FISHERY BULLETIN: VOL. 83, NO. 4 



9B). Haddock CPUE, on the other hand, declined 

 dramatically from 97.3 kg in 1963 to in 1972, re- 

 maining at very low^ levels of abundance in the later 

 years. Yellowtail flounder fluctuated from 15 kg in 

 1963 to a low of 6 kg in 1966, increased from 1966 

 to 1969, and declined through 1978 (Fig. 8B). 



An examination of trends in cumulative percent 

 by weight and number trajectories for the Shallow 

 assemblage highlighted some interesting points. 

 Atlantic cod comprised a fairly constant proportion 

 of the species biomass for all the years except 1964 

 and 1965. Longhorn sculpin, yellowtail flounder, and 

 to a lesser extent winter flounder, made up an in- 

 creasing part of the biomass of this assemblage dur- 

 ing 1966-71 and then all declined in importance (Fig. 

 9C). Haddock, as previously noted, experienced a 

 pronounced decline in abundance from the early 

 1960's and was only present at very low levels from 

 1972 to 1978. Winter skates, little skate, and 

 windowpane flounder accounted for an increasing 

 percent of the biomass in this assemblage from the 

 early 1970's onward (Fig. 9C). 



When cumulative percent by number was inves- 

 tigated, silver hake and red hake became important 

 (Fig. 9D). Silver hake was the numerical dominant 

 through most of the mid- and late 1970's. This trend 

 was due entirely to increased numbers of juvenile 

 silver hake that represented a small amount of 

 biomass. This same phenomenon applies to red hake, 

 which enjoyed several periods of increased abun- 

 dance as a proportion of the total numerical densi- 

 ty from 1963 to 1978. Winter skate numbers remain- 

 ed relatively unchanging from 1963 to 1976 and then 

 rose slightly in the late 1970's. Trends for window- 

 pane flounder, longhorn sculpin, little skate, yellow- 

 tail flounder, and haddock follow the cumulative ab- 

 solute and percent weight data (Fig. 9B, C, D). 



The other Georges Bank assemblages were inves- 

 tigated using the same techniques, but on a much 

 less intense scale TDtal mean catch/tow for the Inter- 

 mediate, Gulf of Maine Deep, and Northeast Peak 

 assemblages is displayed in Figure 6 for the fall 

 surveys 1963-78. The trends in total CPUE follow 

 the same basic patterns for all three groups, a high 

 initial period followed by a decline and subsequent 

 recovery in the mid- to late 1970's. 



General decreases in the catch of throny skates, 

 haddock, and cod were responsible for the downward 

 trend in CPUE for the Northeast Peak assemblage, 

 but the recovery that occurred in the late 1970's was 

 due primarily to increased haddock biomass (Figs. 

 6, 10). The Northeast Peak assemblage is fairly sim- 

 ple in species composition, and although some fluc- 

 tuations in cumulative percent by weight occurred. 



NORTHEAST PEAK ASSEMBLAGE 

 CUMULATIVE PERCENT BY WEIGHT 

 FALL 1963-1978 



100 r 



POLLOCK 



HADDOCK 



1963 



1967 



1971 



1975 



Figure 10.— Reponses of species from the Northeast Peak assem- 

 blage demersal fish community expressed as cumulative percent 

 by weight for autumn 1963-78. 



the same four species remained dominant over the 

 period (Fig. 10.) 



Time sequence cluster analyses were useful as fur- 

 ther indicators of temporal trends in these groups. 

 Species biomass for the Slope and Canyon assem- 

 blage did not appear to follow any clear long-term 

 trend (Fig. 11). Enough fluctuation in CPUE oc- 

 curred to mask any trend, and no clear pattern was 

 established. This same analysis on the Shallow 

 assemblage showed three distinct temporal clusters, 

 composed of consecutive years (Fig. 11). Using this 

 perspective and Figure 9, there appears to have been 

 three periods of significant change in relative abun- 

 dance during the fall time series; an initial period 

 dominated by haddock, intermediate period with 

 high yellowtail, longhorn sculpin, and winter 

 flounder catches, and finally a group with little skate, 

 winter skate, and windowpane flounder as the domi- 

 nant species. 



GRADIENT ANALYSIS 



Gradient analyses of two selected Georges Bank 

 data sets did not prove to be as useful as was hoped, 

 but some information and insight were gained and 

 the dimensionality of the large multivariate data sets 

 involved was much reduced. The data set used in the 

 fall 1976 canonical correlation analysis accounted for 

 about 26% of the variation in species distribution for 

 32 selected species of interest. The variables included 

 in the analysis were latitude, longitude, depth, bot- 

 tom temperature, bottom salinity, and bottom oxy- 



516 



