sign in sign up
<<
Home , Macaroni penguin

MARINE ECOLOGY PROGRESS SERIES Published November 10 Mar. Ecol. Prog. Ser.

Dietary differentiation and trophic relationships in the sub-Antarctic community at Marion Island

N. J. Adams, C. R. Brown*

Percy FitzPatrick Institute of African Ornithology, University of Cape Town. Rondebosch. 7700.

ABSTRACT. We compared the diets of 4 CO-occurringspecies of penguin at sub-Antarctic Marion Island in light of mechanlsn~sthought to result in dietary differentiation. Calculation of overlap indices and correspondence analyses indicated a clear separation in the diets of the 3 penguin genera but considerable similarity between the congeneric pair The pelagic foraging patagonicus consumed mainly myctophid , whereas the near- to offshore foraging macaroni penguin Eudyptes chrysolophus and rockhopper penguin E. chrysocome both consumed predominantly small pelagic , although in different proportions. The inshore foraging papua fed largely on benthic nototheniid fish. Although king rarely take small prey, differences in diet cannot be accounted for solely on the basis of prey size selection. Different diving capabilities may have some role in dietary differentiation, however, we consider prey availability in the apparently distinct feeding zones to explain most of the differences in the diets of the 4 species of penguins at Marion Island. The daily population food requirements of the respective penguin species at the (comprising Marion and Prince Edward islands) increased w~thIncreasing species foraging range. The mainly benthic nature of the prey in the inshore area, compared with the more easily replenished pelagic food stocks, probably explains the differences in food ava11abi.lity that sustain the greater food demands of the large populations of pelagic and offshore foragers.

INTRODUCTION 1987, Adams & Klages 1987, 1989, Adams & Wilson 1987, Brown 1987, Brown & Klages 1987, Ridoux et al. The and islands of the sub- 1988). We compare here the diets of the king penguin Antarctic typically support large populations of up to 5 Aptenodytes patayenicus, macaroni penguin Eudyptes sympatrically breeding species of penguins (Wilson chrysolophus, rockhopper penguin E. chrysocome and 1983). Separation of feeding areas and breeding gentoo penguin Pygoscelis papua at Marion Island and schedules (Trivelpiece et al. 1987), differing diving evaluate factors that may be important in causing capabilities (Trivelpiece et al. 1986) and selection for differences in diet. different prey size (Croxall & Lishman 1987) may be The breeding schedules (Fig. 1) of the species differ important factors in the ecological segregation of sym- markedly. In brief, king penguins take about 14 mo to patnc penguin species. raise a chick, with egg laying occurring from Recent studies at sub-Antarctic Marion Island November to April (N.J.A. pers, obs). Gentoo penguins (46"52'S, 37'51'E) and (46"S,51°E) in are resident at Marion Island throughout the year but the southern have provided descriptions have a more restricted breeding season, with egg lay- of the diets of the 4 different species of breeding pen- ing normally starting at the end of June. However, guins which occur at these localities, and some prelimi- different colonies show considerable asynchrony and nary estimates of various foraging parameters (Adams failed breeders will relay (Williams 1980a). Most chicks fledge by mid-November. In contrast, breeding Present address: Department of Zoology and Entomology, activities and moult of macaroni and rockhopper pen- Rhodes University, PO Box 94, Grahamstown 6140, South guins are highly synchronized and restricted to the Africa months October to May with a separation of some 3 wk

Q Inter-Research/Printed in F. R Germany 250 Mar. Ecol. Prog. Ser. 57: 249-258, 1989

