The Condor 90:157-167 0 TheCooper Ornithological Society 1988

DIVING PATTERNS IN RELATION TO DIET OF GENT00 AND MACARONI AT SOUTH GEORGIA ’

J. P. CROXALL British Survey,Natural EnvironmentalResearch Council, High Cross,Madingley Road, CambridgeCB3 OET, UK

R. W. DAVIS Sea World ResearchInstitute, Hubbs Marine ResearchCenter, 1700 South ShoresRoad, San Diego, CA 92109

M. J. O’CONNELL British Antarctic Survey,Natural EnvironmentalResearch Council, High Cross,Madingley Road, CambridgeCB3 OET, UK

Abstract. The depthsattained on 1,444dives by 14 GentooPenguins ( Papua) and 6,352dives by eightMacaroni Penguins (Eudyptes chrysolophus) were recorded,together with the timing and durationof the foragingtrip and the amountand of prey caught. MacaroniPenguins ate only Antarctickrill Euphausiasuperba. When feedingonly at night they madeno divesdeeper than 20 m; on all-day trips 36% of diveswere between 20 to 80 m. GentooPenguins fed during the day.When theycaught , 77%of diveswere shallower than 54 m; when were taken, 59% of dives were 54 to 136 m, which is consistent with the benthic-demersalhabit of thejuvenile Nototheniaand Champsocephalusfish they eat. The patternof predationon krill by both penguinspecies is consistentwith its vertical migrationto the surfaceat night and dispersalthrough the water column during the day. The food requirements of chick-rearing Macaroni Penguinswould be met by catching at least six adult krill per dive (or 150 juvenile krill or amphipods). For similar Gentoo Penguins,a minimum of 15 adult krill per dive (one every 8 set), or one fish every third dive, is needed. Recorded interannual variations in krill size can treble these rates, which would also be doubled if half the dives were for travelling, not feeding. Key words: GentooPenguin; Macaroni ; PygoscelisPapua; Eudyptes chrysolo- phus;; Euphausiasuperba; fish; diving;diet; foraging; prey capturerate.

INTRODUCTION In none of thesestudies was it possibleto mea- sure accuratelyboth the time that the spent Penguins are the most highly adapted of all birds at sea and also the quantity and identity of food for life under water. However, little is known caught. In this study we obtained the first data about their diving patterns and performance, on the diving patterns of Gentoo (Pygoscelispa- which are fundamental aspectsof their feeding pua) and Macaroni (Eudypteschrysolophus) pen- ecology,influencing the prey they encounter and guins and examine these in relation to foraging their efficient exploitation of these. Recent re- trip timing and duration and to the type and views of feeding ecologyand diving performance amount of food obtained. in penguins (Croxall and Lishman 1987, Kooy- Fieldwork was carried out at Island, South man and Davis 1987) indicate that the depths Georgia (54’S, 38”W), as part of an extensive typically reached during diving by free-living study of the feeding ecology and energy budgets penguins are known for only five .Only ofthese species(Croxall and Prince 1980a; Davis for King Penguins ( patagonicus), et al. 1983; Davis, Croxall, and O’Connell, un- Chinstrap Penguins (Pygoscelisantarctica), and publ.), in the austral summers of 1984 to 1985 JackassPenguins (Spheniscusdemersus) are data and 1985 to 1986. Macaroni and Gentoo pen- other than maximum depth attained available guins are the two commonest speciesof penguin (Kooymanetal. 1982,LishmanandCroxalll983, at South Georgia (with breeding populations es- Wilson 1985). timated at about 5,000,OOOand 100,000 pairs respectively; Croxall et al. 1984) and they are LReceived 20 March 1987. Final acceptance2 Oc- also abundant at Bird Island (breeding popula- tober 1987. tions of ca. 75,000 and 6,000 pairs respectively;

