The Use of Food Samples from Sea Birds in the Study of Seasonal Variation.10 the Surface Fauna of Tropical Oceanic Areas

MYRTLE J. ASHMOLE AND N. PHILIP ASHMOLEI

ABSTRACT: Many parts of the tropical oceans appear to be relatively seasonless,but, because of the difficulty of samplin g mobile and patchily distributed animalsand the cost of oceanographic investigations, few data are available on the extentof seasonal changes. By regularly collecting regurgitations from sea birds, andidentifying and measurin g the food items, seasonal data could be obtained on theavailability, size classes, and perhaps reproductive cycles of the fish and squid char­acteristic of the surface layer of tropical seas. Flying fish (Exocoetidae), juveniletunas (Scombridae) , and squid of the family Ommastrephid ae are especially easilyobtainable.

Experience gained dur ing a recent study of the comparative feeding ecology ofsea birds on Christmas Island (Pacific Ocean) makes it possible to assess the char­acteristics of bird species which affect their suitability for such study. Potentiallyuseful species includ e terns ( especially Sterna fuscata, Anous stolidas, A. tenuirostris,and Gygis alba) and boobies (especially Sula SIt/a) . Samples could be obtained fromseveral bird species in the same period, and a program could include sampling ofinshore waters with the neuston net and making basic oceanographic observations.Investigations of this kind could be carried out economically on any of a large num­ber of tropical oceanic islands.

IN ASSESSING the results of a study of the feed­ing ecology of some sea birds of ChristmasIsland , in the central equatorial Pacific, weattempted to make use of published informationconcerning seasonal variations in the fauna ofthe surface layer of the sea in th is region. Itsoon became evident, however, that this typeof information is very sparse, and that thoseworkers who have made seasonal comparisonshave often had to depend on data collected indifferent areas, in different years, or by differentmethods (see, for instance, King and Demond,1953 ; King and Hida, 1957 ; King and Iver­sen, 1962).

The absence of information on seasonal vari­ation in many tropical oceanic areas resultsmainly from the expense of running oceano­graphic ships, and the conflicting demands ontheir time, which generally prevent adequatesampling in a single area over a period as longas a year. Only in a few areas where there are

1 Peabody Museum of N atur al Hi story, and D epart ­ment of Biology, Yale Un iversi ty, N ew H aven, Con­necticut 06520. Man uscript received January 23, 1967.

1

important commercial fisher ies are good seasonaldata available. Even in these areas, however,there are generall y few data on changes in theabundance of the forage animals which formthe food of most sea birds and of many fishused for hum an food. This is because nets usedfor rout ine plankton sampling do not catchmany nekton animals, and even modern netslike Isaacs-Kidd trawls catch few fast-swimminganimals near the surface (King and Iversen,1962; Pearcy, 1965) .

It would be of considerable inter est to ob­tain more information on the extent to whichseasonal fluctuations in physical and biologicalcharacteristics actually occur in those part s ofthe tropical oceans which appear to be more orless constant th roughout the year. For instance,orni thologists working on trop ical islands inseveral different areas have found that amongthe sea birds breeding on a single island (orgroup) some species have evolved regimes underwhich individuals breed at intervals of less thana year, suggesting that seasonal variati on in theenvironment is not of great impor tance to them.

2

But on the same islands other species haveannual breeding regimes, indicating that forthem seasonal influences are sufficiently strongto outweigh the selective advantages of breed­ing at shorter intervals (Ascension Island: pa­pers in Ibis 103b, 1962-63; Christmas Island:Gallagher, 1960 ; Ashmole, 1965 ; Ashmole andAshmole, 1967 ; Ashmole, in press; GalapagosIslands : Leveque, 1964; Snow, 1965) . Some ad­vances are now being made in understandingthe differences in feeding ecology which leadto dependence of different birds on differentgroups of prey species, but lack of knowledge ofthe importance of seasonal effects in populationsof the various prey species prevents furtherprogress. The reproductive cycles of oceanic fishand cephalopods in relatively seasonless areasare even less well understood, but are of boththeoretical and practical importance.

The purpose of the present paper is to showthat a very economical investigation of seasonal(and year-to-year) variation in the surface faunaof a tropical oceanic area could be carried outby making use of the sampling of the oceanicenvironment which is done by sea birds, andat the same time by collecting certain other data.