King Macerani 6 100 100

Benthic crustaceans 60 60 Pelagic crustaceans

Demersal fish 20 20 2 Myctophid fish - Gentoo X .-.E* l00 100 0m r Chick rearing Q period 5 0 60 60 .ri Summer resident 8 only Fig. 1. Temporal variation in the 20 20 proportion of the main prey classes in the diet of penguins at AJAODF D F Marion Island (redrawn from data l presented in Adams & Klages I* l*' 1987. 1989 and Brown & Klages Months Months 1987) between initiation of breeding in the 2 species (Wil- crustaceans from intact individuals. Proportions by liams 1980b). Both species disperse from Marion Island numbers and within class comparisons of well-digested during winter (Williams 1980b). During November to prey rely on analyses of these hard parts. Unidentified March all 4 penguin species are present at Manon prey was apportioned in the same ratio as identified Island and feed in surrounding waters. prey. Overlap in diet between penguin species was as- sessed using Morisita's Index (Monsita 1959) modified METHODS by Horn (1966) (see Diamond 1983). This index, which varies between 1 (complete overlap) and 0 (no overlap), All diet samples were collected along a 9 km stretch is a relative measure and its value is dependent on the of the east coast of Marion Island. Because there may number of prey categories selected. We did not split be considerable interannual variation in diets of pen- categories into size classes (cf. Diamond 1983) since guins at Marion Island (Brown & Klages 1987), we major taxonomic classes were generally characterized concentrated here on comparisons of diet between the by a specific range of lengths ( > fish > crusta- species based on a comprehensive set of food samples cean). collected throughout 1 yr only (April 1984 to March Dietary diversity between and within penguin 1985). We extracted the proportions of prey identified species was compared using the Shannon index to the lowest possible taxon by both actual and recon- (Tramer 1969). Indices were calcuIated both for pro- stituted mass and number from data of Adams & Klages portions of prey by mass for lumped samples of the 4 (1987, 1989), and Brown & Klages (1987) (see Appendix species (Appendix l), and as a mean of diversity indi- l), to perform correspondence analyses (Underhill ces calculated for individual samples. We present indi- 1981) and calculate overlap indices. Diet collection and analysis procedures are given in detail in Adams & Table 1. Diversity index of diets of penguins at Marion Island Klages (1987) and in Brown & Klages (1987). Briefly, Values are given for lumped samples (H',,, calculated from samples were obtained from adults arriving ashore Appendix 1) and mean of each individual sample (H'lnd)for after foraging or, on occasions, from large chicks the whole year and Decernher to March immediately after being fed (king penguins only) by stomach flushing (Wilson 1984). Samples were sorted Species Whole year December-March within 24 h of collection. Individual samples were sepa- H to, H'md H'tol H'lnd rated into fish, and remains and component parts weighed. Compansons between rela- King penguin 0.570 0.296 0 537 0.337 Gentoo penguin 0.753 0.116 0.765 0 198 tive mass of major prey classes are based on pro- Macaroni penguin 0.810 0.345 portions of recovered wet mass. Fish species were iden- Rockhopper penguin 0518 0.256 tified from otoliths, cephalopods from lower , and Adams & Brown: Trophic relationships anlong penguins 25 1

ces calculated for the whole year and, where appropri- (Table 2). In contrast, there is extensive overlap ate, for samples recovered from December to March between gentoo and rockhopper penguins, and only (the period when all species of penguins are pre- between macaroni and rockhopper penguins. Corre- sent at the island) (Table 1). spondence analysis illustrates the clear separation of Measurements of potential foraging ranges are king penguins from the other 3 species and identifies described in Adams (1987), Adams & Wilson (1987) and myctophid fish and squid as the prey species respon- Brown (1987) and were obtained d.uring the same year sible (Fig. 2). Cructaceans common to gentoo, macaroni as intensive diet sampling. and rockhopper penguins account for the high overlap in diet between these species. Comparison of indices based on percentage mass RESULTS increases the overlap between king penguins and other species, but reduces that between gentoo pen- There were seasonal changes in the proportions (per- gums and macaroni and rockhopper penguins (Table 3). cent wet mass recovered) of major prey types of pen- guins at Marion Island (Fig. 1). In absolute terms, king Table 2. Overlap In the diet of penguins at Marion Island. Prey penguins consumed the largest number of taxa categories selected are as given in Appendix 1 for proportion by numbers. GP, MP, RP: gentoo, macaroni and rockhopper (Appendix 1). However, the diversity index (Table 1) penguins, respectively gives little weight to taxa occurring in small pro- portions and macaroni penguins were consequently identified by the Shannon index as having the most diverse diet, whether indices were calculated from King penguin 0.001' 0.008 0.031 0.003 lumped samples or from the mean of individual sam- Gentoo penguin 0.780 0.943 ples (Table 1). Lumped sample diversity was, not sur- Macaroni penguin 0.920 prisingly, always greater than the mean of individual Comparison based on samples collected over the whole diversity. Little overlap is indicated between the diet of year, remaining comparisons based on samples collected lung penguins and gentoo, macaroni or rockhopper from December to March only penguins when compared on the basis of numbers

