Components of genetic variance in the cultivated strawberry.

F I S H E R (1918) subdivided genotypic variance (the variance due to genetic variation) into three components: (1) a part due to the average effects of genes (now called additive genetic variance) , ( 2 ) a part due to allelic interactions of genes that is called dominance variance, and (3) a part due to nonallelic interactions of genes that is called epistatic variance. A sound plant breeding program will include the use of information on the relative importance of these components. Thus, in strawberries, if it is found that most of the genotypic variance is additive, then a simple breeding program of mating the best with the best over several generation:: would quite likely be a good approach. In contrast, if it is found that the dominance variance is the most important portion of the genetic variation, then it would be worth considering a procedure such as reciprocal recurrent selection (COMSTOCK, ROBINSON and HARVEY 1949). However, if epistatic variance is the major component, then test crosses with evaluation of small samples of plants followed by testing of large progenies of the best crosses would probably be superior to other methods. The important factor to stress when epistasis is important, and when the material can be vegetatively propagated, is the need to identify superior genotypes immediately and maintain them in the population by asexual reproduction. In all three cases, it would be desirable to provide for testing in several locations and years if the genotype x environment interactions were found to be important. Selection of the best breeding procedure based on a knowledge of the components of genetic variance must logically be expected to give the maximum genetic progress towards the objectives of a breeding program. Evidence obtained by COMSTOCK, ELLEHER and MORROW (1958) was interpreted to show that epistatic variance may be very important in strawberries. This conforms with the suggestion of WRIGHT ( 1 956) that in a species such as the strawberry where there .is “a combination of prevailing uniparental reproduction with occasional crossbreeding,” variability is comparable to that under random mating except that the binding of the population to a single adaptive peak is avoided. The resulting peaks of adaptive excellence he interpreted as a consequence of pleiotropy and epistasis, where the effect of pleiotropy on the net worth may be considered statistically as epistasis. He considered the natural breeding system of such a species to be one of violent alternations between the selection of