The Axis of Individual Heterozygosity: Theory
Whether nature ranks and truncates, or approximates this behavior, is an empirical question, yet to be answered.
J. F. Crow and M. Kimura, "Efficiency of Truncation Selection" ( 1979)
For several decades before population geneticists used electrophoresis to estimate the levels of genetic variation in natural populations, empirical geneticists believed that natural populations contained an abundance of genetic variation ( Dobzhansky 1970). Nevertheless, the first quantitative estimates from electrophoretic surveys seemed to surprise many geneticists. One of their first reactions to the empirical estimates was that if the abundant genetic variation were maintained by balancing selection, it would generate an intolerable genetic load ( Lewontin and Hubby 1966). The new data thus forced a sweeping reanalysis of the population genetic models of fitness determination.
Segregational genetic load is defined as the decrement in average fitness within a population in comparison with a population composed solely of the most fit genotype. When population geneticists first discovered that one-third to one-half the proteins in population samples of Drosophila and humans were polymorphic, consideration of genetic load led some biologists to propose that the genetic variation was functionally and adaptively neutral. They proposed neutrality because the estimated level of segregational genetic load was so high that it could be borne only by the most fecund species. Kimura and Crow ( 1964) presented the problem, but when Lewontin and Hubby ( 1966) restated it after estimating the level of genetic variation in Drosophila pseudoobscura, the message hit home. If a representative protein polymorphism were maintained by balancing selection favoring the heterozygote,
then the average fitness would be (1 - s)L, where s is the average selection coefficient over all genotypes and L is the number of loci experiencing this form of balancincy selection. There are more than 1,000 enzymes in a eukaryote but probably fewer than 2,000. If we estimate that the average selection coefficient is .05 and the number of en-