Wells and Rooker: Distribution, age, and growth of young-of-the-year Senola dumerili 



547 



of the fishes and the average correction accounted for 

 9.5% of the actual age estimate. Otolith readings with 

 correction factors accounting for more than 20% of the 

 predicted age were not used for estimates of growth. 

 The following correction factor was used 



Age (d) = 2.88 x otolith radius (jUm) - 0.096 



(r 2 = 0.88, n=20). 



Additionally, all otolith counts were repeated twice 

 to ensure adequate precision. Differences in readings 

 of more than 20% were not incorporated into growth 

 estimates. 



Daily deposition of growth increments on sagit- 

 tal otoliths was validated by using wild S. dumerili 

 (re=14, 136-193 mm SL). Fishes caught in the wild 

 were brought into the laboratory and placed in a cir- 

 cular holding tank (1.71 m diameterx0.75 m depth) 

 for 48 hours. Fishes were then placed in a separate 

 tank containing 80 liters of seawater with 100 mg/L of 

 alizarin complexone for two hours (Thomas et al., 1995) 

 and returned to the circular holding tank. Individuals 

 were fed approximately 10% of their body weight daily. 

 Fishes marked with alizarin were removed from the 

 tank after 5 (;? = 5), 10 (n=5), and 15 (n=4) days. The 

 number of otolith increments between the alizarin mark 

 and outer edge were then counted for daily increment 

 verification. Otolith slides were coded so that all read- 

 ings were blind. 



Hatching dates were determined for all individuals 

 by subtracting daily age from date of capture. An age- 

 specific mortality adjustment was made for individuals 

 because larger S. dumerili have spent more time in 

 the early life stages and hence individuals from these 

 cohorts have experienced greater cumulative mortality. 

 Because of the limited number of individuals in 2001, 

 the mortality correction was calculated only for year 

 2000 collections and applied to hatching-date distribu- 

 tions in 2000 and 2001. Age-specific mortality adjust- 

 ments were made according to the method described by 

 Rooker and Holt (1997). 



Growth and mortality of S. dumerili were estimated 

 by using otolith-derived ages. Daily growth rates were 

 estimated by using the linear growth equation 



SL - slope (age) + y-intercept 



and were reported as mm/d. Length-at-age data were 

 also fitted with curvilinear growth models (von Ber- 

 talanffy, Laird-Gompertz). Percent variation in length 

 explained by age for both curvilinear models was slightly 

 better at times than the percent variation in length 

 explained by age for the linear model; however, certain 

 model parameters (i.e. LJ were biologically unrealistic 

 and thus the linear model was deemed more appropri- 

 ate. Moreover, when possible, L_ values were used to 

 model length-at-age data and the nonlinear models were 

 essentially linear over the limited size range examined. 

 Mortality estimates for year 2000 S. dumerili were 

 determined by using a regression on the decline in log (> - 



transformed abundance on age. A regression coefficient 

 (slope) was used to predict the instantaneous mortality 

 rate: 



\r\N, = ln7V - Zt, 



where N t = abundance at age t (expressed in days); 



N = an estimate of abundance at hatching; 

 and 

 Z (slope) = the instantaneous mortality coefficient. 



Mortality estimates were based upon 10-day cohort 

 groupings. Individuals <40 days old were not included 

 in the mortality regression because of an ascending 

 catch curve and because there were too few individuals 

 >139 days old in our sample — probably owing to gear 

 avoidance or emigration (or both). Therefore, only S. 

 dumerili between 40 and 139 days (45-192 mm) were 

 used to estimate mortality. 



Data analysis 



Effects of location and date on CPUE and size estimates 

 were examined by using a two-way analysis of vari- 

 ance (ANOVA). Levene's test and residual examination 

 established if the homogeneity of variance assumption 

 was met. Normality was evaluated by plotting residuals 

 versus expected values. Abundance data were log (.v+1) 

 transformed when necessary to normalize data and 

 reduce heteroscedasticity. Tukey's honestly significant 

 difference ( HSD ) test was used to determine a posteriori 

 differences among means. Comparisons of spatial and 

 temporal variation in growth were performed by using 

 analysis of covariance (ANCOVA). Prior to ANCOVA 

 testing, the homogeneity of slopes assumption was exam- 

 ined using an interaction regression (Ott, 1993). If no 

 significant interaction was detected, ANCOVA models 

 were used to test for differences in length-at-age (y- 

 intercepts) (Ott, 1993). Statistical analysis was car- 

 ried out by using SYSTAT 8.0 (SYSTAT Software Inc., 

 Richmond, CA), and significance was set at the alpha 

 level of 0.05. 



Results 



Environmental conditions 



Average temperatures from May to July ranged from 

 27.9 to 30.1°C in 2000 and from 24.5 to 30.4°C in 2001 

 (Fig. 2). Mean temperatures over the sampling period 

 were 29.2°C and 27.9°C for 2000 and 2001, respectively. 

 Zonal differences occurred: the inshore zone averaged 

 28.7°C (±0.3) in 2000 and 28.1°C (±0.9) in 2001, and 

 the offshore zone averaged 29.8°C (±0.3) in 2000 and 

 27.6°C (±0.9) in 2001. Similar to temperature trends, 

 mean salinity was higher in 2000 (34.6%< ) than in 2001 

 (31.9%o) (Fig. 2). Average salinity values gradually 

 increased from an average of 31.5%o in May to 37.2%r in 

 July of 2000. A large drop in salinity occurred during 