Axis 1

F R m .H M I K Axis 2

Fig. 2. Correspondence analysis of the diets of penguins at Marion Island. Penguin species with a large proportion of prey species in common plot close together. Prey consumed largely by one penguin species will plot far from the origin, whereas shared items will lie between the common predators. Data for analysis were percentage numbers from December to March only. 1: king penguin; 2: gentoo penguin; 3: macaroni penguin; 4: rockhopper penguin. A: and ; B: Electrona carlsbergi; C: myctophid A; D: Protornyctophum normani; E: Protomyctophun~bolini; F: Gymnoscopelus nicholsi; G: Electrona subaspera; H: Notothenia squamifrons; 1: Notothenia acuta; J: Paranotothenia magellanica; K: Dissos- tichus eleginoides; L: Magnisudis prinosa; M: Channichthys rhinoceratus; N: Kondakovia longimana; 0: oegopsid A; P: Histioteuthissp.; Q: Brachioteuthissp.; R: octopod A; S: Euphausia vallentini; T- Thyssanoessa vicina; U: Nauticaris marionis; V: Themisto gaudichaudi; W: Prjmno sp.; X: Vibilia sp.; Y. Cyllopus sp., Z: Hyperialla sp. Mytophid A, oegopsid A and octopod A are unidentified prey species 252 Mar Ecol. Prog Ser. 57 249-258, 1989

Table 3. Overlap in the diet of penguins at Marion Island. Prey within the major prey classes of fish and cephalopods categones selected are as given in Appendix 1 for proportion (Appendix 1) are based on analyses of otoliths and by mass cephalopod beaks, respectively (Appendix 1). Large cephalopod beaks may remain in the stomachs of G P G P MP RP predators for many weeks and relative proportions be overesbmated (Furness et al. 1984, Jackson & Ryan King penguin 0.014' 0.045 0.183 0.059 Gentoo penguin 0.293 0.312 1986). However large cephalopods, mainly in king pen- Macaroni penguin 0.833 guin diets, form a very small proportion of the diet particularly in summer when all 4 species are on the Comparisons as for Table 2 island (Appendix 1). Most cephalopod beaks in maca- roni, rockhopper and gentoo penguin diets were small These changes can be attributed to the increased (lower rostra1 length < 2 mm) and recovered from undi- importance of fish when diet is analysed on the basis of gested buccal masses. Residence time of otoliths in proportional mass. Increased separation between the stomachs ranged up to 30 h (Jackson & Ryan diets of gentoo penguins and macaroni and rockhopper 1986, Jackson & Cooper 1988). Crustacean exoskele- penguins is again reflected in correspondence analysis tons may remain for longer (Jackson & Cooper 1988). (Fig. 3), which identifies nototheniid fish as being the Individual stomach samples almost certainly represent critical factor. Overlap between the congeneric maca- prey captured during one foraging trip only. That at roni and rockhopper penguins remains high in both least 80 % (by number) of individual diet samples for comparisons. each penguin species was accounted for by only 4 out of a possible 35 prey taxa (Fig. 4) further reduces biases that confound interspecific comparisons. Conse- DISCUSSION quently, we considered interspecific comparisons of diet composition largely justified. There are inherent biases in diet analyses based on Diets of Antarctic and sub-Antarctic penguins have examination of stomach contents, particularly for the been perceived as being dominated by a small number well-digested meals of king penguins, because soft and of prey species (but see Ainley et al. 1984). Therefore hard parts of different prey are digested at different we should expect high dietary overlap, particularly rates. Crustaceans, protected by a chitin , between taxonomically closely related species with are retained in seabird stomachs longer than squid similar foraging methods. Although this may be true of flesh which in turn is retained longer than fish flesh penguins and other seabird communities at particular (Wilson et al. 1985, Jackson & Cooper 1988). Gross localities (e, g. around the ; Croxall et al. analyses of soft parts of prey (Fig. 1) will overestimate 1985),it is not the case at Marion Island and some other such prey in the crustaceans > squid > fish. sites in the Southern Ocean reglon (Anley et al. 1984). Relative proportions by numbers and mass of species Average dietary overlap indices of the sub-Antarctic

Axis 1

R

F "..K Fig. 3. Correspondence analysis ot the I diets of pengulns at Marion Island. Data M for the analysis was percentage mass for December to hlarch only. Species codes Axis 2 as for Fig. 2 Adams & Brown: Trophic relationships among pengulns

Diversity of prey taken by foraging penguins appears Gentoo Penguin Rockhopper Penguin to be a function both their size and foraging range. The intermediate sized macaroni and gentoo penguins are capable of taking both relatively large and small prey and have a high prey diversity compared to king and rockhopper penguins, when the diet of the sample population is considered as a whole (Table 1). The low diversity indices of gentoo penguins when diet is consi- dered on an individual sample basis (Table 1) reflects their short nest relief periods with reduced time to - encounter different prey.

