Prey Delivered to Roseate Tern Chicks in the Azores

j. Field Ornithol., 69(3):419-429

PREY DELIVERED TO ROSEATE TERN CHICKS IN THE AZORES

JAIMEA. RAMOS•, ENCARNACIONSOIJ•, Luis R. MONTEIRO Departamentode Oceanogv'afiae Pescas Universidade dos A fores 9900 Horta (Azores),Portugal

NORMAN RATCLIFFE Royal Societyfor the Protectionof Birds The Lodge, Sandy, Bedfordshire SG19 2DL, United Kingdom Abstract.--Weanalyzed food provisioningof RoseateTern (Sternadougallii) chicks at a colony in the Azores, 16 Jun.-5 Jul. 1995. Prey offered to chicksand feeding frequencywere examined in relation to the age of the chick, fish size, and time of day. Overall, Trumpet fish (Macroramphosusscolopax) predominated in the diet, but sauri (Scomberesoxsaurus and Nanichthyssimulans), mackerel (Trachuruspicturatus), garfish (Belonebelone g*'acilis), and lanternfish(Myctophidae) were importanton certaindays. Mostly epipelagic fish were taken, but mesopelagicfish becamemore common prey as the seasonprogressed. Garfish were deliveredin earlymorning and lanternfishwere ingestedmostly around mid-day, when provisioningof other prey typeswas lower. Prey ingestionby chickswas influenced by age,with youngbirds ingestingthinner prey (sauri,garfish, and mackerel) and not consumingwider prey (trumpet fish and boarfish, Caprosaper). Feedingfrequency increased with chick age and varied significantlywith time of day, being higher in early morning and lower in early afternoon.Mean length of fish consumedby chicksincreased significantly with chick agefor common prey species.Variability of RoseateTern chick diet in the Azorescan be explained by changesin the availabilityof fish speciesand changesin the adults' selectioncriteria as their chicks grew.

PRESAS TRAIDAS A PICHONES DE STERNA DOUGALLII EN LAS AZORES Sinopsis.--Analizamosel aprovisionamientode los pichonesde Sternadougallii en una co- Ionia en Las Azoresen el 1995. Registramosla presa11evada por los adultos,la presatraida a los polluelos,y los pecesdejados caer por los polluelosdiariamente entre junio 16 y julio 5. Las presasofrecidas a los polluelosy la frecuenciaen la alimentaci6n se examinaron en relaci6n a la edad del polluelo, el tamafio del pescadoy la hora del dia. En general,peces de Macroramphosusscolopax predominaron en la dieta, pero individuosde Scomberesoxsau- rus. Nanichthyssimulans, Trachurus picturatus, Belone belone g*'acills y de la familia Myctoph- idae fueron importantesen algunosdias. Principalmente se tomaronpeces epipel•gicos, pero pecesmesopel•tgicos se hicieron presacomfin con el progresode la temporada.Individuos de Belonebelone g*'acilis se 11evarontemprano en la mafiana y los de la familia Myctophidae fueron ingeridosprincipalmente cerca del mediodia, cuando el aprovisionamientode otros tipos de presasera menor. La ingesti6n de los polluelosfu6 influenciada por la edad, con avesj6venes ingiriendo presasm•s finas (Scomberesoxsaurus, Nanichthyssimulans, Belone beloneg*'acilis y Trachuruspicturatus) y no consumiendopresas mgs anchas( Macroramphosus scolopaxy Caprosaper). La frecuencia de alimentaci6n aument6 con la edad de las crias y vari6 significativamentecon la hora del dia, siendomayor temprano en la mafianay menor temprano en la tarde. E1 largo promedio del pescadoconsumido por los pichonesaument6 significativamentecon la edad del pich6n para las especiescomunes de presas.La variabili- dad en la dieta de los pichones de Sterna dougallii en Las Azores puede explicarsepor los cambiosen la disponibilidadde especiesde pecesyen el cambioen loscriterios de selecci6n de los adultos segfin los pichones crecieron.

• Current address:Instituto Piaget, I.S.E.I.T./Mirandela, Av. 25 de Abril, 5370 Mirandela, Portugal.

