ARTICLE IN PRESS

Deep-Sea Research I 51 (2004) 17–31

Antarctic jaws: cephalopodprey of in Kerguelen waters

Yves Cherela,*, Guy Duhamelb a Centre d’Etudes Biologiques de Chize,! UPR 1934 du Centre National de la Recherche Scientifique, BP 14, F-79360 Villiers-en-Bois, France b Departement! des milieux et peuplements marins, USM403 du Museum! National d’Histoire Naturelle, 43 rue Cuvier, 75231 Paris Cedex 05, France

Received1 April 2003; receivedin revisedform 1 September 2003; accepted11 September 2003

Abstract

Only five species of sharks have been recorded in the Southern Ocean, where their biology is essentially unknown. We investigatedthe feedinghabits of the three commonest species from content analysis of specimens taken as bycatches of the fishery targeting the Patagonian toothfish (Dissostichus eleginoides) in upper slope waters of the Kerguelen Archipelago. The three species prey upon a diversity of fishes and . They segregate by feeding on different species of of different sizes. The small lanternsharks (Etmopterus cf. granulosus; 0.3 m on average) feed on small-sized Mastigoteuthis psychrophila, while the large porbeagles (Lamna nasus; 1.9 m) feedon small-sized histioteuthids (Histioteuthis atlantica and H. eltaninae) and on medium-sized juvenile ommastrephids of the genus Todarodes. Finally, the huge sleeper sharks (Somniosus cf. microcephalus; 3.9 m) prey upon large-sizedcephalopods (Kondakovia longimana and ) andgiant squids( Mesonychoteuthis hamiltoni and Architeuthis dux). Thus sleeper is a fish with -like feeding habits and, hence, the second top predator known to science to rely significantly on giant squids. Prey species and biology indicate that porbeagles are pelagic predators in the entire water column, while sleeper sharks are mainly benthic top predators and scavengers. The present study also underlines the diversity and biomass of the poorly known fauna, including giant squids, occurring in outer shelf and upper slope waters surrounding subantarctic islands. r 2003 Elsevier Ltd. All rights reserved.

Keywords: Giant squids; Kondakovia longimana; Lanternsharks; Mesonychoteuthis hamiltoni; Porbeagles; Sleeper sharks; Southern Ocean; Taningia danae

1. Introduction spurdog (Squalus acanthias) anda dogfish ( Cen- troscymnus sp.) plus three relatively common Sharks are not frequent in the Southern Ocean, species that include the porbeagle (Lamna nasus) where only five species have been recorded so far andtwo species of still uncertain taxonomic status, (Compagno, 1990; authors’ unpublisheddata). one sleeper shark (Somniosus cf. microcephalus) They include two rare species, the white-spotted andone lanternshark ( Etmopterus cf. granulosus). In northern waters, the porbeagle is known as a *Corresponding author. Tel.: +33-5-49-09-78-35; fax: +33- large shark, with a maximum total length of more 5-49-09-65-26. than 3.0 m. It is a coastal andoceanic species with E-mail address: [email protected] (Y. Cherel). an amphitemperate distribution in epipelagic

0967-0637/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr.2003.09.009 ARTICLE IN PRESS

18 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 waters. Porbeagles are active andstrong swimmers 1996). Since the mid-1980s, a commercial fishery and are considered as voracious feeders on small targeting the Patagonian toothfish (Dissostichus pelagic schooling fishes andsquids( Compagno, eleginoides) has developed from upper slope to 1984; Joyce et al., 2002). Sleeper sharks are deeper waters surrounding the archipelago (Du- amongst the largest fish, attaining lengths of up hamel, 1992; Cherel et al., 1996). Sharks are taken to 7.3 m andbodymasses of up to 775 kg. In the as uncommon but regular bycatches of the fishery Northern Hemisphere, they are sluggish epibenthic (Duhamel et al., 1997; authors’ unpublisheddata), sharks of coldwaters, which regularly occur thus giving an unique opportunity to collect inshore in the high Arctic but seem confinedto stomach contents to investigate their diet and deeper waters at lower latitudes. They have been feeding habits. described as voracious and indiscrimate predators andscavengers that feedmainly on fishes and marine mammals (Compagno, 1984; Fisk et al., 2. Material and methods 2002). Finally, lanternsharks belong to a large, complex andpoorly known genus. They are small The three commonest species of sharks (lantern- epibenthic sharks (o1 m) living on the outer sharks, porbeagles andsleeper sharks) were caught shelves andupper slopes, where they prey upon during fishery operations aboard commercial squids, small mesopelagic fishes and shrimps trawlers andlongliners from 1997 to 2001 in (Compagno, 1990). Kerguelen waters (mean position 49300S– In the Southern Ocean almost nothing is known 69000E) (Table 1). Fishes were measured, sexed about the biology of sharks. Specimens have been and then dissected to investigate their feeding recorded in Kerguelen waters, which is, to our habits. Fishery observers notedthe main items, knowledge, the only place where all five species are especially the fish species foundin the shark encountered( Compagno, 1990; Duhamel, 1997, stomachs and they discarded them thereafter. unpublisheddata).The Kerguelen Islandsare a They collectedcephalopodbeaks andwhole buccal large archipelago that is locatedin the vicinity of masses andkept them in 70% ethanol andat the Antarctic Polar Front in the Indian sector of À20C, respectively, until analysis in the labora- the Southern Ocean (Park andGamberoni, 1997 ). tory in France. It is one of the most productive areas in the Cephalopodbeaks (both lower andupper Southern Ocean with high chlorophyll concentra- beaks) were identified from their morphological tions occurring consistently during the summer features by comparison with material heldin our months in the north andeast of its extensive shelf own collection andby reference to the available (the so-called‘‘islandmass effect’’) ( Blain et al., literature (Clarke, 1986a). Lower rostral lengths 2001; Moore andAbbott, 2002 ). Consequently, (LRL) of squids and lower hood lengths of secondary production sustains high populations of octopuses were measuredwith a vernier caliper, top predators, including seabirds, marine mam- andthe size of cephalopodswas estimatedfrom mals andfishes ( Duhamel, 1987; Guinet et al., allometric equations between LRL anddorsal

