FISHERY BULLETIN: VOL. 78, NO. 3 



Table 8. — Logistic growth method estimates of maximum bio- 

 mass (finiax in grams ( for Phase I guppy populations. Esti- 

 mates exclude weight of fry and are based on data collected 

 for weeks 18-28, and 30-36. 



200 



Population 

 number 



Population 

 number 



'Correlation coefficient for regression of 1/biomasSf + 2 against 1/biomasSf . 



M 

 E 

 A 

 N 



F 

 E 

 M 

 A 

 L 

 E 



W 



E 



I 



G 



H 



T 



g" 



r 

 a 

 m 

 s 



1.1 



1.0 



0.9 



0.8 



0.7 



0.6 



0.5^ 



I II III- 



50 75 100 



MEAN TOTAL 



125 150 



NUMBERS 



Figure 5. — Density-dependence of guppy growth illustrated by 

 plot of mean female weight (in size categories A 7 -i- A g ) at week 

 36 against mean total numbers during weeks 0-36. 



population numbers, followed by further pulses of 

 increase (Figure 6). No position effect was detected 

 when mean total numbers in blocks A and C were 

 compared by Student's t test. 



The patterns of numerical increase exhibited by 

 these populations were in striking contrast with 

 the control (unexploited) populations maintained 

 by Silliman and Outsell (1958). Behavioral obser- 

 vations and examination of size structure data 

 collected during Phase I indicated that the pres- 

 ence of juveniles in the J4 size category somehow 

 inhibited survival of newly born fry, presumably 

 through some juvenile-fry interaction within the 

 refuge area. Plots of estimated numbers of fry 

 surviving to enumeration during a biweekly in- 

 terval, corrected for observed mortalities ( = ad- 



is 24 



WEEKS 



Figure 6. — Pulsing behavior of numerical population growth 

 illustrated by guppy populations 3 (solid line) and 11 (dots). 



justed population increment), and numbers of J4 

 juveniles present at the beginning of the interval, 

 against time showed distinct offset peaks of abun- 

 dance in many populations. The relations could be 

 further improved by introducing a 2-wk time lag 

 and plotting the number of J4 juveniles at t — 2 wk 

 and the adjusted population increment from t to 

 t + 2 wk against time (Figure 7). The time lag 

 improvement implied that only larger juveniles 

 within the J4 size category were responsible for 

 inhibition of fry survival. Small J4 juveniles at 

 t — 2 wk would have become large J4 juveniles 

 (roughly equal to the J45 size category monitored 

 during Phase II) at time t. 



Strong and statistically significant (P<0.05) 

 negative correlations between biweekly adjusted 

 population increments and the natural logarithm 

 of J4 juveniles (-1-1) present 2 wk before the begin- 

 ning of a sampling interval were found in 8 of 10 

 populations (Table 9). These negative correlations 

 between numerical population growth and juve- 

 nile densities detected at several levels of adult 

 stock density showed that numerical population 

 growth, and thus the recruitment process, in these 

 populations could not be a simple function of adult 

 stock alone. Likely, interactions between newly 



564 



