Vol. 177: 115-131, 1999 MARINE ECOLOGY PROGRESS SERIES Mar Ecol Prog Ser Published February 11 Diet, provisioning and productivity responses of marine predators to differences in availability of Antarctic krill J. P. Croxall*, K. Reid, P. A. Prince British Antarctic Survey, Natural Environment Research Council. High Cross, Madingley Road. Cambridge CB3 OET. United Kingdom ABSTRACT: Knowledge of relationships between prey availability and predator performance is the key to using predators as indicators of the state of marine systems and to assessing potential conse- quences of competition between natural predators and man for common resources. Fluctuations in the abundance of Antarctic knll are beiieved to have a substantial influence on the reproductive perfor- mance of krill-dependent top predator species in the Southern Ocean few quantifications of such inter- actions exist. At South Georgia, for 2 years in which acoustic surveys revealed a major difference in knll abundance, we compared diet, prov~sioning of offspring and breeding success in 4 main predator species ( 2 penguins, 2 albatrosses, with supporting data from Antarctic fur seal) whose dependence on k r d typically ranges from 20 to 90%. The 4-fold difference in krill biomass between 1986 (ca 30 g m-2) and 1994 (ca 7 g m-2) was accompanied by (1) an 88 to 90% reduction in the mass of krill in predator dlets (and some increase in the fish component), ( 2 ) greater prey diversity for most species, (3) reduced diet overlap between species and (4) a switch from krill to amphipods in macaroni penguin but no major dietary change in other species. Rates of provisioning of offspring decreased by 90% in gentoo penguin and 40 to 50% in the other 3 species this was due to reduced meal size in penguins (by 90% in gentoo and 50% in macaroni) and to doubling of foraging trip duration in albatrosses. Breeding suc- cess was reduced by 50% in grey-headed albatross (the species least dependent on krill), by 90% in black-browed albatross and gentoo penguin (only 3 to 4 % of eggs producing fledged chicks) but by only 10% in macaroni penguin, presumably reflecting its ab5ty to switch to small prey unprofitable for the other species. However, all species (except for black-browed albatross), particularly macaroni pen- guin, produced fledglings significantly lighter than usual, probably affecting their subsequent survival. Some effects on adult survival could also be inferred. Our results show a coherent, though complex, pattern of within and between species similarities and differences. These mainly reflect the degree of dependence on krill, the feasibility of taking alternative prey and constraints on trip duration andlor meal size imposed by foraging adaptations (especially relating to travel speeds and diving abilities, whereby flightless divers and pelagic foragers differ markedly). The generality of these principles are explored through comparison with other studies, particularly of Shetland seabirds. K E Y WORDS: Predator performance - Prey availability . Antarctic krill . Albatrosses . Penguins . Fur seals INTRODUCTION cess) are, therefore, potentially vulnerable to fluctua- tions (whether natural or anthropogenic) in availability In marine systems many top predators specialise on of their main prey. Such circumstances are the basis for particular classes or species of prey. Their breeding many long term studies seeking to use variation in populations and especially productivity (breeding suc- predator populations and/or breeding performance as indicators of variations in prey abundance and/or availability (Croxall & Prince 1979, Cairns 1987, Crox- 'E-mail: j.croxall@bas.ac.uk all et al. 1988, Montevecchi et al. 1988, Monaghan et O Inter-Research 1999 Resale of full article not permitted

116 Mar Ecol Prog Ser 177: 115-131, 1999 al. 1989, Montevecchi & Berruti 1991, Furness & Greenwood 1993, Hamer et al. 1993, Ainley et al. 1995, Monaghan 1996). However, in very few studies are estimates of prey abundance, independent of those derived from predator diet or breeding success, avail- able (Anderson et al. 1982, Burger & Piatt 1990, Uttley et al. 1994). Furthermore, the ability even of specialist predators to switch to alternative prey is poorly under- stood. Antarctic krill Euphausia superba is a key food resource for many species of top predators in the Southern Ocean (Croxall & Prince 1980, Croxall 1984, Reid & Arnould 1996, Croxall et al. 1997). This is par- ticularly the case in the austral summer (the maln predator breeding season) and in the Antarctic and sub-Antarctic (especially South Georgia) regions of the