ANNUAL SURVIVAL RATES OF BREEDING ADULT ROSEATE TERNS

JEFFREYA. SPENDELOWt'2 AND JAMESD. NICHOLSt •U.S.Fish and Wildlife Service, Patuxent Wildlife Research Center, Branch of MigratoryBird Research,Laurel, Maryland 20708 USA, and 2LittleHarbor Laboratory, Inc., 69 AndrewsRoad, Guilford, Connecticut 06437 USA

ABSTRACr.--Analysesof the capture-recapturedata on 910 individual RoseateTerns (Sterna dougallii)trapped from 1978-1987as breeding adultson nestson Falkner Island, Connecticut, estimatethe average annual minimum adult survival rate to be 0.74-0.75. There was weak evidenceof year-to-yearvariation in annual survival ratesduring the study period. The Jolly- Sebermodels used to estimatesurvival ratesalso generatedestimates of population size and captureprobabilities. To determine the relative importanceof adult mortality and permanent emigration in contributing to the estimatedannual lossof one-fourth of the breeding pop- ulation will require further study of intercolony movement between all the major colony sites.Assuming that the loss of birds from the Falkner Island colony site is due mostly to mortality rather than permanent emigration, and that the survival rate of this breeding population is typical of the entire North Atlantic breeding population,then the survival rate of this endangeredspecies is low in comparisonto the survival ratesof severalother marine bird speciesin the orders Procellariiformes, Pelecaniformes,and Charadriiformes. Received 23 December1987, accepted18 January1989.

ROSE•TET•N (Sternadougallii) breeding pop- STUDY AREA AND METHODS ulations in the western North Alantic have de- clined since the 1930s.It has been suggested Studysite and field techniques.--Descriptionsof the Falkner Island colony site, located in Long Island that the decline has resulted, in part, from in- Sound about 5 km south of Guilford, Connecticut, creasedmortality on the wintering grounds and someof the general methods used to study the (Nisbet 1980,1981a; Buckley and Buckley1981; Roseate and Common (S. hirundo) terns that nest on Gochfeld 1983; RoseateTern Recovery Team the islandwere given by Spendelow(1982). Starting 1988).Rates of changein the size of any pop- at the beginningof the CommonTern breedingsea- ulation are a function of survival rate, repro- son in early May, we searchedthe main RoseateTern ductiverate, and ratesof immigrationand em- nesting areas on the northern and southern ends of igration. Unfortunately, historicalinformation the islandfor new nestsat 1- to 3-dayintervals through is not available on most of theseaspects of Ro- early Augusteach year. Other partsof the islandwhere seate Tern population dynamics,so the direct the Roseatesnest infrequently or in small numbers were searchedfor new nestsat 3- to 5-day intervals. comparisonof current survival rateswith those Once found and properly identified, RoseateTern of earlier periods is not possible.Another ap- nestswere examineddaily (weather permitting) or at proach to the causesof the RoseateTern decline no greater than 3-day intervals until all eggs had is a comparative one in which characteristicsof either been lost or hatched. If possible,we deter- the tern population are comparedwith thoseof mined the initiation date of nests discovered after the other presumably"healthy" marine bird pop- beginningof incubationby back-dating23 daysfrom ulations.We analyzedone aspect of the capture- the known or estimated hatching date of the first recapturedata from a RoseateTern breeding chick. colony on Falkner Island, Connecticut.We es- During 1978-1980, most adult RoseateTerns were timated survival rates from these data and com- trapped on their nests within a few days before or after the expectedhatching dates.Often both adults pared them with estimates of survival rates of were trapped on the same day. Since 1981 we have other marine birds based on similar mark-and- tried to avoid trapping both membersof a pair on the recaptureor mark-and-resightstudies. We also sameday and, where possible,we waited until their used the capture-recapturedata to estimate chickswere severaldays old before trapping a pair breedingpopulation sizes and comparedthese of adultsto reducethe possibilityof investigator-in- with estimates based on nest count data. ducedmortality of eggsand young (seeNisbet 1981b).