[1571 158 J. P. CROXALL, R. W. DAVIS AND M. J. O’CONNELL

Croxall and Prince 1980b). Both species are FORAGING TRIP DURATION therefore important consumers of marine re- When DHRs were being deployed, birds bearing sources at South Georgia, Macaroni Penguins thesewere watched until they entered the seaand being preeminent among avian species(Croxall a continuous watch was then maintained during et al. 1984, Croxall et al. 1985, Croxall and Prince daylight hours until the bird returned. Birds were 1987) but there are numerous biological and eco- weighed before and after the foraging trip. In logical differences between the two species. In both study seasonsthe attendancepatterns ashore the present context the most important of these of a sample (n = 10 to 15 birds originally) of are that Gentoo Penguins mainly feed inshore breeding birds of each specieswas recorded con- (within 10 km of land; Croxall and Prince 1987; tinuously throughout most of the chick-rearing J. L. Bengtson, unpubl. data) and attempt to rear period using radio transmitters and automatic two chicks, which are fed on adult Antarctic krill recording apparatus. The radio transmitters (15 (Euphausiu superba) and young stagesof mainly mm diameter, 50 mm length, 25 g weight) were benthic-demersal notothenioid fish (Croxall and attachedto the penguins’ back featherswith epoxy Prince 1980a, Croxall and Lishman 1987). In resin and hose clamps or plastic tie wraps. To contrast, Macaroni Penguins feed mainly off- record the presenceof these birds at the colony shore, rarely within sight of land, and rear one a Yagi 4-element antenna was mounted on the chick on a diet of both adult and juvenile E. roof of a small wooden blind and connected to superba, fish being very rarely recorded (Croxall a radio receiver (164 MHz) inside, set to scan and Prince 1980a; Croxall et al., in press). continuously in sequencethe frequencies of the METHODS deployed transmitters for 10 set each. The re- ceiver was connected to a strip chart recorder. DIVING DEPTH This systemprovided control data against which The depth histogram recorder (DHR) described the foraging trip durations of the experimental by Kooyman et al. (1983) was used. In brief, this birds could be compared. It also confirmed that is cylindrical (95.0 mm long by 23.0 mm di- penguins do not arrive ashore at the breeding ameter) and weighs 95 g (approximately 1.6% of colony during the night. Data on the foraging- adult and 2.3% of adult male attendance patterns during postbrooding chick- body weight). The DHR con- rearing of paint-marked birds (Croxall, unpubl. tains eight electronic counters which respond to data), whose presencewas recorded on three to different presetpressure thresholds, each of which four visits (of 1 hr each) daily, are consistentwith is equivalent to a different depth. When the pen- those of birds with radios. Thus paint-marked guin dives below the surface of the water, the Gentoo Penguinsspent about 12 hr at sea(range = increase in pressure with depth is converted to 6 to 24 hr; n = 2 10) and Macaroni Penguins a voltage shift by a pressure transducer in the either 12 hr (range = 10-l 6 hr; n = 140) or 24 DHR. If the voltage exceedsthe threshold of one hr (range = 18-50 hr; n = 72). or more of the counters, it is recorded in each of them. The settingof thesethresholds was checked FOOD in the laboratory against a pressurestation before The contents of the stomachs of experimental and after each deployment. The DHR was at- birds returning ashorewere collectedusing stom- tached to feathersin the middle of the back with ach lavage techniques (Wilson 1984). Birds were two small metal hose clips. Each bird was also flushed repeatedly until no more food was ob- given a distinctive paint mark (number or sym- tained; we have confirmed that this corresponds bol) on the breast to facilitate recapture on its to an empty stomach. In 1985 five complete return to the colony. Each recorder was used for stomach samples and in 1986 10 such samples a single foraging trip only, removed when the were obtained weekly from both speciesthrough- bird was recaptured, and the number of dives out the chick-rearing period. These samples en- registered in each counter was read through a abled us to check that experimental birds were decoding processor which was attached to the taking broadly similar prey and had a similar DHR in the laboratory. Birds were selectedafter mass of food in the stomach to other individuals they had fed a chick and when it wasjudged that engaged in similar parental duties at the same they were about to leave the colony to return to stageof the chick-rearingperiod. For control birds sea. in both seasonsand for experimental birds in DIVING AND DIET IN GENT00 AND MACARONI PENGUINS 159

1986, complete stomach contents were weighed Gentoo Penguin fresh at the time of collection and sorted on the same day. Krill (and other ),fish and Krill Fish were sorted,weighed, and preservedin 10% formalin. Both sagittal otoliths were removed from intact fish crania and these, together with -z any loose otoliths, counted and stored dry in plastic tubes. All material was then returned to the United Kingdom for further processing.All n=862 dives krill in samples were counted, including speci- mens representedonly by pairs of eyes.A sample of 100 krill from each stomach was measured, 0 25 50 either the straightened standard length of per- Percent of dives fectly intact specimensor the carapacelength of damaged ones and sex and reproductive status FIGURE 1. Dive depth histograms (proportion of was recorded for as many specimensas possible. dives terminating in each stratum) for Gentoo Pen- Fish otoliths were identified to as low a taxo- guins feeding on fish and krill. nomic categoryas possible, length, breadth, and thickness measured (to 0.01 mm) and weighed by weight) and amphipods Themisto gaudichau- (to 0.001 g). These measurements were used in dii (ca. 20 mm long; 2% by weight). Gentoo Pen- regression equations to estimate the standard guins ate either krill(9 1% by weight, occasionally length and weight of the original fish specimens with one or two fish present) or fish (87% by (using principally data from Hecht [ 19871 and weight, usually with a few krill). Amphipods (T. unpublished BAS data). Digestion of otoliths was gaudichaudii) were present as a trace in a few not a problem in this study because90% of oto- samples.The fish were mainly Antarctic cod (No- liths were removed from intact fish crania. tothenia rossii and N. neglecta) or ice fish (Champsocephalusgunnari), in roughly equal RESULTS proportions.A few lantern fish Myctophidae were In 1985 eight deployments of DHRs were made present in 1985 to 1986. The Notothenia were on Gentoo Penguins during the chick-rearing pe- mainly juveniles (a few subadults) from 14 to riod. Dietary data and foragingtrip duration were 42 cm in estimated length, equivalent to weights only recorded for four of these. Six deployments of 27 to 1,170 g (mean ca. 130 g). The Champ- were made on sexed adults at a similar stage in socephaluswere smaller, 16 to 20 cm in length, the breeding cycle in 1986. All deployments on 16 to 34 g (mean 22 g) in weight. Krill were all Macaroni Penguins were made on male birds adults; they averaged 52 mm (range = 35-64 (becauseof concern that the DHR might be too mm; n = 3,400) length in 1986 but only 37 mm large a package for the female, which is 20% (range = 28-60 mm; n = 500) in 1985. smaller than the male) in 1986. Six recordswere The diet of the four 1985 Gentoo Penguins for obtained during late chick rearing and two from which it was unknown has been assignedon the birds during the extended trip to sea between basis of the statistically significant similarity be- chicks fledging and the return of adults to the tween their diving patterns and those of birds breeding colony to commence molt (referred to whose diet was known. We feel that this is also here as the premolt period). One of these two biologically realistic becauseof over 100 Gentoo devices was only retrieved after molt had fin- Penguin complete-stomach samples, fewer than ished and therefore the duration of the foraging 10% contained significant quantities of both fish trip was unknown. Neither of these birds was and krill. Excluding thesefour birds changesnone lavaged but krill was the sole food of nonexper- of the interpretations below. Diving data are pre- imental birds at this time. sented in detail in Appendix I. Macaroni Penguins ate krill almost exclusively For the 1986 Gentoo Penguins, there were no (94% by weight of diet) in 1986; all were adults, significant differences between males and fe- averaging 52 mm long (range = 31-62 mm; 12= males, either in the composition of the diet, tim- 4,600). Their only other prey were fingerlings of ing, and duration of foraging trip or in any diving the fish Notothenia rossii (20 to 30 mm long: 4% parameter. The important data from both sea- 160 J. P. CROXALL, R. W. DAVIS AND M. J. O’CONNELL

