Age, Resource Availability, and Breeding Effort in Brandt's Cormorant

AGE, RESOURCE AVAILABILITY, AND BREEDING EFFORT IN BRANDT'S CORMORANT

ROBERTJ. BOEKELHEIDEAND DAVID G. AINLEY PointReyes Bird Observatory, 4990 Shoreline Highway, Stinson Beach, California 94924 USA

ABSTRACT.--Wegathered life-history data on banded Brandt'sCormorants (Phalacrocorax penicillatus)at SoutheastFarallon Island, California, from 1972 to 1984, and documented breedingperformance as affectedby age and annual variation in food availability. Femalesbred at a youngerage than males,but did not live as long. Birdsof both sexes that bred at least once bred the samenumber of years.Mate fidelity was low (9%) because of poor synchronyin the arrival of matesand low site fidelity by females.Prior breeding experiencehad little influence on reproductivesuccess. The most successfulindividuals fledged10-20 chicksover their lifetime, and averaged2.5 chicks/breedingyear over four to eight years. Cormorantsexperienced significant interannual differencesin food availability. "Poor" food yearsoccurred frequently and, consequently,all adultsskipped breeding at leastone seasonduring their reproductivelifetimes. Food availability also affected age at firstbreeding, as well as age-relatedbreeding phenology, reproductiveeffort and success,and return rates of banded juveniles and adults. The reproductive traits of Brandt's Cormorant allow it to exploitan unpredictableenvironment; previously proposed life-history models relegate cor- morantsto stableenvironments. Received 5 May 1986,accepted 24 January1989.

MANYvertebrates can adjustreproductive ef- wind-driven transportof surfacewaters for nu- fort to the amountof energyavailable each year trient input (through upwelling and advec- (Giesel1976). Schaffer (1974) hypothesized that tion). Considerable interannual fluctuation in life-history characteristicsin variable environ- oceanographicclimate, biologicalproductivity, ments depend on when significant mortality and developmentof the food web is character- occursduring the life cycle. If variable condi- istic of these systems(Glantz and Thompson tions lead to variable success,then breeding 1981, Chelton et al. 1982, McLain and Thomas effort should be adjustedto insure survival until 1983, McLain et al. 1985). The most dramatic of the next favorable year. If conditionsinstead the periodicanomalies in productionare known lead to variable adult survival, then breeding as E1Nifio-Southern Oscillations (ENSO). Dur- effortshould be increasedin orderto reproduce ing ENSO, biological productivity decreases immediately, before death occurs. radically and the food web collapses(Glantz In seabirds,which are long-lived and thus and Thompson1981, McLain et al. 1985,Ainley shouldhave ample opportunityto adjustbreed- and Boekelheide 1990). ing effort, many life-history traits are strongly The seabirdsof easternboundary currents are influencedby ageand this relationshiphas been abundantand unique,with a high incidenceof the subjectof much research.Age-influenced endemismin respectiveavifaunas (Brown 1976, traitsinclude the proportionof individualsthat 1980). Brandt's Cormorant (Phalacrocoraxpeni- breed, nest-site characteristics, mate selection, cillatus)is unique to the California Current, but phenology, clutch size, and various compo- is less abundant than its ecological counter- nents of reproductivesuccess (see review by parts,the GuanayCormorant (P. bougainvillii)of Ryder 1980, and more recent studies:e.g. Pu- the Peru Current and the Cape Cormorant (P. gesek 1983, Pugesekand Diem 1983, Ainley et capensis)of the BenguelaCurrent (Ainley and al. 1983, Shaw 1986). Yet, interannual variations Boekelheide1990). The latter two speciesare in breeding behaviorand success,as influenced important commercially becauseof the guano by age, have receivedscant attention. they produceand they have been the subjectof The California Current is one of the world's appreciablestudy. Little work hasbeen directed five major eastern boundary-current systems. toward Brandt's Cormorant. We collected in- These oceansystems are the most biologically formationon the reproductivesuccess of known- productive in the world, and depend upon age Brandt'sCormorants from 1972to 1984,and