Differentiation mechanisms

Prey size King Penguin Macaroni Penguin Although penguins at Marion Island are capable of consuming prey ranging over 2 to 4 orders of mag- nitude in mass (Table S),items are generally small (average length < 100 mm in all cases; Table 6) com- pared to those consumed by sympatric albatrosses of similar mass to penguins (Berruti & Harcus 1978, Brooke & Klages 1986). Minimum prey size of penguins at Marion Island was similar. However, prey of less than 30 mm accounted for only 6 % (by number) of the diet of the large king penguin but comprised a substan-

Table 5. Maximum mass of prey items (all squid) recovered from penguin stomach samples at Marion Island. All penguins 1234567 12345 were capable of taking prey of 0.1 g, the minimum size we NUMBER OF PREY SPECIES estimated Fig. 4. Number of prey species in individual stomach samples as a percentage of total number of samples Species Max. mass (g)

Kinq-- penquin - 807 Gentoo penguin penguin community and some Antarctic seabird com- Macaroni penguin munities, on a prey basis at least, are lower than Rockhopper-- -penquin - or equal to some tropical communities (Table 4).

Table 4. Average overlap indices of diets within seabird communities. Analysis categories are prey family by percent numbers and mass as indicated

Overlap index No. seabird Locality Source

1 No, Mass species

Aldabra Atoll, tropical Diamond 1983 Cousin Island, tropical Diamond 1983 Christmas Island, tropical Diamond 1983 Marion Island, sub-Antarctic This study Farne Island, temperate Diamond 1983 Oceanic, Antarctica Ainley et al. 1984 Continental slope, Antarctica Alnley et al. 1984 Continental shelf, Antarctica Ainley et al. 1984

" Penguins only 254 Mar Ecol. Prog. Ser. 57: 249-258, 1989

Table 6 Mean length of prey species recovered from stomachs of major components in the diets of gentoo, macaroni of penguins at Marion Island Flsh: standard length; and rockhopper penguins cannot be explained by cephalopods: dorsal mantle length; crustaceans: total length selection of prey based on size. Volkman et al. (1980) also rejected the hypothesis that the admittedly simi- Mean length (mm) larly sized Pygoscelis penguins partition euphausiids KP GP MP RP by size.

FISH Diving depths Myctophidae Kreffich fh ys andersson~ Protomyctophum tenisoni If different prey characteristically occur at different Electrona carlsbergi depths, segregation of diet by penguins of different Protorn yctophum normani diving capability (Stonehouse 1967), may occur. Gen- Nototheniidae too penguins at South Georgia (54 "S. 38 "W) feeding Notothenia squamifrons diurnally on benthic fish generally dived deeper (54 to Channichthyidae 136 m) than -feeding macaroni penguins (< 80 m) Channichthys rhinoceratus (Croxall et al. 1988). Data from Marion Island is CEPHALOPODS restricted to measurements of maximum diving depth Kondakovia longimana alone which indicate dives from less than 20 m to Octopoda greater than 70 m (Adams & Brown 1983). The foraging CRUSTACEANS range of gentoo and rockhopper penguins may overlap Euphausiacae extensively during early chick rearing. The presence of Eupha usia vallentini demersal fish in the diet of gentoo penguins but Thysannoessa v~cina absence from that of smaller rockhopper penguins Natantia Nauticaris marion~s probably reflect differing diving ability. Although Croxall et al. (1988) noted that half of the dives of king Themisto gaudichaudi penguins at South Georgia were shallower than 50 m, Average prey item size the presence of mesopelagic fish and squid in the diet Total of king penguins may reflect their ability to dive deep December-March (Kooyman et al. 1982). Culmen length (mm) tial proportion of the prey of the 3 other smaller pen- Travelling speed guin species. Discrimination of prey based on some minimum size may be more important than maximum Travelling speeds of all 4 species of penguins are size in causing differentiation of diets of large and simdar (Table 7). Maximum speeds attained during small penguins at Marion Island. prey pursuit will be higher and may show significant Although ANOVA demonstrated significant differ- differences between species. However, all 4 penguin ences in the mean size of myctophid fish species com- species at Manon Island are able to catch adult mycto- mon to more than one species of penguin (e.g. for phids and juvenile squid. We consider it unlikely that Protomyctophum normani in the diet of king, macaroni any small differences in maximum swimming speeds and rockhopper penguins, F = 164.94, F0.05,3,983 = 2.61, play a significant role in dietary segregation. p < 0.005) (see also Brown & Klages 1987) no consistent relationships between mean fish and cephalopod size and predator size were evident (Table 6). Similarly, Foraging range Croxall & Lishman (1987) have cautioned against drawing conclusions about preferential selection for There are clear differences in the mean maximum specific prey size classes based on significant differ- potential ranges of gentoo, macaroni, rockhopper and ences in length recovered from conspecifics at the king penguins at Marion Island (Table 7). Samples same locality. Maximum size of squid taken was larger sizes on which estimates of foraging range were made than other prey types consumed and increased with are small and, in addition, data are variable. However, penguin size (Table 5) and is probably because of ease the duration of foraging trips of the 4 species at Marion of handling of squid compared to fish (Ashmole & Island (Table 7) and elsewhere support this pattern of Ashmole 1967). zonation (Croxall & Prince 1980, WiIliams & Siegfried Notwithstanding the absence of smaller prey from 1980, Williams 1982, Horne 1985, Adams 1987). Breed- the diet of king penguins, differences in the proportions ing gentoo penguins are restricted largely to inshore Haams & Brown. Trophic relationships among penguins 255