419 420] J. A. Ramoset al. J.Field Ornithol. Summer 1998

RoseateTerns (Sterna dougallii) have a predominantlytropical distribution (Gochfeld 1983), nest in high densities (Ramos and del Nevo 1995), lay smallclutches (Langham 1968, Nisbet 1981), forage in inshore or nearby oceanic waters with particular physicalfeatures during the breeding season (Safina 1990a, Heinemann 1992), dive deeply (Nisbet 1981), and feed on relativelyfew prey species(Safina et al. 1990) in com- parison to tern specieswith a more temperate distributionsuch as the Common Tern (Sterna hirundo). However, delivery of prey to Roseate Tern chicks has been quantified only in Europe and North America (Langham1968, Richardsand Schew1989, Safinaet al. 1990). Preyitems of RoseateTerns in other parts of the world are poorly known, and sea- sonalvariation in prey items and the influence of chick age on diet composition have not been examined thoroughly in this species.The availability of appropriately sized prey fish is crucial for young chicks and important in explaining breeding success(Uttley et al. 1989). Moreover, examining the contributionsof date and chick age on diet composition can help to elucidateprey selectionin the absenceof knowledgeof actual prey availability. We worked on Vila Islet, a mixed Roseateand Common Tern colony in the Azores. Our primary purpose was to examine daily variabilityin prey deliveredand consumedby RoseateTern chicks,and then to relate this variabilityto chick age and time of day.The RoseateTern is a threat- ened seabird (Gochfeld 1983) and its feeding characteristicsin the Azores,which support the largestEuropean population (del Nevo et al. 1993), are unknown.

METHODS We studied prey brought to RoseateTern chickson Vila, an islet off Santa Maria Island, Azores (36.9øN,25øW) in 1995. At this site 236 pairs of RoseateTerns nested on a rocky and moderatelyvegetated area sur- rounded by 500-800 pairs of Common Terns nestingin more open areas. A blind waserected on a smallrocky plateau about 0.6 m abovethe colony overlookingthe densestRoseate Tern breeding area (approximately2 pairs/me). Nestswere individuallymarked in the studyarea and the peak of hatching occurred between 12-22 June. Prey items brought to the colonywere observeddaily from 0700-1900 h by observersworking 3-h shiftsusing 10 x 25 and 10 x 40 binoculars.Three methodswere used to collect data, each focusingon a particular feature of prey. Prey carriedby adults.--Fish carried by adults entering the studyarea were recorded each day between 16 June-5 July. We had a good view overlookingthe studyarea, but a few adults carrying fish at the margins of the studyarea could have flown out of sight and returned later, and thus might have been counted twice.Fish sizewas estimated in relation to the adult RoseateTern bill length (mean culmen -- 38.93 mm, SE = 1.56, n -- 30) and classifiedinto one of five size categories:(4 cm, 4-6 cm, 6-8 cm, 8-10 cm and )10 cm. Preydelivered to chicks.--Sevento eleven single-chickbroods situated 4- Vol.69, No. • RoseateTern Prey in theAzores [421