367 The Auk 106:367-374. July 1989 368 $PENDELOWAND NICHOLS [Auk, Vol. 106

RoseateTerns that nested in the open, in vegetated less than the amount of intercolony movement that areas,under small boards,or inside tires (Spendelow occurred between the birds on Falkner and Tuxis is- 1982)were trappedin box-shapedor circularPotter- lands. Possibleeffects of the intercolony movement style treadle traps (for design,see figure 2.10 in Ca- on parameterestimates also are discussedbelow. nadian Wildlife Service and U.S. Fish and Wildlife Dataanalysis.--Survival rates and populationsizes Service 1977). For situations where nestswere hidden were estimatedfrom the capture-recapturedata with under rocks,logs, or large boards,we constructedjust the Jolly-Seber(Jolly 1965,Seber 1965) and related the front 5-10 cm of a treadle trap. After observing reduced-parametermodels (Brownie et al. 1986)for from a nearbyblind how the adultsapproached these openpopulations. We baseddecisions about model nests,we setthe "fronts" in placeto allow easyaccess appropriatenesson goodness-of-fittests and testsbe- to the eggs,but blockedoff otherpossible avenues of tween models(Pollock et al. 1985,Brownie et al. 1986). escape. Computationsof all capture-recaptureestimates and Capturedadult ternswere bandedwith USFWSsize test statisticswere carried out using program JOLLY 2 bandsand released.If bandedpreviously, they were (Brownieet al. 1986).Reasons for preferring model- released after the old band numbers were recorded. basedcapture-recapture estimators (e.g. see North and Worn bandswere replacedas necessary. We usedalu- Morgan1985) to "returnrate" and other enumeration minum bandson birdstrapped through 1986.In 1987 statisticsoften usedwith bird-banding data were dis- all RoseateTerns were given stainlesssteel (incoloy) cussedin detail by Nichols and Pollock (1983) and bands.It is doubtful that lossof aluminum or incoloy Loery and Nichols (1985). bandsfrom previouslytrapped adults was a problem Estimatesof populationsize alsowere madefrom during the 10-yrstudy period (Nisbet and Hatch 1983). nest count data. Common and Roseateterns usually Time and manpower constraints,as well as nest requirea minimum of 8-10 daysto renestafter a failures that occurred prior to the time a nest was clutch that has been incubated for a week or more considered"trappable" (i.e. after ca. 15 days of in- fails (DiCostanzo 1980, Spendelow 1982). We made cubation),prevented us from trappingadults from all theseestimates by countingthe cumulative total num- the RoseateTern nests on the island each year. We ber of nest initiationsby eachnest census date and concentratedour effortson thoseareas with the great- subtractingthe numberof nestfailures known to have est number of accessiblenests (i.e. those out in the occurred10 or more daysprior to that date.The re- open,in vegetatedareas, or insidetires). In the first suitingvalues were then plottedagainst time and a 3 yr of thestudy (Spendelow and Sibley unpubl. data), curve was fitted by eye to determine an approximate a low percentageof the presumablyolder, more ex- asymptoticmaximum. Under the assumptionthat not periencedbirds (those that made the earliestnests) all pairsrenest after a failureand to allow for the were trapped, which producedsome heterogeneity possibilityof a small amount of intercolonymove- in annualcapture probabilities. Possible effects of such mentalong with latenesting by youngerbirds (Hays heterogeneityon parameterestimates are discussed 1978,Spendelow 1982), we roundedthe asymptotic below. maximumup to the nexthighest 5-pair interval and During 1978-1982,perhaps as many as 50 pairsof arrived at a final populationsize estimate.Unlike RoseateTerns nestedwith a small colony of Common CommonTerns (Hays 1984,Wiggins et al. 1984),Ro- Terns about 6 km away from Falkner on Tuxis Island, seateTerns have not been observedraising a second off Madison, Connecticut. At least 3 of 5 adult Ro- brood in a single season,so the estimatesbased on seatestrapped on Tuxisin 1980were trappedon Falk- the nest count data are of the minimum number of her one or more times from 1980 to 1982 (Spendelow pairsnecessary to producethe totalnumber of nests unpubl. data).We presumesome movement also oc- initiated by a given censusdate. curred in the oppositedirection. The Tuxis colony site was deserted in 1983, when rats invaded the is- land, and few terns have nested there subsequently. RESULTS Other colony siteswhere one or more Roseatesare known to have bred before or after being trapped on Comparisonof models.--From1978 to 1987,we Falkner Island include Bird Island, Massachusetts caught910 different adult RoseateTerns at least (about 170 km away, and about 1,700pairs maximum once on Falkner Island. The capture-recapture sizeduring the studyperiod); Great Gull Island,New data for these birds are summarized in Table 1. York (45 km, 800 pairs);Cedar Beach,New York (90 The testof Model D (survivaland captureprob- km, 100pairs); Gardiner's Island, New York (50 km, abilities constant from year to year, Brownie et 75 pairs);Hicks Island, New York (55 km, 70 pairs); al. 1986) vs. Model A (survival and capture Southold,New York (25 km, 25 pairs),and Waterford Island, Connecticut (40 km, 10 pairs). Other sites probabilities modeled as year-specific,Jolly probablyare used by birdsthat have nested at Falkner 1965) providedstrong evidence that survival Island, but we believe the amount of movement back- and capture probabilitiesvaried from year to and-forth from Falkner Island to all other sites is much year (X2•5 = 140.7, P < 0.0001;Brownie et al. July1989] RoseateTern Survival 369