Table 1 Biological characteristics of sharks caught as bycatches of the Patagonian toothfish fisheries in Kerguelen waters from 1997 to 2001

Species Fishing methods Fishing depths (m) n Total length (m) Sex (n)

Males Females Undetermined

Lanternsharks Bottom trawl 687 12 0.3070.02 (0.26–0.33) 9 3 0 Porbeagles Bottom trawl (n ¼ 18) 459790 (200–582) 20 1.9570.24 (1.24–2.25) 9 8 3 Bottom longline (n ¼ 2) 1320–1410 Sleeper sharks Bottom trawl 6027117 (350–763) 36 3.9171.07 (1.50–5.00) 11 20 5

Values are means7SD with ranges in parentheses. ARTICLE IN PRESS

Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 19 length (ML) andwet mass ( M)(Clarke, 46°00 S 2000 m 1962, 1980, 1986a; Adams and Klages, 1987; 1000 m Rodhouse and Yeatman, 1990; Rodhouse et al., 500 m 1990; Lu andWilliams, 1994 ; Jackson, 1995; Roeleveld, 2000; Piatkowski et al., 2001; authors’ 48°00 S 200 m unpublisheddata). For the few species where no relationships were available, ML and M were estimatedfrom equations for closely related Latitude species or for species with a similar morphology. 50°00 S Dietary data were obtained by three calculation techniques, namely, the frequency of occurrence andpercentages by number andby reconstituted mass of each prey type, as usually done in food 52°00 S analysis for seabirds (see, for example, Cherel et al., 64°00 E 66°00 E 68°00 E 70°00 E 72°00 E 74°00 E 2000). The minimum number of individuals for a Longitude given cephalopodspecies was determinedineach 46°00 S shark stomach content as the maximum number of 2000 m 1000 m lower or upper beaks of that species. These 500 m minimum numbers were subsequently added to calculate the minimum number of individuals for a 48°00 S 200 m given cephalopodspecies in all the samples. Data were statistically analysedwith SYSTAT 9

for WINDOWS (Wilkinson, 1999). Values are Latitude means7SD. 50°00 S

3. Results

52°00 S 3.1. General comments 64°00 E 66°00 E 68°00 E 70°00 E 72°00 E 74°00 E Longitude

Overall 68 sharks containedcephalopodre- Fig. 1. Geographical positions of bycatches of porbeagles mains in their stomachs, including 12 of 32 (upper panel) andsleeper sharks (lower panel) in Kerguelen lanternsharks, 20 of 26 porbeagles and36 of 36 waters. The star in the upper panel shows the location of the sleeper sharks (Table 1). There were highly trawl that caught lanternsharks. significant differences in size among the three species in the increasing order lanternsharks, sleeper sharks andmost of the porbeagles were porbeagles andsleeper sharks (ANOVA, F2;64 ¼ caught in bottom trawls. Two porbeagles were 104:84; po0:0001; all post-hoc Tukey HSD multi- hookedon bottom longlines, but couldhave been ple comparison tests, po0:001). Sex ratio was caught pelagically during setting or hauling close to 1 for porbeagles, but more males andmore operations. All porbeagles but one were taken in females were caught for lanternsharks andsleeper the northern fishing ground, while sleeper sharks sharks, respectively (Table 1). were caught in the northern (n ¼ 15), eastern All the lanternsharks were collectedin the same (n ¼ 17) andwestern ( n ¼ 4) areas (Fig. 1). Sleeper bottom trawl, which contained105 specimens of sharks were collectedin trawls fishedat deeper that species (Fig. 1). The trawl was conducted in depths than those in which porbeagles were found, the eastern fishing groundat about 700-m depth. either when all trawling depths were considered Unlike lanternsharks, most of the porbeagles and (two-sample t-test, t ¼ 4:54; po0:001) (Table 1), sleeper sharks occurredsingly in nets. All the or when the trawling depths in the northern fishing ARTICLE IN PRESS

