DENSITIES OF ANTARCTIC SEABIRDS AT SEA AND THE PRESENCE OF THE KRILL EUPHA USIA SUPERBA BRYAN S. OBST Departmentof Biology,University of California,Los Angeles, California 90024 USA ABSTRACT.--Theantarctic krill Euphausiasuperba forms abundant,well-organized schools in the watersoff the AntarcticPeninsula. Mean avian densityis 2.6 timesgreater in waters where krill schoolsare present than in waters without krill schools.Seabird density is a good predictorof the presenceof krill. Seabirddensity did not correlatewith krill density or krill schooldepth. Disoriented krill routinely were observedswimming near the surface above submergedschools, providing potential prey for surface-feedingbirds. Responsesof seabird speciesto the distribution of krill schoolsvaried. The small to me- dium-sizeprocellariiform species were the best indicatorsof krill schools;large procellari- iforms and coastalspecies were poor indicators.Pygoscelis penguins occurredat high den- sitiesonly in the presenceof krill schools.These responses are consistentwith the constraints imposedby the metabolicrequirements and reproductivestrategies of eachof thesegroups. Krill schoolswere detectednear the seasurface throughout the day. Correlationsbetween seabirddensity and the presenceof krill during daylight hourssuggest that diurnal foraging is important to the seabirdsof this region. Received19 December1983, accepted4 December 1984. RELATIVELY little is known about the factors birds depend on directly or indirectly for food influencing the distribution of seabirdsin the (Haury et al. 1978). These observationssuggest marine habitat. The past decade has produced that relatively small-scalephenomena, such as a number of studiesattempting to correlatepat- local concentrationsof prey, may be of major terns of avian abundance and distribution with importance in determining the patterns of sea- physical featuresof the oceansuch as currents bird distribution within the broad limits set by and convergences,water masses,and temper- featuresof the physical ocean. ature-salinity fronts, features presumed to in- It is well known that seabirdssometimes ag- fluence the distribution of marine prey. Many gregateover concentrationsof prey (seeBrown such studies have demonstrated broad-scale 1980 for a review). But because the number of correlationsbetween these oceanographicfea- potential prey speciesis often quite large and tures and bird distributions or species assem- the movementsof prey within the open ocean blages (e.g. Jehl 1973, 1974; Shuntoy 1974; are complexand difficult to monitor, systematic Brown et al. 1975;Pocklington 1979; Griffiths studies of the influence of prey distribution et al. 1982; Schneider 1982; Gould 1983). upon the density and distribution of seabirds Recently,however, Ainley and Boekelheide in pelagicenvironments are lacking. (1983) concluded that the major, classical The present study documentsthe influence oceanographicboundaries in the south Pacific of a principal prey species,the krill Euphausia do not act as effective distributional barriers for superba,on the densitiesof seabirdsin waters seabirdsinhabiting the region. Furthermore, off the Antarctic Peninsula.The overwhelming where associationsbetween seabirdsand large- dominance of E. superbain this ecosystemand scale oceanographicfeatures have been tested its habit of forming large, well-defined schools statistically, explained variances of seabird (Hamner et al. 1983) facilitated the location of numbers have been low (Abrams and Griffiths this abundant prey organismby SONAR, pro- 1981). Hunt and Schneider (in press) suggest viding a uniquely simple systemin which dis- that "fine-scale" patchinessin bird distribution tributions of seabirdsand their prey could be caneffectively mask differences at larger scales, monitored simultaneously. The relationships e.g. differencesbetween water masses.Patchi- between krill distribution, school size, and ness at scales of meters to hundreds of meters schooldepth and the densitiesof the seabirds is evident for planktonic organismsthat sea- comprisingthis communityare examined. 540 The Auk 102: 540-549. July 1985 July1985] AntarcticSeabirds and Krill 541 SEA V., o Fig. 1. Map of the northern Antarctic Peninsulaand surroundingwaters indicating cruisetracks followed during the study. METHODS elevated and those for penguins somewhat de- Seabird densitieswere determined by performing pressed,relative to the true levels. strip transectsfrom the R.V. 