NATURAL SELECTION 



649 



An accumulation of mutations may be 

 necessary before their combination ex- 

 presses itself in such a manner as to be 

 subject to selection (Wright, 1931; Mather 

 and Wigan, 1942). Selection pressure" is a 

 term used by Wright (1929) for the effect 

 of selection on gene frequency measured by 

 the rate of change in gene frequency per 

 generation that it tends to produce. This 

 rate (as is also true for mutation pressure 

 and immigration pressure) involves a cer- 

 tain constant as well as the variable gene 

 frequency of the population. This constant 

 is the selection coefficient (s), which may 



** Wright (personal communication) illus- 

 trates the terms selection pressure, selection co- 

 efficient, and selective value as follows: "Tak- 

 ing the case of a recessive gene a, assume that 

 whatever the frequencies of A — and aa pheno- 

 types may be, a given number of recessives 

 leave 100 s per cent fewer descendants (that 

 reach reproductive age) than the same number 

 of dominants. Thus, if * = 1.0, 100 s per cent 

 is 100 per cent; and if s = 0.1, 100 s per cent 

 is 10 per cent. 



"The symbol q represents gene frequency (in 

 this case, of gene a). 



"The symbol Ag represents the rate of change 

 of the gene frequency q per generation (i.e., 

 100 A qr would be the percentage change in q). 



Genotype Selective Value (W) Frequency 



AA a il-qy 



Aa a 2^(1 — q) 



aa a(l-s) U v^e,\Ve,6^ > • - 



"Selection pressure = Aq =^ — sq^(l — q), or 

 more accurately 



A, - -f " - ■" 

 1 — sq^ 



"Selection coefficient of genotype aa is — s. 



"Selective value of genotype aa is W = a 

 ( 1 — .s ) , where c is a constant such that the 

 (weighted) average selective value of the 

 population W = a{l — sq^) is the ratio of 

 number of offspring to number in the parental 

 population at the same phase in the life cycle. 



"It should be said that constant selective 

 values and selection coefBcients cannot in gen- 

 eral be expected for genes or even one factor 

 genotypes. They are more likely to be ap- 

 plicable to the genotype as a whole considering 

 simultaneously all pertinent loci. Selective 

 values even of whole genotypes may, how- 

 ever, be functions of the gene frequencies. In 

 this case they are variables involving one or 

 more selection coefBcients which can be defined 

 only as constants in the variable expression for 

 selective value. Somewhat loosely, however, 

 one may deduce a momentary selection co- 

 efficient for a one factor genotype or even 

 gene." 



be positive or negative. Other constants are 

 the mutation or immigration coeflBcients. 

 "Selection pressure can be defined suffi- 

 ciently broadly to include all processes 

 (such as differential mortality, differential 

 fecundity, differential emigration) which 

 tend to change gene frequency systemati- 

 cally without either change of hereditary 

 material itself (mutation) or introduction 

 from without (immigration)" (Wright, 

 1948a, p. 291). Selection and inbreeding 

 tend to reduce heritable variation. Muta- 

 tion pressure or reassortment, or both, are 

 necessary to give a continual supply of 

 variations upon which selection may act 

 (Wright, 1932; Dobzhansky, 1946). 



Isolation is the dividing factor in phy- 

 logeny, while selection is the sorting and 

 survival factor. Selection may act upon a 

 linear sequence chronologically isolated in 

 time (Simpson, 1944, p. 33; p. 626). It is 

 a guiding factor that exhibits both intensit)' 

 and direction. 



Long-term effects are difficult to demon- 

 strate experimentally in the time available 

 to the experimenter. Occasionally short- 

 term effects of selection can be measured 

 (Dobzhansky, 1947, 1948). It is possible to 

 arrange an experimental demonstration of 

 survival in contrasting populations or spe- 

 cies that have presumably long been react- 

 ing to selective agents. We may thus speak 

 of certain characteristics as having survival 

 value. 



Sumner (1934, 1935) placed large num- 

 bers of mosquito fishes (Gamhusia patrue- 

 lis) in two cement tanks painted black and 

 white, respectively. After seven or eight 

 weeks, the fishes accommodated to the 

 color of their background by means of 

 chromatophores. Over 500 of these fishes 

 were then placed together in a tank with 

 a penguin introduced as a predator. In a 

 black tank, 27 per cent black and 73 per 

 cent white fishes were eaten in a few 

 minutes. In a white tank, 62 per cent black 

 and 38 per cent white fishes were eaten in 

 a similarly short time. Fishes that har- 

 monize in coloration with their immediate 

 surroimdings were thus shown to be less 

 likely to be eaten by certain birds than 

 fishes of the same species that do not so 

 harmonize. It should be noted that selec- 

 tion did not act upon different germinal 

 characteristics, but over a long period of 

 time selection might act upon a capacity 