The method which we propose depends onthe fact that tropical sea birds of many speciesare easy to catch and, when caught, will oftenregurgitate the whole or part of their latestmeal. Regurgitations provide, on average, farmore items in good condition than do stomachsof shot or netted birds and their collection doesnot appreciably affect the bird population. Re­gurgitations are stored in formalin and theindividual food items identified and measuredat leisure. Since the feeding methods of eachspecies are reasonably constant, regular samplingon an adequate scale would permit detection ofchanges in the relative availability of differentprey species. Furthermore, each of the bird spe­cies takes prey covering a considerable range insize, so that the data should reflect changes inthe size-frequency distributions of the variousprey species. Deductions could doubtless alsobe made about the timing of reproduction insome of the fish and squid species. The methodcould be used on almost any tropical island onwhich appropriate species of sea birds are avail­able for a large part of the year, but the decisionas to which bird species should be used for

PACIFIC SCIENCE, Vol. XXII, January 1968

sampling depends on considerable knowledgeof the ecology of the available species.

In the course of our study of the food of seabirds on Christmas Island during the periodMarch 1963 to June 1964, we examined 800samples (mainly regurgitations) from eight spe­cies of sea birds. We were mainly concernedwith comparing the food of different species,and we did not have the opportunity to collectvery large numbers of samples from anyonespecies, so that our data are not suitable fordetailed seasonal analysis. However, they dodemonstrate the potentialities of the method.A complete account of the study is available inAshmole and Ashmole (1967), and here weshall discuss only those aspects which are perti­nent to the planning of a more general inves­tigation , and in particular to the choice of birdspecies from which samples should be obtained.

The samples which we collected on ChristmasIsland consisted of entire or partially digestedfish and squid, while two bird species also pro­vided various invertebrates other than squid,mainly water striders (Halobates) and crusta­ceans. Nearly all the squid belonged to a singlegenus (Symplectoteuthis) of the family Om­mastrephidae, but the fish were distributedamong 33 families. The state of the samplesvaried considerably, but there were fairly con­sistent differences among the bird species (Table1). These differences resulted partly from thefact that in some species it was easiest to obtainsamples from young birds which had previouslybeen fed by adults, while in others it was easyto catch adults. The best samples were those ob­tained as the birds arrived to feed their chicks,while those obtained from roosting adults weregenerally more digested and thus the items weremore often unidentifiable . It would be impor­tant, therefore, to plan future work in such away as to obtain the food items as soon as pos­sible after they were caught by the birds.

The usefulness of a particular species woulddepend largely on the ease with which largenumbers of identifiable fish could be obtainedfrom it, so we have calculated (Table 1) thenumber of fish which we were able to identify(to family level) per 100 samples collected. Inthe following short accounts of the most promis­ing bird species we have also commented on theabundance of each species, the ways in which it

Sea Bird Food and Ocean Surface Fauna-ASHMoLE AND ASHMOLE 3

may be caught, the methods and zones which ituses in feeding, and the size of the animalswhich it obtains.

EVALUATION OF THE BIRD SPECIES

Of the eight bird species whose food wassampled on Christmas Island (Table 1), fourwould not be very suitable for marine biologicalinvestigations. It is time consuming to obtainsamples from the Christmas Island Shearwater,Puff inus nativitatis Streets, the Phoenix Petrel ,Pterodroma alba (Gmelin) , and the Red-tailedTropic-bird, Phaethon rubricauda Boddaert, andthe number of identifiable fish which can beexpected in 100 samples (Table 1) is ratherlow for all these species. The Blue-grey Noddy,Procelsterna cerulea (F. D. Bennett), provideslarge numbers of items per sample, but thepopulations on many islands are rather small,and the species is not widely distributed . Fur­thermore, it feeds close inshore on very smallitems which could probably be sampled moreefficiently with a neuston net (David, 1965)towed behind a small boat. The other four spe­cies studied-the Sooty Tern, Sterna fuscataLinnaeus, the Brown Noddy, Anous stolidus(Linnaeus), the Black or Lesser Noddy, Anoustenuirostris (Temminck), and the White orFairy Tern , Gygis alba (Sparrman)-aU havecertain advantages for regular study. Salientnumerical details are given in Table 1.