South Atlantic sector. However, there are clear indica- tions that, from time-to-time, the availability of krill to top predators, either locally or regionally, is substan- tially reduced (Croxall et al. 1988, Priddle et al. 1988, Boyd et al. 1994, Lunn et al. 1994). Much of the evi- dence for this is relatively circumstantial, usually being derived retrospectively from reduced reproductive performance of one or more of the predators of krill. Very seldom have concurrent quantitative data been available on predator diets and on the abundance of krill at appropriate scales. Indeed, in only 2 years in the last decades, 1986 and 1994, has the detailed annual monitoring of breeding population size and perfor- mance of a range of key species of top predators at Bird Island, South Georgia (Croxall et al. 1988, Prince et al. 1994, Boyd et al. 1995, Croxall & Rothery 1995), been accompanied by quantitative research into their diets in conjunction with simultaneous assessment of krill abundance and availability in surrounding waters (Brierley 8 Watkins 1996, C. Goss unpubl. data). In this paper we compare data on predator diet, breeding population size and productivity at South Georgia in the austral summers of 1985/86 and 1993/94 in the light of the results of acoustic echo-inte- gration surveys which indicated that around South Georgia in 1993/94 krill was very scarce and patchy in abundance and distribution (Brierley et al. 1997) and several-fold less abundant than in some previous years (including 1986) when comparable surveys had taken place. In particular we ask (1) How well do predator diets reflect the potential differences in krill availability in 1986 and 1994? (2) Are any predator species able to compensate for low availability of krill by switching to other prey? and (3) How do differences in predator diet between years relate to breeding success in these years? In addition to their intrinsic interest, the answers to these questions also make potentially important contri- butions to the evaluation of results from long-term monitoring studies of breeding population size and success in top predators in general and in the Antarctic in particular. The difficulties and costs of conducting annual surveys of krill distribution and abundance at appropriate spatial scales make the use of indices of predator performance to assess krill abundance partic- ularly attractive. However this requires that the nature of interactions between predator and prey (often referred to as functional relationships-see Butter- worth & Thompson 1995) is adequately understood. METHODS Study species. For critical comparisons between 1986 and 1994 we focused on 4 species: 2 aerial pelagic foragers (grey-headed and black-browed albatrosses Diomedea chrysostoma and D. melanophns) and 2 flightless diving foragers (gentoo and macaroni pen- guins Pygoscelis papua and Eudyptes chrysolophus). In both years we have comprehensive quantitative data on composition of the diet, on the size of krill taken and on breeding population size and breeding success similar data are also available in some other years. In addition, for Antarctic fur seals Arctocephalus gazella we have data on composition of the diet (in terms of frequency of occurrence), on the size of krill taken and the population size and breeding success in both years For Antarctic prions Pachyptila desolata we have data on composition of the diet in both years. Diet. Food samples were collected at weekly inter- vals during February of both 1986 and 1994 from adults of all species returning to feed offspring at Bird Island, South Georgia. Samples from penguins were collected by stomach lavage (Wilson 1984) using the method of CCAMLR (1995), whereas samples from albatrosses were collected by intercepting meals immediately prior to the chick being fed (see Prince 1980b). Antarctic prion samples were collected as spontaneous regurgitations from adults mist netted above breeding burrows (see Prince 1980a). Samples from Antarctic fur seal were obtained by lavaging lac- tating females immediately after they arrived ashore in 1986 and by collecting faeces (scats) in 1994 (see Reid & Arnould 1996). All samples were weighed, then separated into 3 main prey categories (crustacean, cephalopod, fish) the mass of each category, along with any unidentified material was recorded. All cephalopod lower man- dibles and fish otoliths were identified and measured following Clarke (1986), Hecht (1987), Williams & McEldowney (1990) and Reid (1996) and by reference to collections and unpublished data at the British Antarctic Survey (BAS) (see e.g. Reid et al. 1996a).

