SILLIMAN: EXPERIMENTAL POPULATIONS OF TILAPIA MOSSAMBICA 



cies of thickness for immature and mature fish 

 (Figure 7) suggested a thickness at maturity of 

 about 15-20 mm. The two prominent modes in the 

 frequencies seemed to require about 4 mo 

 (63.2-67.1, 69.1-73.2) to reach this size. To this must 

 be added the 2-mo "reproductive lag" mentioned 

 above under "Basic Relations," suggesting a 

 generation length of about 6 mo. The 36-mo period 

 of selective fishing would thus include about six 

 potential generations. Because of irregularities in 

 recruitment, however, the effective number of 

 generations was less, and it was necessary to make 

 an estimate. 



Such an estimate can be derived from the record 

 of recruitment numbers (i^iNT* Table 4, Figure 9). 

 An arbitrary criterion for significant recruitment 

 was established, requiring at least 15 recruits per 2 

 mo. A generation was considered to be such a peak 

 separated from the previous filial generation by a 

 period of at least 6 mo (the parental generation for 

 the test group had been fished selectively at month 

 40). Under the arbitrary criterion the estimated 

 generations (Figure 9) during the selection period 

 were only three for the selectively fished test and 

 four for the unselectively fished control popula- 

 tion. Experiments with other animals, such as 

 those of Robertson with thorax length of 

 Drosophila cited by Falconer (1960), have shown 

 that measurable change in a size character can 

 occur in as few as three generations of selection. 



To test whether genetic response to selection did 



Table 5.-Growth of selected groups of fish. Lengths are from 

 snout to tip of taiL 



200 



160' 



120- 



80- 



3 



o 



lu 



(E 40- 



a: 



CD '-' 



5 80- 



40- 



Test 





Control 



J^^ 





20 



30 40 



MONTH 



50 



80 



Figure 9.-Recruitment numbers from ^jnt in Table 4, with 

 negative values considered zero. Test population was selectively 

 fished; control, unselectively. 



'0^5 January 1973. 



20ne female misclassified as male on day 0. 



30ne female died. 



■•Two females died. 



5Two females misclassified as males on day 0. 



'Two females removed to match mortalities in test group. 



'One female removed to match mortality in test group. 



occur, groups of 46 mature fish as similar in sex 

 and length composition as possible were selected 

 from test and control populations on 5 January 

 1973 at month 77.2 (Table 5). It was not possible to 

 match these fish as closely as desired by total 

 weight, and that of the control group exceeded 

 that of the test group by 19%. These fish were fed 

 the standard diet (Table 1) which furnished them, 

 even at the end of the growth period, with 1.5% 

 (test) or 1.3% (control) of body weight of food per 

 day. This was 2.5 (test) or 2.1 (control) times as 

 much as was received by the 10 kg preexploitation 

 stocks. Any offspring that appeared were removed 

 as soon as possible. 



Growth of the selected fish was measured by 

 determining the lengths of individual fish and the 

 total weight of each group at 55-56, 118-119, and 

 150-151 days after the start of the growth period 

 (Table 5, Figures 10, 11). The length frequencies 

 reveal the general correspondence of the groups at 

 the beginning of the period, in addition to the 

 expected more rapid growth of the males than the 

 females in each group. They also reveal that the 

 males in the unselectively fished control group 

 grew more rapidly than those in the selectively 

 fished test group. 



Growth was further studied by curv.es based on 

 mean lengths and total weights of the selected 

 groups. Gompertz curves fitted to lengths had the 

 equation: 



Lt- = Lo exp[G - G exp{-gt')l 



where L is mean length in millimeters, t' is time in 

 days, and G and g are empirical constants. This 

 curve and all other Gompertz curves were fitted by 



503 