Sterna fuscata (Sooty Tern)

This is the most abundant of all tropical seabirds, breeding in enormous colonies on islandsin the tropics around the world (see Ashmole,1963 for details of its breeding distribut ion andbreeding seasons) . The eggs are laid on theground in the open, and sometimes there areas many as five nests per square meter. Thebirds may desert whole areas if they are dis­turbed too much at the start of nesting, butlater they sit tightly and may be caught with ahand net. However, incubating birds rarely re­gurgitate when caught, and the only way inwhich large numbers of samples can easily becollected is to catch adults (with a long-handledhand net, or mist nets) as they arrive to feedtheir chicks, especially in the late afternoon, orto catch chicks shortly after they have been fed.

Before breeding starts, a few regurgitations canbe obtained by catching birds at night whenthey are roosting on the ground, but the samplesare generally small and largely digested. Onmost tropical islands the species has one fairlyshort breeding season, but on Christmas Islandand a few other central Pacific islands there aretwo breeding seasons each year.

Sterna fuscata catches its prey either whileflying or by plunging to the surface, but it prob ­ably hardly ever submerges completely. It hasbeen recorded as feeding at night (Gould, inpress) , but this probably occurs only when themoon is nearly full. It is an oceanic speciescapable of feeding hundreds of miles from land,even when breeding. However, the actual dis­tance traveled regularly probably varies con­siderably from colony to colony.

Our 242 samples were mostly from adultsfeeding chicks. They contained, on the average,5.6 items each, of which 60% were fish, 40%squid. On the whole, the samples were in goodcondition, and 79% of the fish were identifiedto the family level. This means that about 266identifiable fish could be expected per 100 sam­ples. The identified fish belonged to 21 families,the ones which occurred most regularly beingExocoetidae and Scombridae, Gempylidae, Ser­ranidae, and Emmelichthyidae. The identifiedsquid were nearly all Ommastrephidae . Of thefish, 93% were between 2 and 10 cm in length(measured to the base of the tail), while 98%of the squid were between 2 and 8 em inmantle length . W e obtained sufficient data fromthis species to suggest that important seasonalchanges exist in the availability of certain fishfamilies, but a more intensive sampling programwould be necessary to demonstrate such changesconvincingly.

In summary, the advantages of this speciesare that it is available in great numbers, andsamples can easily be obtained ; the samples areoften in good condition, and contain a highproportion of ident ifiable items ; Scombridaeare especially well represented in the samples;and the feeding range of the species is suchthat it provides samples from a large oceanicarea. However, one cannot determine preciselythe area from which each sample comes, andit is not easy to obtain many samples at timeswhen the birds are not breeding.

TABLE 1

S UMMARY OF D ATA ON THE FOOD OF EIGHT SP ECIES OF SEA BIRDS ON CHRISTMAS ISLAND

~

Phaethon Puffinu s Pterodroma Sterna A nousITEM rubricauda nativitatis alba fuscata stolidus

Mean weight of bird (gm) 665 324 269 173 173Mean number of items per

sample 3.8 10.5 4.4 5.6 5.1

% composition by numberFish 33 63 26 60 71Squid 64 36 48 40 29Other invertebrates 3 0.2 26 0.1 0

% of fish identified tofamily 55 20 1 79 67

% of squid identified tofamily** 100 62 24 88 100

Estimated number of fishidenti fiable to familyper 100 samples 69 132 1 266 243

• Wh en two figures are given for Gyg;s alba the upper refers to food samples carried in th e bill and the lower to regurgit ations .• • Some additional squid were tentatively identified .

A nous Gy gis Procelsternatenulrostris alba* cerulea

90.9 101 45.4

18.3 1.3 41.8-4.3

95 59 494 41 9 'i:i

1 0.3 42 >(J......>-rJ ..

27 87 38......(J

Ul

71 100 74(J......tTlZ(J

470 67 778 ..tTl- <221

~

~~............

