HANKIN: A MULTISTAGE RECRUITMENT PROCESS 



cation achieved by the grading process and size 

 category designations are contained in Table 1. 

 These size category designations will be adopted 

 throughout the paper for brevity. 



After separation into size categories, numbers 

 of males and females in each size category were 

 recorded, and weights of males and females in size 

 categories Ag, A7, Ag (weeks 0-36) and A55, Ag, 

 A7, Ag (weeks 36-58) were obtained separately. 

 Weights of fry were not obtained due to dangers of 

 handling mortality. Weights of remaining size 

 categories were obtained without separation by 

 sex as these groups contained nearly all immature 

 fish. To determine weights, fish were placed in a 

 small nylon net, blotted on paper towels until no 

 further moisture was observed (about 1 min), and 

 then transferred to a previously weighed plastic 

 beaker containing about 40 ml of water. Weights 

 of fish were determined as the difference between 

 the previous weight and the weight obtained after 

 addition of fish. Total population biomass was 

 determined as the sum of all weight measure- 

 ments for a given population but did not include 

 weight of fry. An Ainsworth Model ION analytical 

 balance, accurate to three decimal places, was 

 used for all weighings. Handling mortality was 

 negligible. In 58 wk only five mortalities, all fish 

 <2 mm in "diameter," were recorded as a direct 

 result of biweekly manipulations at enumeration. 

 Approximately 38,000 fish were handled during 

 these enumerations. 



All fish were also examined for external symp- 

 toms of disease at biweekly enumerations. Disease 

 diagnosis and/or confirmation was performed at 



Table 1. — Size-group classifications of guppies at enumerations 

 during weeks 0-36 ( Phase I) and weeks 36-58 ( Phase II). Grader 

 spacing is measured from center to center. 



irregular intervals by Louis Leibovitz, Cornell 

 University Veterinary School. Fish with obvious 

 external disease symptoms (consistently diag- 

 nosed as chronic piscine tuberculosis) but other- 

 wise apparently healthy were treated by a 20-min 

 bath in Formalin (1:4,000). Severely afflicted fish 

 near death were removed from populations and 

 counted as mortalities. Numbers, weights, and 

 sexes of fish showing symptoms and/or treated 

 were recorded for each population at enumeration. 

 Aquaria were examined daily for mortalities. 

 Date of death, size category estimate, and sex were 

 recorded for each observed mortality. 



At termination (week 58) a sample of 41 gravid 

 females ranging from 21 to 39 mm was selected 

 from the populations. Each female was dissected 

 and the number of embryos counted. Data col- 

 lected was used to establish an overall relation 

 between fecundity and female size for the experi- 

 mental populations. 



Experimental Design 



The initial intent of population experiments 

 was to examine dynamics of mixed species popula- 

 tions (guppies, P. reticulata, and southern platy- 

 fish, "platy," Xiphophorous maculatus) from the 

 perspective of stock-recruitment theory. Due to 

 excessive mortality among the southern platyfish 

 and an apparent inability of their fry to success- 

 fully compete with guppy fry, the southern platy- 

 fish were removed from all populations at week 8 

 and experimental goals became limited to analy- 

 sis of single species populations based on simple 

 stock-recruitment theory. Numerical population 

 growth of single species populations, however, 

 was not anticipated on the basis of the sim- 

 ple theory. Numerical population growth often 

 occurred as a series of discrete pulses of increase, 

 followed by periods of roughly static population 

 numbers. Behavioral observations and analyses of 

 collected population data through week 36 indi- 

 cated that the pulsing quality of numerical growth 

 was probably caused by juvenile-fry interactions 

 within refuge areas. Based on this hypothesis, 

 experimental populations were manipulated and 

 exposed to different treatments at week 36. Treat- 

 ment consisted of alteration of original refuge 

 area habitat quality and population growth under 

 these new conditions was monitored through week 

 58 to test the juvenile-fry interaction hypothesis. 



The experiments have been separated into two 

 distinct phases based on the above-described ex- 



559 



