HALES: ROUND SCAD IN THE SOUTH ATLANTIC BIGHT 



ferent time intervals were compared to evaluate size 

 bias that may occur with this method (Jenkins and 

 Green 1977). 



mined from 156 individuals (13-185 mm FL) random- 

 ly selected within 10 mm size classes from all 

 specimens (total = 1047) collected in 1980. 



Age and Growth 



Utricular otoliths (lapilli) of specimens collected 

 in 1980 were used for age determination. Otoliths 

 were stored in 95% ethyl alcohol and prepared for 

 viewing using a modification of the methods of 

 Haake et al. (1981), which resulted in a thin sagit- 

 tal section containing the core of the otolith em- 

 bedded in "Spurr" (Spurr 1969). Otolith length was 

 measured to the nearest 0.1 mm at 100 x with an 

 ocular micrometer. Otolith images were projected 

 on a high resolution television screen with a high 

 resolution camera, which produced a total viewing 

 magnification of 1088 x or 2176 x. Otoliths ex- 

 amined by scanning electron microscopy were pre- 

 pared by the methods of Haake et al. (1981). Two 

 counts of growth increments were made by the 

 author, and an additional count was made by other 

 experienced readers. Mean counts were used in all 

 analyses, and specimens were discarded if individual 

 counts for a specimen differed by more than 10%. 

 Different readers usually showed agreement 

 between counts: percentage difference between 

 readers averaged 8%. 



Counts of otolith increments were obtained from 

 71 juvenile and adult round scad, 13-143 mm FL. 

 Sixty specimens (121-180 mm FL) could not be 

 assigned ages because of the numerous growth 

 interruptions in outer regions of the otolith. Incre- 

 ment formation was validated by examination of the 

 margins of otoliths of juveniles (13-55 mm FL) 

 collected at different times of day. Consistent 

 measurements of the marginal increment could not 

 be made because of the irregular shape of the lapilli; 

 thus, only the occurrence of an incremental or dis- 

 continuous zone (terminology of Mugiya et al. 1981) 

 could be noted. 



The SAS NLIN regression procedure with DUD 

 and Marquardt options (Helwig and Council 1979) 

 was used to determine parameters for the von Ber- 

 talanffy (1957) and Gompertz (Zweifel and Lasker 

 1976) growth equations. Because similar patterns 

 of variation were observed in plots of the residuals 

 of both models, r^ values were used to evaluate 

 model performance (Grossman et al. 1985). Instan- 

 taneous growth rates (%FL d'^ and %WT d^^) 

 were calculated according to Ricker (1979). Weights 

 were converted from lengths by using the least 

 squares regression (Sokal and Rohlf 1981), In wt (in 

 g) = 2.96 In FL (in mm) - 11.2 (r^ = 0.99), deter- 



RESULTS 



Seasonal Distribution and 

 Relative Abundance 



A total of 57,460 round scad were captured at 230 

 of the 739 stations in depths from 11 to 267 m; over 

 99% of the catch came from <92 m. Fish ranged 

 from 2 to 26 cm FL (x = 11.4), with 99% of the 

 fish 6-17 cm FL. 



Round scad were more widely distributed and 

 abundant in summer and fall than in winter and 

 spring. Indices of relative abundance (Fig. 1) were 

 consistently high in summer and fall at shallow 

 depths (<55 m) where D. punctatus were captured 

 at 121 of 220 trawl stations. Indices of relative abun- 

 dance during summer and fall in 56-110 m depths 

 were quite variable, and catches in waters >110 m 

 were rare (3 of 78 trawls), small (52 individuals cap- 

 tured), and occurred only in summer. The highest 

 indices in winter and spring occurred in 19-110 m 

 depths (usually 19-27 m), but were lower than values 

 in summer and fall. Round scad were rarely collected 

 in 9-18 m depths and never collected in waters >110 

 m in winter. 



Differences among un transformed (Xgf), trans- 

 formed (xin), and Bliss (ieiiss) estimates of the 

 stratified mean catch per tow (Table 1) revealed 

 additional seasonal changes in the distribution of 

 round scad. Transformed and Bliss estimates of the 

 stratified mean catch per tow were higher in sum- 

 mer and fall than in winter and spring. However, 

 untransformed values (Xg/) indicated that total 

 catches in winter often exceeded total catches in 

 summer. Such differences in catch statistics resulted 

 from the relatively high frequency and low vari- 

 ability of catches in summer and fall, and the rela- 

 tively low frequency and high variability of catches 

 in winter and spring. This result was generally con- 

 sistent with the coefficients of variation (CV and 

 CVin), which indicated increased clumping in the 

 winter catches. 



The seasonal distribution of round scad appeared 

 affected by temperature (Table 2). Over 97% of 

 winter catches occurred in waters warmer than 

 15°C, over 99% of spring catches were made in 

 waters warmer than 17° C, and over 99% of sum- 

 mer and fall catches occiirred in waters warmer than 

 20°C. 



Habitat affected the distribution of round scad in 



253 