'-<l'O~l:l'O..,

'<.....\00\co

Sea Bird Food and Ocean Surface Fauna-c-Asnxrot. s AND ASHMOLE 5

Anous stolidus (Brown N oddy)

Anous stolidus breeds on many of the tropicalislands where Sterna [uscat« is found, but gen­erally is less numerous. The birds are fairlyeasy to catch with a hand net at night, especiallywhen there is no moon ; mist nets could doubt­less also be used, and unfledged young can becaught by hand. No general account of thebreedin g seasons of this species is available, butin many colonies breeding is probably lesstightly synchronized than that of S. [ascata,while in some areas (i ncluding Christmas Is­land ) ind ividuals roost on the breeding islandseven when they are not breeding.

A. stolidus uses feeding methods similar tothose of S. [uscata, but sometimes also feedswhile swimming on the surface ; there is noevidence that it feeds at night. In contrast toS. [nscat«, A . stolidus apparently does not nor­mally feed more than about 50 miles from th ecolony, although the size of the zone utilizedmay be different in different areas.

N early all our 38 samples from A. stolid/ISwere obtained from roosting adults early in thenight, soon afte r they had return ed from feed­ing. They contained, on the average, 5.1 itemseach, 71% of these items being fish, the re­mainder squid . Over half of the fish wereidentified to the family level, so th at about 243identifiable fish could be expected per 100samples. The identified fish belonged to ninefamilies, of which Exocoetidae, Scombridae,Holocentridae, and Gempylidae occurred mostregularly. All the identified squid were Om­mastreph idae. Of the fish, 94% were between 2and 12 em in length, while all but 1 of the 50measurable squid had mantle lengths between 2

and 8 em.The advantages of A . stolidus as a sampler

of the surface fauna of tropical seas are that thespecies is widely distributed, is available onsome islands even outside the breeding season,and will often provide regurgitations whencaught early in the night; its relatively limitedfeeding range means that animals obtained fromit could normally be assumed to come fromwithin 50 miles of the island. However, it isless abundant than S. [uscat« and rather moredifficult to catch.

Anous tenuirostris (Black N oddy)

Thi s species (i n which we include Anousminutus], although it breeds only on islandswhere bushes, trees, or cliffs are available toprovide nest sites, often occurs in large colonies.Young birds often regurgitate when handled,but more consistent sampling can be carriedout by catching roosting birds with a hand netearly in the night, since on Christmas Island andmany other islands some birds are present atall times of year.

A. tenuirostris feeds by the same methods asA . stolidus and S. [uscata; there is no conclusiveevidence that it feeds at night. This speciesseems usually to feed even closer to its shorebase than does A . stolidus: the Christmas Islandbirds fish very largely within 5 miles of theshore, but in some other areas (for instanceAscension Island , Atlantic Ocean) they gofar ther out to sea.

Our 110 samples, which were mostly regurgi­tations from roosting adults, contained on theaverage 18.3 items each. Of these items, 95%were fish, nearly all the remainder being squid.Th e regurgitations were often in the form oftightly-packed masses, many of the items beingin a rather advanced state of digestion. Weidentified only 27% of the fish to the familylevel, partly because a large numb er of the fishwere fry that we could not identify. Neverthe­less, some 470 fish were identified per 100samples. The identified fish belonged to 17families, among which the most regularly rep­resented were Exocoetidae, Gempylidae, Scorn­bridae, Blenniidae, Holocentr idae, and Emmel­ichthyidae. The few identified squid were allOmmastreph idae. Of the fish, 98% were lessthan 6 em long, and 77% between 1 and 4 em.Of the few squid, 83% were between 2 and 6 emin mantle length.

A. tenuirostris was the only species whichshowed any dramatic differences in the diet atdifferent seasons. In May 1963 and June 1964 wefound in the diet a far higher proport ion of fishless than 2 em long than were present in ourother sampling periods.

This species has the advantages that it isabund ant, easily caught while roosting, and onsome islands is available all the year round ; thesamples provide large numbers of fish, and these

6

are from inshore waters; the small size of itsprey would make study of its food an excellentcomplement to that of one of the larger species,for instance A . stolidus. Its main disadvantageis the difficulty of identifying many of the fishfry.

