OVERHOLTZ and TYLER: DEMERSAL FISH ASSEMBLAGES 



analyses. Sampling frequency averages about one 

 station for every 250 mi^ or roughly 300 locations 

 in a normal survey from Cape Hatteras to Nova 

 Scotia. 



Data from a selected portion of this time series 

 was used in cluster analyses that defined demersal 

 fish assemblages. Specifically, a group of 36 species 

 representing the dominant fishes on Georges Bank, 

 were chosen as the focus for the investigation (Ihble 

 1). This choice was based on a preliminary examina- 

 tion of the data to determine which species were 

 most important in terms of biomass and numerical 

 density. Catches (kg) for each of the species from 

 every station in a particular cruise were organized 

 into a data matrix and processed with an agglomera- 

 tive cluster analysis program (Keniston 1978). lb 

 remove skewness in the species matrices, we trans- 

 formed the data prior to clustering by using an In 

 (x + 1) conversion. Station dissimilarities were 

 calculated by using the Bray-Curtis dissimilarity in- 

 dex, an ecological distance measure that is sensitive 

 to dominant species (Clifford and Stephenson 1975; 

 Boesch and Swartz 1977). 



Table 1.— Species cited by common name in the text. 



Common name 



Scientific name 



Spiny dogfish 

 Winter skate 

 Little skate 

 Smooth skate 

 Thorny skate 

 Atlantic herring 

 Alewife 

 Offshore hake 

 Silver hake 

 Atlantic cod 

 Haddock 

 Pollock 

 White hake 

 Red hake 

 Cusk 



American plaice 

 Summer flounder 

 Fourspot flounder 

 Yellowtall flounder 

 Winter flounder 

 Witch flounder 

 Windowpane 

 Gulf stream flounder 

 Atlantic mackerel 

 Butterfish 

 Bluefish 



Blackbelly rosefish 

 Longhorn sculpin 

 Sea raven 

 Gunner 



American sand lance 

 Atlantic wolffish 

 Ocean pout 

 American goosefish 

 Short-finned squid 

 Long-finned squid 



Squalus acanthias 

 Raja ocellata 

 Raja erinacea 

 Raja senta 

 Raja radiata 

 Clupea harengus 

 Alosa pseudoharengus 

 Merluccius albidus 

 Merluccius bilinearis 

 Gadus morhua 

 Melanogrammus aeglefinus 

 Pollachius virens 

 Urophycis tenuis 

 Urophycis chuss 

 Brosme brosme 

 Hippoglossoldes platessoides 

 Paralichthys dentatus 

 Paralichthys oblongus 

 Limanda ferruginea 

 Pseudopleuronectes americanus 

 Glyptocephalus cynoglossus 

 Scophthalmus aquosus 

 CItharichthys arctifrons 

 Scomber scombrus 

 Peprilus triacanthus 

 Pomatomus saltatrix 

 Helicolenus dactylopterus 

 Myoxocephalus octodecemspinosus 

 Hemitripterus americanus 

 Tautogolabris adspersus 

 Ammodytes americanus 

 Anarhichas luptus 

 Macrozoarces americanus 

 Lopliius americanus 

 lllex illecebrosus 

 Loligo pealei 



The resulting dissimilarity matrix was used in a 

 group average fusion strategy to combine stations 

 with similar species distributions (Clifford and 

 Stephenson 1975). These station combinations were 

 displayed in dendrograms, which were examined and 

 assemblage groups were chosen by two criteria: 

 large-scale separations, as shown in Figure 2, and 

 dissimilarity levels. Stations from these assemblage 

 groups were plotted on cruise maps from the original 

 sampling plan and areas were delineated. This pro- 

 cess was repeated for all spring and fall cruises to 

 provide a consecutive series of maps, which were 

 then examined for continuity (Fig. 3). Finally, data 

 from several consecutive years were pooled to 

 delineate assemblages designated, based on nearby 

 geographic features or depth zones. 



Species lists were prepared for the assemblages 

 outlined in the pooled cluster results, and data were 

 analyzed to further define the structure of each 

 group. Length frequencies from species in the dif- 

 ferent assemblages were used to separate life history 

 stages and catch-per-tow data were used to in- 

 vestigate trends in distribution and abundance Ex- 

 amination of food habit data in the literature and 

 NEFC documents gave further insight into assem- 

 blage structure Trajectories of assemblage CPUE 

 for selected species were plotted and examined for 

 long-term trends. Ibta! assemblage CPUE was also 

 investigated and compared with previous trends 

 reported by other authors for the region. 



Gradient analyses were performed with the objec- 

 tive of explaining species distributions based on a 

 set of location, physical, and chemical variables. 

 Canonical correlations, using information on latitude, 

 longitude, depth, bottom temperature, bottom oxy- 

 gen, and bottom salinity, were employed to define 

 possible gradients that might be useful indicators of 

 species distribution (Pimentel 1979). Data for the 

 autumn cruise were obtained from measurements 

 of bottom temperature and depth made aboard the 

 RV Albatross IV {U.S.A.) 20 October to 5 November 

 1976, and corresponding information on bottom 

 salinity and oxygen from the RV Anton Dohrn 

 (Federal Republic of Germany) 14 November to 1 

 December 1976. Information for the spring cruise 

 was procured from measurements of bottom 

 temperature and depth from the RV Albatross IV, 

 17 April to 3 May 1978, and salinity and oxygen data 

 that was obtained from the RV Argus (Union of 

 Soviet Socialist Republics) from 13 to 28 April 1978; 

 these two data sets were chosen because they cor- 

 responded closely in time to the available station 

 information. 



509 



