Roa et a\: Estimation of size at sexual maturity 



579 



ear combination of parameter estimates 

 and therefore assumes a symmetric in- 

 terval estimate. Both bootstrap and 

 Monte Carlo results indicate that the 

 interval estimate can be quite asymmet- 

 ric. This may account for the better per- 

 formance of Monte Carlo, as compared 

 with Fieller's, in proportion of success. 

 These remarks, along with the facts 

 that Monte Carlo's percentage success 

 and bias are not affected by a natural 

 mortality rate varying between 0.2 and 

 0.8, and that it is very fast on every com- 

 puter platform, allows us to recommend 

 the use of the Monte Carlo method to 

 estimate size at P'/f maturity. 



Acknowledgments 



We would like to thank Bryan F. J. 

 Manly for reviewing an earlier draft of 

 this work and three anonymous review- 

 ers who made several comments and 

 criticisms which greatly improved the 

 manuscript. This work was funded by 

 FONDECYT grant 1950090 to R.R. 



9 10 11 12 

 Standard length (cm) 



Figure 5 



Monte Carlo confidence inten-als for length at 50<7f maturity of female 

 anchovies {Engrauhs r(;igt>ns) sampled from two localities off central Chile. 

 Open circles and dotted line = Talcahuano (36 45'S) (^ sampk''^^'^- 

 A'„^=5000l; Closed circles and solid line 



liV 



San Antonio (33°35'S) 

 =585.iV,„,=5000); Closed squares = 95<7t confidence bounds for/jg,,. 