Table 7 Forag~ngparameters of penguins at Marion Island. Percent time swimming and estimated foraging range for adults feeding small and large chicks are presented where possible

Species Travelling Foraging duration Foraging range Source specd (a) (km) (km h-') Small Large Small Large Range - - Kng penguin 8.7 12.8 4.0 255 301 75-902 Adams (1987) Gentoo penguin 7.9 0.6 14 1-103 Adams & Wilson (1987) Macaroni penguin" 7.5 1.0-2.0 2.5-35 178 59-303 Brown (1987),Willlams (1982) Rockhopper penguinb 7 4 1.0-2.0 2.5-3 5 33 - 2-137 Brown (1987),Williams (1982) " Foraging trip durations estimated to nearest 0.5 d. Estimates of foraging range made during late chick-rearing Foraging trip durations estimated to nearest 0.5 d. Estimates made during early chick-rearing

waters (Table 7). During the remainder of the year from nearer-shore foraging during early chick rearing adults do not have to return regularly to the island to to offshore feeding when chicks are larger. The 3 to 4 feed chicks and foraging ranges may increase. The wk separation period between the breeding peaks of difference in foraging range between rockhopper and macaroni and rockhopper penguins at Marion Island is macaroni penguins reflects the difference in chick of equivalent length to the guard period of small feeding schedules of adults feeding small and large chicks. Thus the separation in timing of breeding chicks at the time estimates were made (Table 7). results in differentiation of foraging zone which is, in Foraging range of rockhopper penguins is probably turn, reflected in the diets of the 2 species. Timing of greater later in chick rearing, and that of macaroni breeding at other localities where 2 congeneric species penguins closer to shore during early chick rearing breed, e.g. adelie Pygoscelis adeliae and chinstrap P. (Brown 1987). King penguins are apparently pelagic antarctica penguins at Signy Island, are also separated throughout chick rearing. by about 4 wk, equivalent to the guard period of the We suggest this zonation accounts for most of the latter species (Lishman 1985, Trivelpiece at al. 1987). major difference in the diets of the 4 penguin species. The staggering of peak food requirements of con- Similarly, Trivelpiece at al. (1987) regarded differences generic species may have an indirect role in causing in foraging ranges of Pygoscelis species at King George dietary differences due to separation of foraging zones Island, Antarctica (62" 10' S, 58" 30') as an important (see Trivelpiece at al. 1987). We are in agreement with factor affecting trophic interactions. Ainley et al. (1984) Trivelpiece at al. (1987) and regard specific differences have demonstrated that diets of Antarctic seabird in diet as reflecting differences in foraging ranges, life species may be different depending on whether history patterns and temporal differences in food were sampled from oceanic, continental slope or conti- availability rather than competitive interactions nental shelf waters, presumably reflecting availability between penguin species. of different prey in these areas. The absence of particu- lar prey species in the diet of one species of penguin at Marion Island but occurrence in others (Fig. 1, Appen- Foraging range, population size and daily population dix 1) suggests foraging zones of the different penguin food requirement species may, to some extent, be nlutually exclusive. Substantial numbers of macaroni and king penguins Ashmole (1963) and Diamond (1978) suggested that radiating away from large colonies (essentially a point population sizes of breeding aerial at oceanic source) will be at relatively high densities in inshore islands are limited by the availability of food. If the waters and potential local depletion of prey resources amount of food is a function of feeding area for surface- may be too great to make foraging profitable. In con- feeding seabirds (or volume for penguins) then pelagic trast, the dispersion of relatively small colonies of rock- feeders should be more numerous than inshore feeders hopper (pers. obs.) and gentoo penguins (Adams & and migrant seabirds more abundant than resident Wilson 1987) along the coastline of Marion Island may species (Diamond 1978). Migration to alternative feed- reduce intraspecific competition in inshore and near- ing areas outside those utilized during breeding shore waters (Croxall & Prince 1980, Adams & Wilson increases effective feeding area and has a more signifi- 1987). cant effect on population size than pelagic feeding Changes in the diets of macaroni and rockhopper (Diamond 1978). Although, as yet, no food limitation penguins over chick-rearing appear to reflect a change has been demonstrated for breeding seabirds at Marion 256 Mar Ecol. Prog. Ser. 57: 249-258, 1989