TABLE1. Breedingadult RoseateTern capture-recapturedata from Falkner Island, Connecticut,1978-1987, summarized in "B-Table" format (Leslie et al. 1953). Lastcaptured Yearof recapture in year 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1978 -- 17 6 4 6 1 0 2 0 0 1979 -- -- 17 18 9 1 2 1 0 0 1980 ------36 11 1 4 3 0 1 1981 .... 44 18 10 9 2 9 1982 ..... 27 14 10 i 5 1983 ...... 24 12 1 5 1984 ...... 32 1 18 1985 ...... 10 44 1986 ...... 19 1987 ...... Total recaptures 0 17 23 58 70 48 54 69 15 101 Newbirds captured 92 126 77 155 67 33 89 108 46 117 Total birds captured 92 143 100 213 137 81 143 177 61 218 % newbirds 100 88 77 73 49 41 62 61 75 54

1986). The test of Model B (survival constant, the actual number of adults that nested on Falk~ captureprobabilities vary from year to year) vs. herIsland during year i. Themodel-based Model A yielded weak evidenceof year-to-year likely overestimatedthe number of birds that variation in survival (X28= 13.0, P = 0.07). The nestedon Falkner Island in any given year be- goodness-of-fittests indicated reasonable fit for causeof the temporaryemigration of Falkner Model A (X222= 30.2, P = 0.11; Pollock et al. Islandbirds that nestedon other islandsin Long 1985) and rejection for Model B (X229= 43.2, P Island Sound(or elsewhere)during other years = 0.04; Brownie et al. 1986),but the goodness- and later returned to Falkner. The last column of-fit probabilitieswere fairly similar for both in Table 3 providesinformation about the de- models.The mostconservative approach would gree to which the model-basedNi may have be to rely primarily on Model A estimates.How- been influenced by temporary emigration. In ever, Model B estimatestend to be more precise general, the ratio values in 1983-1987, larger than those of Model A and we present both than those in 1978-1982, are indicative of the for comparison. reductionin temporaryemigration from Falk- Survivalrate, populationsize, and captureprob- her Island to Tuxis Island after rats invaded the ability estirnates.--ModelA yields annual esti- latter site. The low ratio values for 1980 and mates of survival rate with an arithmetic mean 1986 are discussed below. of 0.75 (Table 2). Model B assumesno year-to- year variation in survival rate and producesa constant survival estimate of 0.74. The Model TABLE2. Survivalrate estimates, $g, for breeding B estimateis somewhat more precisethan those adult Roseate Terns on Falkner Island, Connecti- of Model A because Model B has fewer esti- cut, 1978-1986. mated parameters.These estimatesshould be Model A Model B interpreted as representinga minimal rate of survival as all survival estimatesbased on cap- Year •, •E(•,) • •E(•) ture-recapturedata include both mortality and 1978 0.79 0.14 permanentemigration in their complem,ent. 1979 0.5l 0.07 Estimates of adult population size, N,, ob- 1980 0.86 0.09 1981 0.72 0.08 tained using Model A, Model B, and the nest 1982 0.64 0.08 count data are presentedin Table 3. The ratios 1983 1.02 0.15 of the estimates of N, derived from the nest 1984 0.66 0.10 count data to thoseproduced by using Model 1985 0.83 0.18 1986 -- -- Balso are pr•esented. For practical purposes, the Mean 0.75 0.03 0.74 0.02 nest count Ni is considered our best estimate of 370 SPENDELOWAND NICHOLS [Auk, Vol. 106

T^SLE3. Estimatesofbreeding population size, •i, andcapture probability, P.,for breeding adult Roseate Terns on Falkner Island, Connecticut, 1978-1987.

Nest count data Model A Model B

1978 420 0.22 ...... 1979 360 0.40 583 151 0.23 0.06 560 109 0.25 0.06 0.64 1980 200 0.50 425 82 0.23 0.05 576 92 0.17 0.03 0.35 1981 370 0.58 559 71 0.38 0.05 576 52 0.37 0.04 0.64 1982 270 0.51 432 53 0.31 0.04 449 30 0.30 0.03 0.60 1983 280 0.28 310 42 0.26 0.04 373 28 0.22 0.03 0.75 1984 410 0.35 583 88 0.24 0.04 518 47 0.28 0.04 0.79 1985 470 0.38 529 74 0.33 0.05 535 46 0.33 0.04 0.88 1986 350 0.17 1,027 302 0.06 0.02 958 190 0.06 0.02 0.37 1987 330 0.66 .... 445 40 0.49 0.05 0.74 Estimatedas Ratio= •, (nestcount data)/•, (Model B).