20 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 groundonly were considered(575 770 m, n ¼ 15; identified from 20 of the samples) and 16 contain- and456 791 m, n ¼ 17; respectively; two-sample t- ing fish items. Fishes included a few specimens of test, t ¼ 4:09; po0:001). shelf species in the families Channichthyidae (Channichthys rhinoceratus and Champsocephalus 3.2. Lanternsharks gunnari) andCongiopodidae( Zanclorhynchus spi- nifer) together with a greater number of mesope- Thirty-two lanternsharks were dissected. Only lagic fishes. Most of the latter were lanternfishes 14 stomachs containedprey items, including 12 (Myctophidae) with the giant of the family samples with cephalopodremains. Other items Gymnoscopelus bolini (standard length up to were skin andeye lenses of fishes in one stomach, 28 cm; Hulley, 1990) ranking first andthe large andone large lanternfish (Family Myctophidae; G. piabilis second. Other mesopelagic fishes were Gymnoscopelus microlampas) in another sample. specimens of dragonfish (Stomias sp., Stomiidae) Four species of squids (30 beaks) were identified andsnake mackerels ( Paradiplospinus gracilis, from the stomach contents of lanternsharks. By Gempylidae). far, the main species was Mastigoteuthis psychro- Fifteen species of squids (689 beaks) were phila, which accountedfor 73% of the total identified from the stomach contents of porbeagles number of beaks and85% by mass of the (Table 4). Two cephalopodspecies were important cephalopoddiet( Table 2). The recently described in the diet, the histioteuthid, Histioteuthis atlanti- brachioteuthidsquid, Slosarczykovia circuman- ca, which dominated by number (46% by number tarctica (Lipinski, 2001) rankedsecond,while and9% by mass), andthe ommastrephid, Todar- two other species were foundonly once. Overall, odes sp., which dominated by reconstituted mass the estimatedsize of cephalopodseaten by (33% by number and56% by mass). The latter lanternsharks is small, the largest prey being an species is identical or closely related to T. individual of M. psychrophila with a 121-mm ML angolensis (Cherel andWeimerskirch, 1995 ), but andweighing 67 g ( Table 3). its taxonomic status merits further investigation. Other significant prey were another histioteuthid 3.3. Porbeagles H. eltaninae, the chiroteuthid Chiroteuthis veranyi, the onychoteuthid Moroteuthis knipovitchi, and Twenty-six porbeagles were dissected; 25 sto- another ommastrephid Martialia hyadesi. Finally, machs containedprey items, including 23 samples owing to its large size (Table 5), the onychoteuthid with cephalopodremains (cephalopodprey were Kondakovia longimana accountedfor a significant

Table 2 Lanternsharks: frequency of occurrence, numbers of cephalopodbeaks (both lower andupper beaks), andreconstitutedmass of the cephalopod diet at Kerguelen (total for all 12 samples pooled)

Species Occurrence Number Minimum number Reconstitutedmass of individuals (n)(%)(n) (%) (g) (%)

Brachioteuthidae Slosarczykovia circumantarctica 4 33.3 5 16.7 4 18.49 3.2 Histioteuthidae Histioteuthis atlantica 1 8.3 1 3.3 1 20.29 3.5 Mastigoteuthidae Mastigoteuthis psychrophila 8 66.7 22 73.3 12 485.31 84.8 Batoteuthidae Batoteuthis skolops 1 8.3 2 6.7 1 48.48 8.5

Total 30 100.0 18 572.57 100.0 ARTICLE IN PRESS

Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 21

Table 3 Lanternsharks: measuredlower rostral length (LRL), andestimatedmantle length (ML) andbodymass ( M) of cephalopodprey identified from stomach contents at Kerguelen

Species n LRL (mm) ML (mm) M (g)

Slosarczykovia circumantarctica 3 2.070.3 (1.8–2.3) 5775 (53–63) 4.670.9 (4.0–5.6) Histioteuthis atlantica 1 1.9 27 20 Mastigoteuthis psychrophila 11 3.370.7 (2.0–4.3) 93721 (56–123) 40722 (9–80) Batoteuthis skolops 1 3.6 aa

Values are means7SD with ranges in parentheses. a No allometric equations available.