10 m away from the blind were watched between 20 June-5 July. Nests were fenced with 0.5-m high, 1.5-cm square mesh plastic wire to keep chicksnear the nest site. Dried grassaround the inner, lower portion of the fences helped both to retain and prevent injury to chicks. Fences were about 1 m in diameter and included natural or added cover so chickscould hide and find protection from the sun and rain. Each fence was marked with a numbered flag and chickswere ringed. One or two stone perches were positioned beside the fences to help adults to land and deliver the fish. We observedall prey deliveriesand recordedwhether chicksate or dropped fish and estimatedfish sizeas stated above. Virtually no fish were dropped by the parents or stolen by other adults (though some adultswith large fish were chassed).Prey not consumedby chicks refers to fish that were offered by the parents,but were dropped by their chicksafter attemptsto handle and swallowthe fish. We startedto record prey deliverieson day 1 becausefew deliverieswere observedon the day of hatching (day 0). Prey droppedby chicks.--Freshand dried fish dropped near nestsand chick feeding sites(hiding placesunder rocksor in thick vegetation)were collected between 16 June-5 July from 1900-2000 h, both from within chick enclosuresand in the colony area under observation.All dropped fish collectedwere from RoseateTerns. Although Common Terns nested nearby, their nestswere clearly in open areas and virtually no dropped fish were found in those areas.Entire and fresh fish specimenswere mea- sured (standardlength and width) and weighed.Dropped fish were identified using Whitehead et al. (1984) and served as a reference for fish observed in the field. We were familiar with the fish speciesin the area, having made occa- sional observationsin previousbreeding seasonsin this and other colo- nies, and having collected and identified fish dropped near the nests. Most fish specieswere distinctivein shape or color, and fish carried and deliveredby adultscould be comparedreadily with fish dropped near the nests. During the changeoverof observers,agreement was reached on identification of a few fish species.When positiveidentification was un- clear, only the family waslisted or the fish wasclassified as unidentified. The two speciesof Atlantic saury,Scomberesox saurus and Nanichthyssimu- lans could not be distinguishedand were pooled (=sauri). We quantified daily variationin prey itemsbrought to the colonyas the breeding seasonprogressed and daily proportions of each prey speciesin the diet. Seasonaltrends in diet were quantified using Spearmanrank correlation. We usedchi-square to evaluatevariation in prey consumption with time of day and ANCOVA to quantify dependence of feeding fre- quencyon time of day.The following categoriesof time of day were used: 1 = 0700-1000 h, 2 = 1000-1300 h, 3 = 1300-1600 h, 4 = 1600-1900 h. We also quantified the influence of chick age on diet compositionand assessedseasonal variation in prey consumptionafter controllingfor chick age. Chickswere divided into two age groups,young (1-7 days)and old 422] J. A. Ramoset al. J.Field Ornithol. Summer 1998

(8-20 days), and prey ingestedwere comparedfor each group between early (21-26 June) and late (27 June-3 July) season.

RESULTS

Prey carriedby adults.--Trumpet fish (Macroramphosusscolopax) predominated in the prey items carried by adults (Table 1). There was a significant decreasethrough the breeding seasonin the proportion of sauri (rs-- - 0.69, P < 0.001) and bluejack mackerel,Trachurus picturatus (rs -- -0.72, P < 0.001), whereasthe proportion of lanternfish (Myc- tophidae)increased (r s = 0.87, P < 0.001). Boarfish(Capros aper) and pilotfish(Naucrates ductor) comprised less than 5% of the preyitems with the exceptionof two and three days,respectively (Table 1). RoseateTerns brought mostlyepipelagic fish but, as the seasonprogressed, mesopelagic prey (for information on habitat and depth of fish speciessee Whitehead et al. 1984) increasedin importance (Table 1). Prey droppedby chicks.--Fishwere dropped by chicks at the time of delivery or during handling. Prey fell in crevices,cracks, or vegetation and could not be retrieved by parents. Qualitatively,fresh fish dropped at nestsprovided a fairly similar picture of daily variationin prey itemsas prey carried by adults,though samplesizes were smaller (Table 1). Sam- ple sizesdecreased as the seasonprogressed because large chicksdropped fewer fish. Quantitatively, the agreement was less marked because the percentageof dropped prey made of trumpet fish and boarfishwas much higher and the percentageof dropped prey made of sauri, mackerel,and garfish (Belonebelone gracilis) much smaller (Table 1). Trumpet fish was the most common prey carried by adults, delivered to chicks, and droppedby chicks(Table 2). Boarfishcomprised 12% of the itemsfound at nestsbut lessthan 1% of the deliveries(Table 2). These data suggest that somechicks were unable to handle and/or swallowtrumpet fish and boarfish of certain sizes. Boarfishand trumpet fish, commonlydropped by young chicks,were significantlywider (mean _+ 1 SD -- 19.2 --+2.61 and 11.9 _+ 1.57 mm) than sauri and garfish,which were rarely dropped (9.8 _+2.62 and 4.8 _+ 0.71 mm; t-tests,all P < 0.001). Sauriand garfishwere significantly longer (96.3 + 25.7 and 81.9 _+21.08 mm) than boarfishand trumpet fish (40.2 __+5.78 and 63.4 _+_7.18 mm; t-tests,all P < 0.001), showingthat young chickscan swallowlong but thin prey. Fishwidth and the existenceof an acute dorsal fin should be the main reasonswhy young chicks do not swallowboarfish and trumpet fish. Preydelivered to chicks.•Of the five major prey species,the proportion of sauri, mackerel, and trumpet fish delivered to chicks varied significantlywith time of day (X2 = 11.2, P < 0.01; X2 = 8.7, P < 0.05, and X• = 11.7,P < 0.01, respectively;df -- 3), but variedless than the proportion of garfish and lanternfish delivered (Fig. 1). Garfish decreasedsharply after mid-morning (X• = 25.3, P < 0.001, df = 3) and lanternfishshowed a peak around mid-day (X'• = 18.5, P < 0.001, df = 3). This peak corre- Vol.69, lx'o. • RoseateTern Prey in theAzores [423