389 The Auk 106: 389-401. July 1989 390 BOEKELHEIDœANDAINLœY [Auk,Vol. 106 analyzed patterns in relation to food availabil- SEFI nestedwithin 300 m of the natal colony,but a ity to determine the validity of Schaffer'shy- few nestedas much as 1 km away. On the basisof pothesiswith regard to this species. sitefidelity information,we assumedthat oncea bird nestedwithin a colony it was unlikely to emigrate. In supportof this assumption,Potts (1969) found that STUDY ARF• AND METHODS only 8% of Shags(P. aristotelis)breeding for the first We studied Brandt's Cormorants on Southeast Far- time nestedoutside their natalcolony, and that < 1% allon Island (SEFI), 48 km west-southwestof San Fran- movedafter havingbred once. Consequently, return cisco,California, at 37ø42'N,123ø00'W, the largestof ratesof establishedbreeders more accuratelyreflect the South Farallon Islands. Between 1971 and 1983, true survival rates. cormorantnumbers averaged 16,000 and ranged be- We usedfour principal criteria to rate yearson the tween 4,300 and 23,800individuals (Ainley and Boe- basisof food availability:(1) tendencyof cormorants kelheide 1990). to feed within 5 km of SEFI (vs. as much as 80 km We observedknown-age individuals from a blind away),(2) proportionof juvenilerockfish (Sebastes spp.) overlookinga colony of about500 pairson the north- in the diet of Farallon piscivorousbirds (including west shoreof SEFI (for details,see Ainley and Boe- Brandt'sCormorants), (3) breeding populationsize of kelheide 1990). Bet-ween1970 and 1983,workers from Brandt'sCormorants relative to the 13-yrmean, 1970- Point ReyesBird Observatory(PRBO) banded 5,866 1983,and (4) the occurrenceof significantnest aban- cormorantchicks in the studycolony. From 1971on, donmentduring the breedingperiod (see Ainley and each chick was given two bands:a USFWS individ- Boekelheide1990 for moredetails). Subsequent direct ually numberedmetal band on one leg and a colored, assessmentof prey abundanceand distributionin wrap-around plastic band on the opposite leg. Leg 1985-1988 indicates that these criteria were accurate position, band number, and band color designated (unpubl. data, PRBO and National Marine Fisheries year class.We banded chicks at night to minimize Service,Tiburon Laboratory).Years of "good" food gull predation on nestlingsand eggs.In 1970-1978 availabilitywere 1977,1979, and 1981;1980 and 1982 (except 1971) we used aluminum bands; those used were "moderate"years; and 1976,1978, and 1983(all in 1971 were monel and all were soon lost. After 1978 yearsof oceanographicanomaly; McLain et al. 1985) we usedstainless-steel (incoloy) bands, and through were "poor" years. 1985 none had fallen off. We corrected for loss of RESULTS aluminum bands from estimates made on the double- banding system. We monitored banded birds daily from April to Age of first sightingand first breeding.--Onav- August in 1972-1985.We read band numberswith a erage,female and male Brandt'sCormorants first 20-60x spotting scope, mapped the location and returned to the breeding colony at the sameage movement of each bird, recorded the condition or (female 2.7 + 1.3 yr, n = 96; male 2.9 + 1.2 yr, absenceof metal and colorbands, described plumage, n = 212; t = 0.91, df = 306, P > 0.2; Fig. 1). The and noted sexuallydistinct behaviors. We noted the modal age of first sighting for both sexeswas date when each bird began to occupy a site contin- 2 yr, but < 5% of birds of either sexreturned as uously, date of pairing, egg-layingand chick-hatch- 1-yr-olds. Among all birds eventually sighted, ing dates,and nest condition. From 1979 to 1983,we about90% returned by their fourth year. alsorecorded date of first arrival in spring. Despitesimilar agesof return, malesand fe- We calculatedage at firstbreeding, effects of breeding experience,and lifetime breeding recordsonly malesdiffered in the averageage of first breed- for individuals with completeannual records.Vari- ing (Fig. 1). Femalesfirst bred at 3.5 __1.5 yr (n ation aboutmeans are given as + 1 standarddeviation. = 40), and males at 4.2 + 1.3 yr (n = 70; t = Breeding chronology, clutch size, and reproductive 2.61, df = 108, P < 0.02). The modal age of first successwere analyzedfor similarityof variance(AN- breeding for females was 2 yr, but only 2 out OVA) and between-agedifferences were testedusing of 131 two-yr-old males bred. The modal peak the Student-Newman-Keuls multiple comparison for males occurredat 4 yr. (SNK) (Sokal and Rohlf 1969). The averageage of birds sightedor breeding We sexed cormorants according to criteria dis- for the first time varied annually; many young cussedby Williams(1942). We excludedfrom analyses birds appearedand attempted to breed in years birdsthat we were unableto sex,visited only a few times,or nestedout of sightwithin the colony. of good food availability but few or none did We determined return rates of banded birds after soin poor years.For example,in the 1983ENSO correctingfor band loss.Because of an unknown rate year, no 2-yr-olds (out of 600 chicksbanded in of emigration,return rateswere consideredminimal. 1981) were observed. This contrastswith 1979, The majorityof bandedbirds observed elsewhere on an exceptionallygood year, when we observed July1989] BiologyofBrandt' sCormorant 391

60 of the 592 birds banded as chicks in 1977. In 1979, 64% (n = 11) of 1977-cohort females that eventually bred did so as 2-yr-olds. Breedingphenology.--Spring arrival datesvar- n= 70 ied strongly with age among males (Table 1). 402O t The oldest appearedfirst, and usually by mid- z to late March. Peak arrival for previousbreeders fell between mid-April and early May. Few 4o malesvisiting or breeding for the first time arrived before May, and many, particularly 2-yr- - n-• 40 olds, arrived during June and July. Therefore, 20- youngestmales arrived 2-8 weeks later than older birds. The pattern was altered in yearsof poor food availability. No males occupiedsites I until late April, and all arrived during a brief AGE AT FIRST BREEDING period extending through mid-May. Arrival datesof femaleswere not clearly strat- ified by age except that youngest females arrived last (Table 1). Most females old enough to have bred previously(> 3 yr) arrived slightly n=212 later than similarly aged males. The oldest fe- 20- malesdid not arrive any earlier than othersthat had bred before. Youngestfemales (i.e. 2- and z 3-yr-oldsnearly all of whom were visiting for the first time) did not arrive until late May and June, a pattern similar to that of young males. Laying datesfollowed the pattern of arrival, and younger birds laid later (Table 2). Males n=96 comprisedtwo groups,older individuals who had bred at least once and younger birds who were breeding for their first or occasionallythe secondtime. Youngestfemales began to lay significantly later than older ones;the oldest fe- I males laid earlier than all other age groups. AGE AT FIRST SIGHTING Clutch initiation dates among youngest birds Fig. 1. Age of first sighting and first breeding of of both sexes exhibited a large coefficient of Brandt's Cormorants banded 1972-1977. variation, which indicated little laying syn- chrony. likely due to the early arrival of oldest males, The elapsedtime between arrival and laying which occupied sites in March, weeks before measuredhow quickly a bird could pair, com- females appeared.Early males did not occupy plete nest building, and begin a clutch. Among their sites continuously, and spent less time femalesduring the period 1979-1982,age groups during initial visitsthan did later-arriving males. did not differ (Table 1), and the arrival-laying Nestand mate choice. --The percentageof pairs interval averaged13.6 + 5.8 days(n = 48, range with older males (56%) outnumbered pairs with 5-30 days). According to Grau (1984), Brandt's older females (21%) and those with mates of the Cormorantsrequire 16 days for yolk develop- sameage (24%;n = 34 known-age pairs).In 65% ment, plus additional time for albumen syn- of pairs, mateswere within one year of age. thesis and shell deposition. Apparently, most Among the 34 known-age pairs, only 3 re- females began to form yolks before they first mainedintact in following years,and none per- arrived at the colony. sisted longer than two years. The 3 pairs that Among males,the time between arrival and remated used the same nest site during both the date their mate laid the first egg was longer years.In 24%of the dissolvedpairs (n = 8), both among the oldest individuals. This disparity is mates returned, but in 56%, one failed to return. 392 •OEKELHEmEAllIV AIlll•E¾ [Auk,Vol. 106

TAnI•E1. Mean arrival dates(SD in days) and number of days elapsedbetween arrival and clutch initiation for known-age Brandt's Cormorants, 1979-1982.

