EVOLUTIONARY MECHANISMS 295 



the frequency surface rather slowly. Wright shows that if a large 

 population becomes split up into locaUzed groups, which usually 

 inbreed but between which there is a certain amount of migration and 

 cross-breeding, the conditions for evolutionary change are much better. 

 The individual groups may be small enough for random fluctuation of 

 gene frequencies to carry them rapidly about the frequency surface, 

 but the cross-breeding saves them from degenerating into a condition 

 when all their genes become fixed; they thus have both the rapidity 

 of movement of the small groups and the store of variation of the 

 larger groups, which enables the movement to go on longer. The 

 occurrence of mutations has the same effect as cross-breeding in 

 replenishing the store of variation in a group small enough to undergo 

 rapid changes in gene frequencies. The best conditions for movement 

 from one hill on the frequency surface to another, that is, for evolu- 

 tionary change, therefore comes with a suitable balance between all the 

 evolutionary factors; the group can be large, but if so it should be split 

 into smaller groups, which have a considerable amount of inbreeding 

 but some outcrossing, the mutation rate should be moderately large, 

 and the intensity of selection should be enough to keep random fluctua- 

 tion within bounds but not so severe as to eUminate it altogether. 



It is clear that the rapidity of evolution is very largely influenced 

 by the size of the population. However, the concept of population size 

 in this connection is not at all simple. What is important is not the 

 total number of existing individuals but is the size of the effective 

 breeding population. For instance, in many species the population is 

 almost wiped out every winter, and the only supply of variation on 

 which evolution can be based is that contained in the remnant which 

 survives to begin breeding in the spring. Similarly, inbreeding lowers 

 the available variability. Thus the effective breeding population is a 

 magnitude which can scarcely be measured directly, and we still have 

 very litde knowledge about it. It has been shown, however, that in 

 some natural populations it is small enough to permit quite large 

 random fluctuations in gene ratios. Dobzhansky^ found that in Droso- 

 phila pseudo'ohscura a given lethal factor is often more concentrated in 

 a small isolated and inbreeding race than it is in the whole population 

 of which the isolated community forms a part; this local concentration 

 of the lethal can only be due to a chance fluctuation. 



These conclusions seem to be in general agreement with those of 

 Fisher and Haldane, though these two authors allow rather less im- 

 ^ Dobzhansky and Koller 1938a. 



