XIPHIAS GLADIUS) Lnin AZOREAN WWATERS ATERS

THE DIET OF SWORDFISSWORDFISHH (XIPHIAS GLADIUS) lNIN AZOREAN WWATERS ATERS

M.R. CLARKECLARKE,, D.CCLARKE,D.C.CLARKE, H.R.MARTINS AND H.M.SILH.M.SILVA V A

ARQUIP~LAG~CLCLARKE, ARKE, M.R.,M.R.. D.C. CLARKE,CLARKE. H.R. MARTINS & H.M .SILVA 1995. The diet of swordfish ARQUIPÉLAGO (Xiphias gladius) in Azorean waters. Arquipélago.Arquipe'lago. Life and Marine Sciences l3A:13A: 53-69. Angra do Heroísmo.Heroismo. ISSN 0870-6581.0870-658 1.

A total of 132 stomachs werewere collecte collectedd in 1992-93 from swordfish caught on drifting longlines near theth e Azores. Contents were sorted completely into components which comprised allali stages of digested fish and cephalopods from complete fish to fragments of muscle, bones.bones, otoliths and beaksbealcs of cephalopods. Fish and larger bones which could be identified aI! all belonged to Capros aper. Lepidopus caudalUs,caudatus, Pagellus bogabogaraveo,raveo, myctophids, and the fish used as bait, Trachurus picturatus.piclUratus. Flesh of the squids TodarTodarodesodes sagittatus, Pholidoteuthis boschmai and the squid used as bait Illexlllex sp. was also prescnt.presento 1249 otoliths and 120 cephalopod lower beaksbealcs were identified to genus or species. There was a mean of 3.1 species and 18 animalsanimaIs represented in each stomach. Fish remains occurredoccurred in 98.498.4%% of the s stomachstomachs and contributed 93.4% of animals in the diet. Cephalopod remains occurred in 37.4% and contributed 5.8% of animals.animais. Estimates of thelhe weightsweights of fish and cephalopods showshow that cephalopods provide 50.5% and fish species 49.5% of the stomach contents. The most important constituents of the diet by decreasing % estimated weight are Ommastrephes bartrami (24.4%), LepidopusLepidopus caudatus (17.4%),(17.4%). Pholidoteuthis boschmai (15.9%),(J 5.9%), Capros aper (14.9%),(J 4.9%), Beryx spp. (5.4%), Micromesistius poutassou (4.3%), OnychoteuthisOnychoteulhis sp. (3.9%).(3.9%), Moroteuthis sp. (2.7%) and Pagellus bogaraveo (2.2%). Inln all,ali, there are more than 18 species of fish and 22 species of cephalopod in thelhe diet. Size distributions of otoliths and beaksbealcs are presented. The species include bottom and midwater formsforms from both oceanic and near islandisland water and depths. A number of rare species of cephalopods and species not recorded for thelhe Azores are inin the diet. Diet does not differ according to thelhe size of swordfish sampledsampled or between female and male swordfish. Preliminary calculations were made on the minimum biomass eaten by the swordfishswordfish off thethe Azores as 7280 - 8680 t per year, approximately half fish and half cephalopods.

CLARKE, M.R., D.C. CLARKE, H.R. MARTINS & H.M. SILVA 1995. A dieta do espadarte (Xiphias gladius) em Bguaságuas A~orianas.Açorianas. Arquipe'lago. Arquipélago. CienciasCiências Biol6gicasBiológicas e Marinhas 13A: 53-69. Angra do Heroismo.Heroísmo. ISSN 0870-6581.

Um total de 132 est6magosestômagos de espadarte foram recolhidos durante os anos de 19921931992/93 em individuosindivíduos capturados em palangre derivante pertoperto do arquip6lagoarquipélago dos A~ores.Açores. 0sOs contelidosconteúdos foram completamente separados em componentes que incluiam:incluiam: todostodos os estados de digest20digestão de peixes e cefal6podes,cefalópodes, desde peixespeixes completescompletos at6até fragmentosfragmentos de mdsculo,músculo, ossos e ot6litosotólitos e bicos de cefal6podes.cefalópodes. 0sOs peixes e os ossos maiores,maiores, que podem ser identificados,identificados, pertenciarnpertenciam todostodos aa Capros aper, Lepidopus caudatus, Pagellus bogaraveo.bogaraveo, rnictofidecsmictofidecs e.e, o peixe que serviaservia de isco,isco, TrachurusTrachurus picturatus.picturalUs. MdsculoMúsculo das lulas Todarodes sagittatus,sagillatus, Pholidoteuthis boschmai e a lulalula que servia de iscoisco IllexIllex sp. tambCmtambém foifoi encontrada.encontrada. 1249 ot6litosotólitos e 120 bicosbicos (inferiores)(inferiores) de cefal6podescefalópodes foram identificadosidentificados at6até ao g6nerogénero ou espCcie.espécie. Encontrou-se uma mediamédia de 3.1 esp6ciesespécies e 18 animais representadosrepresentados em cada estbmago.estômago. 98.4% dos est6magosestômagos apresentavem restosrestos de peixes que contribuiam com 93.4% dos animais naoa dieta. Encontraram-seEncontraram-se cefal6podescefalópodes em 37.4% dos estbmagosestômagos que contribuiram com 5.8% dos animais. Estimou-se queque os pesospesos dos cefal6podescefalópodes e peixespeixes

