wide association technique is that it works just as nature does in all its elegance—by detecting, as natural selection does, whether a given allele is correlated with the phenotype.

The methodology of the GWA study has undergone a lot of evolution over the past decade, and it has not been without its critics, but its modern iteration has extraordinary predictive power. While ever far from explaining the full breadth of human behavior, this technique is beginning to look like a powerful tool in the quest to unpack the genetic correlates of real-world behavioral outcomes. A starting point was the landmark study by Cornelius Rietveld and colleagues which appeared in a 2013 issue of Science. Using genome-wide association in over 100,000 participants, the team found three SNPs that were highly replicable predictors of not only years of education but also—in an independent sample of some 24,000 participants—cognitive performance, as captured by a general factor (g) extracted from a diverse battery of tests. Together these SNPs accounted for a very small portion of the overall variance in educational attainment, but their genome-wide significance and reproducibility in a later independent study seems to confirm what twin researchers have always suspected: that cognitive performance and its many correlates, while strongly heritable, are influenced by a vast number of genes, each of small effect. Intelligence, like height, is a polygenic trait.

Evolution: From Genotype to Phenotype

The structure of a weaver’s nest is subject to countless environmental influences: the shape of the branches on which it is built; the stressors of wind, weather, and predators; available resources for building; and the random chance implicit in any macrophysical process. But the environment can only shape what already exists in the bird’s mind. In this way, a set of genetic variants controls the proteins that unfold in development, which control the organization of neurons and brain structures, which then control the preferences and abilities that lead, eventually, to a nest. At any point in the chain, environment and random chance can influence its shape. But all the causal agents, including the genes, are indispensable if a full understanding of the outcome is the goal. 

Evolution: Graph showing that variance in brain tissue is vastly overrepresented by the SNPs related to educational attainment. Redrawn from Extended Data Fig. 8b of Okbay et al. (2016). Graph by James J. Lee, used with permission.
Graph showing that variance in brain tissue is vastly overrepresented by the SNPs related to educational attainment. Redrawn from Extended Data Fig. 8b of Okbay et al. (2016). Graph by James J. Lee, used with permission.

In this year’s May 26 issue of Nature, research carried out by the Social Science Genetic Association Consortium (SSGAC) has built upon Rietveld’s 2013 study to link together part of our own behavioral chain. Using genome-wide association techniques, the team has identified 74 loci associated with years of educational attainment, a well-known predicted outcome of psychometric intelligence. This study replicates many of the SNPs found to influence educational attainment in the previous meta-analysis, and again, the same SNPs that predict variance in IQ also predict years in school. Remarkably, protein-coding regions in these loci are also strongly expressed in the neural tissue of the prenatal brain. This is borne out by the extent of overlap between the 74 loci associated with educational attainment and known clusters of genes that regulate neural development. Significantly enriched clusters include those for neuron migration, axon projection, dendritic sprouting, and synaptic plasticity.

The current estimation for total genetic contribution to variance in educational attainment is at least 20 percent. In the new study, the polygenic scores derived from these loci account for only about 3.2 percent of this variance so far, and each genetic locus on its own has a very small effect. But because of the vast size of the sample, the chance of even such a faint signal being random noise is essentially zero. Additionally, a follow-up study published July 19 found that these polygenic scores explained a whopping 15 percent of the variance in educational achievement scores at age 16—the strongest prediction of polygenic scores for a behavioral trait so far.

And so, by tracking variants in the human genome through neural development, to cognitive ability, and onward to years of schooling and even to achievement scores, behavioral geneticists have elegantly unearthed some of the proverbial genes “for” the human equivalent of nest building itself.

The Strongest Link in the Chain

Few people have a problem with ascribing a genetic basis to the height or eye color of humans, or to the nests and dams of birds and beavers. Yet to many, the idea that a few nucleotides can raise the boat for those who have them—even by a mere few weeks of education—will feel deeply uncomfortable. In the social sciences, one particularly salient paradigm to explain certain behavioral differences is privilege theory, the view that success in the real world is determined only by the unfair advantage society affords some people over others, rather than by innate abilities. Ascribing a genetic underpinning to some of these life outcomes is met with moral revulsion, because it seems to suggest an inflexibility to them—and that there is nothing we can do to help those less fortunate. 

Evolution: The blood of a two-week-old infant being collected for a phenylketonuria, or PKU, screening. Photo published on Wikimedia Commons by Eric T. Sheler; public domain.
The blood of a two-week-old infant being collected for a phenylketonuria, or PKU, screening. Photo published on Wikimedia Commons by Eric T. Sheler; public domain.

Read This