Gygis alba (White or Fairy Tern )

Gygis alba, another widespread tropical spe­cies, generally is less abundant than the preced­ing three species, but many tropical islands havepopu lations of a thousand or more birds. Thenests are dispersed, either in forks or hollows inbranches of trees, on the tops of coral blocks,or on cliff ledges where these are available.G . alba differs from the other species in thatthe adults rarely regurgi tate, and they carryfood for the young in their bills. Hence thetechnique for obtaining food samples is different.The birds usually come quite close to a humanintrud er, even when carrying food, and mayoften be caught with a long-handled net ; thefood items dropped by the birds as they arecaught can then be retrieved from the ground.Our samples were mainly obtain ed from adultsin this way, although we also collected regurgita­tations from juveniles and from adult s.

On Christmas Island G. alba breeds at alltimes of year. In many other areas, however,breeding is more seasonal and, since the individ­uals leave the colonies when they have finishedbreeding, it would be impossible to obtainsamples at all times of year.

This species catches most of its prey in flight,and there is some evidence that it feeds exten­sively in the half light at dawn. It sometimesfeeds close inshore, and most of the prey carriedback to the young in the bill are probab ly caughtfairly close to the colony; however, the speciessometimes ranges hun dreds of miles from land.

Among our 152 samples, those carried inthe bill contained, on the average, 1.3 itemseach, while regurgi tations contained 4.3 items.Fish made up 59% of the items, the remainderbeing squid . The samples were generally inexcellent condition, and 87% of the fish wereidentified to the family level. Collecting onlyfood carried in the bill, one could expect toobtain appr oximately 67 fish identifiab le tothe family level per 100 samples. H owever,all of these would be in excellent condition , so

PACIFI C SCIENCE, Vol. XXII, January 1968

that further iden tification would be much easierthan in most of the other birds. The fish whichwe ident ified represented 22 families, 5 ofwhich were not found in the food of any ofthe other bird species. Blenniidae occurred mostregularly, followed by Exocoetidae, Myctophi­dae, Scombridae, Gempylidae, and Gonostorn­atidae . Squid of the families Ommastrephidaeand Enoploteuthidae were identified. Of thefish, 96% were less than 8 cm in length ;the samples carried in the bill contained alower proportion of small fish than did theregurgitations. Of the squid, 93% were between2 and 6 em in mantle length.

This species ate far more squid dur ing oneof our sampling periods than at other times,and during this same period the squid takenincluded Enoploteuthidae ( Abralia sp.) whichwere not found in samples obtained in the earlierpart of the study.

The main advantages of this species are theexcellent condition of the items carried in thebill and the fact that a wide variety of preyare eaten. However, it is difficult to obtainmany samples, since usually the populations arenot very large, and some of the birds are toowary to be caught when carrying fish, so thata quantitati ve study would probably have todepend primarily on other species.

Other Species

The species just discussed are those withwhich we have most experience, but other spe­cies also might be used for sampling, and wehave tried to indicate some of the criteria whichshould be used in deciding whether a givenspecies would be appropriate. The only largespecies included in our study was Pbaetbonrubricaud« (weights are included in Table 1),which takes large prey, but from which it isgenerally not easy to obtain many samples.The obvious additional candidates, available onmany tropical islands, are the various species offrigate birds,' Fregata spp., and boobies, Sulaspp . We would reject the fr igate birds on thegrounds that they obtain some of their food bypiracy, which complicates interpretation, and alsothe food gener ally is in poor condition. Of thethree widespread tropical boobies, the BrownBooby, Snla leucogaster (Boddaert) , is mainlyan inshore feeder, is rather difficult to catch, and

Sea Bird Food and Ocean Surface Fauna-AsHMoLE AND ASHMOLE 7

is not often available in large numbers. However,either the Red-footed Booby, Sula sula (Lin­naeus), or the Masked (or Blue-faced, orWhite) Booby, Sula dactylatra Lesson, mightbe useful.