Table 8 Estimated breeding populat~onsize (pairs) and daily food requi.rements of penguins at the Prince Edward Islands (hiarion and Pr~nceEdward)

Species Mean Feeding Population size Total Total daily food mass rate" Marion Prince Edward biomass requirements (kg) (g d-'1 pairs palrs (kg X lo3) (kg X 10")

ffing penguin 13.1b 392 215 230' 5 000' Gentoo penguin 6.4b 237 888d 655" Macaroni penguin 4.0' 190 405 084' 17 000' Rockhopper penguin 2.5e 139 157 6OOg 3 500"

" Predicted from: Feeding rate (g d-l) = 0.495 Body Mass (g)' '04 (Eq. 37; Nagy 1987) N.J. Adams unpubl. ' Williams et al. (1979) Adams & Wilson (1987) ' S.R. Henly pers. comm. Wathns (1987) g FitzPatrick Institute unpubl.

Island, we tested these predictions on the penguin Acknowledgements. Research at Marion Island was carr~ed community at the Prince Edward Islands, Manon Island Out under the ausplces of the South African Scientific Commit- and Prince Edward Island (collectively: Prince Edward tee for Antarctic Research. The logistical and financ~alsupport of the South African Departments of Transport and Environ- Islands) are 22 km apart and we considered the ment Affairs are gratefully acknowledged. Richard Knight islands to constitute a single breeding site. Because the kindly performed and in.terpreted the correspondence anal- body mass of king penguins is some 4 times greater yses. Robert Prys-Jones provided useful comments on the than rockhopper penguins we considered daily popula- manuscl-ipt. tions food requirement, which takes account of allometric considerations, to be a more appropriate standard for comparison. There was a consistent trend LITERATURE CITED of increasing daily food demand with increased aver-. age foraging range of penguins at the Prince Edward Adams, N. J. (1987). Foraging range of king penguins Aptenodytes patagonicus dur~ng summer at .Lidnon Islands (Tables 7 and 8). The relatively small difference Island. J. Zool., Lontl. 212: 313-334 in daily food demand between macaroni and king Adams, N. J., Brown, C. R. (1983).Diving depths of the gentoo penguins (Table 8) may reflect the shorter residence penguin, Pygoscelis papua. Condor 85. 503-504 time of the former at the islands. Adams, N. J., Klages, N. T (1987). Seasonal variation in the diet of the king penguin (Aptenodytespatagonicus) at sub- Particularly dunng winter, gentoo penguins feed on Antarctic Marion Island J. Zool., Lond. 212: 475-482 benthic or demersal prey, largely juvenile nototheniid Adams, N. J., Klages, N. T (1989).Temporal variation in the fish and adult shrimp, in the relatively shallow shelf diet of the gentoo penguin @goscelis papua at sub- waters around the island itself. Diets of the other Antarct~cMarion Island Col. Waterbirds 12: in press penguin species, largely euphausiids and myctophid Adams, h'. J.. Wilson M.-P. (1987). Forag~ngparameters of Gentoo Penguins Pygoscelis papua at Marion Island. Polar fish, are mostly pelagic in origin. A pelagic food Biol. 7: 51-56 source, with new stocks moving continuously into Ainley, D. G., O'Conner. E. F., Boekelheide, R. J. (1984). The areas locally depleted by predators, can presumably marine ecology of birds in the Ross Sea, Antarctica. Ornith. support larger populations of these predators than L4onogr. 32: 1-97 Ashrnole, N. P. (1963). The regulation of numbers of tropical can more sedentary demersal organisms utilized by oceanic birds. Ibis 103: 458-473 gentoo penguins. The major current passing Mar~on Ashmole, N P,,Ashmole, hl. J. (1967). Con~pardt~vefccd~ng Island is the eastward flowing West Wind Drift. How- ecology of seahirds on a tropicdl island. Peabody 4Ius. nat. ever, components of the Weddell Drift and Agulhas Hist. Yale Univ Bull. 24: 1-131 Current System may also penetrate to Marion Island. Berrut~,A., Harcus, T. 11978). Cephalopod prey of the sooty albatrosses Phoebetria fusca and P. palpebrata at kfarion These currents provide a feasible mechanism for Island. S. Afr. J antarct. Res. 8: 99-103 introducing pelagic prey into waters surrounding Boden, B. P,, Parker, L. D (1986). Thc plankton of th? Pr~ncc Marion Island (Boden & Parker 1986). The presence Edward Islands. Polar Biol. 5: 81-93 of zooplankton elements, in net hauls and penguin Brooke, M. de L.. Klages, N T (1986). Squid beaks reyrlrgi- tated by Greyheaded and Yelloivnosed 4lbatrosses stomachs, more typical of subtropical and Antarctic Diomed~achrysostorna and D. chloroph!~~~chosat the waters (Brown 1989) suggests this is indeed occur- Prince Edward Islands. Ostrich 57. 203-206 ring. Brown. C. R. (1987) Traveling speed and foraging rdnge of Adaills & Brown Trophic relationships among penguins 257

