Published online by Cambridge University Press: 14 April 2009
1. An experimental study has been made of recurrent selection to an inbred tester. A suitable inbred line is used as a tester parent, and selection is made within a non-inbred population on the individuals crossing performance with the tester line.
It is concluded that there are two situations in which recurrent selection could be profitably applied. Firstly, recurrent selection should, theoretically, be successful when applied to characters closely related to fitness which have little additive genetic variance and secondly, in cases where a character has already been subjected to individual or family selection and has reached a plateau level in that population.
2. The two experiments—i.e. recurrent selection for large litter size in mice and for low bristle number in Drosophila melanogaster—reported here are respectively an example of each of the above situations. In each experiment selection was made between males within the closed non-inbred population on the basis of the performance of their testcross progeny resulting from matings with inbred line females.
3. Initial generation hybrid performance in both experiments was not intermediate between parental performance levels and the divergence from intermediacy was away from the direction of selection.
4. In both experiments there was no evidence to suspect the presence of over dominance.
5. Response to selection was obtained in each experiment but this was close to or less than the expected response calculated on the assumption that all the variance between sires in crossing performance was additive genetic variance.
6. From these experiments it is not possible to draw any firm conclusions about the effectiveness of recurrent selection for exploiting overdominance. It is, how ever, a very inefficient way of exploiting additive genetic variance. It is suggested that more success might be obtained by careful choice of base population material used in recurrent selection.