S. sula nests in colonies in bushes or trees,and regurgitates readily, at least when withyoung. The birds roost on many islands through­out the year; on Christmas Island, where thespecies nests on the main island, there is somebreeding at all times of year. S. sula catchessome of its prey below the surface by divingfrom a height, but has also been recorded catch­ing flying fish in the air ; it is thought some­times to fish at night (Murphy, 1936). S. sulaprobably fishes mainly within a hund red milesof its home island (Murphy, 1936; Royce andOtsu, 1955, who almost certainly mistookS. sula for S. dactylatra). In a small series ofsamples from Oahu, Hawaii (Ashmole andAshmole, in press) we found an average of nineitems per sample, of which 64% were fish,36% squid. Nearly 75% of the fish wereidentified, so that about 400 identifiable fishcould be expected per 100 samples from thisspecies. The fish found most frequently wereExocoetidae and Gempylidae, while all theidentified squid were Ommastrephidae. Halfof the fish were between 8 and 12 em long,while there were appreciable numbers in the12-16 and 16-20 em groups ; however, onlya few squid with mantle length of more than 8em were present in our samples. The mainadvantages of this species are that it is availablein some places throughout the year, and takesfish (and perhaps also squid) larger than anytaken by the terns previously discussed.

We have had little experience with S. dactyl­atra, but relevant data have been presented byDorward (1962). It would have advantagessimilar to S. sula, but probably ranges evenfarther from its breeding colonies.

DISCUSSION

It is clear that the use of analysis of the foodof each of these bird species has certain merits,but also certain disadvantages. We suggest thatvaluable data could be obtained by making useof a group of bird species selected for theparticular problem under investigation. For in-

stance, to investigate the effects of seasons inthe ocean around Christmas Island, it wouldprobably be best to obtain a series of regurgita ­tions, perhaps twice each month , from the threespecies Anous tenuirostris, A. stolidus, and S.sula. These samples would include animalsranging in size from less than 1 em to at least20 em. The regurgitati ons from A. ten uirostrisprobably would represent sampling largelywithin 5 miles of the island, those from A. stol­idus sampling largely within 50 miles, and thosefrom S. sula might have been obtained as muchas 100 miles away. On the other hand, for along-term study concerned primarily with thedetection of differences in the fauna of anoceanic area at the same season in a series of suc­cessiveyears, it would probably be most econom­ical to collect samples from Sterna fuscata,choosing those months when this species hasyoung. In any study, collection of samples fromGygis alba would be useful in providing animalsin excellent condition and wide variety, to formthe basis of a reference collection of the fish andsquid available at the surface in the area.

If one is to use sea birds as samplers of theirmarine environment, it is important to havesome understanding of the part which theyplay in the ecology of the surface layer of thesea. All the terns (including "noddies") con­sidered here obtain their food within a fewcentimeters of the surface of the sea; boobies,Sula spp., tropic birds, Pbaetbon spp., and someshearwaters (Puffinus spp.) , can penetrate alittle deeper. Thus the terns should be samplingthe fauna characteristic of the surface film, or"neuston." In fact, only one of the species forwhich we have data-the Blue-grey Noddy,Procelsterna cerulea- has a diet with a composi­tion conspicuously similar to hauls made withthe "neuston net" in the Indian Ocean (David,1965): both contain large proportions of waterstriders (Halobates), pontellid copepods, andvery small fish larvae. Among the other terns,only A. tenuirostris sometimes took many smallfish larvae ; and small crustaceans and H alobateswere absent from the diet of all the largerspecies. Although there is little informationavailable as to how many larger animals arenormally present within a few centimeters ofthe surface during the day, observations suggestthat, except for Procelsterna cerulea (and per-

8

haps Gygis alba feeding at dawn on animalswhich visit the surface during the night), allthe terns studied are dependent on schools oftuna and other predatory fish to drive fish andsquid to the surface: many of the prey animalsare caught by the birds while they are actuallyjumping out of the water in efforts to escape thepredators below. The larger birds like Phaeth onrubricaud« and S. sula, which can penetrate thewater to a depth of at least 1 meter, may be ableto catch prey which are not disturbed by otherpredators, and so may be parti ally independentof the presence of predatory fish.

Tuna schools are by no means randomly dis­tributed over the tropical oceans, but occurlargely in those areas where the animals onwhich they feed are most concentrated. In anearlier paper (Ashmole and Ashmole, 1967)we have discussed the evidence that in the openocean convergence at " fronts" produces localconcentrations of plankton and nekton, even inareas where the overall density of organi sms isvery low, and that these concentrations attracttuna schools. The existence of large popula ­tions of sea birds-like Sterna fllScata-whichcan range great distances from their islandseven when breeding, probably depends on thepresence of rich feeding areas of this kind.Clearly a knowledge of the areas where the wide­ranging species are feeding will help in inter ­preting the data obtained by the analysis oftheir food . The inshore-feeding sea birds, onthe other hand , make use of the concentrationsof prey animals, and of surface-feeding tunas,which are often present in the lee of islands.

