Marine Biology Research, 2006; 2: 260Á269 ORIGINAL ARTICLE Fish assemblages in the Barents Sea MARIA FOSSHEIM1, EINAR M. NILSSEN1 & MICHAELA ASCHAN2* 1Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway, & 2Institute of Marine Research, N-9291 Tromsø, Norway Abstract Fish species inhabiting the Barents Sea display great seasonal and inter-annual variation in abundance and distribution. This study describes the assemblages and distributions of fish species in the southwestern and central part of the Barents Sea, which includes the polar front. The area has an unpredictable environment due to variable inflow of Atlantic water and fish species composition differs between Arctic and Atlantic water masses. Assemblages can be identified as distinct groups corresponding to different environmental conditions. In the period 1997Á1999, 57 fish species and shrimp were identified and 29 species/species groups were used in the statistical analyses. Cluster and correspondence analyses showed that the fish community consists of four assemblages: a northern, a southern, a deep and a central group. Temperature explained 26.2% of the variation in the species data, and depth 14.5% of the variation. The assemblages are coherent with previous zoogeographical studies from the Barents Sea. To reduce research costs, one might monitor indicator species characteristic for the identified fish assemblages. Fish assemblages may be a good tool when studying ecosystem responses to fishery and marine climate change. Key words: Correspondence analysis, distributions, indicator species, unpredictable environment, zoogeography Introduction indicators of the environment (Horn 1980). Studies on fish assemblages conducted in the Barents Sea The Barents Sea is a highly productive ecosystem have focused on commercial and well-known and an important nursery and nourishment area for several important commercial and non-commercial species. In this study we aimed to include all fish species. Great seasonal and inter-annual varia- identifiable fish species. tion in the abundance and distribution of fish species Exploitation of most marine resources is intrinsi- has been observed (Shepherd et al. 1984; Shevelev et cally multispecies and it is therefore important to al. 1987; Loeng 1989; Nilssen & Hopkins 1992; know whether some groups of species can be Murawski 1993). Understanding these dynamics is a considered ecological entities that respond similarly matter of necessity for sustainable development of to the environment (Tyler et al. 1982; Overholtz & the area. The commercial fish species have hitherto Tyler 1985; Burgos 1989). Spatial patterns of caught the most attention, especially cod (Gadus interaction among different groups of organism (in morhua), herring (Clupea harengus) and capelin this case fish species) are important, as their (Mallotus villosus), but also haddock (Melanogram- co-occurrence suggests common ecological pro- mus aeglefinus), redfish (Sebastes marinus and S. cesses. If different fish assemblages can be identified mentella), Greenland halibut (Reinhardtius hippoglos- in the Barents Sea, this information may help the soides) and, to a limited extent, polar cod (Boreogadus authorities in monitoring and managing the area. saida) (Shevelev et al. 1987; Loeng 1989; Nakken Monitoring the fish assemblages may give informa- 1998). The non-commercial fish stocks can be tion on the biological response to factors such equally important from an ecological point of view, as fisheries or climate change. Surveys are costly for example as prey for commercial species and as and labour intensive and by identifying distinct Correspondence: M. Fossheim, Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway. E-mail: [email protected] Published in collaboration with the University of Bergen and the Institute of Marine Research, Norway, and the Marine Biological Laboratory, University of Copenhagen, Denmark *Present address: Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway. (Accepted 4 May 2006; Printed 12 September 2006) ISSN 1745-1000 print/ISSN 1745-1019 online # 2006 Taylor & Francis DOI: 10.1080/17451000600815698 Fish assemblages in the Barents Sea 261 assemblages of species and the distribution of these is the major transition area of the physical environ- assemblages one might be able to reduce the number ment (Figure 1) (Loeng 1991; Loeng et al. 1997). of stations on surveys and thus minimize costs, or A zoogeographical analysis of the Barents Sea in practicality get more research for a limited fauna demonstrated three geographical areas with amount of money (Weslawski & Kwasniewski different environmental properties (Zenkevich 1983). Identifying assemblages of species that can 1956): (i) the