A key question in genetics is, How much of the variation in a population is due to genetic factors and how much to environmental factors? In the popular press, this question is often phrased in terms of nature versus nurture—that is, what is the influence of innate (genetic) factors compared to external (environmental) factors? Answers to some nature-versus-nurture questions are of practical importance. If high blood pressure is primarily due to lifestyle choices (environment), then changes in diet or exercise habits would be most appropriate. However, if high blood pressure is largely predetermined by our genes, then drug therapy may be recommended.

Quantitative geneticists have developed the statistical tools needed to estimate, with reasonable precision, the extent to which variation in complex traits is due to genes versus the environment. Below, we will describe these tools. At the end of this section, we will discuss the assumptions underlying these estimates and the limits to their utility.

Let’s begin by defining broad-sense heritability (H²) as the part of the phenotypic variance that is due to genetic differences among individuals in a population. Mathematically, we write this as the ratio of the genetic variance to the total variance in the population:

728

The H is squared because it is the ratio of two variances, which are measured in squared units. H² can vary from 0 to 1.0. When all of the variation in a population is due to environmental sources and there is no genetic variation, then H² is 0. When all of the variation in a population is due to genetic sources, then Vg equals V_X and H² is 1.0. H² is called “broad sense” because it encompasses several different ways by which genes contribute to variation. For example, some of the variation will be due to the contributions of individual genes. Additional genetic variation can be contributed by the way genes work together, the interactions between genes, or epistasis.

In Section 19.2, we showed how we can calculate the genetic and environmental variances when we have inbred lines or clones. For the imaginary example of days to pollen shed for maize inbred lines in Table 19-2 (experiment I), we saw that V_g is 12.0 and V_X is 14.67. Using these values, the heritability of the trait is 12.0/14.67 = 0.82, or 82 percent. This estimate of H² tells us that genes contribute most of the variation and environmental factors contribute a more modest share of the variation. Thus, we might conclude that days to pollen shed is a highly heritable trait in maize.

Let’s look at the data for experiment II in Table 19-2. The genotypes are exactly the same as in experiment I; these are the genotypes of the inbred lines A through J. In this case, however, the lines are reared in more extreme environments. If we calculate the variance for the means of the inbred line in experiment II, V_g will be 12.0 days² as in experiment I. Since the genotypes are the same in both experiments, the genetic variance is the same. If we calculate the variance for the means of the different environments (V_e) in experiment II, we will obtain 24.0 days², which is much larger than the value for V_e in experiment I (2.67). Since the environments are more extreme, the environmental variance is larger. Finally, if we calculate H² for experiment II, we obtain

The estimate of H² for experiment II is on the small side—closer to 0 than to 1. Thus, we might conclude that days to pollen shed is not a highly heritable trait in maize.

The contrast between the estimates of the heritability for the same set of maize inbred lines reared in different environments highlights the point that heritability is the proportion of the phenotypic variance (V_X) due to genetics. Since V_X = V_g + V_e, as V_e increases, then V_g will represent a smaller part of V_X and H² will go down. Similarly, if the environmental variance is kept to a minimum, then V_g will represent a larger part of V_X and H² will go up. H² is a moving target, and results from one study may not apply to another.

How can we measure heritability in humans? Although we don’t have inbred lines for humans, we do have genetically identical individuals— monozygotic or identical twins (Figure 19-5). In most cases, identical twins are raised in the same household and so experience a similar environment. When individuals with the same genotypes are reared in the same environments, we have violated the assumption of our genetic model that genes and environment are independent. So, to estimate heritability in humans, we need to use sets of identical twins who were separated shortly after birth and reared apart by unrelated adoptive parents.

The equation for estimating H² in studies of identical twins who are reared apart is relatively simple. It makes use of the statistical measure called the covariance, which was introduced in Box 19-2. As explained in Box 19-3, the covariance between identical twins who are reared apart is equal to the genetic variance (V_g). Thus, we can estimate H² in humans by using this covariance as the numerator and the trait variance (V_x) as the denominator:

729

Here’s how it’s done. For each set of twins, let’s designate the trait value for one twin as X′ and the other as X″. If we have n sets of twins, then the trait values for the n sets could be designated .

Suppose we had IQ measurements for five sets of twins as follows:

730

Using these data and the formula for the covariance from Box 19-2, we calculate that COV_X′,X″ is 119.2 points². Using the formula for trait variance, we would calculate that the value of V_X is 154.3 points². Thus, we obtain

The points² in the numerator and denominator cancel out, and we are left with a unitless measure that is the proportion of the total variance that is due to genetics.

Box 19-3 provides some additional details about estimating H² from twin data, including the derivation of the formula we just used. It also discusses the relationship between the ratio COV_X′,X″/V_X and the correlation coefficient. Quantitative geneticists have developed several means for estimating heritability using the correlation among relatives. Identical twins share 100 percent of their genes, while brothers, sisters, and dizygotic twins share 50 percent of their genes. The strength of the correlation between different types of relatives can be scaled for the proportion of their genes that they share and the results used to estimate the genetic and environmental contributions to trait variation.

Over the last 100 years, there have been extensive genetic studies of twins and other sets of relatives. A great deal has been learned about heritable variation in humans from these studies. Table 19-4 lists some results from twin studies. It may or may not be surprising to you, but there is a genetic contribution to the variance for many different traits, including physique, physiology, personality attributes, psychiatric disorders, and even our social attitudes and political beliefs. We readily observe that traits such as hair and eye color run in families, and we know these traits are the manifestation of genetically controlled biochemical, developmental processes. In this context, it is not so surprising that other aspects of who we are as people also have a genetic influence.

Twin studies and the estimates of heritability that they provide can easily be over- or misinterpreted. Here are a few important points to keep in mind. First, H² is a property of a particular population and environment. For this reason, estimates of H² can differ widely among different populations and environments. We saw this phenomenon above in the case of the days to pollen shed for maize inbred lines. Second, the twin sets used in many studies were separated at birth and placed into adoptive homes. Adoption agencies do not assign babies randomly to the full range of households in a society; rather, they place babies in economically, socially, and emotionally stable households. As a result, V_e is smaller than in the general population, and the estimate of H² will be inflated. Accordingly, the published estimates likely lead us to underestimate the importance of environment and overestimate the importance of genetics. Third, for twins, prenatal effects could cause a positive correlation between genotype and environment. As we saw in the case of Thoroughbreds and jockeys above, such a correlation violates our model and will bias H² upward.

731

Finally, heritability is not useful for interpreting differences between groups. Table 19-4 shows that the heritability for height in humans can be very high: 0.88. However, this high value for heritability does not tell us anything about whether groups with different heights differ because of genetics or the environment. For example, men in the Netherlands today average 184 cm in height, while around 1800, men in the Netherlands were about 168 cm tall on average, a 16-cm difference. The gene pool of the Netherlands has probably not changed appreciably over that time, so genetics cannot explain the huge difference in height between the current population and the one of 200 years ago. Rather, improvements in health and nutrition are the likely cause. Thus, even though height is highly heritable and the past and present Dutch populations differ greatly in height, the difference has an environmental basis.