ESTIMATION OF AVERAGE FITNESS OF POPULATIONS OF DROSOPHILA MELANOGASTER AND THE EVOLUTION OF FITNESS IN EXPERIMENTAL POPULATIONS.

Abstract

In the companion paper (Jungen and Hartl, 1979) we proposed a method for estimating average fitness of entire populations of Drosophila melanogaster by introducing the population into cultures containing compound-autosome flies and measuring, in the next generation, the proportion of wild-type flies among the offspring. (In these papers we operationally define the fitness of a population as the intraspecific competitive ability of that population.) Because compound-autosome flies have both left arms of an autosome attached to one centromere and both right arms attached to another centromere, they are unable to produce viable offspring when mated with wild-type flies. Thus the genomes of the wild-type and compound-autosome flies cannot intermix. Although major components of fitness such as fecundity, viability, and mating activity can be assessed in one-generation experiments (Jungen and Hartl, 1979), and each component is automatically given weight according to its actual importance under the experimental conditions, certain other components of fitness such as longevity and length of fertile period are not amenable to inclusion in one-generation experiments. In the same way, one-generation experiments may give undue weight to certain components of fitness that, in multigeneration tests, are not so important. In this paper we extend the compoundautosome method of Jungen and Hartl (1979) to multigeneration experiments. The method proves to be more sensitive to detecting small differences in fitness than is the one-generation method. Moreover, although fitness rankings of strains are generally consistent with what one would expect based on one-generation tests, the inclusion of more fitness components in the multigeneration test does produce some differences. Although the wild-type strain Swedish-c performs slightly but consistently better than does Oregon R-C in one-generation tests, for example, the use of a multigeneration design shows Oregon R-C to have a higher fitness than Swedish-c. Being the more powerful and inclusive method, multigeneration experiments would seem to provide the best measure of average fitness. Experiments are not as easily replicated as they are in one-generation experiments, however, and there are other problems as well. By recombination in the centromeric heterochromatin in females, the compound autosomes can detach, becoming structurally normal. When an egg carrying a detached chromosome is fertilized by a normal sperm, the resulting zygote will be viable, karyotypically normal, and able to reproduce. Thus, due to detachment of the compounds, the wild and compound-autoI Supported by NSF grant PCM77-00886. D.L.H. is recipient of Research Career Award GM0002301. This paper is dedicated to the memory of Spencer W. Brown.