Age (yr) Arrival date n $1NIKa Days elapsed n SNK• Males 2 19 Jun + 23.2 22 A 0 3-4 31 May _+22.6 66 B 24.7 _+ 9.4 9 A 5-6 11 May _+ 14.3 55 C 23.7 _+ 11.1 28 A 7-8 3 May _+ 18.8 72 D 27.5 _+ 10.1 34 A 9-12 21 Apr _+28.0 39 E 37.8 _+ 17.3 24 B Females 2-3 7 Jun _+ 15.6 18 F 11.2 _+ 6.9 5 C 4-5 18 May _+ 18.2 18 G 12.6 _+4.3 10 C 6-7 9 May _+ 13.4 21 G 14.5 + 5.7 19 C 8-10 9 May + 16.6 17 G 14.2 _+ 6.8 14 C • Student-Newman-Keulsmultiple comparison; values with the sameletter do not differ significantly,P > 0.05.

Males, more than females, used the same nest turning to the colony. When a male returned, site year after year (Table 3). From 1980to 1982, his previous site had to be vacant or he had to 70% of returning males (n = 110) and 32% of successfullyevict the new occupant.Fights be- returning females(n = 28) reoccupiedtheir pre- tween males were common, even when many vious site. The percentageof males that reoc- suitable sites existed nearby. It appeared that cupledthe samesite increased with age.Among suchfights ensuedwhen a previoussite holder 9- to 12-yr-oldbirds, reoccupation was two times encountereda strangerat his spot.Fights ranged more likely than for 3- and 4-yr-olds.We found from minor threats to battles that continued for no pattern among females, but this may have severalhours or sporadicallyover severaldays. been a consequenceof a small sample. If we Femalesoccasionally fought for sitesif theywere include in the sample females that returned to displacedfrom where they had bred the year a nest that was different but within the same before, but their fights were rarely as intense subcolony(usually within 10 m of the previous as those between males. year'snest), then they indeed exhibited no pat- Breedingeffort and success.--Nesting space was tern. not limited during this study (though it likely The majority of Brandt'sCormorants did not was historically;Ainley and Boekelheide1990), reinate in successive seasons. Even if both birds so any bird that so desiredcould have occupied survived the winter, only a synchronousarrival in spring would favor remating, since pair bonding took place quickly among birds re- TABLE3. Fidelity (percentageof returnees)of known- age Brandt's Cormorants to nest sites, 1980-1982.

TABLE2. Mean clutchinitiation datesfor known-age Different site Brandt's Cormorants, 1972-1984. Same Same Different Age (yr) n sitea subcolony subcolony Age (yr) Dates (œ + SD) n SNK • Males b Males 3-4 9 44 33 22 3-4 2 Jun _+ 23.3 60 A 5-6 29 59 31 10 5-6 29 May _+ 15.1 72 A 7-8 41 71 22 7 7-8 22 May _+ 11.9 60 B 9-12 31 87 10 3 9-12 18 May _+ 12.0 43 B Females c Females 3-6 9 56 33 11 3-4 5 Jun _+ 17.7 27 C 7-10 19 21 68 11 5-6 27 May _+ 15.5 27 D 7-8 23 May + 14.1 35 D • Return ratesof malessignificantly greater than females(X • = 13.62, 9-12 12 May _+ 12.3 16 E df = 1, P < 0.001). bReturn ratessignificantly different by age (X• = 8.91,df = 3, P < ' Student-Newman-Keulsmultiple comparison;values with the same 0.O5). letter do not differ significantly,P > 0.05. ßReturn rates no differentby agegroup (X • = 3.33,df = 1, P > 0.05). July1989] BiologyofBrandt' s Cormorant 393

/•= 5 87 52 4,1 21 22 27 14 I0 ß [] OCCUPY I00'BO* FEMALES • NEST

60, [] OBTAINMATE

40' I EGGSLAID

20 I CHICKSFLEDGED I- z uJ

uJ D- I•= 5 I$1 141 134 99 70 57 57 32 23 12 7

IO0t MALES

20 ÷

I 2 3 4, 5 6 7 8 9 I0 II 12 AGE (YEARS)

Fig. 2. Percentagesof known-age male and female Brandt'sCormorants who occupiednest sites,paired (constructeda substantialnest), produced eggs,and fledged at least one chick; n = sample size within each age class.

a site, regardlessof age. Nest-site occupation occupieda site. In fact, the failure of 2-yr-olds increasedwith age, and reached a plateau of to breed was largely due to their inability to 90-100%among males at least5 yr 01d(Fig. 2). obtain a mate.Among older malesthat had bred Only 51% of 2-yr-old malesoccupied sites; the at leastonce, 14-36% (dependingon age class) remainder merely wandered through the col- did not obtain mates. In food-poor years few ony or roostednearby. A sharp increasein site malesobtained mates, regardless of age.In bet- occupationwas evident between ages 2 and 4 ter years, some males did not pair, including yr, and nearly all surviving malesthat had bred individuals that occupiedsites for 2-3 months before occupiedsites in subsequentyears. and advertiseddaily for passingfemales. The percentageof maleswho obtaineda mate, The percentage of males whose mates laid or who remained until nest construction had eggs,and who eventually fledged chicks,also progressed appreciably, increased steadily increased with age, especially among the throughall agesup to 10 yr (Fig. 2). The youn- youngest age classes(Fig. 2). Youngest males gest maleswere unlikely to pair, even if they were unlikely to fledge chicks.Among older 394 ]•OEKELHEIDEAND AINLEY [Auk, Vol. 106