S3 contribuiamcontribuiarn respectivamente respectivamente comcorn 50.5% e 49.5%' ..dosdos conteúdos conteddos estomacais. estornacais. Os0s constituintesconstituintes mais importantesirnportantes da dieta, determinados por depleçãodep1ec;Zo de % peso estimadoestirnado sãos5o Ommastrephes bartramibartratni (24.4%),(24.4%), Lepidopus caudatuscaudatus (17.4%),(17.4%), Pholi-do/euthisPholi-doteuthis boschmai (15.9%),(15.9%), Capros aper (14.9%),(14.9%), Beryx spp.spp. (5.4%),(5.4%), Micromesistius pou/assou poutassou (4.3%),(4.3%), OnycholeuthisOnychoteuthis sp. (3.9%),(3.9%), MorOleuthisMoroteuthis sp.sp. (2.7%)(2.7%) ee Pagellus bogoraveo (2.2%).(2.2%). Ao todo,todo, háhB mais do que 18 espéciesespCcies de peixe e 22 de cefalópodescefaldpodes na dieta. Apresentam-se Apresentarn-se as distribuiçõesdistribuic;Bes por tamanhotarnanho de otólitosotdlitos e bicos. HáHi espéciesespCcies bentónicasbentdnicas e mesope1ágicasrnesopelfigicas de águasBguas oceânicas,oce%nicas, costeiras e profundas. Aparecem, na dieta, algumas espéciesespCcies raras e nãon5o registadas para os Açores.A~ores. As dietas nãon5o variam de acordo comcorn a dimensãodimens50 do espadarte amostradoarnostrado ou entre sexos. CálculosCBlculos perliminares sobre a biomassa mínimaminima consumida pelopel0 espadarte dos Açores &ores apontam apontarn para um númerondmero entre 7280 a 8680 t por ano, sendo aproximadamenteaproxirnadarnente metade peixe e metade cefalópodes.cefaldpodes.

M.R. Clarke & D.e.D.C. Clarke, "Ancarva","Ancarvan, Southdown, Millbrook, Torpoint, Cornwall, PUOPLIO JEZ1EZ u.K.U.K. - H. R. Martins & H. M. Silva, Departamento de OceanografiaOceanograBa e Pescas, Universidade dos &oresAçores University, PT-9900 Horta,Hor/a, Açores,Acores, Portugal.

INTRODUCTION included both fish and cephalopod speciesspecJes by occurrence, number of individualsindividuaIs and volumes.

The present analysis of swordfish stomach Here, the method of voluming was not contents from the Azores is the forth part of a considered of value since very few cephalopods continuing programme to analyse the diets of were represented by flesh and many fish were at large predators in the food web of Azores waters. an advanced stage of dissolution. Instead of only Previously, the diets of the cetaceans Physeter utilising flesh remains, we also identified otoliths and cephalopod lower beaks and used them to macrocephalus (CLARKE et al.aI. 1993), one Kogia and cephalopod lower beaks and used them to indicate occurrence, numbers and wet weights of breviceps (MARTINS et al.a!. 1985) and Loligo thethe species identified. This was also done, for forbesiforbesi (MARTINS 1982) have been described some of thethe prey species by MOREIRA (1990), from thethe Azores. GUERRA et al.a!. (1993) and HERNANDEZ-GARCIA The study of diets of marine fish has (1995). The use of beaks for such studies is well previously been attempted by thethe application of established and has fewer problems (discussed several methods involvinginvolving greater or lesserlesser below and by CLARKE 1986a) thanthan thethe use of complication; the work on swordfish diet is a otoliths which has been much discussed and typicaltypical example of progressprogress inin thisthis field. Work criticised (HYSLOP 1980; JOBLINGJOBLING & BREIBYBRElEY before 1968 comprised notes of thelhe species 1986). Certainly, caution must be exercised inin occurring inin swordfish stomachs,stomachs, sometimes with usingusing otoliths becausebecause theythey dissolve at different referencereference toto thelhe mostmost numerousnumerous species ratesrates accordingaccording toto theirtheir size and species (JOBLING (summarised(summarised byby SCOTT & TIBBOTlEBO 1968). SCOTT & & BREIBYBRElEY 1986; HERNANDEZ-GARCIAHERNANDEZ-GARCIA 1995). TIBBOTlEBO (1968)(1968) werewere thethe firstfirst toto provideprovide However,However, ourour observations suggest that,that, withwith thethe quantitative datadata such asas occurrence,occurrence, numbernumber ofof species occurringoccurring here,here, otoliths can bebe usedused toto individualsindividuaIs andand volumesvolumes ofof thethe component groupsgroups givegive certain valid conclusions.conc1usions. andand thesethese werewere takentaken fromfrom aa largelarge samplesample ofof 514514 IncreasingIncreasing interestinterest inin thethe rolerole of cephalopodscephalopods swordfish.swordfish. WhileWhile thethe fishfish speciesspecies werewere identified,identified, inln foodfood webswebs and thethe useuse of beaksbeaks inin itsits onlyonly thethe mostmost prolificprolific squidsquid (Zllex(I11ex illecebrosus)illecebrosus) elucidationelucidation hashas ledled severalseveral authorsauthors (TOLL(TOLL & HESSHESS waswas identifiedidentified andand thatthat notnot adequatelyadequately 1981;1981; BELLOBELLO 1991;1991; GUERRA etet al.aI. 1993; distinguisheddistinguished fromfrom otherother cephalopods.cephalopods. STILLWELL HERNANDEZ-GARCIAHERNANDEZ-GARCIA 1995) toto concentrateconcentrate onon && KOHLERKOHLER (1985)(1985) alsoalso consideredconsidered onlyonly thethe cephalopodscephalopods andand neglectneglect thethe fishfish inin theirtheir dietdiet identifiableidentifiable specimensspecimens ofof thethe foodfood organismsorganisms andand studiesstudies ofof swordfish.swordfish.

