DeMartini et al Maturity, size, and sex composition o\ Xiphias gladius 



493 



Statistical analyses 



We estimated body length at median sexual maturity (L50) 

 for males and females separately using nonlinear regres- 

 sion. A derivative-free maximum likelihood method (Proc 

 NLIN; SAS, 1989) was used to fit percentage maturity by 

 5-cm length class to the two-parameter logistic model. 



median body length versus latitude after body length dis- 

 tributions were made linear by rank transformation. Anal- 

 yses of sex ratios and size composition were limited to 

 observer-sampled trips and should be considered a first 

 approximation of patterns for the total Hawaii-based long- 

 line catches of swordfish during March 1994— June 1997. 

 Significance was based on an alpha level of 0.05. 



P^ = 100 / ( l-t-exp'" 



-hEFL 



where P = percentage mature at length x; and 

 L^o = (-a/6). 



Sex ratios (females/total) also were related to 5-cm length 

 class by using nonlinear regression. Regressions were 

 weighted by the square root of the number offish examined 

 in each length class. Ninety-five percent confidence inter- 

 vals on L^gS were calculated by using SAS Proc PROBIT. 

 Logistic regression (STATISTIX. vers. 4.1; Analytical Soft- 

 ware, 1994) was used to relate ovarian developmental 

 stage to oocyte diameter and to predict reproductive readi- 

 ness with gonad weight plus body size. Contingency chi- 

 square and unbalanced fixed-factor ANOVAs (Zar, 1984; 

 SAS, 1989) were used to compare sex ratios and the rela- 

 tive body size compositions of swordfish catches, pooled 

 by 1° lat. by 1° long. bins, during spawning (March-July) 

 and nonspawning (August- February: see "Results" sec- 

 tion) periods. Least-squares regression was used to regress 



Results 



Sex and maturity composition of sampled Fish 



During the 40-mo. sampling period, an estimated 142,000 

 swordfish were caught by the Hawaii-based pelagic long- 

 line fishery (2.26 million hooks: Ito and Machado, 1996.^ 

 1997,2 19993). Data foj. ^^e 6639 field-sexed fish examined 

 herein thus represent about 4.8'^ of the commeixial land- 

 ings of swordfish, with the samples reflecting about 4.9*^ 

 of the total longline effort (2.26 million out of 45.7 million 

 hooks) expended during the period. The sex ratio (females 

 as fraction of total) of the field-sexed commercial catch was 

 0.533 (SE=0.006: P<0.001). About 64'7f of the 1336 .sword- 

 fish examined histologically were sexually mature (stages 

 >4, Table 1 ). The two sexes differed in maturity composi- 

 tion based on histological classification (x^=67.4; P<0.001 ): 

 789^ of 506 confirmed males but only 56% of 828 confirmed 

 females were mature. 