Captureprobabilities derived from Model A, were still incubating eggsor had young chicks, Model B, and the nestcount data are presented but after many of the more experiencedadults in Table 3. We obtained the nest count data already had half- to almost full-grown chicks estimateby dividing the number of breeding and were reluctant to enter the traps. The bias birds caughtin year/, n, by the nest count es- that resultsfrom catchinga disproportionately timateof totalpopulation size, largenumber of "new" birds(Table 1) produces The estimatesof populationsize and capture a considerableoverestimate of the population probabilityare relevantto our considerationof size (Table 3). However, unlike population size survival rate becausethey permit insight about estimates,Jolly-Seber survival rate estimatesare permanent and temporary emigration. The robustto suchheterogeneity and any biasshould populationsize estimatesunder modelsA and be small (Carothers 1973, 1979). Power of the B tend to be higher than the estimatesbased on goodness-of-fittest to detectheterogeneous sur- nest counts.Also, the estimatedcapture prob- vival and both heterogeneous survival and abilities under models A and B are lower than captureprobabilities was fair to good (Pollock those based on nest count data (Table 3). et al. 1985: tables 2, 3). Also, the survival rate estimator for a special case of the Jolly-Seber DISCUSSION model has been shown to be robust to small-to- moderateheterogeneity in survival (Pollockand Comparisonof modelsand biasesresulting from Raveling 1982, Nichols et al. 1982). violationsof assumptions.--Assumptionsof the The movementsof birdsbetween nearby Tuxis Jolly-Sebermodel were discussedwith respect Island and Falkner Island during 1978-1982and to avian studiesby Pollock (1981) and Nichols intercolony movement between more distant et al. (1981). The assumptionsof primary con- colony sites throughout the study period re- cernin our studywere that all individualshave quired that we considerthe effectsof violation the same time-specificsurvival and capture of the assumptionthat there was no temporary, probabilities,and that all emigrationfrom the but only permanentemigration. We usedlarge- sampled population is permanent. sampleapproximations similar to thoseof Car- Goodness-of-fittests provide an assessment others (1973) and Nichols and Pollock (1983) to of how well data conform to model assump- examine the bias in survival rate expected to tions. Based on the simulations of Pollock et al. result from this absenceof geographicclosure. (1985: table 1, figure 1), we believe the power As an extremeexample, we assumedthat 300- of the test to detect heterogeneous capture 350 birds bred primarily on Falkner and 100- probabilitieswas low. We notedpreviously that 150birds bred primarily on Tuxisduring 1978- trappingprocedures were likely to producesome 1982. From 1983 to 1987 all birds were consid- degreeof heterogeneityin captureprobability. ered to breed on Falkner. For Falkner birds we For example,in 1986 trapping did not begin assumedcapture probabilities equal to the es- until July when mostof the first-timebreeders timates based on nest-count data (Table 3). We July1989] RoseateTern Survival 371

assumedthat Tuxisbirds had only a smallchance TABLE4. Estimates of annual survival rates (•) of of being capturedon Falkner(0.05) during 1978- marine birdsbased on mark-and-recaptureor mark- 1982,but that their captureprobabilities equaled and-resightingstudies only. those of the Falkner birds for 1983-1987. All Family/ Species • Reference birds were assumed to have annual survival PROCELLARIIFORMES

probabilitiesof 0.75. The expectedvalue of the Diomedeidae

Jolly-Sebersurvival rate estimatorranged from Diomedea exulans 0.91-0.97 Weimerskirch et al. 1987 0.72 to 0.76 under these assumptions,which D. eporaophora 0.97 Richdale 1952 yielded biasesof <3%. Thus, our survival rate D. irrorata 0.96 Harris 1979 D. iramutabilis 0.91 Rice & Kenyon 1962 estimatesshould exhibit only a small bias from D. raelanophris 0.88 Weimerskirch et al. 1987 this degree of intercolony movement and tem- D. bulleri 0.89 Richdale & Warham 1973 D. chlororhynchos 0.91 Weimerskirch et al. 1987 porary emigration during the early yearsof the Phoebetriafusca 0.95 Weimerskirch et al. 1987 study. P. palpebrata 0.97 Weimerskirch et al. 1987