I- • I• I © I © I © I• I © I © I 424] J. A. Ramoset al. J.Field Ornithol. Summer 1998

TABLE2. Comparisonof the three methods used to analyze diet of RoseateTern chicks. Fish dropped by chicks includes fresh and dry fish.

Prey carried by Prey delivered to Prey dropped by adults (16 June-5 chicks (20 June-5 chicks (16 June-5 Method July). n = 3894 July). n -- 1468) July). n = 521 Prey species No. records(%) No. records(%) No. records(%) Trumpetfish 2122 (55) 870 (59) 341 (66) Sauri 359 (9) 147 (10) 26 (5) Mackerel 391 (10) 133 (9) 28 (6) Garfish 191 (4) 61 (4) 7 (1) Lanternfish 613 (16) 180 (12) 43 (8) Boarfish 59 (2) 3 (<1) 61 (12) Pilotfish 81 (2) 32 (2) 3 (<1) Others and unidentified 78 (2) 42 (3) 12 (2) spondedwith the time when provisioningof other prey types,especially trumpet fish, waslower (Fig. 1). Provisioningrate varied with time of day (ANCOVA; F3.s• = 3.22, P < 0.02). Terns delivered more food between 0700-1000 h (mean = 1.05 feeds/h, SD = 0.558) than between 1300-1600 h (mean -- 0.84, SD -- 0.585; Tukey test, P < 0.04). There were no significantdifferences between 1000-1300 h (mean = 1.04, SD = 0.711) and 1600-1900 (mean = 1.02, SD = 0.664). Diet varied with age of chicks.The proportion of trumpet fish in the diet increasedup to 10 days(5 = 0.988, P< 0.01) and remained generally constantthereafter (Fig. 2), while the proportion of trumpet fish offered

• 40

• 30

• 2o

0700-1000 h 1000-1300 h 1300-1600 h 0 Sauri Mackerel GarfishLanternfish Trumpet fish [• 1600-1900h FIGURE1. Diurnal variation in proportionsof major prey consumedby RoseateTern chicks, pooled data 21 Jun.-5 Jul. 1995. Numbers abovebars indicate samplesize. Vol.69, No. • RoseateTern Prey in theAzores [425

8O 75 7O 65 • 60

• 45 • 3• • 30 • 25 • 20 -o- Sauri • 15 -O- Mackerel 10 -&- Garfish 5 -•- Lanternfish 0 .... 1-4 5-8 9-12 13-16 17-20 -&- Truml• fish Chickage class (days) FIGURE2. Relationshipbetween chick age and the proportion of severalmajor prey consumedby RoseateTern chicks.Sample size varied between3 and 9 chicks(median = 6). but not consumeddecreased (5 = -0.870, P < 0.01, Fig. 2). Conversely, the proportion of mackerel and garfish ingesteddecreased up to the age of 10 days (r•, = -0.82 and -0.94, P < 0.01) and the proportion of mesopelagicfish was nearly constant over the same period (Fig. 2). In comparisonwith trumpet fish, fewer sauri, mackerel, garfish, and lanternfishwere not consumed(3, 6, 5 and 7 items,respectively). These data suggestthat: (1) trumpet fish were commonlyavailable but not consumed by young (0-7 days) chicks, (2) sauri, mackereland garfishwere taken mostlyby young chicks,and (3) chicksof all agesate lanternfish equally. Chicksless than 7-d old alsowere offered trumpet fish, but 31.5% (n = 181) was not consumed whereas chicks 8-10-d old did not consume 13.4% (n = 187, X• = 16.4, P < 0.001, df = 1 with Yates correction). Trumpet fish offered to young chickswere grouped into <4 cm, 4-8 cm and >8 cm. The proportionof fish rejected (1.7%, 21.5% and 8.3%) and accepted (12.7%, 51.4%, 4.4%; n -- 181) differed significantlyamong thesecategories (X•= 17.18, P < 0.001, df = 2)' Youngchicks preferred fish <4 cm and avoidedfish >8 cm (observed/expected- 23/17.8 and 15/7.2, respectively).Mean length of fish ingestedby chicksincreased significantlywith age for trumpet fish (rs = 0.88, P < 0.001, n -- 20), mackerel (rs = 0.64, P < 0.01, n = 18), garfish (rs = 0.84, P < 0.001, n = 14), and lanternfish (rs = 0.51, P < 0.02, n -- 20) but not for sauri (r• -- 0.06, NS, n = 14). The proportion of sauri, mackerel, and lanternfish delivered to young chicksdecreased from early to late in the season,whereas the proportion of lanternfish and trumpet fish delivered to old chicksincreased (Table 3). Within each breeding-seasonperiod, parentswith young chicksdelivered mainly sauri and mackerelwhile parentswith old chicksdelivered trumpet fish. If prey brought to the colony depended largely on their 426] J. A. Ramoset al. J.Field Ornithol. Summer 1998

