Rooker et al.: Variability in growth, mortality, and recruitment of Sciaenops ocellatus 



583 



(18%), only one of the lapilli was available owing to 

 loss, breakage, or staining. 



Growth rates were determined by using otolith- 

 derived estimates of age (size-at-age plots) in 1994 

 (n=249) and 1995 (n = 140). Daily instantaneous 

 growth coefficients were calculated from an exponen- 

 tial model described as 



where L^ = length (mm SL) at time t; 



Ly = the estimated length at hatching; 



g = the instantaneous growth coefficient (/d); 



and 

 t - theotoKth-derivedage(daysafterhatching). 



Ages of red drum not assessed with otolith-based 

 techniques ( 1994 [« = 1,057], 1995 [n=8004] ) were es- 

 timated by using age-length relationships. 



Weight-specific instantaneous growth coefficients 

 (G) were calculated by using the equation 



W^ = W^e^', 



where W^ = the wet weight (mg) at time ^ 



Wg = the estimated weight at hatching; 



G = the weight-specific instantaneous 



growth coefficient (/d); and 

 t = the otolith-derived age. 



Red drum lengths ( mm ) were converted to wet weight 

 (mg) by using a polynomial equation (third order) 

 based on measurements from laboratory-reared red 

 drum ( range; 6-30 mm ): W = -7.745 -t- 2. 122L - 0.205L- 

 -t- 0.024L'^ (n=200, r"=0.97), where W and L represent 

 wet weight and standard length, respectively. 



Mortality rates were estimated from regressions 

 of the decline in log, -transformed abundance on age. 

 Although interannual variation in mortality was 

 examined by pooling individuals from each year (all 

 cohorts combined), cohort-specific rates were based 

 on regression plots of log^.-transformed abundance 

 of individuals from 10-d cohorts determined from 

 hatching-date analysis. Hatch dates of individual red 

 drum were determined by subtracting the otolith- 

 derived age from the date of collection. Hatching 

 dates were then used to separate individuals into 

 specific cohorts, defined as individuals hatched 

 within a 10-d period. Each cohort was designated by 

 a letter (A to F): 1-10 Sep (A); 11-20 Sep (B); 21-30 

 Sep (C); 1-10 Oct (D); 11-20 Oct (E); and 21-30 Oct 

 (F). Densities for the first and last cohort (A, F) were 

 low and thus mortality regressions were not fitted to 

 these data. Daily instantaneous mortality rates were 

 calculated fi-om the exponential model of decline: 



N, = N^e-^', 



where A^^ = abundance at time t; 



Nq = the estimated abundance at hatching; 

 Z = the instantaneous mortality coefficient 



(/d); and 

 t = the otolith-derived age. 



Owing to incomplete capture (i.e. ascending limb 

 of catch curve) of small red drum (<25 d, <8 mm), 

 these individuals were not included in mortality re- 

 gressions. Also, size-based gear avoidance was sus- 

 pected, which placed constraints on the upper age 

 (or size) of individuals used in regression analysis. 

 In preliminary trials, the capture efficiency of the 

 epibenthic sled was compared with a bag seine (seine 

 dimensions: 7 m length x 1 m height; mesh size: 

 3 mm) and size-specific differences in capture effi- 

 ciency (individuals/m''^ of habitat sampled ) between the 

 two gears were not detected for sciaenids <25 mm 

 (Rooker, 1997). However, density estimates of sciae- 

 nids >25 mm were higher with the seine. This sug- 

 gested that individuals >25 mm are capable of avoid- 

 ing the sled and that densities calculated for these 

 individuals would be underestimated. Consequently, 

 only red drum between 8 and 20 mm (25-40 d ) were 

 used to estimate mortality rates. 



Our assessment of early life mortality was based 

 on two assumptions: 1 ) individuals entering seagrass 

 sites remained in these habitats (i.e. settle and stay) 

 during the time period when mortality rates were 

 estimated, and 2) immigration and settlement (i.e. 

 late settlers) to the designated study sites from other 

 locations or habitats, or both, was negligible. Al- 

 though postsettlement movement (emigration or 

 immigration, or both) can be important for certain 

 species and failure to account for such movements 

 can severely bias mortality estimates (Frederick, 

 1997), our assumptions are reasonable because red 

 drum appear to settle and stay in seagrass meadows 

 at least through the early juvenile stage (Rooker et 

 al., 1998b). Furthermore, no salient increases or de- 

 creases in density were observed in length-frequency 

 profiles which would be expected in the presence of 

 postsettlement emigi'ation or immigration activity. 



Since total mortality (Z) is the sum of both natural 

 (M) and fishing mortality (F), it is important to ac- 

 count for the fishing mortality iF) caused by our sam- 

 pling (Ricker, 1975). Most studies evaluating early 

 life mortality do not estimate F (^sampling mortal- 

 ity ) and assume F to be inconsequential. In this study, 

 seagrass meadows of limited size (ca. 25,000 m-) were 

 sampled repeatedly and, as a result, the effect of F 

 was considered potentially significant. Therefore, 

 mortality regressions were run on abundance data 