Table 4 Porbeagles: frequency of occurrence, numbers of cephalopodbeaks (both lower andupper beaks), andreconstitutedmass of the cephalopod diet at Kerguelen (total for all 20 samples pooled)

Species Occurrence Number Minimum number Reconstitutedmass of individuals (n)(%)(n) (%) (g) (%)

Ommastrephidae Martialia hyadesi 5 25.0 17 2.5 11 2890 4.9 Todarodes cf. angolensis 11 55.0 224 32.5 120 33321 56.5 Onychoteuthidae Moroteuthis ingens 4 20.0 8 1.2 5 1050 1.8 Moroteuthis knipovitchi 4 20.0 19 2.8 13 1856 3.1 Kondakovia longimana 4 20.0 12 1.7 7 11504 19.5 Brachioteuthidae Brachioteuthis linkovskyi 2 10.0 4 0.6 3 41 o0.1 Gonatidae Gonatus antarcticus 2 10.0 7 1.0 4 312 0.5 Histioteuthidae Histioteuthis atlantica 12 60.0 317 46.0 167 5484 9.3 Histioteuthis eltaninae 5 25.0 41 6.0 24 839 1.4 Neoteuthidae Alluroteuthis antarcticus 1 5.0 2 0.3 1 244 0.4 Nototeuthis dimegacotyle 1 5.0 1 0.1 1 138 0.2 Mastigoteuthidae Mastigoteuthis psychrophila 1 5.0 1 0.1 1 40 o0.1 ?Mastigoteuthis A (Clarke) 1 5.0 2 0.3 1 120 0.2 Chiroteuthidae Chiroteuthis veranyi 6 30.0 33 4.8 17 1174 2.0 Cranchiidae Galiteuthis glacialis 1 5.0 1 0.1 1 6 o0.1

Total 689 100.0 376 59019 100.0

percentage of the diet by reconstituted mass (2% estimatedML and M of 175–196 mm and154– by number and19% by mass). 208 g, respectively. The mode corresponds to Porbeagles prey mainly upon one size class of juvenile squids, as indicated by the undarkened Todarodes cf. angolensis, with a mode at 4.5– wings of the lower beaks, thus contrasting with the 5.0 mm LRL (Fig. 2), which corresponds to wholly darkened beaks of two adult specimens ARTICLE IN PRESS

22 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31

Table 5 Porbeagles: measuredlower rostral length (LRL), andestimatedmantle length (ML) andbodymass ( M) of cephalopodprey identified from stomach contents at Kerguelen

Species n LRL (mm) ML (mm) M (g)

Martialia hyadesi 9 4.771.3 (3.5–7.3) 241737 (205–316) 2687156 (137–600) Todarodes cf. angolensis 105 5.171.0 (4.2–11.2) 198740 (163–452) 2397226 (128–2033) Moroteuthis ingens 3 4.470.6 (3.8–4.9) 187720 (165–203) 210763 (141–265) Moroteuthis knipovitchi 10 4.770.6 (4.0–6.1) 186736 (141–277) 156792 (77–406) Kondakovia longimana 5 11.473.1 (6.0–13.6) 4047115 (200–484) 18057938 (197–2642) Brachioteuthis linkovskyi 2 4.0–4.4 97–106 12–14 Gonatus antarcticus 2 3.2–6.6 93–240 24–281 Histioteuthis atlantica 155 2.170.8 (1.0–4.9) 33717 (9–95) 33728 (5–192) Histioteuthis eltaninae 20 2.470.3 (1.8–2.9) 3978 (25–51) 37712 (18–59) Alluroteuthis antarcticus 1 4.3 145 244 Nototeuthis dimegacotyle 1 3.5 aa ?Mastigoteuthis A (Clarke) 1 5.0 aa Chiroteuthis veranyi 16 5.270.6 (4.0–6.1) 138715 (110–161) 69721 (34–105)

Values are means7SD with ranges in parentheses. a No allometric equations available.

indicating that sharks fed on different size classes. 50 Porbeagles Todarodes sp. Overall, histioteuthids (both H. atlantica and H. 40 n = 105 eltaninae) were small with an estimatedML and M of 30–40 mm and30–40 g, respectively ( Table 5). 30 The largest cephalopodprey was an adultof K. longimana with a 48 cm ML andweighing 2.6 kg. 20 Fr (%) equency 10 3.4. Sleeper sharks

