People often want to know the extent to which a trait is genetic versus environmentally determined (e.g., “nature” versus “nurture”). This distinction is not nearly as clear cut as is usually assumed. Let’s consider the obvious example of height in a population, a trait that’s well known to be purely hereditary.
Many causes of a population’s height distribution are not hereditary. For example, a population’s height is also determined by economic factors, like whether there was malnourishment among children or pregnant mothers. Cultural causes may also be at play, such as the extent to which height was valued in mate selection or whether the population has a gender imbalance. Even smaller cultural elements, like whether it is common for children to be enrolled in grueling, height halting athletics, whether height altering surgery is used to satisfy aesthetic preferences, and whether growth hormone is administered to children, can have an effect. Height is also influenced by long term trends, such as the percentage of people in each age group.
So, to what extent is height genetic? There is no single answer to that question. It depends on what you mean, and it may be that none of the answers satisfy what you think you mean by the question.
You could ask the extent to which one individual person’s height (as measured at one particular point in time) was “caused” by genetics. Genetics did cause their height, in the sense that, if you swapped some of their genes, they would have had a different height. Their height was effected by their nutrition as an infant (e.g., protein, vitamin A, vitamin D, and calcium). If they had better or worse nutrition, their height could have been different. Their height was also caused by their age since if you’d measured their height when they were one year old, they would have been a lot shorter.
Suppose you have an apple tree that has seven apples. What caused it to have seven apples as opposed to fewer or more? Was it the amount of sunlight it got? The amount of water? The average temperature? The soil it was planted in? This question doesn’t make sense to ask. The number of apples that this tree has is a complex function of all of these variables (plus many more), and even if we knew this complex function, at best all we could say (by taking derivatives with respect to one variable at a time) is how the number of apples would have changed with a small change in one of the inputs (holding the others constant), but even that isn’t the question we are attempting to ask.
Okay, perhaps it is more meaningful to talk about groups rather than individuals. We can ask, within a particular group, the extent that height is genetically determined.
The first problem we encounter is that the answer may depend a lot on the group we pick. If we are talking about a group of males and females who are all the same age and who all received the same nutrition when young, then we’ll find that a larger percentage of the variation in height is explainable from genetics than if we’re talking about a group of males of varying ages and varying early nutrition. The more variation that happens to have been caused by non-genetic factors like nutrition, the less heritable we’ll find the trait to be.
For instance, according to Scientific American [1], a study of Finnish twins found the heritability of height to be 78% for men and 75% for women. Another study found the heritability of height in China to be 65%. Hence, it technically makes no sense to say, “How much of height is determined by genetics? (full stop)” We have to pick a group G and then ask, “How much of height for group G is determined by genetics?” We’ll get a different answer for Finland than we will for China, and we’ll get a different answer for Finland in 2017 than we will for Finland in 1990. The distribution of genes changes over time in a population, as do the strength of environmental impacts, which, of course, also change the extent to which traits seem to be caused by genetics.
Okay, so suppose we’ve picked a group G. How then would we calculate how heritable height is within that group G? Well, ideally, we’d look at identical twins within G that were separated at birth and raised by different families. Because the age of these twins is identical and the genetic factors are close to identical (save for chance mutations and differeing gene expression), if the environmental factors are selected purely at random (with likelihood representative of how those factors affect members of G), we can assume that differences in the height of identical twins are mostly caused by non-genetic factors. Hence, we can look at the ratio of the amount height varies among twins raised in different families (the environmental component) to the amount that height varies among unrelated people raised in different families (which includes both genetic and environmental components). A similar sort of analysis can be done by comparing identical twins to fraternal twins.
Unfortunately, even these sorts of analyses make a variety of assumptions. They assume that the hormonal environment of the womb (which is shared by identical twins but which may not be mainly genetic) has no effect on height. They assume that the families that give birth to identical twins and the families that adopt them are representative of the broader population (e.g., if twins are adopted only by families that can provide for their children proper amounts of nutrition whereas many other families don’t, then the calculations will be skewed). And they assume that there are no hidden non-linear interactions between genetics and environment (for instance, if some genes cause a person’s height to be more sensitive to early nutrition, we may falsely chalk up any such height differences purely to nutrition differences when genetics are partly responsible). There is also, of course, sampling error. A large number of twins is hard to come by, and if we only manage to get 100 of them, the error in our heritability estimates could be +- 10% from just noise alone, not taking into account the other potential issues.
While the measurement of height is quite straightforward, for other variables of interest, we may find different heritability depending on the exact definition of the trait used. For instance, this remarkable chart [3] showing happiness heritability estimates that range from 18%-47% (mean=32%) for life satisfaction measures and 23%-59% (mean=36%) for wellbeing measures. The wide range of values occurs in part because of the variability in which measures of happiness were used [4], but also because of the different populations studied, different techniques used to make the estimates, and ordinary sampling error [5].
There are enough challenges with computing heritability that, at least occasionally, when the assumptions of the models are accidentally violated, the results of such analyses are hilariously non-sensical [2]: “‘Did you have your back rubbed’ came out to 92% [heritable] for males and 21% for females. Many of the resulting [narrow heritability estimates] exceed 100%. For example, the HRs of ‘Did you wear sunglasses after dark?’ Are 130% for males and 103% for females.”
What is perhaps most pernicious about calculations of heritability, though, is that people assume a high level of heritability implies a trait cannot be changed through environmental interventions, and more generally, people assume that the proportion that is heritable cannot be changed. You hear this, for instance, when people investigate the heritability of happiness and assume that the heritable portion is fixed, a predestined part of our fate that we societally have no control of.
The idea that the proportion attributed to genetics is unchangeable is false. For instance, you may have a group where CURRENTLY nearly all of the variation in height is determined by genetics, but height could still be dramatically changed via interventions such as better childhood nutrition or growth hormone. Heritability only tells us about the fraction of the CURRENT variation that seems to be due to environmental differences, not the fraction that COULD BE if we were to change the environment.
[1] https://www.scientificamerican.com/…/how-much-of-human…
[2] On Models and Muddles of Heritability https://link.springer.com/…/10.1023%2FA%3A1018358504373…
[3] https://www.ncbi.nlm.nih.gov/…/PMC4346667/figure/Fig2/
[4] https://www.ncbi.nlm.nih.gov/…/PMC4346667/table/Tab1/
[5] Genetics of Wellbeing and Its Components Satisfaction with Life, Happiness, and Quality of Life – https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4346667/
Comments