FISHERY BULLETIN: VOL. Sfi, NO. 4 



Table 1 .—Parameters chosen for the simulations. 



previous ones, but with stochastic recruitment, and 

 random recruitment intensities assigned to random- 

 ly selected months (situation 4). 



The choice of these recruitment patterns was 

 based on the knowledge that tropical species have 

 different types of recruitment periodicities with 

 varying temporal and spatial variation (Thresher 

 1984). The same species may have different patterns 

 (Sale et al. 1984). Two principal peaks and year- 

 round recruitment have been reported in coral reef 

 fish from Hawaii (Walsh 1987) and Curasao (Luck- 

 hurst and Luckhurst 1977). One single recruitment 

 per year has also been reported (Li 1960; Gladstone 

 and Westoby 1988). 



Mortality values, both natural and fishing, were 

 chosen to have small standard deviations (Table 1) 

 because the main objective was to examine the ef- 

 fects of different recruitment patterns, and we 

 wanted to keep the possible effects of variations in 

 other factors as small as possible. 



The standard deviations of length-at-age used 

 (Table 1) are representative of values for species 

 with life history characteristics similar to those of 

 the cases considered for this study (K. Erzini, work 

 in progress). 



The Length-Frequency Analysis 



The two techniques chosen to represent length- 

 frequency analysis were ELEFAN (Electronic 

 Length Frequency Analysis) (Brey and Pauly 1986) 

 and the package entitled "Length Frequency Based 

 Fish Stock Assessment Microcomputer Programs" 

 (LFSA package) by Sparre (1986; adapted to MS 

 DOS by K. Erzini). ELEFAN has been widely used 

 in the analysis of tropical fish stocks, and its non- 

 parametric basis for determination of K and L^ 

 makes it a unique methodology for the analysis of 

 length frequencies. 



In ELEFAN, the length-frequency samples are 

 restructured in order to emphasize peaks. Details 

 of the restructuring methodology are given in Brey 

 and Pauly (1986). Growth curves are generated for 

 values of K and L^ within specified ranges and fit 

 to the reconstructed length-frequency data. The best 

 curves are considered to be the ones that pass 

 through the most peaks and the least troughs. 



The LFSA package uses a method of a different 

 nature— the Bhattacharya method (Bhattacharya 

 1967), to separate normal curves, under the assump- 

 tion that the length distributions for each age are 



650 