Thinking that there might be a close corre­spondence between the food of surface-caughttuna s and the food of the birds, we studiedrecords of stomach contents of Yellowfin Tuna,N eoth lt111Z11S macropt erus (Temminck andSchlegel) , kindly made available by the Hono­lulu Biological Laboratory of the U.S. Fish andWildlife Service. The data used were for Yel­lowfin caught at the surface within 10 miles ofChristmas Island and neighboring islands, but indifferent years from our sampling of the foodof the birds. We found that there were strik­ing differences between the diets of the tunasand of the birds, even if the comparison wasrestricted to the birds which fed close inshore.The tunas took a far higher proportion of reef-

PACIFIC SCIENCE, Vol. XXII, January 1968

originating fish than did any of the birds,and presumably this reflects the fact that tunascan quickly and easily change from feedingright at the surface to feeding at considerabledepths or around reefs, while the birds mustwait for their prey to come to the surface.After they have completed larval life , typicalreef-inhabiting fish such as Acanthuridae evi­dently do not often come to the surface evenwhen pursued, and so are rarely eaten by bird s ;however, they were eaten by the tunas in con­siderable numbers . (The pelagic larvae werealso eaten by tunas; they may also sometimesbe taken by birds, and it must be rememberedthat we were able to identify only a smallproportion of the partly digested larval fishfound in samples from A . tenuirostris at certaintimes of year. ) In contrast, Exocoetidae wereunimportant in the diet of the tunas, but are ofgreat importance in the diets of all the seabirds studied ; probably the ability of these fishto escape submarine predators by flying rendersthem especially vulner able to aerial predation bybirds.

Attempts to sample forage animals near thesurface of the central Pacific dur ing the daytime,by means of oblique hauls with Isaacs-Kiddand other trawls, yielded so little material thatthey were abandoned in favor of night hauls,which produced catches containing far moreorganisms of generally larger size (King andIversen, 1962) . Although the greater numberof organisms in night hauls partly reflects themovement to the . surface layers at night ofanimals which are mesopelagic during the day,it is of interest that very few Exocoetidae,Scombridae, or Gempylidae, or squid of thefamily Ommastrephidae, were identified ineither the day or night samples studied byKing and Iversen. Since members of all thesegroups are commonly caught by birds feedingat the surface, it may be that they are such fastswimmers that they can generally dodge thenets (ef. Pearcy, 1965:266) . Alternatively, theymay be extremely patchily distr ibuted, or sostrictly confined to the surface that they areonly rarely encountered by nets used in obliquehauls. In any case, it is clear that members ofthese groups are much more efficiently sampledby the birds.

In view of the commercial importance of tuna

Sea Bird Food and Ocean Surface Fauna-ASHMoLE AND ASHMOLE 9

fisheries, and consequent special interest in thebiology of tunas, the effectiveness of tropicalsea birds in catching juvenile tunas is worthemphasizing . King and Iversen (1962:301)said, " It was our hope that by means of mid­water trawls we would capture juvenile tunasof lengths above 12 mm. which were able toelude the plankton nets. In 274 hauls made withthe four midwater trawls described in this re­port we captured only six juvenile tunas, whichranged from 18 to 60 mm. in length ." In con­trast, in the 800 food samples which we ob­tained from birds on Christmas Island we found247 young scombrids of which 166 were in243 samples from Sterna fu scata. Of the lattergroup, 77 were examined in detail, and 69proved to be Yellowfin Tuna. The scombridsfrom S. fu scata ranged from about 2 cm to 11 cmin length (measured to the base of the tail) ,but over 70% were between 4 and 8 cm. Thoseobtained in samples from A. tenuirostris in­cludedabout a third which were less than 2 cmin length.

