HANKIN: A MULTISTAGE RECRUITMENT PROCESS 



Alteration of refuge fence spacing during Phase 

 II was believed to have influenced both the inten- 

 sity and duration of juvenile-fry interactions. 

 Behavioral observations of a 5.0 mm refuge popu- 

 lation provided a partial explanation of mecha- 

 nisms involved. A female was observed delivering 

 four fry and the subsequent immediate fate of the 

 fry was followed. Two of the fry were pursued and 

 consumed by two different adult females. Two 

 other fry successfully entered a refuge area after 

 vigorous pursuit by several males and females, 

 including the female giving birth. Once w'ithin the 

 refuge area the fry were chased and nipped by 

 larger juveniles. Although the juvenile chase 

 shortly ceased and the fry settled to the aquar- 

 ium bottom apparently uninjured, alternative 

 outcomes seemed possible. Fry could have been 

 chased outside the refuge area where they would 

 once more be vulnerable to adult predation or the 

 juvenile contact could have resulted in death and 

 perhaps cannibalism. The observations indicated 

 that survival of newly born fry was determined in 

 an extremely short period of time. From release to 

 death or apparent survival in the refuge area 

 never occupied more than perhaps 2 min. Since 

 guppy fry are capable of rapid swimming within 

 minutes after birth, it is unlikely that juvenile 

 pursuit within the refuge area would be successful 

 except during the first few minutes after birth. 



The intensity and duration of juvenile-fry inter- 

 actions, reflected in patterns and magnitude of 

 numerical population growth among treatment 

 groups, was measured and characterized (in pa- 

 rentheses see Table 10) in several ways: 



1) by the number of biweekly intervals in 

 which adjusted population increments were 

 «0(#PI«0); 



2) by the length in biweekly intervals of the 

 longest period without a positive population 

 increment (Long. PI); 



3) by the median adjusted population incre- 

 ment (Med. PI); 



4) by the median population numbers (Med. N); 



5) by the total number of data observations 

 with zero individuals per any size category 

 (#0/CAT);and 



6) by the mean percentage of population num- 

 bers in size categories ^ J5.0 (% adult). 



Measures 1) and 2) were designed to evaluate the 

 prominence of pulsing. Intervals between succes- 

 sive pulses of increase should be greatest when 



Table 10. — Comparison of measures of numerical population 

 growth for treated guppy populations during Phase II. See text 

 for explanation of column entries. 



longest periods of juvenile presence occur within 

 refuge areas. The 5.5 mm group should exhibit 

 strong pulsing behavior while the 4.5 mm group 

 should show little if any pulsing (Silliman and 

 Outsell 1958). Measures 3) and 4) were designed to 

 evaluate effects of juvenile-fry interactions on 

 population numbers. Although the duration of 

 juvenile-fry interactions might not be reflected in 

 total population size at any given time, it should 

 be reflected in median population numbers and in 

 median population increments during the experi- 

 mental period. Measures 5) and 6) were designed 

 to evaluate size structure smoothness and relative 

 dominance by adults. The characteristic patterns 

 of growth produced under different refuge envi- 

 ronments should be reflected in the age structure 

 of treated populations and, although less distinct- 

 ly, in the size structure. The 4.5 mm group should 

 have a finely graded size structure with juveniles 

 nearly always present, while the 5.5 mm group 

 should have a fluctuating size structure perhaps 

 including distinct "size classes" corresponding to 

 separate pulses of increase. 



Comparison of these measures compiled for all 

 treated populations showed clear separation for 

 each measure between 4.5 mm and 5.5 mm groups. 

 In no case was the 5.0 mm group clearly separated 

 from the other groups, although means of all 

 measures for the 5.0 mm group fell between means 

 for 4.5 mm and 5.5 mm groups. Orders of means 

 were in the directions expected on the basis of the 

 juvenile-fry interaction hypothesis (Table 10). 



The above comparisons do not, however, allow a 

 "test" for differences in numerical population 

 growth patterns among treatment groups, in part 

 because the several measures are not independent 

 of one another. Rather, comparison of these quan- 



567 