54 Previous work was for swordfish caught in the animal remains. Volumes of the major western North Atlantic (SCOTT & TIBBO 1968; components of the stomach contents were not STILLWELL& KOHLER1985; TOLL& HESS l98l), measured because the variation in condition made the coast of Portugal (MOREIRA 1990), the eastern such measurements worthless. General North Atlantic (GUERRAet al. 1993; HERNANDEZ- identification of flesh and bone remains was GARCIA1995), the eastern Mediterranean (BELLO followed by examination of otoliths and 1991) and the North Pacific (summarised by cephalopod beaks. Where almost complete MARKAIDA 1995). The results show distinct animals were present their lengths (fork length in differences depending on the region studied and the case of fish) were measured, they were on the type of analysis employed (see weighed and their otoliths were removed for Discussion). comparative studies. Collection and identification The present work should have value for of fishes was aided by help from fish specialists in understanding the role of fish and cephalopods in the Department of Oceanography and Fisheries, the foodweb of the Azores as well as for a greater University of the Azores, Horta, Faial (DOP, see understanding of the diet of the swordfish. Acknowledgements). Identification of otoliths (DCC) was achieved by comparison with a collection removed from a variety of local fishes MATERIAL AND METHODS both from catches and from stomach contents and, for myctophids, from the literature (FITCH1969; GRENFELL1984; HECHT& HECHT 1987; NOLFE Stomachs were collected from 132 swordfish, & CAPETTA 1988). Two species were identified caught by drifting longlines near the surface by fish and otolith specialists (see during commercial (105) and research fishing Acknowledgements, Smale and Merrett). 1249 (27) off the islands of Pico and Faial during the otoliths in reasonable condition were identified. period 25.10.92 - 25.12.93. Line setting began at 50 (3.8%) were too digested for identification to about 17:OO-18:OOh and hauling started at about be possible. 08:OOh the following morning. The sampling area for the research fishing, carried out R.V. Identification of cephalopod beaks (MRC) "ArquipClago" of DOP, was between the three was carried out with a large reference collection islands Faial, SHo Jorge and Pico to the Azores and methods prescribed by CLARKE (1986a). Bank and Princesse Alice Bank. Commercial Seven small lower beaks could not be identified catches were from the same general area. but were probably in three species of small Of the 132 swordfish, 113 were collected in cranchiids. October-December 1992 and 19 in February to December 1993. The monthly distribution of the Estimates of the weights and lengths of Reryx samples was February (2), August (2), October species were made from unpublished size (53), November (33), December (33) which relationships supplied by E.Isidro of the D.O.P. reflects the seasonal presence of the species in the Estimates for Micromesistius poutassou were area. taken from Ln Wt(in g) = 2.83 x Ln OL(in rnrn.)- ~owerjaw fork lengths of the swordfish were 1.29 (MOREIRA1990) and SL = 23.17 x OL-25.54 measured from the tip of the lower jaw to the fork (JOBLING & BREIBY 1986), for Pagellus of the tail. Stomachs were removed at sea = immediately after capture. After being deep bogaraveo Wt(g) EXP(Ln(39.4177) + (1.189 x frozen, the contents of each stomach were thawed Ln(0Radius))) and Wt(g) = EXP(Ln(0.0124) + out in the laboratory and weighed; the average (3.137 x LN(FL in cm))) (derived from KRUG weight of the stomach contents was found by 1989), and for Trachurus picturatus FL (in cm) = deducting the mean of empty stomachs from the (OL-1.759) 10.224 and Log Wt(g) = Ln 0.00819 mean of stomachs with food. Stomach contents + 3.1 1 x Ln(FL in cm) (ISmRO 1990). A sufficient were carefully sifted and sieved for all identifiable number of Capros aper were present in the stomachs to relate otolith size to fish weight. RESULTS This shows considerable variation but several stomachs were dilated with Capros in good condition and the total weight and number of fish present showed that Capros taken by the General description of stomach contents swordfish averaged near to 5 g. This agreed with estimates made on fewer fish from otolith length relationships. No weights for myctophids were Nine of the total samples were empty except for a available but their total weight was calculated few fish eye lenses. The 123 swordfish containing from a mean volume of 8.5 ml obtained by data of food remains other than fish eye lenses averaged SCOTT & TIBBO (1968). Other estimates were 107 cm in fork length (SD. = 37 cm, range: 49- based on relationships given in SMALEet al. 240 cm) which we calculate to equal 13.4 kg from (1995). From the overall totals of food items the published relationships (see PALKOet al. 1981). sum totals of each species represented in the diet An attempt to categorise the condition of the were calculated by adding the number of fish stomach contents is shown in Fig. 1. The identified from flesh to half the number (rounded categories show the state of the least digested up to the next even number) of otoliths, taking food in the stomachs; some of the stomachs also account of instances where two otoliths in a contain food in other states of digestion e.g. some have otoliths and debris together with fresh stomach only represented a single fish. specimens but are placed in the 'good ID' category. The stomach weights plotted against the Estimates of the weights and mantle lengths swordfish fork lengths (Fig. 2) shows that the (ML) of cephalopods were made from known weights of the stomachs with no identifiable flesh relationships for the lower beak rostra1 length or are broadly spread over the size range but there is hood length (LRL, LHL) of each species or family a suggestion that more swordfish below 100 cm found (CLARKE1986a). (35.6%) ate less recently than swordfish over 100 cm (25.6%). The stomachs containing remains averaged 518 g in weight. If only the Eye lenses of cephalopods can be stomachs containing identifiable remains (other distinguished from those of fish by being easily than otoliths and beaks) are included, the mean is fractured into two unequal hemispheres. This 739 g. A swordfish of 100 cm (close to the mean allows digestives fluids to enter between the FL=lO7 cm) had a stomach weighing 1660g onion-like layering and the lenses break up under which might be taken as an indication of the mechanical pressures so that they disintegrate and maximum weight for the mean fish and the mean disappear from the stomachs much more quickly fish then had about 50% of its stomach capacity than the fish eye lenses. The ratio between fish filled. Subtraction of the mean weight of stomachs with no contents from the mean weight of and eye lenses does not, therefore, indicate the stomachs with contents gives a mean weight of relation of cephalopods to fish in the diet, as was 334 g for the stomach contents. formerly supposed (CLARKE1986b; HERNANDEZ- GARCIA1995). Stomach contents were often bones with macerated fish attached and could be identified as Two items removed from the stomachs, the Capros aper, Lepidopus caudatus, Pagellus horse mackerel, Trachurus (12 in 5 stomachs) and bogaraveo or myctophid remains. In total, 2189 the squid Illex (one only) were in an undigested animals were represented by either flesh, bones, state and were almost certain to have been bait otoliths or beaks in the stomachs (Table 1). A used in catching the swordfish. These have been mean of 18.0 animals (SD. 18.0, range: 1-108) excluded from the calculations. However, where (fish, cephalopods or crustaceans) were Trachurus otoliths alone were present, they were represented in each stomach. Thirty-nine species taken to be indicative of Trachurus in the diet. of fish and cephalopods were identified from Percentage (%) myctophids. Cephalopod flesh comprised two heads and a body of Todarodes sagittarus in one sample and a tentacular crown (with beaks) of Pholidoteurhis boschmai in another. Fish remains occurred in all but two stomachs containing food (98.4%). Cephalopod remains occurred in 37.4% and crustacean remains in only 12.2%. Two samples (1.6%) 'contained cephalopod but no fish remains. The total number of fish represented by remains was 2045 and Identification I Fig. 1. The percentage of stomachs with contents in cephalopods was 127, a ratio of 16:l. The ratio of a particular stage of digestion. Only the least fish lenses to cephalopod lenses was 472:1, digested state is shown for each stomach. reflecting the faster digestion and loss of cephalopod lenses compared with those of fish. Stomach & contents weight (g) 5050 , I The number of fish otoliths was 1299 and the 4550 - w ldffltfiiedprey x NO identifiedremsins d flesh number of cephalopod beaks was 263 giving a 4050 - :,Mean ratio of 5: 1. 3550 w 3050 W Fish comprised 93.4% by number of all animals represented in the stomachs. cephalopods 5.8% and crustaceans 0.8%. Fish averaged 2.2 species and 16.9 individuals per stomach containing fish and 2.2 species and 16.8 per stomach containing food. Cephalopods averaged 1.81 species and 2.81 individuals per ~orkLenqth (mm) stomach containing cephalopods and 0.78 species Fig. 2. Weight of stomach including contents plotted and .21 individuals per sample with food. against the lower fork length of the swordfish. The &an stomach length is plotted against the mean fork length. Species contributions otoliths, beaks, bones and flesh remains (including an unidentified myctophid and Forty-two species were present (including three excluding 3 very small cephalopod species, cephalopod and one fish species which could not probably cranchiids). The state of the crustaceans be identified) including 18 species of fish and 24 prevented specific identification but 4 penaeid species of cephalopod and there was an average shrimps, 3 'shrimps', 1 euphausid and 3 isopods of 3.2 species per stomach (SD. 1.8, 1-1 1). The (which may have been parasites off fish) were number of individual animals represented in the collected. Crustaceans were too damaged and stomachs averaged 17.9 per stomach. their weight was too small to be included in calculations of weight. Fish 'flesh' (including bones with flesh Fish attached and pieces of muscle) occurred in 60.2% of samples and cephalopod flesh in only 2 Numerical composition of the fish and estimates samples (1.6%) containing food remnants. The of the wet weights are shown in Table 2. Otolith fish 'flesh' included recognisable, complete length distributions for the commonest species Capros aper and the heads and bones of (Fig. 3) show that all species except Lepidopus caudatus, Pagellus bogaraveo and Micromesistius poutassou and Diretmus