TABLE 1. Diving data (2 f SD, rangebeneath) for GentooPenguins.

Depth reached (%) Diet n Trip duratmn (hr) No. of dweskip No. of dives/hr <54 m >54 m Fish 6 11.3 f 1.7 97.0 & 26.2 9.8 -+ 1.1 41 59 (57-123) Krill 8 10.7 + 1.3 107.8 * 31.0 9.3 i 2.5 77 23 (69-169)

sons are summarized in Table 1 in relation to or more at sea averaged 45% (range = 31-54%) the highly significant differences in dive depth dives to depths greater than 20 m. The depth distribution of birds which had been eating fish histogramsof premolt birds and of chick-rearing and krill (Fig. 1). In essence,foraging trips that birds in this latter category are not significantly resulted in capture mainly of fish included a sig- different; other comparisons are difficult because nificantly higher proportion of deeper dives (i.e., the duration of the foraging trip by the second to > 54 m) than those which resulted in the cap- premolt bird was unknown. The single bird with ture of mainly krill (x2 = 161; P < 0.001). There full data might appear to have made relatively were no associatedsignificant differencesin trip fewer dives per hour than the chick-rearing birds. duration or number and frequency of dives. For- However, if the other premolt bird, which made aging trip duration was longer, but not signifi- 3,374 dives, had been absent for the mean du- cantly so (t = 0.47; P > 0. l), than trips made by ration of the premolt period at sea (14 days) then birds without DHRs (9.8 f 2.7 hr, IZ = 128). All it would have averaged 10 dives per hour, very but two of the foraging trips of instrumented similar to the values for chick-rearing birds. Gentoo Penguins were completed during day- The foraging trip durations of chick-rearing light hours. The two exceptionswere the trip with birds without DHRs (i.e., either with radios or the most dives (169), whose exact duration was individual marks) are complex (Croxall, unpubl. not recorded (it was between 12 and 16 hr) and data). In brief, during the brooding period only the longest (12.1 hr) of the trips made by birds females forage, delivering meals daily as a result which caught krill. Neither dive profile was of trips almost equally divided between 12 hr anomalous. and 24 hr duration, the frequency of longer trips For Macaroni Penguins (Table 2) the main dif- increasing as the chick grows. During the early ferences in dive depth distribution (Fig. 2) and crecheperiod femalescontinue this routine; males other parameters relate to the duration of the bring meals to the chick every 2 to 3 days. There- foraging trip. Thus birds that were away from after (and corresponding to the main study pe- the colony only overnight (12-hr trips, including riod here) both sexesshare duties approximately 7 to 8 hr of darkness) made significantly fewer equally and trip durations have three modes, of dives in total, but more per hour, than birds in ca. 12 hr, 24 hr, and 48 hr with a few (~5%) either of the other categoriesand made no dives lasting 48 to 72 hr. Consequently we have no exceedingthe shallowestdepth threshold (20 m). evidence that birds with recorders make longer In contrast, all birds which spent a complete day trips than those without.

TABLE 2. Diving data (Z f SD; rangebeneath) for MacaroniPenguins.

Depth reached (%) status Time n Trip duration (hr) No. of dives/trip No. of dives/hr <20 In >20 m Chick rearing Night 3 12.6 i 0.9 203 i 14 16.1 -+ 0.1 100 0 (190-2 18) Night and day 3 40.0 i 28.1 392 i 247 10.1 * 0.8 56 44 (243-677) Premolt Night and day 2 264 1,195-3,374 4.5 54 46 DIVING AND DIET IN GENT00 AND MACARONI PENGUINS 161

Macaroni Penguin

Night Night & Day Premoult

60 I 12.6h 11 days 80 n=608 dives n=1175 dives n=4569 dives N=3 birds N=3 birds N=2 birds 1 I I I I I I r I I 0 50 100 0 30 60 o 30 60

Percent of dives FIGURE 2. Dive depth histogramsfor Macaroni Penguinson foragingtrips at night, for a complete day during chick rearing and during the premolt period.