We are not as confident in our model-based Procellariidae estimatesof population size and capture prob- Fulraarusglacialis 0.95-0.97 Dunnett & Ollason 1978 abilities,and suspectthat they may exhibit rel- Puffinusgriseus 0.93 Richdale 1963 P. puffinus 0.90 Perrins et al. 1973 atively large biases during some years. Tem- P. lherrainieri 0.93 Harris 1979

porary emigration leads to overestimationof Hydrobatidae population size (Balser 1981), as does a per- Oceanites oceanicus 0.91 Beck & Brown 1972 manent "trap-shy" responseby some individ- Oceanodroma leucorhoa 0.94 Morse & Buchheister 1977 uals(Nichols et al. 1984).If we regardthe nest- PELECANIFORMES count Ni as our best estimates of the number of Sulidae birds that breed on Falkner Island, then they Sula bassana 0.95 Nelson 1978 provide a basisfor evaluating the model-based CHARADRIIFORMES estimates.During 1978-1982,the yearsin which Laridae-Larinae temporary emigration was believed to be most Larusargentatus 0.91-0.93 Kadlec& Drury 1968 L. ridibundus 0.82 Clobert et 81. 1987 important becauseof the existenceof the Tuxis Creagrusfurcatus 0.94 Harris 1979 colony,the model-based • were substantially Rissa tridactyla 0.81-0.86 Coulson& Woollet 1976 largerthan the nest-countNi (seeratios in Table Laridae-Sterninae 3). The differences between model-based and Sternadougallii 0.74-0.75 This study nest-count N• were smaller during 1983-1987 S. hirundo 0.92 DiCostanzo 1980 S.paradisaea 0.87-0.88 Coulson& Horobin1976 (exceptas already noted for 1986), and we be- Alcidae lieve that the model-based estimates exhibit less Alca torda 0.89-0.92 Lloyd & Perrins 1977 bias during this period. Uria aalge 0.88-0.92 Birkhead& Hudson 1977 Other factorsmay producehigh model-based Fratercula arctica 0.95-0.96 Ashcroft 1979, Harris 1983 estimates of population size and low model- basedestimates of captureprobabilities relative to estimates based on nest count data. For ex- ample, some birds may simply not breed in a of temporal variation in survival probabilities. particular year rather than emigrate temporar- Survival probabilitiesprobably vary to somede- ily to a different colony site and return to their gree from year to year, but such variation does former site at a later time. The relatively large not appear to be large, and it may be reasonable differences in the nest-count and model-based to model survival as a constant for estimation estimatesof population sizesand captureprob- purposes. abilitiesin 1980may have been due, in part, to Comparisonsto otherspecies of marinebirds.- this phenomenonof a skipped year (perhaps Our Roseate Tern survival estimates are sub- resulting from low food-resourceavailability) stantially lower than publishedestimates based rather than, as in 1986, the bias resulting from on capture-recaptureor capture-resightdata for investigator-inducedheterogeneity in the cap- other speciesof marine birds (Table 4). Most of ture rates of old vs. first-time breeders. these estimates were not based on good esti- The between-modeland goodness-of-fittests mation models (except Dunnett and Ollason (Brownie et al. 1986)test for temporal variation 1978, Clobert et al. 1987, Weimerskirch and Jou- in captureand survival probabilities.The Mod- ventin 1987, and Weimerskirch et al. 1987). el B vs. Model A test provided some evidence Many of the methods used in the past under- 372 SPENDELOWANDNICHOLS [Auk,Vol. 106 estimated survival rate (see Nichols and Pollock ford Food Center; Guilford High School Birdathon; 1983), so the conclusion of low survival in Ro- Little Harbor Laboratory,Inc.; MenuncatuckAudu- seate Terns is valid. bon Society;National Audubon Society,Northeast Possiblecauses of low survivalrates of Roseate RegionalOffice; New Haven Bird Club; Royal Print- Ternsat FalknerIsland.