630 



Fishery Bulletin 92(3). 1994 



In addition to the "All Years" analysis, individual 

 surveys were classified, revealing moderate agree- 

 ment among surveys. Of the two methods, RGA (pres- 

 ence-absence) produced more consistent results 

 across individual surveys. Changes in the structure 

 of the three assemblages in any given year usually 

 involved the addition or omission of one species or 

 the shifting of one of the more ubiquitous species into 

 another group. 



On the other hand, assemblage patterns across 

 surveys were less discernible with the biomass-ori- 

 ented CA because of the high variability in rockfish 

 CPUE's, particularly among shelf species. Usually, 



50 100 150 200 25C 300 350 400 



Depth (m) 



Figure 6 



The frequency of occurrence by depth of the 

 three rockfish assemblages determined from 

 cluster analysis. Group 1 consists of shortspine 

 thornyhead. Pacific ocean perch, darkblotched 

 rockfish, and splitnose rockfish. Group 2 con- 

 sists of yellowtail, canary, and greenstriped 

 rockfish. Group 3 consists of sharpchin. 

 redstripe, and rosethorn rockfish. See Table 1 

 for scientific names. 



at the highest dissimilarity level (Fig. 5), CA parti- 

 tioned the offshore community into a deep water, 

 upper-slope group and a shallower water, mid-shelf 

 group. The deeper species consisted of shortspine 

 thornyhead, Pacific ocean perch, darkblotched rock- 

 fish, and splitnose rockfish. Two additional species, 

 redbanded and rougheye, S. aleutianus, rockfish, 

 were often closely associated with this group. In 1983, 

 sharpchin, S. zacentrus, rockfish was included. The 

 shallower species comprised two core assemblages 

 (most easily seen in the "All Years" dendrogram). The 

 first of these consisted of canary, yellowtail, and to a 

 lesser extent, greenstriped rockfish. This assemblage 

 was clearly identified from 1977 through 1986. How- 

 ever, in 1989 and 1992, it was not defined, most likely 

 because of the shift in survey design that de-empha- 

 sized canary and yellowtail rockfish as target spe- 

 cies, thus reducing the sampling effort in areas where 

 they were most likely to have been found together. 

 The other core assemblage among shelf rockfishes 

 consisted of sharpchin, rosethorn, and redstripe rock- 

 fish, S. proriger. However, since these species were 

 taken along with a variety of other species, such as 

 silvergray rockfish, bocaccio, S. paucispinis, 

 yelloweye, S. ruberrimus, canary, and yellowtail rock- 

 fish, CA dendrograms showed considerable variabil- 

 ity between years. 



Discussion 



Survey data have provided the unique opportunity 

 to study broad-scale community composition of off- 

 shore rockfishes, including smaller specimens typi- 

 cally discarded in commercial operations. These data 

 are useful to both ecologists interested in describing 

 the biological associations of our ocean's resources 

 and to resource managers concerned with commer- 

 cial catch and bycatch issues. 



For a variety of reasons however, survey-defined 

 assemblages may differ from assemblages deter- 

 mined through commercial fisheries data. Most of 

 these revolve around the strict adherence to scien- 

 tific design of most surveys as opposed to industry's 

 opportunistic approach driven by economic needs. For 

 example, the assemblages identified in this study 

 would probably not be detected if sampling were lim- 

 ited to catches made while targeting strong hydro- 

 acoustic signals i Richards et al., 1991). Pelagic and 

 semipelagic aggregations, such as widow, shortbelly, 

 and in some cases yellowtail rockfish are under-rep- 

 resented in this study because they were sampled 

 less intensively during the surveys. Differences in 

 catch composition is also gear dependent. Our sur- 

 vey gear and methods of deployment were standard- 