TABLE3. Comparisonof the numberof preyitems consumed by young(1-7 days)and old (8-20 days)chicks between two periodsduring the breeding season.

Young chicks Old chicks 21-26 27Jtme- 21-26 27June- June 3 July x2 P< June 3 July x2 P<

Sauri 95 17 52.9 0.001 24 12 3.4 Mackerel 45 27 4.0 0.05 21 22 0 Garfish 26 17 1.5 6 5 0 Lanternfish 31 16 4.2 0.05 21 101 56.3 0.001 Trumpet fish 46 78 7.8 0.05 146 414 127.3 0.001

availability,this suggestsimportant changesin the availabilityof prey species along the breeding season.

DISCUSSION Six prey specieswere important in the diet of RoseateTern chicksin the Azores. In North America, RoseateTern chicksand fledglingsrely on 3-4 species,of which the most important is sand lance, Ammodytessp. (Nisbet 1981, Richards and Schew 1989, Safina et al. 1990, Heinemann 1992, Shealer and Kress1994). Like terns in other parts of the world, RoseateTern chicksin the Azores consumedmostly epipelagic fish, although mesopelagicfish were an important dietary component. Meso- pelagicprey have not been documentedfor other RoseateTern popula- tions,but are taken by Common Terns (Granadeiroet al. 1995) and Yel- low-leggedHerring Gulls (Larus cachinnansatlantis) in the Azores(Ham- er et al. 1984). We compared three methods of assessingprey delivered to Roseate Tern chicks in the Azores. The observationof prey delivered and consumed should be the most accurate method of assessingchick diet, but it is time consuming.Also, becausethe speciesof fish delivered varies from morning to evening,observations should be spreadthroughout the day to avoid biases.Observation of prey carried by adults reflects prey selectedby adultsand has the advantagethat a large samplesize can be obtainedduring a shortperiod, but itemsdropped by the chickswill also be registeredin this type of sample.Records of fish dropped at nests overestimatesthe proportion of wider fish deliveredto smallchicks, and underestimatesthe proportion of preferred thinner fish delivered.Most dropped fish were found at nestswith young chicks;up to ten trumpet fish and boarfish were found daily at nestswith chicksof 1-5-d old. The method to be used depends,obviously, on the objectivesof the study.It is noteworthythat none of our methodsmissed any important prey types. Prey carried by adults and prey delivered to chicksgave similar results. Fish dropped at nestsmay be a quickerway of assessinggeneral prey types capturedby adultsin a particularyear, but they give a highly biasedview of what the chicksare ingesting(Randall and Randall 1978). However,it Vol.69, No..• RoseateTern Pr¸, in theAzores [427 will be usefulif the aim is to obtain a generalpicture of inter-colonyand inter-annualvariations in prey items. Changesin prey compositionthrough the breeding seasonobserved in our studyare common for terns (Frank 1992) including RoseateTerns (Richards and Schew 1989, Safina et al. 1990, Heinemann 1992). Echo sounderequipment has shownthat abundanceof smallprey fish is high- estwhen terns are feedingyoung chicks and declinessharply later in the season,which may coincidewith the arrivalof predatoryfish (Safinaand Burger 1985). We showedthat prey selectionis highlyinfluenced by chick age, with young birds being fed a greater diversityof smaller and/or thinner prey (sauri,garfish and mackerel) and rejectinglarger prey (es- peciallyboarfish and trumpet fish). We also detected an important sea- sonaldecline in the intake of sauriand mackerelafter controllingfor the effect of chick age,which seemsto agreewith