Appendix 1. Percentage composition of food samples from penguins at Marion Island. Proportions for all samples and samples collected from December to March are given for species resident all year Myctophid A and oegopsid A are unidentified prey species. Krefftichthys anderssoni and Protomyctophum tenisoni were not separated in analysis of king penguin diets

Prey species King penguin Gentoo penguin h4acaroni Rockhopper penguin penguin Annual Summer Annual Summer No Mass No Mass No Mass No hlass No. Mass No Mass 111 #>y0 L! > Ot<, % O/o

FISH Myctophidae Krefftich thys andersson~ Protomyctophun? tenisoni Electl-ona carlsbergi Myctophld A Protomyctophum normani Protomyctophum bolini Gymnoscopelus nicholsi Electrons subaspera Notothenlidae Notothenja squamifrons Notothenia acuta Notothenia magellanica Dissostich us eleginoides Magnisudis prionosa Channichthyidae Channichthys rhinoceratus Muraenolepidae Muraenolepis sp. CEPHALOPODS Kondakovia longimana Oegopsid A Histioteuthis sp. Alluroteuthis sp. Moroteuthis sp. Martialja hyadesi Gonatus an tarcticus Galiteuthis glacialis Brachioteuthis SD. Octopoda CRUSTACEANS Euphausiacea Euphausia vallentini Thyssanoessa vicina Natantia Naut~cansmarionis Amphipoda Themisto ga udichaudi Primno spp. Vibilia spp. Cyllopus spp. Hyperialla spp.

Macaroni and Rockhopper Penguins at Marion Island. J. patterns in relation to diet of Gentoo and Macaroni pen- Field Om. 58: 118-125 guins at South Georgia. Condor 90: 157-167 Brown, C. R. (1989). Energy requirements and food consump- Croxall, J. P,, Lishman, G. S. (1987). The food and feeding tion of Eudyptes penguins at the Prince Edward Islands. ecology of penguins. In: Seabirds: feeding ecology and Antarct. Sci. 1: 15-21 role in marine ecosystems. Croxall, J. P. (ed.) Cambridge Brown, C. R.,Klages, N. T.(1987). Seasonal and annual variation Un~versityPress, Cambridge, p. 101-133 in diets of macaroni and southern Rockhopper penguins at Croxall, J. P.,Prince, P. A. (1980). Food, feeding ecology and Sub-Antarctic Marion Island. J. Zool., Lond. 212: 7-28 ecological segregation of seabirds at South Georgia. Biol. Croxall, J. P., Davis, R. W., O'Connell, M. J. (1988). Diving J. Linn. Soc. 14: 103-131 258 Mar Ecol Prog. Ser. 57: 249-258, 1989