Diretmus N=148

- Beryx splendens Beryx decadactylus N=18

Notoscopelus N=47 I

Other Myctophids N=175

Micromesistus N=55

Otolith length (mrn) Otolith length (rnm)

Fig. 3. Otolith lengths of fish identified from swordfish stomachs. argenteus have distinct unimodal peaks. The been in the stomach some time before capture, smallest species, Notoscopelus sp. and Diaphus otoliths are assumed to represent natural food and spp., have skewed distributions suggesting that are used in calculations. However, the heads from the swordfish selects for the larger fish of these which they came may also have been taken off small species. hooks. All the fish in good condition and most of the identifiable poor specimens were Capros aper Cephalopoda and these had a fork length distribution with a peak at 10-20 mm (Fig. 4). The pelagic octopods Argonauta argo, Ocythoe tuberculara and Haliphron atlanticus Percentage (%) (=Alloposus mollis) make a surprisingly large numerical contribution to the cephalopods in the diet (40.1%) but the larger, strong-swimming squids of the Ornrnastrephidae and the Pholidoteuthidae together contribute 79% of the cephalopods and 41.6% of all the food by weight (Table 3). Another 12.3% of cephalopod weight and 6.5% of the weight of the food is contributed by onychoteuthids, one being Moroteuthis sp. which has not been recorded for the North Atlantic but is known to occur here from Fork Length (mm) unpublished data of one of the authors (MRC). Fig. 4. Histogram of the fork lengths of complete Beak measurements (LRL's and LHL's) are Capros aper from stomachs of swordfish. too few to give well-based distributions (Fig. 5) but they do show some grouping in each species. Very rarely collected genera which are Diretmus argenteus has previously been represented are Discoteuthis, Valbyteuthis and regarded as rare in this area. Grimalditeuthis. Four small Lepidopus caudatus measured 320-380 mm in length but other specimens were General too digested for measurement. The myctophid otoliths included Lumpadena A histogram of estimated mean weights of all the sp., Lampanyctus sp., Notoscopelus sp. and species (Fig. 6) shows that most are less than several species of Diaphus. Although it is not 100 g in weight and only three species, all possible, with material available to us, to estimate cephalopod, are over 400 g. their contribution by weight from otolith size, a Figure 7a shows the % weight contributed, of rough approximation afforded by applying a mean species contributing more than 1%, arranged in derived from SCOTT & TIBBO (1968) is probably descending order of importance. The steep part of not too far from reality. the curve includes five fish genera, Lepidopus, No flesh of Micromesistius poutassou was Capros, Beryx, Micromesistius and Pagellus and found and otoliths were identified by comparison four cephalopod genera, Ommastrephes, Pholido- with material of J. Castro (see Acknowledge- teuthis, Onychoteuthis and Moroteuthis. ments). The bimodal peak to the length Figure 7b gives the cumulative curve of the distribution (Fig. 3) may suggest a second species same species and shows that these nine species is included. contribute over 91 % to the food of the swordfish. The thirteen complete Trachurus picturatus in This is in contrast to the cumulative curve of the stomachs are considered to be bait swallowed numbers in which Capros comprises 75% of the at the time of capture and their weights are not number of animals in the diet and only another six included in calculations. Because they must have species contribute more than 1%. 50 Ommastrephes -- I Pholidoteuthis 40 N=10

Chiroteuthis N=6 I

Brachioteuthis Onychoteuthys N=7

2 3 4 5 6 7 8 9101112 Beak length (mm) Fig. 5. Lower rostra1 lengths of the beaks of the commoner cephalopods present in the stomachs of swordfish