DISCUSSION to the chicks was proportional to the distance EFFECT OF INSTRUMENT travelled. In our case (Table 3) the mass of prey brought ashorewas not significantly different be- Although DHRs weigh less than 3% of the mass tween instrumented and control birds, nor were of the experimental penguins, they are relatively the durations of foraging trips significantly dif- bulky and, considering the highly streamlined ferent (although this would not be easy to detect nature of penguins (Nachtigall and Bilo 1980) with Macaroni Penguins) but we cannot exclude likely to causedrag. Wilson et al. (1986) tried to the possibility that instrumented birds usedmore estimate the importance of this effect on Jackass energy. Penguins. They found that travelling speed,y (m set-I) was related to device cross-sectionalarea, expressedas a percentageof penguin cross-sec- DIVE DEPTH AND PATTERN tional area, x, by the equation y = 2.14 - 0.063x, There is no previous published information on with device size explaining about half the vari- Macaroni Penguin diving. The difference be- ance in mean speed.With our DHRs, the cross- tween the dive pattern from night and night-and- sectionalarea (4 15 mm*) is about 3% of the cross- day trips suggests that they rarely dive below 20 sectional area of a male Macaroni Penguin (ca. m at night and that the 36% of dives terminating 14,000 mm2). This would represent a 9% reduc- below 20 m are probably made during the day. tion in travelling speed. For the larger Gentoo This would be consistent with Macaroni Pen- Penguins speed reduction would only be about guins exploiting the natural vertical migration of 5 to 6%. The devices that Wilson et al. (1986) Antarctic krill (virtually their exclusive prey at used in their experiments, however, were not South Georgia), which rise towards the surface streamlined. DHRs are tapered anteriorly, which at night and sink during the day (Kalinowski and should help reduce drag. Wilson et al. (1986) Witek 1980, Everson 1982). The same suggestion found no significant difference in the energy con- was made to account for the very similar die1 sumption of JackassPenguins with and without changesin diving patterns (but based on contin- devicesbut noted that the mass of prey delivered uous records) of Antarctic fur seals (Avctoceph- 162 J. P. CROXALL, R. W. DAVIS AND M. J. O’CONNELL

TABLE 3. Comparisons between penguins with dive recorders (DHR) and those without (controls). Values are K ? SD, with sample size in parentheses.

Gentoo Penguin Macaroni Pengum DHR Control DHR Control

Foraging trip duration (hr) 11.0 + 1.5 (9) 9.8 4 2.7 (128) 12.6 ~fr0.9 (3) 11.8 * 2.1 (130) 40.0 + 28.1 (3) 28.2 + 5.5 (52) Stomach content mass (g) 801 +- 251 (10) 878 +- 262 (40) 408 + 210 (6) 448 +- 259 (40) Diet compositiom (% by wt) K F K F Krill 92 15 89.8 13.1 95 93.5 Fish 8 85 10.2 86.9 : 5.0 Other 0.1 n (6) (4) (28) (6) a K: samplescontaining mainly krill; P: samplescontaining mainly fish alus gazella) from Bird Island, South Georgia birds were foraging only during the day, and krill (Croxall et al. 1985a). tends to be away from the surface at this time, For Gentoo Penguinsthe rangeof diving depths it is not surprising that most dives (63.2%) ter- recordedhere encompassesthe few previous data. minate below 28 m. If anything, our data will Thus Conroy and Twelves (1972) captured a underestimate the proportion of feeding dives to Gentoo Penguin at , South Orkney greater depths, because any travelling dives are Islands at 100 m in a trammel net. Adams and certain to be shallower than 28 m. Brown (1983) used capillary tubes to record the Gentoo Penguins taking fish usually also have maximum depth attained over one or more for- some krill present. If catching these krill (and aging trips at Marion Island. Of 19 birds, 16 did any travelling dives) accounts for most of the not exceed20 m, two attained 40 m, and one bird 41% of dives shallower than 54 m (and 77% of exceeded70 m. This is rather different from our dives by krill-catching birds do not exceed this results where all exceeded20 m and five, eight, depth) then fish are probably mainly caught on and one bird respectively exceeded54 m, 8 1 m, the dives terminating between 54 and 109 m. and 109 m. We cannot explain why Marion Is- The main fish prey in both years were juvenile land birds appear to make shallower dives than Antarctic cod (Notothenia rossiiand N. neglecta) South Georgia ones; there is no obvious differ- andjuvenile ice fish (Champsocephalusgunnari), ence in the inshore water depths around the is- in roughly equal proportions. Such fish are usu- lands, the 100-m contour generally being within ally benthic, sometimes benthic-demersal, and 1 km or so of both sites. live mainly in the seaweed (kelp) beds fringing Gentoo Penguins which mainly catchkrill tend the coast (Kock 1981, Burchett et al. 1983). We to have rather few other prey in their stomachs do not know the depths of water in which these (Table 3). Becauseall but two ofthe experimental fish typically live around Bird Island. The steeply

TABLE 4. Daily food and prey consumption of Gentoo and Macaroni penguinsat South Georgia.

