354 



Fishery Bulletin 100(2) 



We constructed fishery-independent keys from samples 

 taken in 1979-94 (Harris and McGovern, 19971 by the 

 Marine Resources Monitoring, Assessment, and Predic- 

 tion (MARMAP) Program, a fishery-independent sampHng 

 progi'ani of the SC Department of Natural Resources. In 

 that study, 8660 red porgy were collected, primarily with 

 hook and line and various traps, and preserved sagittae 

 were used for age estimation. As a compromise between 

 estimating annual keys based on smaller sample sizes and 

 an overall key, which would disguise growth variability 

 over time, we gi-ouped MARMAP aging data into 3-,yr pe- 

 riods, with the last four years divided instead into two 2-yr 

 periods: 1979-81, 1982-84, 1985-87, 1988-90, 1991-92, 

 1993-94. Potts and Manooch (2002) aged 111 red porgy 

 collected during 1996-97 by the MARMAP program (pre- 

 dominantly younger fish, maximum age of 6). We used an 

 age-length key from those data for 1995-97. Where fewer 

 than 10 fish occurred in a 25-mm total-length interval, we 

 pooled the data over longer time periods. 



No aging data are available from fishery-independent 

 sources before 1979. We constructed a key. used for 

 1972-74, from fishery samples taken in those years (Ma- 

 nooch and Huntsman, 1977). For 1975-78, we interpo- 

 lated linearly between that key and the earliest key de- 

 rived from fishery-independent data (1979-81). Thus, the 

 fishery-independent (primary) catch matrix described be- 

 low reflects some fishery-dependent data in the earliest 

 years. 



Catch-at-age matrices Equation 1 was applied to each 

 fishery-gear combination and catch-at-age estimates were 

 accumulated for each year to obtain estimates of annual 

 catch in numbers at age (the catch matrix). 1972-97. This 

 was done twice, first by using the fishery-independent age- 

 length keys to obtain the primary catch matrix, and then 

 by using fishery-derived age-length keys to obtain the 

 alternate catch matrix. (Although the designation of "pri- 

 mary" and "alternate" matrices is somewhat subjective, 

 it was based on the more extensive and continuous qual- 

 ity of fishery-independent data, as discussed later Despite 

 the designation, most analyses were conducted twice, once 

 with each matrix.) Final catch matrices each contained 

 numbers of fish caught at ages 1 through 8-i- in fishing 

 years 1972 through 1997; partial recruitment for age-0 red 

 porgy was essentially zero. 



We judged coherence of catch matrices by examining 

 pairwise correlations between ages as a cohort progressed 

 through the fishery. For fully recruited ages and with F 

 varying only moderately from year to year, we expected to 

 be able to follow a strong or weak cohort through the catch 

 matrices. Both matrices generally showed significant cor- 

 relations (P<0.1) between lagged catches at adjacent ages. 

 A few significant correlations were found among lagged 

 catches for nonadjacent ages. 



Growth in length and weight 



Von Bertalanffy ( 1938) growth models were fitted from data 

 on total length (mm) and age obtained from fishery-inde- 

 pendent and fishery-dependent sources (Table 2). Disaggre- 



gated aging data could not be located for fishery-dependent 

 data from 1972 to 1974 and from 1986; therefore observed 

 midintei-val lengths at age from keys were used for those 

 years. For fishery-dependent data for 1989-98 and fishery- 

 independent data for 1996-97, back-calculated length at 

 oldest age was used, a procedure that avoids potential bias 

 by the use of multiple measurements per fish (Vaughan 

 and Burton, 1994). For fishery-independent data, 1979-94, 

 obsei-ved length at age was used, adjusted for month of col- 

 lection. The particularly high standard error associated with 

 the estimate of L . for fishery-independent data, 1996-97, 

 was associated with a lack of fish older than 6 years of age 

 in the sample (Table 2). 



We estimated weight W (in kg) from total length L (in 

 mm) by using the relationship W = aV\ Parameter es- 

 timates were as follows: for fishery-dependent modeling, 

 we used values from Manooch and Huntsman (1977) (o = 

 2.524 X 10-''> and 6=2.894) for the 1970s and 1980s; and 

 from Potts and Manooch (2002;a=8.85 x 10-6 ^^d 6=3.060) 

 for the 1990s. For fishery-independent modeling, we used 

 values for 1972-78 from Manooch and Huntsman (1977), 

 and we estimated values for 1979-97 from MARMAP da- 

 ta. 1979-94, (a=3.064 x lO-^ and 6=2.865). 



Abundance indices 



Fishery-independent data on length, weight, age, and CPE, 

 1979-97 (Table 1) were obtained through the MARMAP 

 program (Collins and Sedberrv, 1991; Harris and McGov- 

 ern, 1997). In computing fishery-independent indices of 

 abundance from those data, we used CPE from hook-and- 

 line and trap gears only. We extended data from chevron 

 trap, 1988-97, back to 1980 with data on Florida snapper 

 trap, using a conversion factor determined by MARMAP 

 (Collins, 1990; Vaughan et al., 1998); we refer to the result- 

 ing series as "extended" chevron trap. To obtain age-spe- 

 cific indices of abundance for hook-and-line, 1979-97, and 

 extended chevron trap, 1980-97, we applied Equation 1, 

 using CPE in place of/;, and using annual MARMAP esti- 

 mates of CPE, age-length keys, and length-frequency data 

 for each gear 



Like the catch matrices, matrices of fishery-indepen- 

 dent CPE estimates were examined for coherence through 

 correlations among lagged CPE. The hook-and-line index 

 showed greatest coherence among ages 1 through 4, where- 

 as the extended chevron-trap index appeared quite coher- 

 ent over a wider range of older ages (3 through 7 ). 



Mortality estimation 



Instantaneous total mortality rates (Z) were estimated 

 from cohort-specific catch-cui-ve analyses (Ricker, 1975) of 

 the two catch matrices by using only ages that appeared 

 fully recruited (4 through 7). Empirical approaches were 

 also used to estimate total mortality rate (Z). The more 

 complex of two methods presented by Hoenig (1983) is 

 based on age of the oldest fish obsei-ved, sample size, and 

 age of recruitment to the sampling procedure. Assuming 

 those quantities to be 18 yr, 10,000 fish, and 1 yr, we esti- 

 mated Z = 0.58/yr, a likely upper bound on M. 