Size and habits of prey targeted measurement for expressing size of a prey item in the way fork lengths are and a better dimension is the standard length from the tip of the arms to the Estimates of the mantle and standard lengths of tip of the mantle. Head and arm lengths can be the cephalopods (Table 3) and the fork lengths of between 0.4-3.75 times the mantle length some of the fish (Table 2) range between 7 mm according to the species. If the estimates of mean and 627 mm. Mantle lengths are not a good standard lengths from the beaks and fork lengths Weiaht f044 25 nOmmastrephes 0Fish mcephalcpods

~e'yx~cmmesistius 4\ Onychoteuthis

50 250 450 650 850 IDSO 12% 1450 1650 Estimated Mean Weight (g) Fig. 6. Histogram showing the estimated mean weights Species of fish and cephalopods from swordfish stomachs. Fig. 7a. The estimated % total weights of species with more than 1% arranged in descending values. from the otoliths are plotted in decreasing values Cumulative % for the fish and cephalopods separately (Fig. 81, it 1W1 I is evident that the mean of no fish exceeds the means of the 7 largest cephalopods. Six of the eight largest cephalopods are muscular, fast swimming and negatively bouyant squids. Seven cephalopods with means between 125 mm and 280 mm are all neutrally bouyant and slow swimming except for Brachioteuthis. Five of the 6 smallest genera are negatively buoyant, fast jetters. Ocythoe has an air-bladder to control I bouyancy. Species Fig. 7b. Cumulative % weights of the species in 7a Influence of swordfish sex on food and cumulative numbers of species in descending value showing Capros aper as the first species. Forty-one male and 33 female swordfish were and 608 g respectively for the smallest to the sexed and contained food: 49 were not sexed. The largest fish. As one might expect from this, the males averaged 102 cm in fork length and the weight of Capros aper flesh (the main component females 121 cm. Stomachs with contents weighed by bulk of the contents) increased from a mean of an average of 574 g in the males and 838 g in the 4.4 g through 8.8 g to 10 g but no other trends females. Females had a mean of 13.6 animals and were evident in fish or cephalopods. 2.3 species while males had 21.5 animals and 3.0 species represented in the stomachs. No significant differences between the food species DISCUSSION of the two sexes could be detected. Methodology Influence of swordfish length on food Although otoliths in diet studies must be used To examine differences in diet attributable to the with care, we believe that the present work sizes of the swordfish we compared the samples demonstrates their value and justifies their use. If from 21 fish in each of three size groups with fork only identifiable flesh were used for lengths less than 80 cm (mean 69 cm), 90-120 cm identification, only five instead of 18 species of (mean 100 cm) and over 130 cm (mean 147 cm). fish and two instead of 21 species of cephalopod Weight of stomach contents were 207 g, 191 g would have been recognised. If size and number Table 3

Numbers and estimated sizes of cephalopods represented in swordfish stomachs by lower beaks. (No.- number; ML- mantle length; Std 1- standard length; 0cc.-occurrence) Family Genus No. Mass ML Std I Occ. No % % all total % % all Mean Min. Max. Mean Mean % Wt Argonautidae Argonaura 45 35.4 2.05 297 1.04 0.55 16 58 36 80.64 Ocythoidae Ocyrhoe Alloposidae Huliphron Sepiolidae Hereroteurhis Loliginidae Loligo Ommastrephidae Otntnustrephes Todurodes Onychoteuthidae Onychoteuthis Moroteuthis Pholidoteuthidae Pholidoteuthis Brachioteuthidae Brachioreurhis Enoploteuthidae Abraliopsis Pyroteuthidae Pyroteurhis Histioteuthidae Hisrio A Hisrio B Discoteuthidae Discoteuthis Chiroteuthidae CI7iroreuthis Valbyteuthis Grimalditeuthidae Grbnalditeuthis Mastigoteuthidae Masrigoteufhis Cranchiidae Tuonius 3 unidentified SP TOTALS UDD~~.. beaks 136 Total animals 2189 56336 of otoliths and beaks had been ignored, Capros stomach was appreciably longer than the time aper would have been considered the taken to dissolve otoliths; the counts would be overwhelmingly most important food species biassed towards cephalopods. instead of providing only14.9% of the food. Such The conclusion drawn below that the food has computations from beak and otolith size must be been in the stomachs for less than 12 hours makes considered rough estimates but are much better it unlikely that many of the otoliths have dissolved thari relying on weights of partially and variably completely in stomach juices (JOBLING & BREIBY digested stomach contents. 1986; HERNANDEZ-GARCIA1995) and this is In determining the relative importance of fish supported by there being few halves of Capros and cephalopods from flesh, or beaks and otoliths, otoliths which are small and easily break into two. various problems arise. The flesh of many Thus, we consider the ratios between the otoliths cephalopod species is digested much more rapidly are meaningful for determining diet. than fish flesh; cephalopods have no bones, which As pointed out by JOBLING & BREIBY(~~~~) take longer to digest than flesh. On the other reduction of the lengths of otoliths after a short hand, beaks are not digested in the stomach while time in the stomach could materially affect some otoliths take only a few hours to digest estimates of body lengths and weights of the fish. (JOBLING & BREIBY 1986; HERNANDEZ-GARCIAFor the commonest fish, Capros aper, the 1995). These factors would clearly make estimates from the better otoliths gave similar comparisons meaningless if the time in the results to direct weighing and measuring of intact Mean Length (mm)

Moroteuthis

A f~sh(fork length) muscular cephalopods (standard length) I Ch~roteuth~s lammonlocal cephalopods Phol~doteuthis 0 gelatmous cephalopod Taonu~s I rn Om mastrephe Loligo a r Onychoteuthis Coelorhynchus

Todarodes f@ Haliphron 0 Discoteuthis & Epigonus sp. A r Histioteuthis A. A A Beryx dec Mastigoteuthis .a Grimalditeuthis Pagellus A Be'YXsPl. Brachioteuthis Histioteuthis B. Epigonus :p, A Micromesistius i; I Pyroteuthis Lepidopk A Macrorhamphosus Ocythoe Argonauts Lampanyctus A A A A Notoscopelus ~braliopsis' mVa~yteuthis Diretmus A A Hoplostethus Diaphus Heteroteuthis