the fact that smallprey fish decline later in the season(Safina and Burger 1985). Therefore, the variabilityin RoseateTern chick prey found in our studycan be explained by changesin the availabilityof prey speciesand changesin the bird's selectioncriteria as chicksgrew. Our study also showsdaily variation in prey composition.Some prey speciesmay have their own diet rhythms (or influencing factors) that make them differentially susceptibleto tern predation on a daily scale (Safina et al. 1990). Our resultssuggest that garfish were probablyeasy to capture in the morning, whereasthe availabilityof trumpet fish remained fairly constantthroughout the day.The capturerate of one prey type alsomay be influencedby the relativeavailability of other prey types. Lanternfishwere consumedmore often around mid-dayand later in the season,when deliveriesof other prey typeswere at their lowestand decreasedsharply, respectively. Although the abundanceof lanternfishmay have increasedas the seasonprogressed, these two points suggestthat it is taken when other prey were lessavailable. Two facts suggestthat lan- ternfishmight be more costlyto capture:(1) RoseateTerns were observed bringing lanternfishat the same time as Common Terns and thus they might have been feeding in mixed flocks;in this situationRoseate Terns are at a competitivedisadvantage (Safina 1990b); (2) RoseateTerns carrying lanternfishwere seen arriving from the open ocean and thus may have been feeding farther from the colony. Becauseyoung chicks cannot swallowlarge prey, the availabilityof smaller prey will determine the timing and successof breeding (Safina et al. 1988). In Shetland, low availabilityof appropriately-sizedsandeels, Ammodytesmarinus, has been the main causeof recent widespreadbreed- ing failure in the Common Tern and the Arctic Tern, Sternaparadisaea (Uttley et at. 1989). RoseateTerns may be better able than Common Terns to exploit smaller prey. Observationson three Common Tern broods close to our RoseateTern fencesrevealed that parents brought mainly trumpet fish and boarfish (86% and 9%, n -- 58). The seven chicks of these broods could not swallow fish this wide and died after 4- 6 days.Such differencesbetween Roseate and Common Terns could be 428] J. A. Ramoset al. J.Field Ornithol. Summer 1998 related to prey distribution;Roseate Terns feed more successfullyin smaller groupsover more dispersedprey than other terns (Duffy 1986, Safina 1990a, Shealer and Burger 1993) and may be better adapted to exploit prey over the subtropicalAzores seas.For both species,predatory fish may be important in driving prey to the surface (pers. obs.), similar to the situationin North America (Safina and Burger 1985, Safina 1990a). The smaller clutch size of Roseate Terns at Vila Islet (the clutch size has been monitored since 1990 and the annual mean has been around 1.1- 1.2 eggs/clutch) and other subtropicalcolonies when compared to that in more temperateareas (seeNisbet 1981) may be explainedby scattered foragingresources in thoseareas. These are interestingpoints for further research.

ACKNO•VLEDGMENTS

This studywas funded by the European Commision (contractLIFE B4-3200/95/351). We are indebted to Filipe Porteiro for fish identification and suggestionsrelated to fish biology. We thank alsoJos• Maria and his family for accessto Vila islet and Leslie Batty for com- menting on a draft. Earlier versionswere improved by the commentsof Dr. Jeffrey Spen- delow,Dr. Carl Safina, Dr. David Shealer and an anonymousreferee.