--It is possiblethat a sub- ing Company;The Nature Conservancy,Connecticut Chapter;U.S. Fishand Wildlife Service;Valley Shore stantial proportion of the loss of breeding in- Waterfowlers,Inc.; and Yale University'sPeabody dividuals from the Falkner Island colony site Museum of Natural History. Through 1985 the U.S. was due to permanent emigration. However, CoastGuard, and, sincethen, the Division of Refuges evidence based on the immigration rates to of the U.S. Fish and Wildlife Servicegranted per- FalknerIsland of birdsbanded as young at other mission to work on the island. We thank these or- colony sitessuggests that this was probably not ganizations,and the tern wardensand volunteerswho the case, and that most of the loss was due to assisted in the fieldwork. For much additional advice, mortality that occurredwhen birds were off the support,and assistance,we thank severalanonymous breedingcolonies. The nearestlarge breeding donors,Frank Gallo, Helen Hays, ScottHopkins, Pat colony of RoseateTerns to Falkner Island is Lynch, Fred and Sally Richards,Bill Schew, Fred Sib- ley, and Howard and Ruth Spendelow.We alsothank GreatGull Island,New York,about 45 km away. JoannaBurger, Jay Hestbeck,Barry R. Noon, Harry Only a small percentage(usually <2%) of the M. Ohlendorf,Carl Safina,and two anonymousref- Roseate Tern chicks banded at Great Gull Island ereesfor their helpfulreviews of the manuscript.James each year between 1970 and 1984 have been E. Hines assistedus in designing and running the trapped as breeding adults at Falkner Island, computerprograms, and Lynn A. Hungerbuhler,Amy whereas ca. 10% of chicks banded on Falkner Cyphers,and Marian J. Kreamer typed various drafts Island during this same time period have re- of the manuscript.We dedicatethis paperto the mem- turned to breed as adults at their natal colony oriesof Linda Spendelowand Howard R. Spendelow (Spendelow unpubl. data). This suggeststhat Jr. only a small amount of dispersaland intercolo- LITERATURE CITED ny movement takes place between these two ASHCROFt:,R. E. 1979. Survival rates and breeding colonies.Corroboration of this hypothesiswill biologyof Puffinson SkomerIsland, Wales. Ornis require a knowledge of how many chicksfrom Scandinavica 10: 100-110. Falkner Island settle as adults on Great Gull BALSER,J.P. 1981. Confidence interval estimation Island. and testsfor temporaryoutmigration in tag-re- Nisbet (1980, 1981a)suggested that there was capture studies. Ph.D. dissertation, Ithaca, Cor- strong circumstantialevidence that low surviv- nell Univ. al of terns on the wintering grounds in the BECK,J. R., & D. W. BROWN.1972. The biology of 1970s may have been due to increased human Wilson's Storm-Petrel, Oceanitesoceanicus (Kuhl), predation during this period, but this hypoth- at Signy Island, South Orkney Islands.Brit. Ant- esis has not been tested. We have little direct arctic Surv. Sci. Rep. 69: 1-54. BIRKHEAD,T. R., & P. J. HUDSON. 1977. Population evidence regarding RoseateTerns on their win- parametersfor the CommonGuillemot Uria aalge. tering grounds, but, like Nisbet, we feel that Ornis Scandinavica 8: 145-154. little adult mortalityoccurs during the breeding BROWNIE,C., J. E. HANES,& J. D. NICHOLS. 1986. Con- season. We lack historical or current informa- stant-parameter capture-recapture models. Bio- tion from other colonies on survival rates of metrics 42: 561-574. adults or immatures,reproductive rates,or rates BUCKLEY,P. A., & F. G. BUCKLEY. 1981. The endan- of immigration and emigration, and are forced gered statusof North American RoseateTerns. to view the low survival estimates of the birds Colon. Waterbirds 4: 166-173. CANADIAN WILDLIFE SERVICE • U.S. FISH AND WILDLIFE at Falkner Island as cause for concern. SERVICE.1977. North American bird banding techniques,volume II. ACKNOWLEDGMENTS CAROTHERS,A.D. 1973. The effectsof unequalcatch- ability on Jolly-Seber estimates. Biometrics 29: In the past ten years the work on Falkner Island 79-100. has been generouslysupported by the Connecticut --. 1979. Quantifying unequalcatchability and Audubon Society;Connecticut Department of Envi- its effect on survival estimatesin an actual pop- ronmental Protection; Foundation for the Preserva- ulation. J. Anita. Ecol. 48: 863-869. tion of Natural Areas and Wildlife; Great Gull Island CLOBERT,J., J. D. LEBRETON,& D. ALLAINE. 1987. A Project(American Museum of Natural History); Gull- general approachto survival rate estimationby July 1989] RoseateTern Survival 373