Since the birds eat mainly animals more than0.5 cm in length, a program using birds to detectseasonal or longer-term changes in the environ­ment could profitably be combined with othermethods (such as use of the neuston net) ofsampling the smaller organisms inhabiting thesea surface. If a version of this net could bearranged to fish clear of the bow wave of asmall boat, it would permit quantitat ive sam­pling of the waters immediately around anoceanic island, to complement the sampling oflarger animals, and a larger area, by the birds.The use of a small boat would of course alsomake it possible to obtain basic physical andchemical data requisite to an und erstanding ofchanges detected in the biota.

ACKNOWLEDGMENTS

The work on Christmas Island and in H awaiiwas supported by a Yale University-BishopMuseum fellowship awarded to the secondauthor. Grateful acknowledgment is made toH. Laurence Achilles, S. Dillon Ripley andRoland W . Force, and the others who helped tomake this arrangement possible. For assistancein identifications of fish we are especially in­debted to Gareth J . Nelson and James W . Atz;

Malcolm R. Clarke kindly undertook the identi­fication of the cephalopods. Garth I. Murphyand Richard S. Shomura generously allowed usto make use of their unpublis hed paper on "Theabundance of tunas in the Central EquatorialPacific in relation to the environment." Wethank Donald W . Strasburg and Philip Helfrichfor their helpful comments on the manuscript.

REFERENCES

ASHMOLE, MYRTLE J., and N . PHILIP ASH­MOLE. (In press) Notes on the breedingseason and food of the Red-footed Booby(SIt/a Stlla ) on Oahu, Hawaii. Ardea .

ASHMOLE, N. P. 1963. The biology of theWideawake or Sooty Tern Sterna fuscataon Ascension Island. Ibis 103b(3) :297-364.

--- 1965. Adaptive variation in the breed- .ing regime of a tropical sea bird . Proc. Nat.Acad. Sci. 53 :311-318.

--- (In press) Breeding and molt in theWhite Tern (Gygis alba) on ChristmasIsland, Pacific Ocean. Condor.

- -- and MYRTLE J. ASHMOLE. 1967. Com­parative feeding ecology of sea birds of atropical oceanic island . Peabody Mus. N at.H ist., Yale Univ. , Bull. 24.

DAVID, P. M. 1965. The Neuston net : a devicefor sampling the surface fauna of the ocean.J. Mar. BioI. Assoc. UK. 45 :313- 320.

DORWARD, D. F. 1962. Comparative biology ofthe White Booby and the Brown BoobySIt/a spp. at Ascension. Ibis 103b ( 2) :174­220.

GALLAGHER, M. D. 1960. Bird notes fromChristmas Island, Pacific Ocean. Ibis 102 :489-502.

GOULD, PATRICK J. (In press) Nocturnal feed­ing of Sterna fuscata and Plt/finus pacificus.Condor.

KING, JOSEPH E., and JOAN D EMOND. 1953.Zooplankton abundance in the Central Pacific.US. Fish and W ildl ife Serv., Fish. Bull. 54(No. 82 ) :111-144.

--- and THOMAS S. HIDA. 1957. Zooplank­ton abundance in the Central Pacific, Part II.US. Fish and Wildlife Serv., Fish. Bull. 57(No. 118) :365-395 .

10

--- and ROBERT T . B. IVERSEN. 1962. Mid­water trawling for forage organisms in thecentral Pacific 1951-1956. U.S. Fish andWildlife Serv., Fish. Bull. 62 (No. 210):217-321.

LEVEQUE, RAYMOND. 1964. Notes sur la repro­duction des oiseaux aux Iles Galapagos.Alauda 32 :5-44, 81-96.

MURPHY, ROBERT CUSHMAN. 1936 . OceanicBirds of South America. American Museumof N atural History, New York. 2 vols.

PA CIFIC SCIENCE, Vol. XXII, January 1968

PEARCY, WILLIAM G. 1965. Species composi­tion and distribution of pelagic cephalopodsfrom the Pacific Ocean off Oregon . PacificSci. 19 :261- 266 .

ROYCE, WILLIAM F., and TAMIO OTSU. 1955 .Observations of skipjack schools in Hawaiianwaters, 1953. U.S. Fish and Wildlife Serv.,Spec. Sci. Rept. , Fisheries No. 147. 31 pp.

SNOW, D . W . 1965 . The breeding of Audubon'sShearwater (PufJinus lherminieri) in theGalapagos. Auk 82:591-597.