Another way of estimating heritability with phenotypic data alone is through the response to selection. This approach is widely used in agriculture. Keeping the environmental factors as constant as possible across the generations, plant and animal breeders select some segment of the parental population on the basis of their phenotype (e.g., the cows that produce the most milk and the plants with the highest yields) and breed these selected individuals only. Obviously, the point of all this is to alter the phenotypic distribution in the next generation along a desired direction. The Fisherian model given above allows us to predict what the response to selection will be. From equation 9.18 or Figure 9.1, we see that the offspring phenotype can be regressed upon the midparent values. Now suppose that only some parental pairs are selected to reproduce. Assume that the selected parents have a mean phenotype /s that differs from the mean of the general population, /, as shown in Figure 9.2. The intensity of selection S = /s — / is the mean phenotype of the selected parents minus the overall mean of the total population (the selected and nonselected individuals). We can then use the regression line to predict the phenotypic response of the offspring to the selection, as shown in Figure 9.2, to be
where h2 is the heritability of the phenotype being selected and response to selection R the is measured by the mean phenotype of the offspring (/o) minus the overall mean (selected and nonselected individuals) of the parental generation, that is, R = /o
Equation 9.22 shows that a population can only respond to selection (R) if there is both a selective force (S) and heritability (h2) for the trait. If there is no additive genetic variation, heritability is zero, so there will be no response to selection no matter how intense the selection may be. Hence, the only aspects of the genotype-phenotype relationship that are important in selection (and hence in agricultural breeding programs and, as we will see in Chapter 11, adaptation) are those genetic aspects that are additive as a cause of phenotypic variation.
If the heritability of a trait is known, say from studies on the phenotypic correlations among relatives, then equation 9.22 allows the prediction of the response to selection. However, if the heritability is not known, then the heritability can be estimated by monitoring the intensity and response to selection as h2 = R/S. For example, Clayton et al. (1957) examined variation in abdominal bristle number in a laboratory population of the fruit fly Drosophila melanogaster. They first estimated the heritability of abdominal bristle number in this base population through offspring-parent regression (equation 9.18) to be 0.51. In a separate experiment, Clayton et al. selected those flies with high bristle number to be the parents of a selected generation. The base population had an average of 35.3 bristles per fly, and the selected parents had a mean of 40.6 bristles per fly. Hence, the intensity of
selection is S = 40.6 - 35.3 = 5.3. The offspring of these selected parents had an average of 37.9 bristles per fly, so the response to selection is R = 37.9 - 35.3 = 2.6. Fromequation 9.22, the heritability of abdominal bristle number in this population is now estimated to be R/S = 2.6/5.3 = 0.49, a value not significantly different from the value estimated by offspring-parent regression.
Was this article helpful?