Fischer et al.: Demographic structure of Lutjanus campechanus in the northern Gulf of Mexico 



595 



and MSB) without knowledge of either date of capture 

 or morphometric data. The appearance of the otolith 

 section edge condition was coded as opaque or translu- 

 cent after Beckman et al. (1989). Annuli were counted a 

 second time when initial counts disagreed. In instances 

 where a consensus between the two readers could not be 

 reached, annulus counts of the more experienced reader 

 (AJF) were used. Between-reader differences in annulus 

 counts were evaluated with the coefficient of variation 

 (CV), index of precision (D) (Chang, 1982), and average 

 percent error (APE) (Beamish and Fournier, 1981). The 

 periodicity of opaque zone formation was verified for 

 each sampling location with edge analysis after Wilson 

 and Nieland (2001). Ages of red snapper were estimated 

 from opaque annulus counts and capture date with the 

 equation described by Wilson and Nieland (2001): 



Day age= -182 + (opaque increment count 

 ((»;-!) x 30) + d, 



365) + 



where m = the ordinal number (1-12) of month of cap- 

 ture; and 

 d = the ordinal number (1-31) of the day of the 

 month of capture. 



The 182 days subtracted from each age estimate are to 

 account for the uniform hatching date assigned for all 

 specimens (Render, 1995; Wilson and Nieland, 2001). 

 Age in years was assigned by dividing day age by 365. 



Fork length-TW relationships were fitted with lin- 

 ear regression to the model FL = a TW b from log 1(l - 

 transformed data for Alabama, Louisiana, and Texas 

 specimens. Analysis of covariance (ANCOVA) was used 

 to compare slopes and intercepts among sampling lo- 

 cations (SAS, 1985). Variability in age. FL, and TW 

 frequency distributions of red snapper were compared 

 among states with the Komolgorov-Smirnov two-sample 

 test (Tate and Clelland, 1957). 



Growth of red snapper was modeled for FL and TW 

 with the von Bertalanffy growth equations. Because 

 of differences in sample population size among states, 

 weighted mean FL and mean TW at age were fitted for 

 each state with nonlinear regression in the forms: 



FL, = LJ1 -e'-*" 11 ) 

 TW, = Wjl-el-'''"'!)''. 



where FL, = FL at age t\ 

 TW, = TW at age /; 

 L„ = the FL asymptote; 

 W„ = the TW asymptote; 

 k = the growth coefficient; 

 t = age in years; and 

 b = exponent derived from our length-weight 



regressions (SAS, version 5, 1985, SAS 



Inst, Cary, NO. 



Because of a lack of smaller individuals in all sample 

 populations, no y-intercepts for t were specified and 

 models were forced through 0. Larger individuals and 



juveniles selectively sampled by size were excluded from 

 the models to more accurately reflect a random sample. 

 Likelihood ratio tests (Cerrato, 1990) were used to test 

 for differences among states in models and in growth 

 parameter estimates. Differential growth was evalu- 

 ated for red snapper in the first 10 years of life when 

 somatic growth is most rapid (Szedlmayer and Shipp, 

 1994; Patterson et al., 2001; Wilson and Nieland, 2001). 

 Linear regressions of mean FL and mean TW at age for 

 fishes aged 1 to 10 years were compared among states 

 with analysis of covariance (ANCOVA) and tested for 

 homogeneity of slopes. 



Results 



During the three-year study period, 5192 red snapper 

 were sampled from the recreational harvest of the north- 

 ern GOM (Table 1): 642 individuals from fishing tourna- 

 ments, 71 undersize fish from tagging cruises, and 4479 

 random samples from recreational catches. The samples 

 included 2502 males, 2568 females, and 122 individu- 

 als of undetermined sex. The resultant male-to-female 

 ratios were 0.96:1 for Alabama, 1:0.99 for Louisiana, 

 0.94:1 for Texas, and 0.97:1 for all states combined. A 

 chi-square test indicated no significant difference in 

 the number of males to females (j 2 =0.78, P=0.38). Fork 

 lengths ranged from 237 to 916 mm (Fig. 2A). Speci- 

 mens from Alabama ranged from 237 to 916 mm FL, 

 Louisiana specimens ranged from 282 to 913 mm FL, 

 and Texas specimens ranged from 266 to 846 mm FL. 

 The FL frequency distributions of the random samples 

 were different among all states (AL and LA, maximum 

 difference (MD)=5.26; AL and TX, MD = 51.86; LA and 

 TX, MD = 51.77)(Fig. 2A). 