Daily requirements FMR’ Energy Food (9)’ Species Acuvity Food (W.kg-? Wb Adult Chick Total Gentoo Chick rearing Q-ill 7.4 3,708 1,008 + 880 = 1,888 Fish 7.4 3,708 976 + 880 = 1,856 Macaroni Chick rearing’ Krill 12.1 4,077 1,108 + 450 = 1,558 Macaroni Premolt Izrill 20.1 7,294 1,982 1 Field metabolicrate, from Davis, Croxall, and O’Connell, unpubl. bCalonfic c”“tent of krill = 4.6 kJ g-’ W.W.(Clarke 1984); for fish = 4.75 kJ g-’ (averageof 4.08 k.Jg- ’ for Nolothenrane&cta (Crawford 1972) and 5.4 kJ g-’ for N. magellanica(Hecht and Cooper 1986). r Assimilation efficiency0.75 (Davis, unpubl. data);food for chicksbased on meanweight of stomachcontents brought ashore by adults I” 1985 to 1986. * Krill averaged52.4 mm long in 1986; 37.0 mm long in 1985; fish averageweight estimated (from otolith data) as 74 g. r Mean with rangein parentheses. ‘Average of the five birds that madetrips lasting ca. I2 hr and ca. 24 hr. DIVING AND DIET IN GENT00 AND MACARONI PENGUINS 163 shelving topography and proximity to shore of same site, where foraging trips averaging 11.7 the 100-m depth contour are consistent with the and 13.1 hr (as at Bird Island) were reported in penguin dive data, given the habits of the fish two different seasonsby Trivelpiece et al. (1987). they are exploiting. Some of these fish probably These and other differences cannot readily be also make vertical migrations, tracking the explained but sample sizes are small and there planktonic elements of their diet, otherwise dif- are considerable topographic and hydrographic ferences in dive profile between krill and fish- differences between the shallow, land-locked eating birds would probably be accentuated. waters of Admiralty Bay at 62”s and the rela- tively steeplyshelving, open seasurrounding Bird DIVE FREQUENCY Island at 54”s. Overall the Gentoo Penguins averaged 103 dives per day (range = 57-169, 12 = 14) and 10.9 PREY CONSUMPTION AND CAPTURE RATES (range = 8.8-13.1, y1= 9) dives per hour at sea. We can make a rough estimate of potential prey The three birds radio-tracked by Trivelpiece et capture rates by combining estimatesof the daily al. (1986) at Admiralty Bay, Ring George Island, food requirement of adults (and their chickswhere , were estimated to have appropriate) at the relevant stageof the breeding averaged 193 (range = 157-231) dives per trip cycle with information on the averagesize of prey and 31.7 (range = 28.5-33.7) dives per hour at captured. The results (Table 4) indicate that the sea, on trips lasting 6.1 hr (range = 5.5-7.1 hr). food requirements of the various categories of Some of the differences between these two sets birds that we used are not greatly different of results may be attributable to the different (range = 1,558-1,982 g). The number of prey methods and locations of the two studies.We set that this representsand hence the average num- our DHR upper threshold at a pressure corre- ber of prey that would need to be caughtper dive, spondingto 2 m depth in to avoid counting dependscritically on prey size. Thus when Gen- submersions during porpoising, which might at- too Penguins take fish they need to catch only tain 1 m depth, as dives. We would, therefore, about 25 to satisfy their daily requirement and have missed any very shallow dives, but it is this could be attained by capturing one fish on inconceivable that these could have represented one-third of all dives. When feeding on krill, nearly half the total number of dives. The for- however, several individuals, on average, have aging trips of Gentoo Penguins in Admiralty Bay to be caughton every dive. Furthermore, changes during the Trivelpiece et al. (1986) study aver- in krill standard length are accompanied by sub- aged only 60% of the duration of those at Bird stantial changesin body mass. Thus the 15-mm Island, which, given the larger number of dives reduction in the mean length of krill caught by by the Admiralty Bay birds, accounts for the penguins in 1985 and 1986 is accompanied by nearly three-fold higher dive frequency. The for- a change in mean mass from 1.18 to 0.36 g and aging trip durations of the three experimental a consequent tripling of the number of prey re- birds, however, were very short compared with quired. Macaroni Penguins at South Georgia oc- other data recorded for the same speciesat the casionally take significant quantities of Thysan- oessamacrura, a euphausiid crustaceanof mean length 20 mm and of the amphipod Them&o TABLE 4. Extended. gaudichaudii of similar size. Such individuals weigh only about 0.1 g and Macaroni Penguins Daily requirements would therefore need to catch ca. 15,000 prey No. of preyd Prey per dive’ per day during chick rearing and 20,000 per day 1986 1985 1986 1985 during the premolt period. This is equivalent to 1,600 5,244 15 (9-23) 49 (3 l-76) 150 to 200 individuals per dive. 25 0.3 (0.2-0.4) All these very rough calculations, in addition 1,320 3,667 6 (5-7) 17 (14-19) 1,680 5,506 15 51 to assuming that prey are caught on all dives, also assume that all the dives we recorded were feeding dives. Trivelpiece et al.