•+•u

•+•<

+•< July 1989] Biologyof Brandt'sCormorant 395

males,less than 50% fl•dged chicks. Curiously, TABLE5. Incidence of relaying and successof re- the percentageof malesthat succeededdropped placement clutches by known-age Brandt's Cor- for both 9- and 12-yr-olds,in part becauseof morants, 1972-1984. Sample sizes in parentheses.

the large proportion that reachedthese ages % lost 1st % relaid Mean no. chicks during the strong 1982-1983 ENSO. Age clutcha after loss fledged/nest Among females,the youngestbirds (ages2- Male 4) differed distinctly from older birds in the 3-4 52 14 0.0 _+ 0.0 (2) proportion that paired, laid eggs,and fledged 5-6 26 38 1.2 _+ 1.6 (5) chicks (Fig. 2). Youngestbirds were usually 7-12 19 42 1.0 _+ 0.9 (8) either visiting the island or breeding for the Female first time. Lessthan 40% paired and approxi- 2-3 36 30 0.5 _+ 0.7 (2) mately 25% laid eggs.By 5 yr of age, breeding 4-5 27 33 1.0 _+ 1.4 (2) effort becameconsistent across age classes,but 6-9 20 80 2.3 _+ 0.9 (8) even among the oldest females >20% did not • On the basisof data from Table 4 (no. nestswith eggs). lay eggsand ca. 30% did not fledge chicksbe- causeof the general reproductive failure of all birds in food-poor years. of eggsthat disappearedat expectedhatch dates, All females (ages grouped) were more suc- or that never hatched despite full-term incu- cessfulthan malesin pairing, obtainingeggs, bation, did not differ among age or sex classes. and fledging chicks(Fig. 2). This was particu- In contrast to hatching success,the propor- larly evident between ages 5-9 yr, when fe- tion of chickssuccessfully fledged did not differ maleswere much more likely to fledgechicks. relative to parental age (male X2 = 2.56, df = 3, This patternmay relate to the tendencyof males P > 0.25; female X2 = 2.00, df = 3, P > 0.25; to pair with femalesyounger than themselves, Table 4). Youngestbreeders lost a higher per- asnoted previously. Perhaps more importantly, centageof chicks0-10 daysold, suggestingdif- our sampleof malesincludes a greaterpropor- ficulty with the transition from incubation to tion of individuals presentduring food-poor chickcare. Chicks 0-10 daysold of older parents years when few females returned. As a result fared no worse than chicks 11-20 days old. most of the males present then remained un- Chicks older than 20 days experienced little paired. mortality regardlessof their parents' age. Clutchsize.--Clutches laid by 2- and 3-yr-old Overall breedingsuccess also differed among femaleswere smallerthan thoselaid by females age classes(male X2 = 21.21, df = 3, P < 0.01; older than 5 yr (Table 4). No 2- or 3-yr-old fe- female X2 = 10.00, df = 3, P < 0.05). The number males laid four-egg clutches,whereas 10% of of chicksfledged per nestwas significantly low- thoselaid by 4- and 5-yr-olds,36% of thoselaid er for youngestbreeders, but thereafter perfor- by 6- and 7-yr-olds,and 43%of thoselaid by 8- mancedid not improve with age (Table 4). to 10-yr-oldswere four-egg clutches.We ob- Older birds were much more likely to relay servedonly one five-eggclutch and it was laid if their first clutch was lost, and were similarly by a 6-yr-old. moresuccessful at fledgingchicks from replace- Among known-age males, clutch sizes of the ment clutches (Table 5). Older breeders laid first oldestbirds (9-12 yrs)were significantly larger clutchesearly in the laying periodand had more than clutchesof younger birds (Table 4). Mates time to relay if their first clutch was lost. Only of the oldestmales produced clutches equal in one known-age female, a 9-yr-old, laid two re- size to thoselaid by older females,confirming placementclutches in one season.This occurred that older males are likely to pair with older in the good-foodyear of 1979. females. Breedingsuccess (interannual variability).-- Breedingsuccess (all yearscombined).--Hatch- Comparisonsamong years of differing food ing successdiffered by age for both males and availability reveal that age classesdid not fare females (male X2 = 25.86, df = 3, P < 0.01; female equally (Table 6). Youngestbirds bred only in X2 = 10.53,df = 3, P < 0.02;Table 4). Younger the most favorable years. The proportion of birds had poor hatchingsuccess because they young females that laid eggs declined from lost eggs,either through nest abandonmentor >33% of thosepresent in good yearsto < 10% incidental lossduring incubation. The number in moderate to poor years. The proportion 396 •œr,•aDE ANDAINLEY [Auk,Vol. 106

TABLE6. Breedingeffort and successof Brandt'sCormorants in years of different food availability.

Males Females

Age: 3-4 5-6 7-12 2-3 4-5 6-10 Good years (1977, 1979, 1981) No. birds observed 95 66 67 34 30 37 % attempting breedinga 31 47 66 35 57 81 Mean clutch size 3.0 + 0.7 2.8 + 0.7 2.9 + 0.6 2.6 + 0.5 2.9 + 0.3 3.3 + 0.7 n 22 26 36 9 ! 1 24 % fledging chicksb 17 35 49 15 47 73 Mean fledglings/pairc 1.1 _+1.1 1.3 + 1.0 1.6 _+ 1.2 1.0 + 1.3 1.4 + 1.0 1.9 _+0.9 n 29 31 44 12 17 30 Moderate years (1980, 1982) No. birds observed 40 43 66 15 6 26 % attempting breedinga 13 65 67 7 83 81 Mean clutch size 2.0 _+ 0.0 3.1 + 0.5 3.2 + 0.6 2.5 _+ 0.7 3.3 + 0.5 3.4 + 0.5 n 2 22 37 2 4 16 % fledging chicksb 0 47 55 7 50 77 Mean fledglings/pairc 0.0 1.2 + 1.0 1.4 _+1.0 1.0 1.0 + 1.2 1.7 + 0.9 n 5 28 44 1 5 21 Poor years (1976, 1978, 1983) No. birds observed 95 32 26 35 17 12 % attempting breedinga 16 28 12 9 12 25 Mean clutch size 2.7 + 0.7 2.6 + 1.0 3.0 + 0.0 2.3 + 0.5 1.0 3.0 + 1.4 n 11 7 2 4 1 2 % fledging chicksb 4 16 0 3 0 17 Mean fledglings/pairc 0.3 + 0.6 0.6 + 0.5 0.0 0.7 + 1.2 0.0 0.7 + 0.6 n 15 9 3 3 2 3

