An experimental study of neutrality at the malic dehydrogenase and esterase-5 loci in Drosophila pseudoobscura

Polymorphism and the Red Queen: the selective maintenance of allelic variation in a deteriorating environment

Although allelic variation is ubiquitous in natural populations, our theoretical models are poor at predicting the existence and properties of these observed polymorphisms. In this study, inspired by Van Valen's Red Queen hypothesis, we modeled the effect of viability selection in a deteriorating environment on the properties of allelic variation in populations subject to recurrent mutation. In Monte Carlo simulations, we found that levels of polymorphism consistently built up over time. We censused the simulated populations after 10,000 generations of mutation and selection, revealing that, compared with models assuming a constant environment, the mean number of alleles was greater, as was the range of allele numbers. These results were qualitatively robust to the addition of genetic drift and to the relaxation of the assumption that the viabilities of phenogenotypes containing a new mutation are independent of each other (i.e. incorporating a model of generalized dominance). The broad range of allele numbers realized in the simulated populations-from monomorphisms to highly polymorphic populations-more closely corresponds to the observed range from numerous surveys of natural populations than previously found in theoretical studies. This match suggests that, contrary to the views of some writers, selection may actively maintain genetic variation in natural populations, particularly if the selective environment is gradually becoming harsher. Our simulations also generated many populations with heterozygote advantage, a mismatch with real data that implies that this selective property must arise extremely rarely in natural populations.

The maintenance of single-locus polymorphism by maternal selection

Population geneticists have long been interested in the ability of natural selection to maintain the levels of standing variation observed in natural populations. Here, we study the polymorphism-maintaining properties of maternal selection, in which the fitness of an individual is a function of its own and its mother’s genotype. Using a model proposed by Gavrilets, we first estimate the proportion of parameter/state space that preserves allelic variation, before investigating the construction of polymorphism over time through the joint action of mutation and selection. These two methods, the “parameter-space” and “constructionist” approaches, respectively, enable us to draw some general conclusions. We argue that, even though the proportion of parameter-state space allowing multiallele polymorphism is greater under maternal selection than under the standard model of constant viability selection, the former is, in fact, less likely to maintain large numbers of alleles. Nevertheless, variation that is balanced by maternal selection is likely to show elements of heterozygous advantage and be resistant to depletion by genetic drift. We observe that the population mean fitness frequently decreases after the successful invasion of a new mutation, but such declines are usually temporary.

The role of gene conversion in determining sequence variation and divergence in the Est-5 gene family in Drosophila pseudoobscura

Nucleotide sequences of eight Est-5A and Est-5C genes corresponding to previously sequenced Est-5B genes in Drosophila pseudoobscura were determined to compare patterns of polymorphism and divergence among members of this small gene family. The three esterase genes were also sequenced from D. persimilis and D. miranda for interspecific comparisons. The data provide evidence that gene conversion between loci contributes to polymorphism and to the homogenization of the Est5 genes. For Est-5B, which encodes one of the most highly polymorphic proteins in Drosophila, 12% of the segregating amino acid variants appear to have been introduced via gene conversion from other members of the gene family. Interlocus gene conversion can also explain high sequence similarity, especially at synonymous sites, between Est-5B and Est-5A. Tests of neutrality using interspecific comparisons show that levels of polymorphism conform to neutral expectations at each Est-5 locus. However, McDonald-Kreitman tests based on intraspecific gene comparisons indicate that positive selection on amino acids has accompanied Est-5 gene duplication and divergence in D. pseudoobscura.

Perturbation-reperturbation test of selection vs. hitchhiking of the two major alleles of Esterase-5 in Drosophila pseudoobscura

A perturbation-reperturbation tests selective neutrality of 100/100/100/100/100 and 106/100/100/100/100, the two most common alleles at the highly polymorphic X-linked locus Esterase-5 in Drosophila pseudoobscura. A total of 22 replicate populations are set up in cages, 11 start at a high frequency of 76% (U) and 11 at a low frequency of 21% (N) of the 106 allele. Allele frequencies change directionally and decrease in both U and N populations as groups and reach equilibria of 60 and 14%, respectively, after 200-300 days. These changes suggest natural selection. A hypothesis of balancing selection accounts for the pattern and predicts a dynamic equilibrium. A rival neutral hypothesis accounts for the pattern equally well by postulating hitchhiking and breakup of linkage leaving the Est-5 variants to drift at neutral equilibria. A reperturbation of allele frequencies in each population, creating 22 additional reperturbed populations EN and EU, with the original populations as controls, directly addresses the question of balancing selection or hitchhiking and breakup of linkage effects. Allele frequencies do not change directionally among the reperturbed populations as a group. The hypothesis of balancing selection is rejected in favor of the hypothesis of initial hitchhiking and dissipated linkage effects. The power of the experimental design to detect selection is studied by simulation. Within the limits of power set by the design, it is concluded that the 100 and 106 are iso-fitness alleles of Est-5 under the environmental conditions of the laboratory populations. The requirements of a method of perturbation and reperturbation are discussed.