‘s (1986) analysis of penguin foraging distinguished three types of dives (as representedby periods of lack of radio signal, presumably when the penguins were be- 164 J. P. CROXALL, R. W. DAVIS AND M. J. O’CONNELL low the water surface)according to the length of 1982) and Chinstrap Penguins (which eat only time the birds spend below the surface(or, more krill) at Signy Island, precisely, were unable to transmit a receivable (Lishman and Croxall 1983) and can be con- signal) and the interval before the next dive. Un- structed from the continuous records of diving derwater swimming, consisting of 50-set dives activity for female Antarctic fur seals(eating only followed by 12-set surfaceperiods, was taken to krill) at South Georgia (Croxall et al. 1985a, be the main travelling mode and distinguished Kooyman et al. 1986). These are summarized from searching behavior (alternating periods of (Fig. 3) so as to allow comparison with the Gen- long dives with one or more short ones) and too and Macaroni penguin data. This shows a stationary diving (a consistentratio of subsurface broadly consistentpattern among the krill-eating to surface periods, the bird usually moving in a species,with shallower dives at night than during restricted area). If South Georgia birds behave the day and, overall, shallower dives than when like Admiralty Bay birds, then we would have fish prey are sought. The stands recorded the subsurface periods during under- apart by virtue of its very deep dives but the fact water swimming as dives. Such dives comprise that halfthe dives of this 12-kg bird are shallower 44% of all dives in Trivelpiece et al’s (1986) than 50 m is perhaps equally noteworthy. Its study. Therefore only just over half the dives we main squid prey at South Georgia is Martialia recorded might represent opportunities for prey hyadesi(formerly identified as Todarodes;Crox- capture. This would nearly double the required all and Lishman 1987) which, although regarded prey capture rates given in Table 3, with mean as a mesopelagicspecies, is commonly caught at values of 27 and 87 per dive for chick-rearing the surface by albatrosses(Prince 1980, Prince Gentoo Penguins and 11 and 29 and 28 and 90 and Morgan 1987). Fish may be more important per dive for chick-rearing and premolt Macaroni in its diet than squid and recent studies in the Penguins respectively. If the average dive for (Adams and Klages 1987) showed Gentoo Penguins lasts 128 set (Trivelpiece et al. that lantern fish predominate. The species in- 1986) then one krill needs to be caught on av- volved (of the genera Kreflichthys and Elec- erage each 1.5 to 5.0 set during each dive. trona) are essentially mesopelagic but occur at The limited data from watching penguins feed the surface at night. The long, 4- to 6-day for- in shallow water and in aquaria suggestthat they aging trips of King Penguins presumably result typically catch prey individually and ingest small in a dive frequency distribution comprising shal- prey with no visible pause in their swimming low dives at night and deeper daytime ones. (Croxall, pers. observ.). Providing, therefore,that The Gentoo Penguin is the only speciesstudied krill are sufficiently aggregated,such prey capture which has a mixed diet involving fish and krill. rates should not be difficult to achieve. For Mac- We do not know why individual birds tend to aroni Penguins feeding on Thysanoessa and catch either krill or fish, rather than a mixture. Them&o however, prey capture rates of two to The habit is not confined to South Georgia, be- three per second would be necessary.Macaroni causeof 46 birds examined at the South Shetland Penguin bills, however, are a very different shape Islands, 76% contained only krill, 15% only fish, from Gentoo Penguins, being much shorter, and the rest a mixture (Jablonski 1985). This broader, and deeper and it is possible that this study also showed that the same individual did facilitates capture of several small prey at a time. not catch the same type of prey in successive It is notable that while some Macaroni Penguin seasonsbut some degreeof specialization within stomachs may contain large numbers of Thy- a season cannot be ruled out. Data on overall sanoessaor Themisto, such prey are rarely ob- diet composition (Croxall and Prince 1980a, served in Gentoo Penguin stomachs and never Volkman et al. 1980, Jablonski 1985) all indicate in large numbers. that krill are the main prey (at least in the ), the 32% contribution of fish at South Geor- COMPARISON WITH OTHER gia being the highestrecorded. Therefore, despite ANTARCTIC SPECIES fish being theoretically more economical prey Dive depth histograms, obtained from the same (larger than krill, fewer need to be caught), the type of DHR, are available for King Penguins greaterabundance of krill and its swarming habit (which eat fish and squid but not krill or other presumably result in its being more often en- crustaceans)at South Georgia (Kooyman et al. countered and more easily caught than the DIVING AND DIET IN GENT00 AND MACARONI PENGUINS 165