0 Thirty-six sleeper sharks were dissected; all the 024681012 stomachs containedcephalopodflesh and/or 20 accumulatedcephalopodbeaks of large size. Histioteuthis Stomach contents regularly containedwhole speci- 15 atlantica mens of Patagonian toothfish, and discarded fish n = 105 heads of that species (fishery waste), which were 10 most likely eaten directly in trawls and scavenged at the bottom, respectively. Other prey items were demersal fishes (mainly the skates Bathyraja irrasa Fr (%) equency 5 and B. eatonii, andthe nototheniids Notothenia rossii and Lepidonotothen squamifrons), benthic 0 (e.g. sea stars, ophiurids) and car- 0 123456 rion. The stomach of nine sharks containedbig Lower rostral length (mm) chunks of flesh (up to 30 kg), which, in 2 cases, Fig. 2. Porbeagles: frequency distribution of LRL of Todarodes were identified as belonging to fur seals (probably cf. angolensis and H. atlantica eaten in Kerguelen waters. the Antarctic fur seal Arctocephalus gazella). Nineteen species of cephalopods (553 beaks) (9.7 and11.2 mm LRL). Most histioteuthidswere were identified from the stomach contents of juveniles (undarkened wings) and subadults (dar- sleeper sharks; they include 18 species of squids kening wings), with no well defined modes (Fig. 2), andone species of an unknown large cirrate ARTICLE IN PRESS

Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 23

Table 6 Sleeper sharks: frequency of occurrence, numbers of cephalopodbeaks (both lower andupper beaks), andreconstitutedmass of the cephalopod diet at Kerguelen (total for all 36 samples pooled)

Species Occurrence Number Minimum number Reconstitutedmass of individuals (n)(%)(n) (%) (g) (%)

Architeuthidae Architeuthis dux 7 19.4 14 2.5 8 324644 14.9 Ommastrephidae Todarodes cf. angolensis 2 5.6 4 0.7 2 2801 0.1 Onychoteuthidae Moroteuthis ingens 1 2.8 2 0.4 1 1021 o0.1 Moroteuthis robsoni 3 8.3 5 0.9 3 7213 0.3 Moroteuthis knipovitchi 1 2.8 1 0.2 1 804 o0.1 Kondakovia longimana 24 66.7 274 49.5 154 458519 21.0 Brachioteuthidae Slosarczykovia circumantarctica 1 2.8 11 2.0 11 128 o0.1 Gonatidae Gonatus antarcticus 2 5.6 8 1.4 7 1251 o0.1 Octopoteuthidae Taningia danae 16 44.4 69 12.5 40 235678 10.8 Histioteuthidae Histioteuthis atlantica 1 2.8 6 1.1 4 1583 o0.1 Neoteuthidae Alluroteuthis antarcticus 1 2.8 3 0.5 2 467 o0.1 Cycloteuthidae Cycloteuthis akimushkini 5 13.9 11 2.0 6 6800 0.3 Mastigoteuthidae Mastigoteuthis psychrophila 1 2.8 24 4.3 24 1249 o0.1 Mastigoteuthis B 1 2.8 1 0.2 1 234 o0.1 Cranchiidae Taonius sp. B (Voss) 1 2.8 6 1.1 6 1147 o0.1 Galiteuthis glacialis 2 5.6 22 4.0 17 903 o0.1 Mesonychoteuthis hamiltoni 22 61.1 89 16.1 49 1133621 52.0 sp. C 1 2.8 1 0.2 1 234 o0.1 Cirrata sp. A 1 2.8 2 0.4 1 2237 0.1

Total 553 100.0 338 2180535 100.0

octopod( Table 6). Four species dominated the Sleeper sharks fedon adult K. longimana,as cephalopoddiet,accounting together for 81% of indicated by the large size and level of darkening the total number of beaks and99% of the of the beaks. Fig. 3 shows a mode at 14–15 mm reconstitutedmass. They were K. longimana, LRL, which corresponds to estimated ML and M which was the main prey by number (50% by of 50–54 cm and2.9–3.6 kg, respectively. On the number and21% by mass), the cranchiid Meso- other hand, sharks prey upon juveniles, subadults nychoteuthis hamiltoni, which was the main prey andadult M. hamiltoni, T. danae and A. dux,as by mass (16% by number and52% by mass), plus indicated by the large range in size in LRL and in the octopoteuthid Taningia danae (12% by num- the level of darkening of the beaks (Fig. 3, ber and11% by mass) andthe architeuthid Table 7). The mean ML and M of T. danae were Architeuthis dux (3% by number and15% by 65 cm and6.1 kg, respectively, while both M. mass). hamiltoni and A. dux were much larger on average ARTICLE IN PRESS

24 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31

30 specimens of the different species of cephalopods Sleeper sharks (Fig. 4). The two modes in ML of cephalopods eaten by porbeagles correspondto histioteuthids 20 Kondakovia longimana (o50 mm ML) and Todarodes cf. angolensis (175– n = 123 200 mm ML), while the markedmode(400– 600 mm) observedin the size of cephalopods 10 caught by sleeper sharks corresponds to K. long-