LITERATURE CITED

DEL NEVO, A.J., E. K. DUNN, F. M. MEDEIROS,G. LE GRAND,19. AKERS, M. I. AVEP,Y, AND I. MONTEIRO.1993. The statusof RoseateTerns Sternadougallii and Common Terns Sterna hirundo in the Azores. Seabird 15:30-37. DUFFY,D. 1986. Foragingat patches:interactions between Common and RoseateTerns. Ornis Scandinavica 17:47-52. FRANI•,D. 1992. The influence of feeding conditionson food provisioningof chicksin Com- mon terns Sternahirundo nesting in the German Wadden sea.Ardea 80:45-55. GOCHFELD,M. 1983.World Statusand distributionof the RoseateTern, a threatenedspecies. Biol. Conserv. 25:103-125. GRANADEIRO,J.P., L. R. MONTEIRO,R. W. FURNESS,AND M. C. SILVA. 1995. The food of Common terns Sternahirundo in the Azores.Proceedings of the 5th International Sea- bird Group Conference,Glasgow, U.K., March 1995. SeabirdGroup, U.K. HAMER,K. C., D. R. THOMPSON,A.J. RUNDLE,S. A. LEWIS,AND F. M. STE•NART.1994. Me- sopelagicfish eaten by Yellow-leggedHerring Gulls Larus argentatusatlantis in the Azores. Seabird 16:30-33. HEINEMANN,D. 1992. Foragingecology of RoseateTerns breedingon Bird Island, Buzzards Bay, Massachusetts.Final Report to the United StatesFish and Wildlife Service,Newton Corner, Massachusetts. HOLZLOHNER,S., ANDA. OP,LOWSKI. 1986. Hydroacousticstock estimation and biological resultson the JosephineBank in August1984. International Council for the Exploration of the Sea 57:27 pp. LANGHAM,N. P. 1968. The comparativebiology of terns Sternaspp. Ph.D. thesis.University of Durham, Durham, United Kingdom. NISBET,I. C. T. 1981. Biologicalcharacteristics of the RoseateTern Sternadougallii. Report to U.S. Fish and Wildlife Service, Newton Corner, Massachusetts. Massachusetts Audubon Society,Lincoln, Massachusetts. RAMOS,J. A., ANDA.J. DEENEVO. 1995. Nest-siteselection by RoseateTerns and Common Terns in the Azores. Auk 112:580-589. RANDALL,R. M., ANDB. M. RANDALL.1978. Diet of RoseateTern during the breeding season at St. Croix Island, Algoa Bay. Cormorant 5:4-10. RICHARDS,S. W., ANDW. A. SCHEW.1989. Speciescomposition of food brought to Roseate Tern chicks on Falkner island, Connecticut in summer 1984. Connecticut Warbler 9:1-5. Vol.69, No. 3 RoseateTern Pr¸, in theAzores [429

SAFINA,C. 1990a. Foraginghabitat partitioning in Roseateand Common terns.Auk 107:351- 358. ß 1990b. Bluefish mediation of foraging competition between Roseateand Common terns. Ecology71:1804-1809. , •'•D J. BU•tCER.1985. Common Tern foraging:seasonal trends in prey fish densities and competitionwith bluefish. Ecology66:1457-1463. , J. BuRcEg,M. GOCHV•LD,^ND R. H. W^•N•g. 1988. Evidenceof prey limitation of Common and Roseatetern reproduction. Condor 90:852-859. , R. H. WAGNER,D. A. WITTING,AND K.J. SMITHß1990. Prey delivered to Roseateand Common tern chicks;composition and temporal variability.J. Field Ornithol. 61:331- 338. SHEALER,D. A., ANDJ. BURGER.1993. Effectsof interference competitionon the foraging activityof tropical Roseateterns. Condor 95:322-329. , ANDS. W. KRESS.1994. Post-breedingmovements and prey selectionof RoseateTerns at Stratton island, Maine. J. Field Ornithol. 65:349-362. UTTLEY,J., p. MONAGHAN,AND S. WHITE. 1989. Differential effects of reduced sandeel availability on two sympatricallybreeding speciesof tern. Ornis Scandinavica20:273-277. WHITEHEAD,P.J.P., M. L. BAUCHOT,J. C. MUREAU,J. NIELSEN,A•ND E. TORTONESE.1984. Fishes of the North-eastern Atlantic and the Mediterranean. Vols. 1-3. U.N.E.S.C.O. Paris. Received18 Dec. 1996; accepted27 May 1997.