SILLIMAN: EXPERIMENTAL POPULATIONS OF TILAPIA MOSSAMBICA 



Table 6.-Constants of Gompertz curves for growth of selected 

 groups of fish. 



Note: Lg = Length at time zero. 



Loo^ Asymptotic limit of length. 



W„ 



Weight at time zero. 



Wqo^ Asymptotic limit of weight. 

 g and G ^ Empirical constants of the Gompertz curve. 



Thus it was assumed that at day one of the two 

 males at maximum length in the test group was 

 misclassified as a female and, similarly, for the two 

 smallest males in the control group. Resulting 

 mean lengths in millimeters comparable to those 

 for day in Table 5 are: test male, 149.4; test 

 female, 141.5; control male, 151.2; control female, 

 141.1. Percentage differences are 1.7, 0.6, 2.2, and 

 0.2, respectively. The means under the "most con- 

 trary assumption" are indistinguishable from the 

 values used on the scale of Figure 12. It is evident 

 that substitution of the most contrary values 

 would not change the conclusion of greater male 

 growth in the controls. 



Gompertz curves were also fitted to biomasses of 

 the two groups (Table 5, Figure 13). Here the 

 equation was: 



W,= Woexp[G-Gexp(-gt')l 



where W is total weight in kilograms, t' is time in 



T mossombico , growth in weight 



Figure 13. -Gompertz curves fitted to total weights of groups 

 from selectively fished test population and unselectively fished 

 control population. Line branching upward from test curve in- 

 dicates control curve moved over to same starting point as test 

 curve. 



days, and G and g are empirical constants. Con- 

 stants are given in Table 6. In weighings, fish were 

 not separated as to sex, and only a single growth 

 curve was available for each population. Because 

 of the initial difference in total weight, the curve 

 for the control population is shown moved over to 

 the time when weight of the test group had grown 

 to the initial weight of the control group. Even so 

 treated, the control group exhibits markedly 

 greater growth than the test group, reflecting 

 greater growth of the males in it. This is even 

 more striking when it is considered that the 

 amount of food per weight of fish was somewhat 

 less in the control than in the test group. The ob- 

 served growth in biomass supports the conclusion 

 of diminished genetic growth in males as a result 

 of selective fishing. 



CONCLUSIONS 



Analyses presented above have revealed sig- 

 nificant differences of responses to exploitation 

 between the selectively fished test population and 

 the unselectively fished control population. These 

 differences were demonstrated both in catches 

 obtained and in genetic growth patterns. 



Yield models fitted demonstrated marked 

 differences in the catches obtained under selective 

 and unselective fishing. It is clear that, with the 

 particular populations studied and under the as- 

 sumptions of stability made, weight of yield under 

 unselective fishing was greater than that under 

 selection. This yield included a large proportion of 

 fish below the selection point, however. To the ex- 

 tent that one may generalize from this 

 experiment, it appears that unselective fishing 

 would be preferable if maximum physical yield 

 were the sole objective. If selection is required to 

 secure fish that are of appropriate size for the 

 market, the objective may be achieved only at the 

 sacrifice of part of the weight of the catch. 



That three generations of selective fishing 

 caused a change in the genetic growth pattern of 

 males, resulting in slower growth than in the con- 

 trols, seems certain from the results. It is neces- 

 sary to explain, however, why a similar change did 

 not occur in the females. This may have resulted 

 from the phenomenon of epistasis. In this it is 

 considered that a single gene may control the 

 hormone which permits males to grow to larger 

 ultimate size than females. Since females have less 

 of this hormone than males, they are unable to 



505 