No. with eggs/no.occupying sites x 100. No. fledgingchicks/no. occupying sites x 100. Includesonly thosethat had eggs. amongyoung males with eggsdropped by 50%. to be "experienced."Though our sampleswere In contrast,the percentageof older birds that small, we found no definitive increase in clutch laid eggs did not decline between good and size or breeding successwith increasedexpe- moderateyears. The percentageof middle-aged rience (Table 7). birdsthat laid eggseven rosein moderateyears. Individualperformance.--The average bird bred The trends in clutch sizes were consistent with 2-3 seasonsand fledged 2-4 chicksduring its those discussedabove (see Clutch size), but sam- lifetime (Table 8). This was due not only to ple sizeswere too small to test statisticallyin mortality,but alsoto the frequencywith which any but the good-foodyears. Then, only the birds skippedbreeding. Unfavorableyears oc- youngestfemales laid clutchessmaller than curredoften enoughthat 75%(n = 20) of cor- those of the oldest females (SNK test, P < 0.05). morantsthat bred four or more years skipped The data for moderateand poor years indicate breeding at least once following their first the possibility that between-age-classdiffer- breedingyear. Nearly all birds failed to breed encesmay increaseas resourcesdiminish, but during 1978and 1983,and all known-agebirds more data are required to test this hypothesis. that laid eggsduring those years later aban- Older birds fledged equivalent numbers of doned their efforts. youngin both goodand moderateyears, where- Somebirds greatly outperformed the average asyoungest birds fledged young in goodyears values. The stellar male reared 20 chicks to in- only. In poor years,few birds of any age at- dependence,an averageof 2.5 chicks/yrfor the temptedto breedand they failed. 8 yr in which he bred.Several other males bred Effectof breedingexperience on success.--We 4-7 yr and reared9-12 chicks.The two females considered those birds that had attempted with the bestrecord bred in 4 and 5 yr, fledged breeding(i.e. producedeggs) in a previousyear 10 and 12 chicks, and averaged 2.5 and 2.4 July1989] Biologyof Brandt'sCormorant 397

T^nI•E7. Clutch sizes and breeding successof known-age Brandt'sCormorants with different levels of experience.

Clutchsize agePercent-eggs age Percent- chicks Fledglings/nest Sex Age Year as breeder Mean n hatched fledged Mean n Male 2 First 2.0 + 0.0 2 25 100 0.5 + 0.7 2 3 First 2.7 + 0.6 20 42 67 0.7 + 0.9 24 4 First 3.0 + 0.6 23 37 77 0.8 + 1.1 25 Second 2.9 + 0.6 8 61 50 0.9 + 0.8 8 5 First 2.8 + 0.7 21 61 78 1.3 + 1.0 21 Second 3.1 + 0.6 10 46 74 1.2 + 1.1 12 Third 3.0 + 0.0 2 83 60 1.5 + 0.7 2 Female 2 First 2.7 + 0.5 15 45 174 0.8 + 0.7 17 3 First 2.0 + 0.6 6 30 83 0.6 + 1.1 8 Second 3.0 + 0.0 2 67 75 1.5 + 0.7 2 4 First 2.6 + 0.5 7 67 63 1.1 + 0.9 9 Second/third 2.5 + 1.3 4 40 100 1.0 + 1.2 4 5 First 3.0 + 0.7 5 55 80 1.6 _+ 1.1 5 Second/third 3.0 + 0.0 4 55 67 1.0 _+ 0.8 4

chicks/yr,respectively. One femalelaid 26 eggs males from those cohorts, however, were still during 7 yr, including two replacementclutch- alive in 1988. es, but she fledgedonly 8 chicks.Birds of both In contrast,only slight decreasesin return sexesaveraged the same number of breeding ratesoccurred following the 1976ENSO for birds years despite the earlier recruitment of females of the 1972 and 1973 cohorts(Fig. 3). Return (Table 8). This phenomenonappeared to be re- ratesof adultsin 1979,following the 1978warm- lated to a longer reproductivelife span, more water event, were amongthe highestobserved, years skipped by males, or both (see Discus- even though nearly all Brandt's Cormorants sion). abandonedtheir breeding effortsmid-season in Returnrates of banded birds.--We subsequently 1978 (Ainley and Boekelheide 1990). As noted observed 17% of the 2,876 chicks banded from in the latter report, the Farallon population re- 1972 to 1977 as adults (i.e. birds ->2 yr of age). bounded immediately in both 1977 and 1979 The 1976 cohort returned at a much lower rate when prey populationsreturned in abundance. thanthe five othercohorts (Fig. 3). A weakENSO These data reveal that catastrophicadult mor- occurred in 1976, and an aberrant warm-water tality did not occurduring theseyears, in con- year occurredin 1978 (McLain et al. 1985). Food trast to 1982-1983. availabilitywas poor during bothyears (Ainley and Boekelheide 1990), and this must have af- fectedpostfledging and prebreedersurvival. A TABLE8. Number of years in which known-age stronger ENSO occurred in 1972-1973, but it Brandt's Cormorants bred. ended by the fall of 1973and was followed by Males (n = 73) Females (n = 38) 2 yr of good food availability. Years All age and sex groups exhibited lowest re- bred No. % No. % turn rates during 1982-1984, with the sole ex- ! 32 44 17 45 ceptionof 1975 males (Fig. 3). Theselow rates 2 12 16 9 24 coincidedwith the 1982-1983ENSO, a period 3 16 22 5 13 4 4 5 3 8 when nearly all Farallonspecies failed at breed- 5 4 5 3 8 ing and dispersedwidely from the Gulf of the 6 3 4 Farallones(Ainley and Boekelheide1990). Mor- 7 1 1 1 3 tality during this period was exceptionallyse- 8 1 vere for older females,as all 1972-to 1975-year- Median 2 2 class females failed to return after 1983. A few Mean 2.4 + 1.7 2.2 + 1.4 398 BOEKELHEIDEANDAINLEY [Auk,Vol. 106

I00 / • olds survived than did those first visiting as 2-yroolds,but we cannot samplebirds that do not visit the island to prove this.

