202 



Fishery Bulletin 98(1) 



from the core to the first annulus was made for a 

 subsample of 23 specimens that were age 1 and com- 

 pared with the measurements from the three YOY to 

 estabhsh the position of the first annulus. 



Age-length keys were formed by obtaining a matrix 

 of numbers at age by length interval for each gear 

 type (longline and snapper reel ) in two periods ( 1980s 

 and 1990s). Besides the differing selectivities of long- 

 lines and snapper reels, another reason for develop- 

 ing keys by gear type was the difference in sampling 

 area in 1993-94 (Fig. 1) and potential differences in 

 sampling depth due to the restriction (SAFMC, 1991) 

 that limits the use of longlines to waters deeper than 

 91 m. Additional keys were generated to address 

 two questions: 1) Do data from specimens with age 

 estimates of lower precision affect the accuracy of 

 the key for specimens caught with longlines in the 

 1990s? and 2) Are there differences in the keys for 

 specimens caught with longlines or Kali poles during 

 1982-85 that were examined in the present study 

 and in an earlier study by Waltz?^ To address the 

 first question, keys based on all specimens and spec- 

 imens for which there was a difference of 0-1 incre- 

 ments between readers were compared. 



All analysis of age and growth data was conducted 

 with SAS software (SAS Institute, Inc., 1990). Fish- 

 er's exact test ( Siegel, 1956 ) was used to compare the 

 age distributions of two age-length keys by 25-mm 

 length intervals. The FREQ procedure was used to 

 run this test. A comparison was made only if each 

 key had >6 specimens in an interval because of 

 the low power associated with small sample sizes 

 (Bennett and Hsu, 1960). The large number of tests 

 required to compare age-length keys necessitated 

 compensating for experimentwise error by comput- 

 ing an adjusted significance level (a*) using the for- 

 mula presented by Hayes (1993). 



Nonlinear regression analysis with Marquardt's 

 algorithm (NLIN procedure) and the NLIN weight 

 statement were used to fit the von Bertalanffy growth 

 model to observed length at age data (von Berta- 

 lanffy, 1938). Lengths were weighted by the inverse 

 of the number of fish at each age to moderate the 

 effect of large and small sample sizes on the esti- 

 mates of growth parameters. Age-length keys were 

 applied to length data collected through the Trip 

 Interview Program (TIP) in the Carolinas to gener- 

 ate an age-frequency distribution. TIP is a commer- 

 cial fisheries data collection program funded by the 

 National Marine Fisheries Service (NMFS). 



landed whole by fishermen (Table 1). Ninety gonad 

 samples from the headboat fishery during 1973-1981, 

 collected in association with the study of Matheson 

 and Huntsman ( 1984), were obtained from the Beau- 

 fort Laboratory of the NMFS. The posterior portion 

 of the gonad was fixed in 10*^ seawater-formalin for 

 1-2 weeks and transferred to 509^ isopropanol for 

 1-2 weeks. Gonad samples were processed, sectioned, 

 and stained with double-strength Gill hematoxylin 

 and eosin-y by using the methods of Schmidt et al. 

 (1993). 



Sex and reproductive state were assessed primar- 

 ily by one reader using histological criteria (Table 2), 

 without reference to body length or date of capture. A 

 second reader examined sections from 75 specimens 

 to ensure accurate interpretations. If the assessments 

 of the two readers differed, the section was viewed 

 simultaneously by the readers and rejected if agree- 

 ment could not be reached. Specimens with develop- 

 ing, ripe, spent, or resting gonads were considered 

 sexually mature. For females, this definition of sexual 

 maturity included specimens wdth oocyte development 

 at or beyond the cortical granule stage and speci- 

 mens with beta, gamma, or delta stages of atresia (see 

 Hunter and Macewicz, 1985). To ensure that females 

 were correctly assigned to either the immature or 

 resting categories, the length-frequency histogram of 

 females with evidence of certain maturity (e.g. those 

 that were developing, ripe, or spent) was compared 

 with the histograms for immature and resting females. 

 Females of uncertain maturity (Table 2) were excluded 

 from data analyses. To estimate length at SO'X matu- 

 rity (L50) and age at 50*7^ maturity (Ajg), the PROBIT 

 procedure (SAS Institute, Inc., 1990) was used to fit 

 gompit, logit, or probit models to maturity data in 

 25-mm length intervals or one year increments. The 

 LOGISTIC procedure was used to determine which 

 model to use in the PROBIT procedure. 



Females with hydrated oocytes or postovulatory 

 follicles were considered to be in spawning condition. 

 Macroscopic observations of snowy grouper caught 

 in June 1978 during trawls (Fig. 1) of an explor- 

 atory squid cruise conducted jointly by the govern- 

 ment of Spain and the Northeast Fisheries Center 

 of NMFS in Woods Hole, Massachusetts, on the RV 

 Pescapuerta Segundo were also used to define the 

 area and timing of spawning. 



Results 



Reproduction 



Gonads were obtained during 1979-95 from 870 spec- 

 imens collected on research cruises and from fish 



Age and growth 



An age was assigned to 91.6'^ of 1937 otoliths that 

 we examined (Table 1). Otoliths were rejected if the 