Soecies

Fig. 8. Estimated mean fork lengths for fish and standard lengths for cephalopods arranged in descending values, fish and cephalopods plotted separately. Square points are ammoniacal squids. fish from the stomachs. In other species the Todarodes and Pholidoteuthis were eaten directly weights were estimated from the better otoliths or, by the swordfish but that is also highly likely for where appropriate the mean, and multiplied by the all species but the smallest such as Hereroteuthis, total number of otoliths present. Cephalopod Abraliopsis, Pyroteuthis. Flesh was identified estimates depend upon beaks whose rostra1 from the smallest fish, Capros aper and the lengths do not reduce with digestion. The main myctophids. limits to accuracy for cephalopod estimates are imposed by the limited accuracy of body weight Regional comparisons and length to LRLs relationships. For the species occurring here, most relationships are reasonably Comparisons with other studies on swordfish from accurate except for Ocythoe, Haliphron, the Atlantic are sumrnarised in Table 4. Oceanic Discoreuthis, Valbyteuthis and Grirnalditeuthis, samples showed a numerical dominance of together only comprising 9.6% of the number of cephalopods in the diet and samples near or on beaks. the Continental shelves showed a dominance of Other possible errors in this kind of analysis fish. The difference in this respect between the are acquisition of food organisms from the results of SCOTT& TIBBO (1974) and STILLWELL stomachs of the food targeted by swordfish. This & KOHLER(1985), from much the same region could be the source of the smallest beaks, and was explained by the latter as a result of the should be borne in mind when species are increase in squid in the time between the sampling considered, but this is not likely to be a cause of years. It could also partly be explained by the error in considering the commonest and larger difference in latitude and depth of water of the cephalopods. Evidence from flesh showed that main areas of sampling. Table 4 Comparison of general results with the literature. (No.-number; 0cc.-occurrence: Vol-volume: wt-weizht:- Sw- swordfish: s.-same: F-Fish: C- Cephalopod: f-family: p- genera). Region group No.% Occ.90 VoVwt% Main food No.Sw LJW Sw % empty S. taxa No spp Author as here Azores fish 93.1 98.4 Lepidopus 132 107cm 7 14 This paper cephalopods Omtnustrephes 22 crustacea N.E.Atl. fish 13 GUERRAet al. (oceanic) cephalopods Ommastrephids 1ocg 17 crustacea Equator fish selected for C (oceanic) cephalopods Sthenofeuthis crustacea Portugal fish Mlcrotnesisrius

cephalopods Illex crustacea Str.Gib, fish Micromesistius

cephalopods Todarodes crustacea W.Afr. fish

cephalopods Sthenoteuthis crustacea N.W.Atl fish Scomber

cephalopods Illex (Mar-Oct) crustacea N.W.Atl. fish

cephalopods Illex; Histioteuthis crustacea N.W.At1. fish Merluccius

cephalopods Illex crustacea E.Med fish cephalopods Todurodes

crustacea W.Baja. fish Calif cephalopods Sthenoteuthis crustacea N.Baja fish Trachipterus Calif cephalopods 38 16

