Weinberg: Rockfish assemblages off Oregon and Washington 



621 



conducted to identify assemblages using trawl data 

 covering extensive geographic areas. Nagtegaal 

 (1983) studied both annual and seasonal interrela- 

 tionships of some rockfish species off British Colum- 

 bia based on commercial catch data. However, these 

 catch statistics do not allow a full description of the 

 effects of selective harvest on the entire rockfish com- 

 munity, as the landed species are those of highest 

 economic value allowed for harvest at the time and 

 reflect only a portion of the overall rockfish commu- 

 nity exposed to trawling. Rogers and Pikitch (1992) 

 defined several groundfish assemblages based on 

 prediscard data from the commercial trawl fishery 

 off Oregon and Washington. That study included a 

 variety of groundfish families, but only the most 

 abundant rockfish species were considered. Similarly, 

 Gabriel (1982) included a wide variety of ground- 

 fishes in a 1-year assemblage study that utilized 1977 

 survey data from California to Washington. All of 

 these researchers recognized the value offish assem- 

 blage identification as a tool for fisheries manage- 

 ment and emphasized the need to verify assemblage 

 persistence. 



Since 1977, the Alaska Fisheries Science Center 

 (AFSC) of the National Marine Fisheries Service 

 (NMFS) has been conducting controlled bottom trawl 

 surveys aimed at assessing and monitoring ground- 

 fish resources off the west coast of the United States. 

 Now that several of these surveys have been per- 

 formed, a unique opportunity exists to monitor per- 

 sistence in fish assemblages. In this paper I describe 

 and summarize these surveys and their rockfish 

 samples and use standard numerical classification tech- 

 niques to identify the major rockfish assemblages. 



Methods 



cific hake, Merluccius productus. The survey design 

 was patterned after rockfish distributions, deter- 

 mined by fisheries catch data and the results of a 

 pilot rockfish survey (Gunderson and Nelson 2 ). The 

 1980 survey was specifically redesigned to better 

 assess the canary and yellowtail rockfish populations, 

 in addition to Pacific hake (Coleman, 1986). Thus, 

 the sampling effort was divided among three depth 

 strata: 55-183 m, 184-220 m, and 221-366 m. The 

 1983 survey repeated the work conducted in 1980 

 with the addition of some stations in the northern 

 U.S. -Vancouver area (Weinberg et al., 1984). Based 

 on the results of the previous surveys and in an at- 

 tempt to further reduce the variance of canary and 

 yellowtail rockfish catch rates, station allocation was 

 changed again in 1986 (Coleman, 1988). In that year, 

 sampling was apportioned among four depth strata: 

 55-91 m, 92-183 m, 184-219 m, and 220-366 m. 

 Almost three times the effort was applied in the U.S.- 

 Vancouver area, most of which was off northern 

 Washington (lat. 48°00'-42 J 23'N). However, having 

 not been able to improve rockfish estimates signifi- 

 cantly, the 1989 and 1992 AFSC surveys shifted away 

 from rockfish concerns of past surveys and concen- 

 trated on abundance estimation of Pacific hake and 

 young sablefish, Anoplopoma fimbria. Consequently, 

 the high density rockfish strata were abandoned and 

 sampling was allocated within only two depth strata, 

 55-183 m and 184-366 m (Weinberg et al., 1994). 



Samples were collected with standardized Nor'eas- 

 tern high-opening rockfish bottom trawls rigged with 

 roller gear. In general the gear's horizontal and ver- 

 tical openings measured 13 and 9 m, respectively. 

 Towing was controlled by fishing along depth con- 

 tours for one-half hour at about three knots. Catches 

 were sorted by species, weighed, and counted. 



This study utilizes rockfish catch data from six AFSC 

 multispecies groundfish assessment surveys con- 

 ducted triennially over a 16-year period from 1977 

 to 1992. Only data from bottom trawling in the Co- 

 lumbia and U.S. -Vancouver INPFC areas (43°00'N 

 to the U.S. -Canada border) were examined. Trawl- 

 ing occurred during August and September between 

 the depths of 55 and 366 m. All surveys employed 

 stratified random sampling designs, apportioning 

 towing sites according to various geographic strata 

 and depth intervals. While the overall multispecies 

 assessment goal remained unchanged from one sur- 

 vey to the next, many of the specific objectives did 

 not. Objectives of the 1977 survey included deter- 

 mining the distribution and abundance of several 

 commercially important rockfishes (Gunderson and 

 Sample, 1980) and the on-bottom component of Pa- 



Assemblage analyses 



I examined rockfish associations using two tech- 

 niques: recurrent group analysis and cluster analy- 

 sis. These two methods provide somewhat different 

 characterizations of species distribution and 

 cooccurrence and, when used together, can enhance 

 our understanding of rockfish communities. 



Recurrent group analysis (RGA), a nonhierarchical 

 technique, addresses the question of which rockfish 

 are likely to be caught together, thus reflecting their 



2 Gunderson, D. R., and M. O. Nelson. 1977. Preliminary report 

 on an experimental rockfish survey conducted off Monterey, 

 California, and in Queen Charlotte Sound, British Columbia, 

 during August-September 1976. U.S. Dep. Commer., NOAA, 

 Natl. Mar. Fish. Serv., Northwest and Alaska Fish. Cent., 2725 

 Montlake Blvd. E.. Seattle. WA 98112. Unpubl. manuscr., 82 p. 