1973 Among2- to 7-yr-olds,males averaged slight- t7=4•4 (0.22) ly higher return ratesthan females,but the dif- 50t ferencewas not significant(male 0.82 + 0.06, female0.79 + 0.06;Mann-Whitney U test,P > 0.05).After 7 yr of age,females exhibited a sub- stantialdrop in return rates;only a few banded femalesin the cohortsunder investigation lived beyond 10 yr. Males also exhibited a decrease in return ratesat 9 yr of age.In large part, the 1975 precipitousdecline in return ratesamong older /7=676(0-20 birds is due to the effectsof the 1982-1983 ENSO, asreported above. Whether significant numbers of BrandifsCormorants survive beyond these agesmust remain for future study.

DISCUSSION

Brandt's Cormorants are endemic to the Cal- 1977 ifornia Current. Consequently,they display /• = 592 (0.18) severallife-history traits appropriatefor maxi- mizing lifetime reproductivesuccess in the an- nuallyvariable environment. For a seabird,they YEAR have a relativelyyoung age at first breeding, Fig. 3. Annualreturn ratesof Brandt'sCormorants which enablesyoung birds to respond to fa- >-2 yr old for 1972-1977 cohortsin 1975-1985;solid vorableconditions. Second, there is a largein- lines = males, dashed lines = females (no banded terannualvariation in the age of first breeding, 1976 females seen). which resultsfrom youngbirds breeding in favorable years but delaying breeding in poor years. They demonstratelow mate fidelity, All years combined, females returned at a which permits immediateremating if former slightly lower rate than malesbetween the sec- mates are not available. Females show low site ond and third years (2% and 5% decreasesin fidelity, which similarlypermits an immediate rates,respectively). Among 3-yr-old females, low switch to a new male if the former mate does return rates occurred in the 1974 and 1977 co- not return on time or at all. The inconsequential hortswhere only 56%and 64%of 2-yr-oldsen- role of breedingexperience relative to environ- tered their third year (Fig. 3). The 1974 birds mental experiencemeans that adequatesocial were possiblyhandicapped by poor conditions and breedingskills are attainedat an early age. in 1976,similar to thosethat affected 1976 fledg- Finally,they abandonor skipbreeding in poor lings. Femalesof the 1977 cohort, however, ex- years,which insuressurvival to a later year. perienced exceptionallygood feeding condi- Their plasticityin reproductiveeffort is further tions in 1979;that cohortexhibited the highest illustratedby the smallerclutches laid in years proportionof femalesfirst breeding as 2-yr-olds when breedingwas delayed--the mean clutch (64%,n = 11).Lower return ratesin 1980suggest sizes varied by more than one egg between thatthese females incurred a substantialcost by years--and by the marked asynchronyin egg breedingas 2-yr-olds,but data for all 2-yr-old laying in better years.That is, the population femalessuggest otherwise. Return rates of fe- egg-layingperiod could extend over more than males that bred as 2-yr-olds were no different 3 months (Ainley and Boekelheide 1990). In than those that visited but did not breed (X2 = more southerlypopulations, this speciesmay 0.379,df = 1, P > 0.5).A possibilitystill remains lay eggsfrom mid-winterto earlysummer (Mi- that morefemales who firstvisited as 3- or 4-yr- chael1935, Williams 1942). In general,the above July1989] BiologyofBrandt' sCormorant 399 traits are typical of most seabirds,and in total Snow (1960) and Potts (1969) recorded little an- they combineelements from both of Schaffer's nual variation in either the number of chicks (1974) reproductive models, but unlike most fledged or in breeding population size during seabirds, Brandt's Cormorants are capable of 11 sampleyears. On the basisof mathematical balancinginvestment in reproductionand sur- simulation, Aebischer (1986) proposed that a vival depending on annual conditions. marked3-yr declinein Shagnumbers at another No information is available on the breeding sitewas due to extensivenonbreeding by adults effort of known-age Guanay or Cape cormo- and low recruitment of young breederswhen rants, the Brandt's Cormorant analogs in the feeding conditionswere poor. Our resultsshow Peru and Benguela systems,respectively. Like that this is indeed feasible for cormorants. Fi- the California Current, both systemsdisplay nally, in the Blue-eyedShag (P. atriceps)of the strong interannual variation in oceanographic Antarctic, where the vagaries of pack-ice con- climate and prey abundance (Shannon et al. centration determine access to food, mate re- 1984, McLain et al. 1985). Breeding populations tention has been found to be very low (Shaw of both the Guanayand Cape cormorantschange 1985, see also Cuthbert 1985), and little effect dramatically from year to year depending on on breeding performance by a change in mate food supply (Murphy 1925; Rand 1960, 1963; is evident (Shaw 1986). As in the Brandt's Cor- Jordan1967; Glantz and Thompson1981; Duffy morant, Shaw (1986) also found that breeding 1983; Duffy et al. 1984). For example, Guanays experiencehad little influence on success.Blue- in Peru dropped from 28 million in 1955 to 6 eyed Shagsfirst breed when 3 yr old. million during the 1957-1958 ENSO. By 1959 Lack (1968) and Nelson (1983) theorized that numbershad returned to 11 million and by 1963, the breeding strategy of membersof the Phal- 18 million (Jordan 1967). At the Gal•pagos Is- acrocoracidaeis related primarily to their in- lands, which are strongly affectedby the Peru shore feeding habits. Compared with tropical Current, the Flightless Cormorant (Nannopte- pelagic pelecaniforms and other pelagic sea- rum harrisi),a close relative of the Brandt's (Sie- birds, the cormorants' restriction to coastal hab- gel-Causey 1988), displays even greater ex- itats, large brood size, and rapid chick growth