Feeding ecology of three inshore fish species at Marion Island (Southern Ocean) w.o. Blankley School of Environmental Studies, University of Cape Town, Rondebosch The diets, morphological features and habitats of 258 Three species of fish occur in the shallow inshore waters specimens of the three inshore fish species Notothenia of Marion Island. Notothenia macrocephala Gunther 1860 coriiceps, N. macrocephala and Harpagifer georgianus from and a subspecies of N. coriiceps Richardson 1844 are Marion Island are described and compared. Correspondence analysis of the three diets shows the existence of three Antarctic cods of the family Nototheniidae. The third clearly defined feeding niches despite the occurrence of species, Harpagijer georgianus subsp. georgian us Nybelin some common prey species. Inter- and intraspecific 1947, which was previously described as Harpagijer similarities and differences in the diets of small and large bispinis subsp. marionensis, is a member of the plunder­ size classes of each species are also displayed by fish family Harpagiferidae. correspondence analysis. Size-limited predation by N. coriiceps of the limpet Nacel/a delesserti is described. While there are many studies on Antarctic fish Differences in the habitats occupied by the fish appear to be (Holloway 1969; Everson 1970; Meier 1971; Permitin & important in determining the species composition of their Tarverdieva 1972; Richardson 1975; Targett 1981), few diets. detailed reports on the feeding of sub-Antarctic fish exist S. Afr. J. Zool. 1982, 17: 164 - 170 except that of Hureau (1966) who examined the diet of . ) 0 Notothenia macrocephala and two other species of Die diete, morfologiese kenmerke en habitat van 258 1 0 eksemplare van die drie vis-spesies, Notothenia coriiceps, N. Nototheniidae at Kerguelen Island. De Villiers (1976) 2 macrocephala en Harpagifer georgianus, wat naby die kus described the major prey of the three species of fish at d e t van Marion-eiland voorkom, word beskryf en vergelyk. Marion Island without providing any quantitative data. a Vergelykende analise van die drie diete toon die bestaan van d The aim of the present study was to provide baseline ( drie duidelik gedefinieerde voedingsgebiede ten spyte van die r quantitative data on the diets and interrelationships of the e voorkoms van sommige algemene roofspesies. Inter- en h three species mentioned above. s intraspesifieke ooreenkomste en verskille in die diete van i l klein en groot klasmonsters van elke spesie is ook aangetoon b u deur vergelykende analise. Grootte-beperkte roof deur N. Materials and Methods P coriiceps op die klipmossel Nacel/a de/esserti word beskryf. e Specimens were obtained through numerous collections h Verskille in die habitat van die verskillende spesies blyk t between May 1979 and May 1980, made at various sites y belangrik te wees in die vasstelling van die b in Transvaal Cove, a relatively sheltered bay close to the spesiesamestelling van hulle dil:ite. d e S.-Afr. Tydskr. Dierk. 1982, 17: 164 - 170 research station on the north-east coast of Marion Island. t n Notothenia macrocephala was captured on hook and line a r at depths ranging from 20 cm in the intertidal zone to 20 m g e at the base of the offshore Macrocystis belt. Notothenia c n coriiceps was caught mainly at depths of 1 - 4 m by using e c i a hand-net whilst snorkelling, although a few specimens l r were obtained on hook and line in the shallow subtidal e d zone. Harpagijer georgianus was found under boulders and n u amongst rubble or algal turf in the intertidal and in shallow y a subtidal zones, and all specimens were caught by hand. w e Specimens were examined and dissected in the t a laboratory _ The standard length of each fish was recorded G t to the nearest millimetre and mass measured to the nearest e n 0,1 or 1,0 g. Stomachs were removed and the wet mass i b of contents recorded. Prey from each stomach were a S sorted to species level and then counted and weighed to y b w.o. Blankley the nearest 0,01 g. Lengths of selected prey species from d e Present address: S.A. Association for Marine Biological Research. each stomach were also recorded. Ingested seaweeds were c u Oceanographic Research InstilUte. P.O. Box 10712. Marine Parade 4056. classified as rhodophytes, chlorophytes or phaeophytes and d Republic of South Africa o r wet mass recorded. Intestinal contents were examined p e Received 15 December 1981; accepted 15 April 1982 although they were not used in the final analysis. All gut R S. Afr. J. Zool. 1982, 17(4) 165 contents were preserved in 1011,70 formalin. Stomach con­ forward