The specific composition of fish and et al. 1993) which share 9-11 species (or genera) cephalopods in the samples varies greatly between of 28 named specieslgenera recorded in the two areas, particularly for the fish. The closest studies. The most notable difference, similarity to the Azores region is, not surprisingly, Sthenoteuthis pteropus, which is the most the area between the Azores and Spain (GUERRA numerous in the Spanish samples and does not occur in the Azores samples has a usual northern aper, Micromesistius poutassou and Trachurus limit to the South of the Azores and this is picturatus, near bottom species such as Pagellus reflected in the samples. The abundance of bogaraveo and slope species such as Argotzauta argo in the present samples may Coelorhynchus and midwater slope species such reflect an unusual abundance in the year or as Lepidopus caudatus. The myctophids are possibly the time of year (the months of sampling oceanic shoaling species and share the same water were not recorded by GUERRAet al. 1993) between 200 m and 1000 m with many of the sampled off the Azores as they did not occur in neutrally bouyant squids also found in the the Spanish samples. However, they did occur in swordfish stomachs. Most of these have die1 swordfish off Portugal (MOREIRA1990) showing vertical migrations bringing them into the top that they are probably a regular part of the diet. In 200 m at night so that their presence in the diet all regions, species in the Ommastrephidae were does not contradict the conclusion that the the most important cephalopods by number and, swordfish is a nocturnal feeder. Heteroteuthis is where measured, by weight or volume. particularly associated with sea mounts. Thus, the swordfish appears to search all the sea around the Predator comparisons islands and the adjacent sea mounts for a wide diversity of prey, perhaps having little preference Twelve of the 40 species of cephalopods from except for size. stomachs of sperm whales caught near the Azores It would clearly be to the swordfish's (CLARKEet al. 1993) also occurred in the advantage to select for the largest prey items but swordfish diet of the area which comprised 21 its adaptability in accepting a large variety of identified species. Genera eaten by swordfish but species, large and small, arnmoniacal as well as not found in stomachs of sperm whales in the muscular, mesopelagic as well as epipelagic region are Argonauts, Ocythoe, Moroteuthis, assures it of an ample food supply when large Loligo, Hereroreuthis, Brachioteuthis, Abraliop- species are scarcer. The mean weights of the sis, Pyroteuthis, Valbyreurhis and Grimaldireu- cephalopods in the diet is 224 g and for the fish it this. Squids of the last six genera are all too small is 13.6 g. to be likely targets of the whales. The first two genera may live too near the sea surface to fall Food consumption within the whale's normal sampling depth range. As explained above, Moroteuthis is a new record Finding the time taken for a swordfish to eat the for the area and may be extremely rare here. estimated mean stomach contents cannot be done Loligo lives close to the edge of what little shelf is directly. If food is eaten more or less continuously present off the Azores islands and may be too and leaves the stomach continuously, we might close to the islands to be a regular prey of the expect to find all stomachs to be in more or less whales. Considering the importance of the same state of balance between the octopoteuthids to sperm whales (5.9% by number) constituents. This is clearly not so (Fig. 1). Such a it is surprising they were not found in the diet of system would have the problem of clearing bones swordfish although they are eaten elsewhere in the and beaks from the stomach in a regular manner eastern North Atlantic (MOREIRA1990; GUERRA with the fine breakdown products, without losing et al. 1993; HERNANDEZ-GARCIA1995). recently eaten food at the same time. More likely, is that a feeding bout of some duration is Prey distribution accompanied by, and followed by, reduction of all the muscle food to a particulate state followed by The fish eaten by swordfish include species from expulsion of the harder tissues before the next all the ecological niches near these oceanic feeding bout begins. The variation of our samples islands. Species living in island-influenced waters supports this. Such a natural break could be night such as the near surface shoaling species Capros and day and work by CAREY& ROBISON(1981) shows that swordfish spend nights near the continuous process but is more an explosive surface and days at considerable depths in cold process after one feeding and before the next. water where they may be far less active and may, indeed, feed less. As the present samples were all We believe the estimated weight of the collected from fish caught at night and many have stomach contents are most likely to be the product fresh fish and identifiable, deteriorating fish, they, of only one night's feeding because, if the food at least, must be assumed to be that night's food remains had been obtained in two nights, one intake. We might conclude that the night's feeding might expect the otoliths, beaks and the bout, is followed by a daylight period of digestion unidentifiable slurry to be the product of the first and an emptying of the stomach prior to the night's feeding and to have, on average, much the swordfish's migration to shallower depths at dusk. same constituents as the second night. This is JOBLING& BREIBY(1986) suggested that marine clearly not so. In addition, the work on mammals and birds may have an interdigestive disappearence of otoliths due to digestion is migrating myoelectric complex (IMMC) of sufficiently fast to remove all traces of the muscular activity as in terrestrial mammals. To smallest otoliths, Caprus and myctophids, from a quote from them: "This cyclic pattern of previous night's feeding period. contractions includes a series of powerful bursts As the mean swordfish weight, calculated of contractions which pass from the stomach to from the mean fork length is 13.4 kg, in one day the upper intestine and serve to sweep any the average swordfish would eat 2.8-3.4% of its indigestible remains from the stomach." The body weight per day and 10.4-12.4 times its body present work suggests a similar mechanism in weight annually. This is higher than the estimate swordfish and that beaks and otoliths are therefore of 3.4-5.8 times derived by STILLWELL& KOHLER not held up in the stomach. However, we cannot (1985) using digestion time intervals of sharks, rule out the possibility that the clearout of food although it is below estimates for some other fish. remains is not totally effective and some beaks and otoliths may remain from the previous night's From knowing the annual requirement of food feeding. and the proportion of the components of the diet, it is possible to estimate the minimum mass of If some holdup takes place a clue to the each component eaten by swordfish in various magnitude of this may be obtained by reviewing regions (STILLWELL& KOHLER 1985; HERNAN- the contents of stomachs with no flesh but only DEZ-GARCIA1995). In the Azores region about otoliths and or beaks. From these stomachs 200 344 t of swordfish are landed annually. As these otoliths (15.5% of all otoliths) and 29 lower beaks are probably three years old (OVCHINNIKOV1970) (22.8% of all lower beaks) were collected. These and the catch has been sustained over a number of otoliths represent 1360 g and the lower beaks years, our sample might be expected to represent 6489 g of the food represented in the stomachs. If a population at least 3 times as many and the total, 7849 g, is subtracted from the total food weighing at least twice as much. However, the represented in the stomachs, 56336 g, the weight catches suggest they may only spend three months left for the average feed is 382 g. This compares in the region. The consumption of the 700 t of with 458 g if one assumes no otoliths or beaks are swordfish would consume while in the Azores carried over from one feeding period to