Frequency (%) imana, and, to a lesser extent, T. danae (Fig. 4). Overall, sharks fed on cephalopods of different 7 ¼ 7 ¼ 0 sizes (ML=98 30 mm, n 16; 110 93 mm, n 15 330; and681 7425 mm, n ¼ 219; for lanternsharks, Mesonychoteuthis hamiltoni porbeagles andsleeper sharks, respectively; Krus- n = 42 kal-Wallis test statistic=379, po0:0001). How- 10 ever, sleeper sharks preyedupon larger items than the two other fishes, but no significant difference was foundbetween the size of cephalopodseaten by lanternsharks andporbeagles (Mann-Whitney 5

Frequency (%) U test statistic=2680, p ¼ 0:918). When relatedto predator size, again, the huge sleeper sharks fed on large to giant prey, but the large porbeagles, on 0 30 average, caught relatively small prey, andthe small lanternsharks preyedupon small cephalopodsonly Taningia danae n = 28 (Fig. 5). Since sharks fedon differentspecies of cepha- 20 lopods, a direct comparison of the size of a given species of squideaten by different sharks was possible for K. longimana only (Tables 5 and7 ). 10

Frequency (%) There was a tendency for porbeagles to prey upon K. longimana of smaller size than those eaten by sleeper sharks (Mann-Whitney U test statis- 0 515253545 tic=159.5, p ¼ 0:069). Lower rostral length (mm) Fig. 3. Sleeper sharks: frequency distribution of LRL of K. 4. Discussion longimana, M. hamiltoni and T. danae eaten in Kerguelen waters. 4.1. General comments and resource partitioning

(136 cm and24 kg, and110 cm and39 kg, respec- The three species of sharks segregate both by tively). The largest cephalopodprey was an adult their foodandfeeding ecology in Kerguelen of A. dux with a 220 cm ML andweighing 96 kg. waters. All occur in upper slope waters, but porbeagles were caught on average in shallower 3.5. Relationship between predator size and waters than sleeper sharks. The three sharks prey prey size upon different species of squids, which can be relatedto the size of the predators. The small To compare the size of cephalopods eaten by the lanternsharks feedon small-sized M. psychrophila, three species of sharks, we estimatedML from the large porbeagles on small-sizedhistioteuthids LRL and, for a given shark species, we pooled and (H. atlantica and H. eltaninae) andmedium-sized plottedtogether the estimatedML from all the juveniles Todarodes cf. angolensis, while the ARTICLE IN PRESS

Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 25

Table 7 Sleeper sharks: measuredlower rostral length (LRL), andestimatedmantle length (ML) andbodymass ( M) of cephalopodprey identified from stomach contents at Kerguelen

Species n LRL (mm) ML (cm) M (kg)

Architeuthis dux 6 13.973.1 (10.4–18.1) 110769 (45–220) 39.0734.9 (7.6–95.7) Todarodes cf. angolensis 2 9.6–10.0 39–40 1.3–1.5 Moroteuthis robsoni 3 9.470.2 (9.3–9.7) 6273 (61–65) 2.470.2 (2.3–2.7) Kondakovia longimana 123 13.772.0 (8.4–18.9) 4977 (29–68) 2.971.3 (0.6–7.6) Gonatus antarcticus 2 5.6–5.8 20–21 0.2 Taningia danae 28 16.072.7 (9.4–20.9) 65720 (15–101) 6.173.2 (0.9–13.6) Histioteuthis atlantica 2 6.3–7.1 13–14 0.3–0.4 Alluroteuthis antarcticus 1 4.2 14 0.2 Cycloteuthis akimushkini 5 13.872.7 (10.3–16.1) 4378 (32–50) 1.170.4 (0.6–1.5) Galiteuthis glacialis 5 4.071.5 (1.6–5.1) 1776 (8–22) o0.1 Mesonychoteuthis hamiltoni 42 22.377.2 (10.1–38.8) 136744 (61–237) 24.4722.1 (2.1–91.2) Cirrata sp. A 1 13.8ab b

Values are means7SD with ranges in parentheses. a Lower hoodlength (mm). b No allometric equations available.

120 Lanternsharks 30 Porbeagles 100 Porbeagles n = 330 lower beaks Sleeper sharks 80

20 60

40

Frequency (%) 10 20

0 0 Cephalopod mantle length (cm) 0 100 200 300 400 500 600 0 100 200 300 400 500 Shark total length (cm) Fig. 5. Relationship between prey size (dorsal ML of squids) 30 Sleeper sharks andpredatorsize (total length of sharks) in Kerguelen waters. n = 219 lower beaks

20 gigantic sleeper sharks feedon large-sizedand giant squids (see below). Little is known about the foodof lanternsharks elsewhere. Specimens from 10 Frequency (%) Kerguelen prey upon squids, and, to a lesser extent upon mesopelagic lanternfishes (myctophids). This is in broadagreement with previous results 0 0 500 1000 1500 2000 2500 obtainedon various species of Etmopterus (Com- Mantle length (mm) pagno, 1984; Ebert et al., 1992), including E. Fig. 4. Frequency distribution of dorsal MLs of squids eaten lucifer and E. granulosus, which are probably by porbeagles andsleeper sharks in Kerguelen waters. Note the closely relatedto the species occurring in Kergue- different scales of the x-axis on upper andlower panels. len waters (Compagno, 1990; Duhamel, 1997). ARTICLE IN PRESS