Croxall et a1 . Responses of marine predators to availability of krill 117 Crustaceans were identified in association with appro- priate BAS specialists using extensive reference col- lections from the South Georgia area. The sex and maturity stages of krill were assessed using the method of Makarov & Denys (1981) following the nomencla- ture of Hill (1990). The total length (AT) of krill was measured using the removed carapace length (RCL) and the appropriate regression relationship for each maturity/sex stage as given in Hill (1990). The occurrence of each prey taxon is represented by its frequency of occurrence (the total number of sam- ples in which the taxon was recorded/total number of samples) and proportionate mass (the total mass of the taxon/mass of all samples). Only frequency of occur- rence data are available from the fur seal samples. The mass of euphausiids and the amphipod Themisto gau- dichaudii was obtained by direct weighing the mass of each cephalopod and fish taxon was estimated using the contribution of that taxon to the reconstructed mass of the prey group. Diet diversity was quantified using Shannon-Weaver (1949) indices, and dietary overlap indices were calculated by a modified version of Morisita's equation adapted from Diamond (1983) the overlap index C is given by : S 2 c x i y i C = '=l i x i ? + k y i 2 ) = I 1=I where S is the number of categories represented in the diets of species X and y. Population size and breeding success. For alba- trosses and penguins the number of pairs laying eggs and the proportion of these pairs successfully rearing chicks has been recorded at the same selected study colonies on Bird Island in each year since 1976 (see e.g Croxall et al. 1988, Prince et al. 1994). For Antarctic fur seals the number of females giving birth and the sur- vival of pups to 6-8 wk of age has been recorded annually at the same study site on Bird Island since 1981 (see Croxall et al. 1988, Lunn et al. 1994, Boyd et al. 1995). Prey availability. Acoustic estimates of krill abun- dance were conducted using standard techniques cruise and survey details are given in Brierley et al. (1998) most details, especially relating to survey design, area, equipment and methodology is also pro- vided in Hunt et al. (1992) and Brierley & Watkins (1996). RESULTS Diet Full details of the composition of the diets of alba- trosses and penguins in 1986 can be found in Croxall et al. (1997 Appendix 1 therein) equivalent details for 1994 are given here in Appendix 1. Overall diet com- position in the 2 yr is sunlnlarised in Table 1 and com- pared with the average diet for all other years for which we have data. In broad terms, these data indi- cate that the 4 species selected for the main study can be placed in a graded sequence of increasing depen- dence on krill, at least with respect to data for 1986 and 1976 to 1995. Thus grey-headed albatrosses are least Table 1 Composition (percentage by mass) of the main prey groups in the diet of predators during the breeding seasons of 1986 and 1994 at South Georgia, with comparison with data from other years. The chi-squared statistic and the level of significance ("p c 0.01) of differences from equivalent proportional composition of the diet in both 1986 and 1994 are also shown Species Year % Fish % cephalopod "h krill "h amphipod x 2 P Grey-headed albatross 1986 13.6 70.5 15.9 0.0 5483.9 . . 1994 60.8 37.7 1.5 0.0 Otherd 26.8 54.1 18.0 0.0 Black-browed albatross 1986 29.5 31.1 39.4 0.0 4382.6 . . 1994 72.4 22.9 4 .? 0.0 Otherd 36.1 26.8 34.3 0 0 Gentoo penguin 1986 31.0 0.0 69.0 0.0 4499.7 . . 1994 85.9 1.2 9.0 3.6 Otherb 36.3 0 0 63.2 3.0 Macaroni penguin 1986 2.3 0 0 95.0 2.7 19527.9 . . 1994 15.0 1.0 13.1 67.3 OtherC 4.0 1.3 89.8 10.5 "Mean from 3 other years between 1976 and 1980 bMean from l 1 other years between 1977 and 1995 'Mean from 9 other years between 1977 and 1995