the next. region over 7280-8680 t. This would include, for A rough weight for the mean stomach contents is example, something in the region of 1387 t of derived from subtracting the mean weight for Lepidopus caudatus and 435 t of Beryx, tonnages empty stomachs from the mean weight of the considerably exceeding those taken commercially stomachs containing food and is 334 g. The (=255 t and 176 t respectively). The 2045 t of difference between this and the estimates of 382 g ommastrephids also eaten by swordfish shows the and 458 g can be accounted for by loss of importance of these squids in a region where they digestive products from the stomachs but this is so are not caught commercially. Such computations small as to suggest that most loss is not a should not be carried through to other years since we have no idea of the inter-year stability of the History Museum identified Epigonus otoliths and diet until the work is repeated. confirmed several other species identifications. Dr. Vicente Hernandez-Garcia and Dr. Jose Castro of the University of Las Palmas, Canary Islands kindly let us Future examine their collections of otoliths. To all the above, we are most grateful. Our special thanks are given to While several papers now show the considerable Carmelina Leal whose energy and thoroughness in the regional variation in the species eaten by unenviable task of sorting the total stomach contents swordfish, there is no evidence for annual changes into Classes and sometimes Genera, counting, or their magnitude in any one place. Assessing the measuring and note-taking has made the final analysis possible. stability and reproducibility of the results of such analyses is of crucial importance in assessing their value for prediction and must involve collections REFERENCES made in as precisely similar ways as possible at different times. Comparisons should not only involve species composition of whole specimens BELLO,G. 1991. Role of cephalopods in the diet of but should include both flesh, otoliths and beaks swordfish, Xiphias gladius, from the eastern Mediterranean Sea. Bulletin of Marine Science 46 and their size distributions. (1 -2): 3 12-324. As more species are treated in the same way as CAREY,F.G. & B.H. ROBISON 1981. Daily patterns in here, it is hoped that well based modelling of the the activities of swordfish, Xiphias gladius, foodwebs in this area will be possible. Seeing observed by acoustic telemetry. Fishery Bulletin 79 changes in the ecostructure may then be practical (2), 277-292 by monitoring the changes in diet composition of CLARKE, M.R. (Ed.) 1986a. A Handbook for the a few selected, high predators in the system. It is Identification of Cephalopod Beaks. Clarendon our conviction that such monitoring will be much Press, Oxford. 273pp. more effective and a much cheaper way of CLARKE,M.R. 1986b. Cephalopods in the diet of odontocetes. Pp. 281-321 in: Research on observing change than collection with nets from dolphins. BRYDEN,M.M. & R. HARRISON(Eds) oceanographic vessels; such will give direct Clarendon Press, Oxford. information of the interaction of animals by CLARKE,M.R., H. MARTINS& P. PASCOE1993. Sperm means of rather few samples including a good whale diet of the Azores. Philosophical spread of species and sizes. Transactions of the Royal Society of London B 339: 67-82. FITCH, J.E. 1969. Fossil lanternfish otoliths of California, with notes on fossil Myctophidae of North America. Contributions to Science, Los This paper has involved the help of many people Angeles County Museum (1 73): 1-20 involving the sampling of many species of fish for GRENFELL,H.R. 1984. Early Miocene teleost otoliths otolith comparisons as well as swordfish for stomach from Parengarenga Harbour, New Zealand. New contents. The authors extend their warmest thanks to Zealand Journal of Geology & Geophysics 27: 51- all the fishermen and the Captain and crew of R./V. 96. "Arquiptlago" as well as to various members of the GUERRA,A., F. SIMON & A.F. GONZALEZ1993. DOP staff including Norberto Serpa, Filipe Porteiro, Cephalopods in the diet of the swordfish, Xiphias Humberto Rodrigues, Patricia Sim6es for diligently gladius, from the northeastern Atlantic ocean. Pp. sampling from the swordfish and other fishes. Helena 158-164 in: OKUTANI,T., R.K. O'DOR, & T. Kmg kindly gave her collection of otoliths to form the KUBODERA(Eds), Recent Advances in Fisheries basis for the reference collection used in the present Biology, Tokai University Press. Tokyo. study. Eduardo Isidro and Gui Menezes very kindly HECHT,T. & A. HECHT1987. A guide to the otoliths of made available their unpublished curves of otolith to Southem Ocean Fishes. South African Journal of body weight relationships. Dr. Malcolm Smale of the Antarctic Research 17(1): 1-87. Port Elizabeth Museum, KSA identified Diretmus HERNANDEZ-GARCIA,V. 1995. The diet of the otoliths and Nigel Merrett of the London Natural swordfish (Xiphias gladius Limaeus, 1758) in the central East Atlantic with an emphasis on the role NOLFE,D. & H. CAPPETTA1988. Otolithes de pliocenes of ccphalopods. Fishery Bulletin 93(2): 403-41 1. du Sud-est de la France. Bulletin de l'lnstitut Royal HESS. S.C. & R.B. TOLL 1981. Methodology for des Sciences Naturelles de Belgique. 58: 209-271. specific diagnosis of cephalopod remains in OVCHINNIKOV,V.V. 1970. Swordfishes and billfishes stomach contents of predators with reference to the in the Atlantic Ocean. Ecology and functional broadbill swordfish, Xiphias gladius. Journal of morphology. Israel Programe For Scientific Shellfish Research l(2): 161-1 70 Translations 1971: 1-75. HYSLOP. E.J. 1980. Stomach contents analysis- a PALKO,B.J., G.L. BEARDSLEY& W.J. RICHARDS198 I. review of methods and their application. Journal of Synopsis of the biology of the swordfish, Xiphias Frshery Biology 17: 41 1-429. gladius Linnaeus. NOAA Technical Report NMFS ISIDRO,H.A. 1990. Age and growth of Trachurus Circular 441. 25 pp. picturatus (Bowdich, 1825) (Teleostei: Scorn, W.B. & S.N. TIBBO 1968. Food and feeding Carangidae) from the Azores. Arquipelago. Life habits of swordfish, Xiphias gladius, in thc western and Earth Sciences 8: 45-54 north Atlantic. Journal of the Fisheries Resesearch JOBLING.M. & A. BREIBY1986. The use and abuse of Board of Canada, 25: 903-9 19. fish otoliths in studies of feeding habits of marine SCOTT, W.B. & S.N. RBBO1974. Food and feeding piscivores. Sarsia 7 1 : 265-274. habits of swordfish, Xiphias gladius, in the KRUG, H. 1989. The Azorean blackspot seabream, northwest Atlantic ocean. Pp. 138-141. in: Pagellus bogaraveo (Bninnich, 1768) (Teleostei: Procedings of the International Billfish Symposium Sparidae): Age and growth. Cybium 13(4): 347- Kailua-Kona, Hawaii, 9-12 Aug 1972 Part 2 355. Review and contrib. papers, U.S. Department of MARKAIDA.U. 1995. Caracterizacion de la dieta del Commerce, NOAA Technical Report NMFS SSRF- pez espada (Xiphias gladius Linnaeus. 1758) en la 675.335 pp. costa occidental de Baja California, Mexico. SMALE,M.J, G. WATSON& T. HECHT1995. A Otolith 1 19pp. Atlas of southern African marine fishes. MARTINS, H.R. 1982. Biological studies of the Icthyological Monographs. J.L.B. Smith Institute of exploited stock of Lolrgo forbesi (Molluscs: Ictltyology. 253pp, 149pls. Cephalopoda) in the Azores. Journal of Marine STILLWELL,C.E. & N. E. KOHLER1985. Food and Biology Association of the U.K. 62: 799-808. feeding ecology of the swordfish Xiphias gladius in MARTINS,H.R., M.R. CLARKE,M.R., R. REINER & R.S. the western North Atlantic Ocean with estimates of SANTOS 1985. A pygmy sperm whale, Kogia daily ration. Marine Ecology Progress Series 22: breviceps (Blainville.1838) (Cetacea: Odontoceti) 239-247. stranded on Faial Island, Azores, with notes on TOLL,R.B. & S. C. HESS1981. Cephalopods in the diet cephalopod beaks in the stomach. Arquipe'lago of the swordfish, Xiphias gladius, from the Florida SCrie Ciencias da Nature~a6: 63-70. straits. Fi~heryRulletin U.S. 79(4): 765-774. MOREIRA,F. 1990. Food of the swordfish Xiplzias gladius Linnaeus, 1758 off the Portuguese coast. Journal of Fishery Biology 36: 623-624. Accepted 12 December 1995.