Although there is little experimental information on the effect of simultaneous selection for two quantitative characters on the magnitude of the genetic correlation between them, it is apparently generally expected that such selection for the two characters in the same direction will cause a negative change in the genetic correlation, and selection in opposite directions a positive change. Selection using independent culling levels was done for each of the four combinations of high or low third coxal bristle number with high or low sternopleural bristle number in Drosophila melanogaster for 22 generations. To estimate changes in the genetic correlation, realized genetic parameters were estimated from single-trait lines started from the base population, and from the two-trait lines after 10 and 22 generations of selection. Changes in the genetic correlation in individual two-trait selection lines were variable and unpredictable. At generation 22 concurrent two-trait selection had resulted in significantly larger realized genetic correlations than divergent two-trait selection, so that results were contrary to the generally accepted expectation.

Selection using independent culling levels (20% overall selection intensity) was done for the four combinations of high or low third coxal bristle number with high or low sternopleural bristle number in Drosophila melanogaster. Single-trait lines from the base population were selected at 20 % selection intensity for 10 generations for high or low bristle number. Realized heritabilities and genetic correlations were estimated after 10 and 22 generations of selection in the two-trait lines, and also from the single-trait lines. There was considerable variation among replicate lines within treatments in selection responses, and in realized heritabilities and genetic correlations. Although average estimates of the realized parameters were all less than those made from a diallel analysis of the base population, the average base population genetic correlation and the average realized genetic corre-lations from single-trait and two-trait selection lines were reasonably similar. Predicted selection responses in both single-trait and two-trait lines using genetic parameters derived from a diallel analysis of the base population were greater than average realized responses. However, predicted selection responses in the two-trait lines using realized genetic parameters from the single-trait lines were in reasonable agreement with average realized responses.