'Hero' during 3 cruises Euphausiid distribution was monitored via a con- between 12 January and 12 February 1983. I per- tinuously reading SONAR echosounderthat pro- formed 226 10-min transects(48 from 12-17 January, duceddiscrete, easily recognizable tracings as the ship 83 from 25-30 January,and 125 from 2-12 February) passedover krill schools.The identity of the organ- in the waters to the northwest of the Antarctic Pen- ismsproducing these tracings was verified by netting insula,including the BransfieldStrait, Gerlache Strait, within the schools;over 99% of the tracings were and Drake Passage(Fig. 1). One 10-min transectwas attributable to E. superba.Several problems are asso- carried out during each 30-min period that the ship ciatedwith the useof hydroacousticsto monitorprey. was in transit and visibility permitted. Only birds First, becausethe SONAR systemused was vertically seenwithin a quadratdefined by a line out from the oriented, only schoolsdirectly under the ship were ship's bow and a second,perpendicular line off the detected. Second, because the transducer was mount- beam out to 300 m were counted. The area of each ed in the ship's hull, the upper 3 m of the water transectwas calculatedusing ship'svelocity, time of column (the ship's mean draft) were not monitored. travel, and the width of the transectstrip, and bird Finally, vertical SONAR does not provide an abso- numberswere convertedto units of density. lute measure of a school's size because it bisects it Many antarctic seabirds,including most procellar- only in a singleplane. However, the frequencywith iiforms,skuas, and larids,are inveterateship follow- which distinct krill schools appear on the echo- ers. Becauseseabird numbers typically were low, ship sounder, as well as their duration, provides an index followers usually could be recognized and moni- of the general abundance of krill in the waters cov- tored throughout the 10-min census;such individu- ered during a transect.Each transectwas assigneda als were not recorded. However, the presence of a ranking (0-10) corresponding to this apparent prey ship may attract such speciesinto the transectlimits abundance.A value for krill depth correspondingto even if they do not follow, thereby increasingtheir the shallowest school detected also was assigned to densityartificially. On the other hand, penguinscan each transect. be quite difficult to detectduring a shipboardcensus, Finally, SCUBA and blue-water diving techniques particularly in rough seas,because they float low in (Hamner 1975) were used to observe shape and be- the water and may dive upon approach.Thus, den- havioral characteristics of several selected, shallow sity estimatesfor volant seabirdsmay be somewhat (10-30 m) krill schoolslocated by SONAR. 542 BRYANS. OBST [Auk,Vol. 102 TABLE1. Comparisonof avian .tensityand biomassfor waterswith and without Euphausiasuperba schools. Transects without krill Transects with Factorial schools krill schools increase pa Mean density (birds/km 2) 8.8 22.8 2.6 <0.001 Mean biomass(kg/km 2) 23.2 45.4 2.0 <0.001 n 56 167 Significanceof the difference between means. RESULTS oceanicus),were found in transects with krill Krill schoolswere presentin 167 (75%)of the significantly more often than expected by 226 transects.Mean avian density was 2.6 times chancealone (Chi-squarepaired test of associ- greaterin waterswhere krill schoolswere pres- ation, P < 0.05 for each). Both regularly feed ent, and mean avian biomass was double that on krill during the breeding season. One found in waters without krill schools(Table I). species,the South Polar Skua (Catharactamac- The differencebetween these meanswas sig- cormicki),showed a significantnegative associ- nificant in each case (t-test, P < 0.001). ation with krill schools(P < 0.05). These skuas When seabirddensity (all speciescombined, regularly include krill in their summer diet, rounded to the nearestbird/km 2) was plotted but are restrictedprimarily to coastalwaters againstthe probability that krill schoolswere during this season. The Black~browedAlba- present(i.e. the fractionof transectswith a giv- tross (Diomedeamelanophris) tended toward a en bird density in which krill schoolswere negative associationwith krill schools,but this present), a regular, increasingtrend was ob- tendency was not statisticallysignificant (P = vious (Fig. 2). Of the transectswhere avian 0.07). densitywas 1-10 birds/km2, 68% also had krill; Densities of individual specieswithin the of thosetransects with an avian density of 31- seabird community varied in relation to the 40 birds/km 2, 92% had krill schools. Where ob- distribution of krill schools. Three species, served seabird density was greater than 40 SouthernFulmar, Cape Petrel (Daption capense), birds/km2, krill schoolswere always present. and Wilson'sStorm-Petrel, showed clear posi- This suggeststhat seabirdsare effectivelycon- tive correlations between their densities and centratingtheir activitiesin responseto E. su- the probabilitythat krill were present(Fig. 3). perba,and that seabird density is a good pre- Thesespecies