tremes in breeding effort (Harris 1979, Tindie require that they nest near abundant, reliable 1984). This speciesbreeds year-round if con- food. This may be true for many cormorant ditions permit, and the age of first breeding species,but although eastern boundary cur- averages 2.5 yr. Females may breed when as rentsare high in productivity,food availability young as17 months,but during the 1972ENSO, is not reliable. Yet cormorantsare principal avi- no young birds bred. Also like Brandt's Cor- an components of such systems. All inshore morant, individuals are not faithful to either habitats, thus, are not alike. Those of eastern matesor nestsites, and breedingeffort and suc- boundarysystems are clearlynot similar to those cess fluctuate greatly between years. During where most seabird studies have been con- 1972, few adults attempted to breed but there ducted (i.e. shallow-water, high-latitude sys- was no subsequentdecrease in adult survival tems;see Ainley and Boekelheide1990). Lack (Harris 1979). and Nelson based their model of the cormorant Cormorantsbreeding in more stable environ- breeding strategy on the average minimum of ments exhibit less variability in effort than food availability. It is apparent though that cor- Brandt's or Flightless cormorants. Kortlandt morants can cope with extreme variability in (1942) found that 2-yr-old Great Cormorants(P. the abundance of food, and that they do so by carbo),in Holland, may pair and build nestsbut beingexceedingly plastic in their breedingtraits. rarely lay eggs.Most individuals bred at 3 yr, In this way they are far better able than other but others at 4 or 5 yr of age. (He did not sep- seabirdsto exploit periods of food abundance, arate sexes.) Coulson et al. (1969) found that and like other seabirdscan forgo breedingwhen most female Shagsstart to breed at 3 yr, a few conditionsare poor (seealso Boerstoa 1978, Ain- breed at 2 yr, and some delay until 4 or 5 yr. ley and Boekelheide1990). No Shagsin Snow's (1960) small samplebred Cormorants are major components of the as 2-yr-olds. Snow (1963) also found that 93% highly variable easternboundary currents,and of marked males and 41% of marked females one could perhaps considerthem to be bound- reused the same nest in successiveyears. Both ary current specialists.An ENSO occurson av- 400 BOEKELHEIDEAND AINLEY [Auk, VoL 106 erage every 5.4 yr, with a strong event every BROWN, R. G. B. 1976. Seabirds of South America 12.3 yr (Quinn et al. 1978, Chelton et al. 1982, and the northwest Atlantic. Proc. 16th Int. Or- McLain et al. 1985). Years when food is super- nithol. Congr., Canberra,Australian Acad. Sci. abundant, in the California Current at least, 1980. Seabirdsas marine animals. Pp. 1-39 in Behavior of marine animals, vol. 4: marine birds come at a frequencysimilar to that of ENSOs (J.Burger, B. L. Olla, and H. E. Winn, Eds.).New (Ainley and Boekelheide1990). Each adult cor- York, Plenum Press. morant, therefore,will likely experiencethe ex- CHELTON,D. B., P. A. BERNAL,•r J. A. McGOWANß 1982. tremes of food availability, including at least Large-scaleinterannual physicaland biological one, and likely more, food-poorand food-rich interaction in the California Current. J. Mar. Res. yearsduring its lifetime. It is stronglypossible 40: 1095-1125. that such extreme events have much more im- COULSON,J. C., G. R. POTTS,& J. HOROBIN. 1969ß Vari- portant effectson populationsize, age structure, ation in the eggsof the Shag (Phalacrocoraxaris- and ecologicaladaptation in cormorantsof east- totelis). Auk 86: 232-245. ern boundary systemsthan do the "average" CUTHBERT,F.J. 1985. Mate retentionin CaspianTerns. Condor 87: 74-78. years (Murphy 1981, Duffy 1983, Ainley and DUFFY,D.C. 1983. Environmental uncertainty and Boekelheide 1990). commercialfishing: effectson Peruvian guano birds. Biol. Conserv. 26: 227-238. ACKNOWLEDGMENTS --, A. BERRUTI,R. M. RANDALL,•r J. COOPERß1984. Effects of the 1982-83 warm water event on the Many Farallonstaff and volunteerscontributed to this study,in particularS. Morrell and T. J. Lewis,as breedingof SouthAfrica seabirds. South African well as C. Strong, H. Huber, T. Penniman, H. Carter, J. Science 80: 65-69. P. Henderson, J. Penniman, J. Nusbaum, B. Bain- GIESEL,J.T. 1976. Reproductivestrategies as adap- bridge,C. Swarth,K. Schafer,J. Higbee,R. LeValley, tationsto life in temporarilyheterogeneous en- S. Peterson,B. Lewis,L. Astheimer,P. Abbott, J.Young, vironments. Annu. Rev. Ecol. Syst. 7: 57-79. and A. Rovetta.Assistance by personnelof the San GLANTZ, M. L., & J. D. THOMPSON(Eds.). 1981. Re- FranciscoBay National Wildlife Refuge (which ad- sourcemanagement and environmental uncer- ministersthe FarallonRefuge) was indispensable,as tainty: lessonsfrom coastalupwelling fisheries. were the valiant efforts of the Farallon Patrol of the New York, Wiley & Sons. San FranciscoBay Chapter,Oceanic Society, and the GRAu,C.R. 1984. Egg formation.Pp. 33-58 in Sea- U.S. CoastGuard for transportationof people and bird energetics(G. C. Whittow and H. Rahn,Eds.). supplies.S. Goldhaber,M. Simonds,and L. Tuomi New Yorkß Plenum Press. assistedin manuscriptpreparation; K. Hamilton and HARRIS,M.P. 1979. Populationdynamics of the I. Gaffheyprepared the figures.A. H. Brush,F. Cuth- FlightlessCormorant, Nannopterum harrisi. Ibis 121: 135-146. bert, N. Bernstein,D. McCrimmon, L. Spear,and B. Sydemanprovided helpful comments on earlierdrafts JORD.&N,R. 1967. The predation of guano birds on of the manuscript.We particularlythank Point Reyes the Peruvian anchovy (Engraulisringens Jenyns). Bird Observatorymembers and donorsfor their sup- California Coop. Oceanic Fish. Invest. Rep. 11: 105-109. port.This is contributionnumber 337 of PointReyes Bird Observatory. KORTLANDT,A. 1942. Levensloop,samenstelling en structuur der Nederlandse Aalscholverbevolk-