Macaroni Antarctic Gentoo Penguin Fur Seal King Night Night & Day Night Day Penguin Krill Fish Penguin

~~~~~~~~~~~~~~

” 0 50 100 0 50 100 OS0 0 0 0 00050

Percentage of dives FIGURE 3. Dive depth histograms for Macaroni, Gentoo, King, and Chinstrap penguins and Antarctic fur seals.Sample sizes refer to number of dives; trip durations are mean values.

scarcer, deeper-dwelling fish. It may be that only from the South Orkney islands. Bull. Br. Antarct. when krill are not readily encountered do birds surv. 30:106-108. CRAWFORD,R. E. 1972. Effect of starvation and ex- make deeper dives to look for fish. These also perimental feeding on the proximate composition seem mainly to be exploited when in shoals, be- and calorific content of an Antarctic teleost No- cause Notothenia and Champsocephalus are rare- totheniacoriiceps neglecta. Comp. Biochem. Phys- ly found mixed together in Gentoo Penguin iol. A. Como. Phvsiol. 62:321-326. CROXALL,J. P., L EVE&ON, G. L. KOOYMAN,C. RICK- stomachs. ETTS,AND R. W. DAVIS. 1985a. Fur seal diving behaviour in relation to krill vertical distribution. ACKNOWLEDGMENTS J. Anim. Ecol. 54:1-8. CROXALL.J. P.. AND G. S. LISHMAN. 1987. The food We thank the members of the Bird Island station for and feeding ecology of penguins, p. 101-133. In their help in the field, C. S. Harcourt, H. J. Hill, and J. P. Croxall [ed.], :feeding ecology and A. W. North for assistancewith identification, mea- role in marine ecosystems.Cambridge Univ. Press, surement and interpretation of krill and fish material Cambridge. and two anonymous referees for their helpful com- CROXALL,J. P., T. S. MCCANN, P. A. PRINCE,AND P. ments on the manuscript. ROTHERY. In press. Reproductive performance of seabirdsand sealsat South Georgia and Sinnv LITERATURE CITED Island, South Orkney Islands, 197611986:imph- ADAMS, N. J., AND C. R. BROWN. 1983. Diving depths cationsfor SouthernOcean monitoring studies.In of the Gentoo Penguin PygoscelisPapua. Condor D. Sahrhage[ed.], Antarctic Ocean and resources 85:503-504. variability. Springer-Verlag,Berlin. ADAMS, N. J., AND N. T. KLAGES. 1987. Seasonal CROXALL,J. P., AND P. A. PRINCE. 1980a. The food variation in the diet of the King Penguin Apte- of the Gentoo PenguinPygoscelis Papua and Mac- nodvtespatagonicus at sub-Antarctic Marion Is- aroni Penguin Eudytes chrysolophusat South land. J. %oolr(Lond.) 2 12:313-334. Georgia. Ibis 122:245-253. BURCHETT.M. S.. P. J. SAYERS.A. W. NORTH,AND M. CROXALL.J. P, ANDP. A. PRINCE. 1980b. Food, feed- G. WHITE. 1983. Some biological aspectsof the ing ecologyand ecologicalsegregation of seabirds nearshorefish populationsat South Georgia. Bull. at South Georgia. Biol. J. Linn. Sot. 14:103-l 3 1. Br. Antarct. Surv. 59~63-74. CROXALL,J. P., ANDP. A. PRINCE. 1987. Seabirdsas CLARKE,A. 1984. Lipid content and composition of predators on marine resources,especially krill, at Antarctic Krill Euphausiasuperba Dana. J. Crus- South Georgia, p. 347-368. In J. P. Croxall [ed.], tacean Biol. 4, Spec. No. 1:283-292. Seabirds:feeding ecologyand role in marine eco- CONROY,J.W.H., AND E. L. TWELVES.1972. Diving systems.Cambridge Univ. Press, Cambridge. depths of the Gentoo Penguinand Blue-eyed Shag CROXALL,J. P., P. A. PRINCE, AND C. RICKETTS. 166 J. P. CROXALL, R. W. DAVIS AND M. J. O’CONNELL

1985b. Relationshipsbetween prey life-cyclesand 1983. Two recently developedrecorders for mon- the extent, nature and timing of seal and itoring diving activity of marine birds and mam- in the Scotia Sea, p. 5 16-533. In W. R. mals, p. 197-214. In A. G. MacDonald and I. G. Siegfried, P. R. Condy, and R. M. Laws [eds.], Priede [eds.], Experimental biology at sea. Aca- Antarctic nutrient cyclesand food webs. Springer- demic Press,New York. Verlag, Berlin. KOOYMAN,G. L., AND R. W. DAVIS. 1987. Diving CROXALL,J. P., C. RICKETTS,AND P. A. PRINCE. 1984. behaviourand performance,with specialreference The impact of seabirds on marine resources,es- to penguins,p. 63-75. In J. P. Croxall [ed.], Sea- pecially krill, at South Georgia, p. 285-3 18. In G. birds: feeding ecology and role in marine ecosys- C. Whittow and H. Rahn [eds.],Seabird energetics. tems. Cambridge Univ. Press, Cambridge. Plenum Publishing Corporation, New York. KOOYMAN,G. L., R. W. DAVIS, AND J. P. CROXALL. DAVIS, R. W., G. L. KOOYMAN,AND J. P. CROXALL. 