AbstractA population of paradise fish (Macropodus opercularis) was subjected to five generations of bidirectional selection for social dominance in order to investigate correlated changes in agonistic behaviors. In addition to the high and low dominance lines (HD and LD respectively), a third dominance line was created through selection for individuals closest to the median in dominance success. The MD line provided standard opponents for members of the other two dominance lines and served as a control for inbreeding effects. The response to selection was asymmetrical. After five generations of selection the LD line had diverged significantly from the MD line, but the HD line had not. None of the three factors known to cause an asymmetrical response to selection (directional gene frequencies, directional allelic dominance and inbreeding depression) can account for the response to selection found here. The divergence of the HD and LD lines with respect to dominance success was not accompanied by any changes in the measures of aggressiveness. The F-5 generation of the HD and LD lines did not differ significantly for any of the measures of aggressiveness used here, indicating that social dominance and aggressiveness are not intimately related. An alteration in the sex ratio of the HD line was an unanticipated byproduct of selection. The F-5 generation of this selection line contained an overwhelming preponderance of males. Interpretation of these results is complicated by the fact that environmental factors such as temperature and the degree of crowding probably influence sexual differentiation in this species. The social crowding factor, in particular, was found to be important and was investigated in a preliminary way. For all three dominance lines it was found that as crowding increased, so did the proportion of females in a brood. Fish raised in physical isolation showed a strong tendency to become males. These results suggest that sexual differentiation is socially mediated. Within a given brood, fish of high dominance status tended to become males and fish of low dominance status tended to become females. The results from the isolation-reared fish indicate that it is not dominance per se that promotes maleness, but rather, the condition of being undominated. Under uncrowded conditions an individual is, on average, dominated by fewer fish than under crowded conditions. Above and beyond this density related trend in the sex ratios, however, there were significant between-the-line differences. The HD line showed a greater proportion of males than the other dominance lines, over all density conditions. It appears that selection for high social dominance constituted selection for maleness. This response to selection indicates a continuum of genetic sex in this species and a polyfactorial mode of inheritance. The alteration in the sex ratios might also explain the asymmetrical response to selection for social dominance, in that the HD line did in fact exhibit a greater proportion of dominant fish, though not a greater proportion of dominant males, than the other two dominance lines. As such, the asymmetrical response to selection would be an artifact of the measuring procedures.