directed eyes and sharply pointed jaw. The species tent data were pooled for each fish species and the con­ was abundant in the intertidal zone and many specimens tribution of each prey species compared by four methods: were found in residual pools of water under boulders at (i) as a percentage of the total wet mass of stomach con- low tide. A few individuals were encountered amongst algal tents (11,70 mass); turf at depths of 2 - 3 m. Like N. coriiceps, this species (ii) as the percentage of stomachs in which it appeared also spends much of its time lying motionless on the bot­ (11,70 occ.); tom. Up to three H. georgian us could be found in close (iii) the number of prey specimens (N); proximity to one another although most specimens were (iv) the ranking index method recommended by Hobson found singly. (1974). Ranking index (R.I.) values for each prey species were calculated from the formula: R.1. = 11,70 mass x 1I,7°1~c. and then expressed as a percentage of the sum of all R.1. values for each species of fish. Diets of the three species were compared by cor­ respondence analysis, a relatively recent technique developed by French statisticians. An early review of the Figure lA Notothenia coriiceps (length = 260 mm) technique is given by Benzecri et al. (1973). Greenacre (1978) provides a more recent description and Underhill (1981) describes the computer program used. Cor­ respondence analysis was used for inter- and intraspecific comparisons of the species composition of diets of small and large fish (which were defined as those fish less than, or more than, the median standard length for each species, respectively). The lengths of all shells of the limpet Nacella delesserti retrieved from the stomachs or intestines of . Notothenia coriiceps were measured with vernier calipers ) 0 to the nearest 0,1 mm. 1 0 18 = 2 Figure Notothenia macrocephala (length 166 mm) Results d e t Size, morphology and habitats a d ( Notothenia coriiceps was the largest of the three fish and r e the 31 specimens studied had a mean standard length and h s i standard deviation of 304 ± 60 mm. Maximum length and l b mass were 444 and 1800 g. Colour was always dark blue­ u P black with a yellow to white ventral surface. Notable e h features were the squat head, wide mouth and fleshy pelvic t y fins (Figure lA). Solitary individuals were always seen ly­ b ing on the bottom, between boulders or on rocky ledges, d e t usually in association with the abundant limpet Nocella = n Figure Ie Harpagijer georgian us (length 57 mm) a delesserti. Swimming was poorly developed and specimens r g usually attempted to escape capture by moving into gaps e c between boulders rather than swimming away. n Diets e The 129 specimens of Notothenia macrocephala (Figure c i l IB) had a mean standard length and standard deviation Notothenia macrocephala stomachs contained the widest r e of 166 ± 60 mm. Maximum length and mass recorded were range of prey types of which Dynamenella huttoni, d n 294 mm and 546 g. Colouration patterns were varied, and Platynereis australis, and rhodophyte algae had the highest u younger specimens were usually dark red with white to percentage R.I. values of 27,811,70,20,311,70 and 17,911,70 respec­ y a orange bellies, whilst larger ones were dark brown dorsal­ tively (Table 1). Algae, isopods, polychaetes and am­ w e t ly, with orange and white ventral markings. Small phipods formed the bulk of the diet. Notothenia coriiceps a specimens of N. macrocephala were regularly sighted under stomachs contained mostly rhodophytes (53,111,70) and the G t water at depths of 1 - 4 m, either singly or in loose aggrega­ limpet Nacella delesserti (32,211,70). Chlorophyte algae, e n i tions of up to 30 fish. Large specimens inhabit deeper water isopods and other fish were also eaten (Table 1). Har­ b a and were mostly caught in water 10 - 30 m deep. pagifer georgian us, the smallest of the three fish, was the S y The 98 Harpagifer georgian us (Figure 1C) had a mean most carnivorous species and algae had a low percentage b standard length and standard deviation of 48 ± 8 mm. R.1. value ofO,711,70. The three amphipods Shakeltonia sp., d e Hyale hirtipalma and Jassa falcata formed 76,811,70 of the c Maximum length and mass were 69 mm and 6,5 g. Col­ u ours were cryptic and specimens were usually mottled diet but isopods and polychaetes were also preyed on. d o r brown and red with pale ochre undersurfaces. Notable Table 1 shows that only a few prey species were con­ p e features were the two pairs of defensive opercular spines, sumed in significant numbers by all three species of fish R 166 S.-Afr.