1986. Diving behaviour of the 1983. Water flux and estimated metabolism of Arctocephalusgazella,p. 115-125. In R. L. Gentry free-ranging Gentoo and Macaroni penguins at and G. L. Kooyman [eds.], Fur seals: maternal South Georgia. Polar Biol. 2:41-46. strategieson land and at sea.Princeton Univ. Press, EVERSON,I. 1982. Diurnal variations in mean vol- NJ. ume backscatteringstrength of an Antarctic krill KOOYMAN,G. L., R. W. DAVIS, J. P. CROXALL,AND Euphausiasuperba patch. J. Plankt. Res. 4: 155- D. P. COSTA. 1982. Diving depths and energy 162. requirementsof King Penguins.Science 2 17:726- HECHT,T. 1987. A guide to the otoliths of southern 727. ocean . S. Afr. J. Antarct. Res. 17:l-87. LISHMAN,G. S., AND J. P. CROXALL. 1983. Diving HECHT,T., AND J. COOPER. 1986. Length/mass re- depths of the Chinstrap Penguin Pygoscelisant- lationships, energetic content and the otoliths of arctica. Bull. Br. Antarct. Surv. 61:21-25. Antarctic cod Paranotothenia manellanica (No- NACHTIGALL,W., ANDD. BILO. 1980. Stromungsan- totheniidae: Pisces) at sub-Antarctic Marion Is- passung des Pinguins beim Schwimmen unter land. S. Afr. J. Zool. 21:294-296. Wasser. J. Camp.. Physiol. Al 37: 17-26. JABLONSKI,B. 1985. The diet of penguins on King PRINCE.P. A. 1980. The food and feedine ecoloev of GeorgeIsland, South ShetlandIslands. Acta Zool. Grey-headedAlbatross Diomedea chry&tom&d Cracov. 29: 117-l 86. Black-browedAlbatross D. melanophris.Ibis 122: KALINOWSKI,J., AND Z. WITEK. 1980. Diurnal ver- 476-488. tical distribution of krill aggregationsin the west- PRINCE,P. A., AND R. A. MORGAN. 1987. Diet and em Antarctic. Pal. Polar Res. 1:127-146. feeding ecology of Procellariiformes, p. 135-l 7 1. KOCK, K-H. 1981. FischereibiologischeUntersuch- In J. P. Croxall _led.1. _, Seabirds: feedine ecoloev-. ungen an drei Antarktischen Fischarten: Champ- and role in marine ecosystems.Cambridge Univ. socephalusgunnari Lijnnberg, 1904, Chaeno- Press, Cambridge. cephalus aceratus (LGnnberg, 1905) und TRIVELPIECE,W. Z., J. L. BENGTSON,S. G. TRIVELPIECE, Pseudochaenichthysgeorgianus Norman, 1937 AND N. J. VOLICMAN.1986. Foraging behaviour (Notothenioidei, Channichthyidae). Mitt. Inst. of Gentoo and Chinstrap penguinsas determined Seefisch.32: l-226. bv new radiotelemetrv techniaues.Auk 103:777- KOOYMAN,G. L., J. 0. BILLUPS,AND W. D. FARWELL.

APPENDIX 1. Number of dives terminating in each depth stratum, diet, and trip duration of Gentoo and Macaroni penguinsat South Georgia.

Gent00 Penguin

Year 85 85 85 85 85 85 85 85 86 86 86 86 86 86 Sex ---- - _ - - M F F M M M Trip duration 6r) - - - 9.7 11.2 Dieta (F) (F) F F (K, &) 8;(” 1;‘6 13.1F 11.2F 10i 8 12i 1 g4 K Depth (m) 2-28 15 23 5 1 48 20 14 19 18 15 49 55 70 29-54 23 5 13 27 65 49 25 40 47 46 52 40 41 :: 55-8 1 22 11 11 11 38 38 43 10 54 60 13 18 9 6 82-109 51 43 28 24 18 4 5 3 2 1lo-136 24 136+ Total 111 82 57 87 169 111 87 69 122 123 114 113 120 79 * F: fish; K: krill: parenthesesindicate diet assumed(see text). DIVING AND DIET IN GENT00 AND MACARONI PENGUINS 167

TRIVELPIECE,W. Z., S. G. TRIVELPIECE,AND N. J. VOLK- WILSON,R. P. 1985. The JackassPenguin @hen&us MAN. 1987. Ecological segregation of Adelie, demersusas a pelagic predator. Mar. Ecol. Progr. Gentoo, and Chinstrap penguinsat King George Ser. 25:219-227. Island, . Ecology 68:35 l-36 1. WILSON, R. P., W. S. GRANT,AND D. C. DUFFY. 1986. VOLKMAN,N. J., P. PRESLER,AND W. Z. TRIVELPIECE. Recording devices on free-ranging marine ani- 1980. Diets of pygoscelidpenguins at King George mals: does measurement affect foraging perfor- Island, Antarctica. Condor 82:373-378. mance?Ecology 67:1091-1093. WILSON,R. P. 1984. An improved stomach pump for penguinsand other birds. J. Field Omithol. 55: 109-l 12.

APPENDIX 1. Extended.

MacaroniPenguin 86 86 86 86 86 86 86 86 M M M M M M M M

12.5 13.5 11.8 24.0 23.5 72.5 264 9 K K K K K K (K) W Depth (m) 2-20 200 218 190 127 117 467 611 1,934 21-40 56 60 117 446 1,079 41-60 48 57 71 129 349 61-80 12 20 21 9 12 81-100 1 1 100+ 200 218 190 243 255 677 1,195 3,374