northernmost part of the sea is be managed adaptively as similar entities of produc- characterized by a severe temperature regime and tion has also been suggested to alleviate overfishing drift ice during a lengthy period and is considered a in the trawl fishery (Tyler et al. 1982; Jay 1996). deep Arctic area of the Arctic region; (ii) the main, To manage such ecological entities they need to be central part of the sea is considered a sub-Arctic area stable [in the sense of ‘‘resilience’’ used by Grimm & of the Arctic region; and (iii) the southwesternmost Wissel (1997)]. Especially, the assemblages need to part of the sea is a section receiving most of the consist mainly of the same species between years, Atlantic heat, with the bottom water temperature even if their combined geographical distribution being above /18C and the fauna belonging to the varies considerably. It is also important to know boreal type. Ekman (1953) classified the fish fauna where faunal discontinuity between assemblages into four zoogeographical groups: Arctic, ArcticÁ arises and, of course, why. In the Barents Sea, it is boreal, boreal and warm waterÁboreal [not included expected that a faunal discontinuity area will be by, but south of the area described by Zenkevich concurrent with the position of the polar front, as it (1956)]. Figure 1. The Barents Sea with main surface currents. Atlantic currents (*/), Arctic currents (---/) and the mean position of the polar front (+++). The study area is indicated by the grey square. 262 M. Fossheim et al. Since the middle of the twentieth century, more and one species of shrimp (Table I). Due to data on several species have emerged and the uncertain identification, some species were merged computer has offered an important tool in exploring in groups of lowest taxonomic level. Rare species, community data of a multivariable nature. defined as species with less than six individuals in Multivariate methods such as cluster analysis and total or represented on less than five stations each ordination have proven very helpful in exploring year, were excluded from the analysis because they patterns in large data sets from community sampling represented less than 5% of the station catch (Gomes et al. 1995; Greenstreet & Hall 1996; Farina (Høines et al. 1998). Pelagic species (i.e. herring, et al. 1997; Gaertner et al. 1998; Jacob et al. 1998; capelin and blue whiting) were excluded from the Ungaro et al. 1998). Ordination is also able to analyses. There were 29 species/species groups in the explore complex environmental gradients in an final analyses (Table I). Abundance data were intelligible way. Burgos (1989) used cluster analysis standardized to 20 min trawl hauls (or 1 nautical and ordination to explore the fish community of the mile) when necessary and log10(a/1) transformed southern part of the Barents Sea. Our study area prior to cluster and correspondence analyses with comprised the polar front and the data set included the purpose of downscaling very abundant species both commercial species, previously explored by and reducing skewness (ter Braak 1997). Nilssen & Hopkins (1992), as well as non-commer- Cluster analysis was used to study the concurrence cial species. of species and a hierarchical method was chosen, The main questions are thus: exploring the results as a dendrogram for each year. The cluster analysis was based on a Spearman rank (1) Can the fish community in the southwestern correlation matrix and Ward’s method was chosen to Barents Sea be divided into different assem- minimize the variance within clusters. Groupings of blages? species for the 3 years combined and their relation- (2) Is there a faunal discontinuity across the polar ship with the environmental parameters were ex- front? plored by correspondence analysis (Greenacre (3) Are the assemblages stable? 1984). Correspondence analysis ‘‘extracts’’ the ordi- (4) Can this pattern be explained by abiotic nation axes from the species data alone. Species factors such as temperature and different appearing close to one another in the ordination water masses with certain characteristics? diagram have a more similar distribution than (5) Can the assemblages be considered entities species further apart. The environmental variables suitable for management? are added afterwards and are represented as arrows that point in the direction of maximum change. The Material and methods fraction of variance accounted for by the regression Data on fish assemblages and distributions were indicates whether the environmental variable is collected during three surveys in the southwestern sufficient to predict the variation in species composi- Barents Sea in spring 1997Á1999 (Figure 1). The tion that is represented