LITERATURE CITED ing. Ardea 31: 175-280. LACK,D. 1968. Ecologicaladaptations for breeding AEBISCHER,N.J. 1986. Retrospectiveinvestigation of in birds. London, Methuen. an ecologicaldisaster in the Shag, Phalacrocorax McLA•N, D. R., R. E. BRAINARD,& J. G. NORTONß 1985. aristotelis:a general method basedon long-term Anomalous warm events in eastern boundary marking. J. Anita. Ecol. 55: 613-629. current systems.California Coop. Fish. Invest. AINLEY, D. G., & R. J. BOEKELHEIDE.1990ß Seabirds Rep. 26: 51-64. of the Farallon Islands: ecology, structure and ß & D. H. THOMASß1983. Year-to-yearfluc- dynamicsof an upwelling-systemcommunity. tuations of the California countercurrent and ef- Palo Altoß Californiaß Stanford Univ. Press. In fects on marine organisms.California Coop. press. Ocean.Fish. Invest. Rep. 23: 165-181. , R. E. LERESCHE,•r W. J. L. SLADEN. 1983. MICHAEL,C.W. 1935. Nesting habitsof cormorants. Breedingbiology of the Ad•lie Penguin.Berke- Condor 37: 36-37. leyßUniv. California Press. MURPHY,R.C. 1925. Bird islands of Peru. New Yorkß BOERSMA,P.D. 1978. Breedingpatterns of Gal•pagos Putnam. penguinsas an indicatorof oceanographiccon- ß 1981. The guanoand the anchovettafishery. ditions. Science 200: 1481-1483. Pp. 81-106 in Resourcemanagement and envi- July1989] BiologyofBrandt's Cormorant 401

ronmental uncertainty (M. H. Glantz and J. D. ing biologyof colonialnesting seabirds. Pp. 153- Thompson,Eds.). New York, Wiley & Sons. 168 in Behavior of marine animals, vol. 4: marine NE•.soN,J. B. 1983. Contrastsin breeding strategies birds (J. Burger,B. L. Olla, and H. E. Winn, Eds.). betweensome tropical and temperatemarine Pel- New York, Plenum Press. ecaniformes. Stud. Avian Biol. 8: 95-114. SCn^F•R, W.M. 1974. Optimal reproductive effort Potts, G. R. 1969. The influence of eruptive move- in fluctuatingenvironments. Am. Nat. 108:783- ments,age, populationsize and other factorson 790. the survival of the Shag (Phalacrocoraxaristotelis S•^N•ON, L. V., R. J. M. CR•WFORr•,& D.C. Du•. L.). J. Anim. Ecol. 38: 53-102. 1984. Pelagicfisheries and warm events:a com- PVCESE•C,B. H. 1983. The relationship between pa- parativestudy. South African J. Sci. 80: 51-60. rental age and reproductive effort in the Cali- S•Aw, P. 1985. Age-differenceswithin breedingpairs fornia Gull (Laruscalifornicus). Behav. Ecol. So- of Blue-eyedShags Phalacrocorax atriceps. Ibis 127: ciobiol. 13: 161-171. 537-543. •, & K. L. DIEM. 1983. A multivariatestudy of 1986. Factorsaffecting the breedingperfor- the relationship of parental age to reproductive manceof AntarcticBlue-eyed Shags Phalacrocorax successin California Gulls. Ecology64: 829-839. atriceps.Ornis Scandinavica17: 141-150. QvINN, W. H., D. O. ZoPF, K. S. SNORT,& R. T. W. SI•½•L-CAvsE•,D. 1988. Phylogeny of the phalacro- Kuo YANG. 1978. Historical trends and statistics coracidae. Condor 90: 885-905. of the southern oscillation, E1 Nifio, and Indo- SNow,B.K. 1960. The breedingbiology of the Shag nesian droughts. Fish. Bull. 76: 663-678. Phalacrocoraxaristotelis on the island of Lundy, RANts,R. W. 1960. The biologyof guano-producing Bristol Channel. Ibis 102: 554-575. seabirds: 3. The distribution, abundance, and 1963. The behavior of the Shag. Brit. Birds feeding habits of the cormorantsPhalacrocora- 56: 77-102, 164-186. cidaeof the southwestcoast of the Cape Province. SOK^L,R. R., & F. J. RonrF. 1969. Biometry. San Invest. Rep. Div. Fish. Union of South Africa 42: Francisco, W. H. Freeman & Co. 1-32. TINDLE,R. 1984. The evolution of breeding strate- 1963. The biology of guano-producingsea- gies in the FlightlessCormorant (Nannopterum birds: 4. Composition of colonies on the Cape harrisi)of the Gal•pagos.Biol. J. Linn. Soc. 21: Islands. Invest. Rep. Div. Fish Union of South 157-164. Africa 43: 1-32. WIrLmMS,L. 1942. Display and sexualbehavior of RYDER,J.P. 1980. The influenceof age on the breed- the Brandt's Cormorant. Condor 44: 85-104.