Selection in experimental populations. II. Components of selection and their fluctuations in two populations of Drosophila melanogaster.

POPULATION biologists have long followed gene frequency changes in model populations in attempts to verify various aspects of the mathematical theory of natural selection. The selection has been assumed to be constant in many of these studies, although it is widely recognized (e.g., TEISSIER 1954) that the selection in fact fluctuates with the physical and biological elements of the environment. In only a few instances have experiments been specifically designed to reveal how the selection varies from generation to generation (KOJIMA and YARBROUGH 1967; ANDERSON 1969). Likewise, most experiments on breeding populations have been designed to estimate the overall selective values of the genotypes, and our knowledge of how the selection is partitioned into components such as viability and fertility is notably incomplete. TEISSIER (1942) first showed how to carry out such a partition of selection, by scoring the frequencies of the genotypes after selection by viability but before selection by fertility. WILSON (1968) and ANDERSON (1969) performed similar analyses. All three of these workers assumed constant selection and partitioned into components the selective values estimated over many generations. The purpose of this article is to show how the selection in each generation of an experimental population may be divided into components of viability and fertility. We shall use this information to demonstrate that selection has significantly fluctuated over a period of thirteen generations in two experimental populations of Drosophila melanogaster.