26 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31

However, the small sample size of lanternsharks presence of cephalopodremains in all the samples, together with the fact that all specimens were thus indicating that squids are of primary im- caught in the same trawl preclude any definite portance in the diet of sleeper sharks in Kerguelen. conclusion about the feeding ecology of the species Elsewhere, cephalopods are minor prey, but the at Kerguelen. giant octopod dofleini accountedfor Previous studies on the feeding ecology of the a small but significant percentage by mass (5%) of porbeagle were largely anecdotal, but a recent the diet of sleeper sharks in the north Pacific article details its food in the northwest Atlantic Ocean (Yang andPage, 1999 ). Ocean. Fishes dominated by far in its diet, fresh In summary, sleeper sharks are benthic top remains of cephalopods being much less important predators and scavengers, while porbeagles are (Joyce et al., 2002). In contrast, stomach content pelagic predators in upper slope waters of the analysis made onboard by fishery observers has Kerguelen archipelago. Moreover, sleeper sharks shown that both fish andcephalopodswere the are sluggish fish at andnear the bottom and main fooditems of porbeagles in Kerguelen porbeagles active swimmers throughout the water waters. Interestingly, ommastrephids are the main column (Compagno, 1984). Prey species and squidprey both in the Atlantic (where the only biology also suggest that, unlike a recent estimate identified cephalopod was Illex argentinus, Joyce in a review indicating an identical trophic level et al., 2002) andat Kerguelen ( Todarodes cf. (4.1) for Somniosus microcephalus, S. pacificus and angolensis). At the latter locality, the diet was porbeagles (Cortes, 1999), the trophic position of dominated by species that are known to be sleeper sharks is higher than that of porbeagles in primarily pelagic, i.e. myctophidfishes of the Kerguelen waters. genus Gymnoscopelus (Hulley 1990), Todarodes squids (Dunning andWormuth, 1998 ) and Histio- 4.2. The sleeper shark: a fish with a sperm teuthis squids (Voss et al., 1998). Other prey whale-like diet include a few specimens of the demersal shelf fishes C. rhinoceratus and Z. spinifer andthe semi- The main fooditems of sleeper sharks were two pelagic icefish C. gunnari (Duhamel andHureau, large species of squids K. longimana (ML up to 1985; Heemstra andDuhamel, 1990 ). Hence, the 0.9 m) (Nesis, 1987) and T. danae (ML up to 1.7 m) porbeagle at Kerguelen is mainly a pelagic (Nesis, 1987), together with the largest inverte- predator, thus contrasting with specimens from brates living on earth, the endemic Antarctic the northwest Atlantic where both pelagic and cranchiid M. hamiltoni (ML up to 2.3 m) (Klumov demersal fish are important components of their andYukhov, 1975 ; Nesis, 1987) andthe giant diet (Joyce et al., 2002). architeuthid A. dux (ML up to 5 m) (Nesis, 1987). The sleeper shark from Kerguelen belongs to the K. longimana is an endemic from Antarctic subgenus Somniosus andis therefore closely andsubantarctic waters ( Nesis, 1987; Cherel and relatedto the two giant species S. microcephalus Weimerskirch, 1999) that is rarely caught by from the north Atlantic and S. pacificus from the research trawls. It was foundin stomach contents north Pacific (Compagno, 1990). In the Northern of many top predators in the Antarctic (Cherel and Hemisphere, these sharks are voracious, indiscri- Weimerskirch, 1999; Santos et al., 2001a), includ- minate predators and scavengers that feed mainly ing the largest flying birdthe wandering albatross on the bottom on a wide variety of pelagic and (Diomedea exulans)(Cherel andKlages, 1998 ) and demersal fishes, as well as on marine mammals and the deepest diving odontocet, the sperm whale various invertebrates (Compagno, 1984, 1990; (Physeter catodon)(Clarke, 1980). Other large and Yang andPage 1999 ; Fisk et al., 2002). Accord- giant squids were not previously reported to be ingly, stomach contents of specimens from Ker- important in the diet of marine predators, except guelen containedlarge benthic fishes (skates, the sperm whale. M. hamiltoni is a leading member Patagonian toothfish), fishery offal andcarrion, of the Antarctic teuthofauna by biomass (Nesis, including fur seals. A consistent feature was the 1987) andis typically present in the dietof sperm ARTICLE IN PRESS

Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 27 whales in the Antarctic (Klumov andYukhov, 4.3. Predator indication of key cephalopod species 1975; Clarke, 1980; Yukhov, 1982). T. danae and on the Kerguelen Plateau A. dux are cosmopolitan cephalopodspecies (Nesis, 1987; Roper andVecchione, 1993 ; Forch,. Almost nothing is known about the cephalopod 1998) that were foundto be important fooditems fauna andits biomass on the Kerguelen Plateau in sperm whales in various parts of the world (Lu andMangold,1978 ; Guerra et al., 2000). (Clarke, 1986b). This makes the sleeper shark the Recent works on the diet of top predators breeding only fish with feeding habits like sperm whale and, at Kerguelen andat the closely relatedHeard hence, the secondtop predatorknown to rely to a Islandpoint out the importance of cephalopodsin significant extent on giant squids. To our knowl- the nutrition of some fishes, seabirds and marine edge, no other shark species target giant squids mammals. Predators’ diet underlines the key role (Smale, 1996), but A. dux was occasionally of at least 17 species of cephalopods, including 16 reportedin the dietof sharks from the North squids and one benthic octopod in the area Atlantic (Clarke andMerrett, 1972 ; Gonzalez (Table 8). Noticeable is the importance of ommas- et al., 2002), off South Africa (Smale andCliff, trephids and onychoteuthids. M. hyadesi is the 1998) andoff eastern ( Dunning et al., main prey by mass of king penguins during the 1993). austral winter (Moore et al., 1998), andjuveniles Comparison of the size of squids eaten by of Todarodes cf. angolensis are important items for sleeper sharks andsperm whales ( Clarke, 1980) porbeagles (present study), and for the black- shows that both species prey upon adult K. browedandgrey-headedalbatrosses( Cherel et al., longimana andupon differentsize-classes (from 2000, 2002). Among onychoteuthids, owing to juveniles to adults) of T. danae, M. hamiltoni and their large size, adults of K. longimana are key prey A. dux. The mean estimatedsize of cephalopods for sleeper sharks (present study), wandering eaten by sleeper sharks was 68 cm ML and8.2 kg, albatrosses (Cherel andWeimerskirch, unpub- which is slightly larger than the average mass lisheddata),southern elephant seals ( Slip, 1995) (7.2 kg) estimatedfor cephalopodseaten by sperm andsouthern bottlenose whales ( Slip et al., 1995). whales in the Antarctic (Clarke, 1980). The two Moroteuthis ingens, which is an abundant bycatch predators thus target the same species of cephalo- species on shelf andupper slope fishery grounds pods with about the same size and also in the same (Cherel andDuhamel, 2003 ), makes up a signifi- location, since Kerguelen was a whaling ground cant part of the diet of elephant seals (Green and for sperm whales in the last century (Yukhov, Burton, 1993), andto a lesser extent of king 1982). The mean size of squids eaten by sharks is, penguins in winter (Moore et al., 1998). moreover, likely to be underestimated because the Among other squidfamilies, cranchiids,mainly smallest individuals were probably secondary Galiteuthis glacialis, are important in the foodof items ingested with primary prey. Indeed, the sleeper sharks, albatrosses andmarine mammals main foodof sleeper sharks—large squidsand (Slip et al., 1995; Cherel et al., 2000, 2002; present the Patagonian toothfish—are known to feed study). An unexpected finding was the abundance on cephalopods (Rodhouse and Nigmatullin, of M. psychrophila in the diet of lanternsharks, 1996; Xavier et al., 2002). ML—the standard elephant seals andbottlenose whales ( Table 8). To length for cephalopods—does not take into our knowledge, mastigoteuthids were not pre- account the size of arms andtentacles. By doing viously foundto be key items in the nutrition of so, the largest squids had a much bigger size any squidpredator( Clarke, 1996), except prob- than that of sharks. For example, a specimen of ably the Cuvier’s beakedwhales in the North A. dux with a ML of 2.2 m reacheda total length Atlantic (Santos et al. 2001b). This emphasizes of more than 12 m. How sharks catch such giant how poorly known are trophic interactions of squids remains unknown because they may either many cephalopodtaxa in the Southern Ocean and prey on live or scavenge on dead elsewhere. The present study also underlines the individuals. diversity and biomass of cephalopods, including ARTICLE IN PRESS

28 Y. Cherel, G. Duhamel / Deep-Sea Research I 51 (2004) 17–31 76) and c ¼ Green and b 9 21 Hyperoodon planifrons Slip (1995) n ( ?In b c + 50 11+ + Slip et al. (1995) Mastigoteuthis . One of us (YC) examined81 c b 4 Mirounga leonina Green and Burton (1993) Mastigoteuthis psychrophila Diomedea chrysostoma Cherel et al. (2002) Green andBurton (1993) a , thus possibly also identified as andin Cherel et al. (2002) Slip (1