Croxall, J. P., Prince, P. A., Ricketts, C. (19851. Relationships Trivelpicce. W. Z., Bcngston, J. L.. Trivelp~ece,S. G., Volk- between prey life-cyclesand the extent, nature anci t~mingof man. N. J.(1986). Foraging bchavior of gentoo and chin- seal and smbird preddtion in the Scotia Sea. In: Antarctic strap penguins as determined by new radiotelemctn nutrient cyclesand food wrbs. Siegfried, W. R , Condy, P. R.. techniqt~es.Auk 103: 777-781 Laws. R. h.1. (cds.) Springer-Verlag, Berlin, p 516-533 Trivelplece. LY Z.,Trivelpiece, S. G., Volkman, N. J (1987). Diamond. A. W. (1978). Population size and feeding strategies Ecologir-a1 segregat~on of adelie, gentoo and chinstrap in tropical seabirds. Am. Nat. 112: 215-223 penguins at Ing George Island, Antarctica Ecology 68: Diamond, A. W (1983). Feeding overlap in some tropical and 351-361 temperate seabird communities. Stud. avian Biol. 8: 26-46 Underhill, L. G. (1981). An ASCII FORTRAN program with Furness, B. L.,Laugksch, R. C., Duffy, D. C. (1984).Cephalopods dynamic core allocation for correspondence analysis. and studies of seabird diets. Auk 101. 619-620 Technical Report. Department of Mathematical Statistics, Home, R. S. C (1.985).Diet of royal and rockhopper penguins University of Cape Town at MacQuarie Island. Emu 85: 150-156 Volkman, N. J.,Presler, P., Trivelpiece, W. (1980). Diets of Horn, H. S. (1966). Measurement of 'overlap' in comparative pygoscelid penguins at King George Island. Antarctica. ecological studies Am. Yat. 100: 419424 Condor 82 373-378 Jackson. S., Cooper, J. (1988). Use of fibre-optic endoscopes In Watkins, B. P. (1987). Population sizes of King, Rockhopper studies of gastric digestion in carnivorous vertebrates and Macaroni Penguins and U'andering Albatrosses at the Comp. Biochem. Physiol. 91A: 305-308 Prince Edward islands and Gough Island, 1951-1985. S. Jackson, S., Ryan, P. G. (1986). Differential digestion rates of Afr. J. antarct. Res. 17. 155-162 prey by bvhlte-chinned petrels Auk 103: 617-619 Williams, A. J (1980a). Aspects of the breeding biology of the Kooyman, G. L., Davis, R. W., Croxall, J. P., Costa, D. P. (1982). Gentoo Penguin Pygoscelis papua. Gerfaut 70: 283-295 Diving depths and energy requirements of King Penguins. Williams, A. J. (1980b). The breeding biology of Eudyptes Science 217: 726727 penguins wth particular reference to egg slze dimorph- Lishman, G. S. (1985). The comparative breeding biology of ism. Ph. D. thesis, University of Cape Town Adelie and Chinstrap Penguins Pygoscelis adeliae and P Williams, A. J (1982). Chick feeding rates of macaroni and antarctica at Signy Island, . Ib~s127: rockhopper penguins at Marion Island. Ostrich 53. 84-99 129-134 Morisita, M. (1959).Measuring of interspecific association and Williams, A. J . Siegfried, W. R. (1980) Foraging ranges of similarity between communities. Mem. Fac. Sci. Kyushu krill-eating penguins. Polar Rec. 20: 159-162 Univ., Ser. E, Biol. 3: 65-80 Williams, A. J., Sieqfried, W. R. Burger. A. E.. Berruti, A. Nagy, K. A. (1987).F~eld metabol~c rate and food requirement (1979). The Prince Edward Islands: a sanctuary for sea- scaling in mammals and birds. Ecol. Monogr. 57. 111-128 birds in the Southern Ocean. Biol. Conserv. 15: 59-71 Rdoux, V., Jouventin, P,, Stahl, J. C.. Weimerskirch, H. (1988) Wilson, G. J. (1983). Distribution and abundance of Antarctic Ecologie alimentalre comparee des manchots nicheurs aux and sub-Antarctic penguins: a synthesis of current know- fles Crozet. Revue d'Ecologie (Terre et la Vie) 43: 345-357 ledge. Biomass Sci. Ser. 4: 146 Stonehouse, B. (1967) Feedlng behaviour and diving rhythms Wilson, R.P. (1984).An improved stomach pump for penguins of some shags, Phalacrocoracidae. Ibis 109: and other seabirds. J. Field Ornith. 55: 109-1 12 600-605 Wilson, R. P., La Cock, G. D., LVilson, M-P., Mollagee, F. Tramer, E. J. (1969). species diversity: components of (1985). Diffcrentldl digestion of fish and squ~din Jackass Shannon's formula. Ecology 50: 927-929 Penguins Spheniscus demersus. Ornis Scand. 16: 77-79

This article was submitted to the editor ~llanuscriptfirst rece~ved:January 10, 1989 Revised version accepted: July 17, 1989