recapturesor resightingsof marked birds. Ardea Additional commentson the assumptionof ho- 75: 133-142. mogeneoussurvival rates in modern bird band- COIJLSON,J. C., & J. HOROItIN. 1976. The influence ing estimationmodelsß J. Wildl. Manageß46: 953- of age on the breeding biology and survival of 962ß the Arctic Tern Sternaparadisaea. J. Zool. London ß J. E. HINES, & K. H. POLLOC•C.1984ß Effects 178: 247-260. of permanenttrap responsein captureprobabil- ß& R. D. WOOLLER. 1976. Differential survival ity on Jolly-Sebercapture-recapture model esti- rates among breeding Kittiwake Gulls Rissatri- matesßJ. Wildl. Manageß48: 289-294. dactyla(L.). J. Anim. Ecol. 45: 205-213. NISBET,I. C.T. 1980. Status and trends of the Roseate DICOSTANZO,J. 1980. Population dynamics of a Tern Sternadougallii in North America and the CommonTern colony.J. Field Ornithol. 51: 229- Caribbean. Unpubl. rep. U.S. Fish Wildl. Serv., 243. Office of Endangered SpeciesßNewton Cornerß D•JNNETT,G. M., & J. C. OLLASON. 1978. The esti- Massachusetts. mation of survival rate in the Fulmar, Fulmarus ß 1981a. Biologicalcharacteristics of the Ro- glacialis.J. Anim. Ecol.47: 507-520. seateTern Sternadougallii. Unpubl. rep. U.S. Fish GOCHFELD, M. 1983. The Roseate Tern: world dis- Wildl. Serv.,Office of EndangeredSpecies, New- tribution and statusof a threatenedspecies. Biol. ton Cornerß Massachusetts. Conserv. 25: 103-125. 1981b. Behavior of Common and Roseate HARRIS,M.P. 1979. Survivaland agesof firstbreed- Ternsafter trappingßColonß Waterbirds 4: 44-46ß ing of Gal•pagosseabirds. Bird-Banding 50: 56- ß & J. J. HATCH. 1983. Band wear and band 61. loss in Roseate TernsßJ. Field Ornithol. 54:90ß ß 1983. Biologyand survival of the immature NORTH, P.M., & B. J. T. MORGANßEds. 1985. Statistics Puffin Fratercula arctica. Ibis 125: 56-73. in ornithologyßNew YorkßSpringer-Verlag. HAYSßH. 1978. Timing and breedingsuccess in three- PERRINSßC. M., M.P. HARRIS, & C. K. BRITTON. 1973. to seven-year-old Common Terns. Ibis 120: 127- Survival of Manx ShearwatersPuffinus puffinus. 128. (Abstract.) Ibis 115: 535-548ß 1984. CommonTerns raise young from suc- POLLOCK,K. H. 1981. Capture-recapturemodels: a cessive broods. Auk 101: 274-280. review of current methods,assumptions and ex- JOLLY,G.M. 1965. Explicitestimates from capture- perimental designßStudß Avian Biol. 6: 426-435ß recapturedata with both deathand immigration- ß & D. G. RAVELINGß1982ß Assumptionsof stochastic model. Biometrika 52: 225-247. modern band-recoverymodels with emphasison KADLEC,J. A., & W. H. DRURY. 1968. Structure of the heterogeneoussurvival rates.J. Wildl. Manageß New EnglandHerring Gull population.Ecology 14: 88-98ß 49: 644-676. ßJ. E. HINESß& J. D. NICHOLS. 1985. Goodness- LESLIE, P. H., D. CHrrr•, & H. CHITTY. 1953. The of-fit tests for open capture-recapturemodelsß estimation of population parametersfrom data Biometrics 41: 399-410ß obtained by means of the capture-recapture RICE,D. W., & K. W. KENYON.1962. Breedingdis- method: III. An exampleof the practicalappli- tribution, history and populationsof North Pa- cations of the method. Biometrika 40: 137-169. cific albatrossesß Auk 79: 365-386ß LLOYDßC. S., & C. M. PERRINS. 1977. Survival and RICHDALEßL.E. 1952ßPost-egg period in albatrossesß ageat first breedingin the Razorbill (Alcatorda). Biol. Monogr. 4: 1-166. Bird-Banding48: 239-252. 1963. Biologyof the SootyShearwater Puf- LOERY,G., & J. D. NICHOLSß1985. Dynamicsof a finusgriseus. Proc. Zool. Soc. London 141: 1-117. Black-cappedChickadee populationß 1958-1983. ß& J. WARHAM. 1973. Survival, pair bond re- Ecology 66: 1195-1203. tention and nest-sitetenacity in Buller'sMolly- MORSE,D. H., & C. W. BIJCHHEISTER.1977. Age and mawk. Ibis 115: 257-263ß survival of breeding Leach's Storm-Petrelsin ROSEATE TERN RECOVERY TEAM. 1988ß Draft Roseate Maine. Bird-Banding48: 341-349. Tern Recovery Plan--Northeastern Population. NELSON,J. B. 1978. The Sulidae:gannets and boo- Newton Cornerß MassachusettsßUßSß Fish Wildl. bies. Oxford, EnglandßOxford Univ. Press. Servß NICHOLS,J. D., & K. H. POLLOCK. 1983. Estimation SEBER,G. A. F. 1965ßA note on the multiple-recap- methodologyin contemporarysmall mammal ture censusß Biometrika 52: 249-259ß capture-recapturestudies. J. Mammal. 64: 253- SPENDELOW,J. 1982. An analysisof temporalvaria- 260. tion in, and the effects of habitat modification on, --, B. R. NooN, S. L. STO•CES,& J. E. HINES. 1981. the reproductive successof RoseateTernsß Colonß Remarkson the use of mark-recapturemethod- Waterbirds 5: 19-31. ology in estimatingavian population size. Stud. WEIMERSKIRCH,H., & P. JOUVENTIN.1987. Population Avian Biol. 6: 121-136. dynamicsof the WanderingAlbatrossß Diomedea -, S. L. STOKES,J. E. HINES,& M. J. CONROY. 1982. exulans, of the Crozet Islands: causes and conse- 374 SeENr)ELOWAND NICHOLS [Auk,Vol. 106

