188 



Fishery Bulletin 97(1), 1999 



Figure 4 shows the timing of discard rates for scup 

 in area 613 between 30 and 40 fm and in area 621 

 between 6 and 17 fm. For area 613, the highest rates 

 at those depths occurred in November and December 

 of each of the four years studied. Such a persistent trend 

 for high scup discards was not evident for area 621, 

 where highest rates occurred in May, August, and Sep- 

 tember 1991, and in September and October 1992. 



Figure 5 shows the mean number of pounds re- 

 tained and discarded for the most common species 

 caught in all tows sampled in area 613, between 30 

 and 40 fm, in the months of November and Decem- 

 ber each year. Because of inconsistent identifications, 

 data for squid and skates include data for all such 

 species. Squid were by far the main retained species 

 with an average of 507 Ib/h. Scup and whiting, 

 Merluccius bilinearis, were also retained in signifi- 

 cant quantities at 227 and 223 Ib/h respectively. The 

 main discarded species was dogfish, Sqiioliin 

 acanthios, (672 Ib/h), followed by scup i319 Ib/h) and 

 skates (236 Ib/h), and lesser quantities of butterfish, 

 Peprilua triacanthus, (96 Ib/h), whiting ( 72 Ib/h), and 

 red hake, Urophycis chuss, (51 Ib/h). 



Discussion 



The above treatment of a four-year subset of the 

 NMFS sea sampling database indicated that trawl- 

 ing in a particular area (area 613), depth (30-40 fm), 

 and time of year (November to December), consis- 

 tently led to high discard rates for scup. Before dis- 

 cussing this result, however, it is necessary to con- 

 sider the problems inherent in the data analyzed. In 

 particular, the design of this observer program in- 

 volved a nonrandom, nonindependent allocation of 

 sampling effort that was uneven in space and time 

 (see Fig. 2). These problems precluded 1) the use of 

 conventional statistical analyses to detect trends and 

 2) the identification of patterns for all areas for all 

 months. For example, the uneven observer coverage 

 may have resulted in the identification of only 9 of 

 the 20 statistical areas in the region as having high 

 discard rates for scup (Fig. 2). Although adequate 

 obsei^ver coverage seemed to occur across most of the 

 depths in many of these nine areas (Fig. 3), not all 

 areas had all depths sampled, precluding the con- 

 clusion that other depth-related areas of high scup 