quencesof the populationdecline. Oikos 49: 315- WIGGINS, D. A., R. D. MORRIS, I. C. T. NISBET, & T. W. 322. CUSTER.1984. Occurrenceand timing of second •,J. CLOBERT,AND P. JOUVENTIN.1987. Survival clutches in Common Terns. Auk 101: 281-287. in five southern albatrossesand its relationship with their life history. J. Anim. Ecol. 56: 1043- 1055.

From "A bird wave" by Philip Cox (1889, Auk 6: 241-243):

"Early one morning in April, 1885, I startedfrom chirp. This call would be answered by one or more Newcastle,New Brunswick,for a day'sduck shooting of those on the wing, and then the loiterer would on the Miramichi River, which was then free of ice. rise and join them. Snowwas falling when I left my house,the tumbling "The storm increasing,I abandonedthe idea of flakes forming a strange contrastwith the blossoms, looking for Ducks that day, and seekingthe refuge bursting buds, and catkinsof the trees and shrubs. of an adjacent house, for more than two hours I Presentlybirds were seenflying eastward,and upon watched this bird wave as it rolled along. There was lookingupward, through the snowwhich wasby this no gap,no cessation,neither wasthere deviation from time falling thick and fast, I saw hundredsof Robins the line of the river bank. As the time passedthe ( Merula migratoria), Song Sparrows( Melospizafascia- smaller birds displayed evidence of growing more ta), and Juncos(Junco hyemalis) mingled together in and more weary. Increasednumbers alighted, and an unbrokencolumn and passingnoiselessly on. Some thesetook longer rests,and mademore energeticde- of the birds were only a few feet above the tops of mandsfor a generalhalt. Abouteight o'clock,and as the tallest trees, while others were higher up, the if by the commandof a leader, or by magic,the mov- column extending so far skyward that the topmost ing host vanished. line could with difficulty be outlined amid the falling "Previousto this morning only an occasionalearly flakes. The width of the column--from flank to flank-- bird of thesespring migrants had beenobserved, but appearedto averageabout twenty-five yards. Outside now as I returned homeward I found every bush and of these flanks few birds were to be seen--either to- fenceswarming with birds.As snowhad fallen to the ward the centre of the river, or over the meadow depth of some four or five inches, little food could through which I was walking; the bulk were massed be obtained,and by noon great flockshad gathered in this narrow column and kept directly over the in the farmyards, and that afternoon many a kind margin of the shore,apparently guided by the line hand strewed crumbsand seedsupon the snow for of strongcontrast between the whitenedmeadow and these little friends--heralds of warm days and smil- the dark watersof the river. They movedon in perfect ing fields. silence, save for the flutter of the myriad wings,- "How was this wave formed? What brought this not a note was heard from them. Their flight wasslow throngof birdstogether? I cannotthink that they had and suggestedweariness, but they displayedno in- wintered within a limited area and begun the move- clination to rest, though the tree-topswere thrust so ment northward at the same hour. I am inclined to temptingly toward them. However, in abouthalf an the opinion that such flocksare comparativelysmall hour from the time when they were first observed at the start, and increaseby attracting similar small some individuals showed a disposition to halt. An companiesas they move along. Often, in the early occasionalSong Sparrow or Juncowould alight on spring, I hear on soft mild evenings,faint bird calls the top of a tall tree, and after remaining at rest for from the sky, which are answeredfrom brush and a few seconds--never longer than half a minute-- tree, and these, in my opinion, are the trumpeters would grow uneasyand utter a rather faint cry or who call togetherthe winged armiesof the air."