J. Field Ornithol., 68(3):400-412 THE MOLT CYCLE OF THE ARCTIC TERN, WITH COMMENTS ON AGING CRITERIA GARY VOELKER BurkeMuseum and Departmentof Zoology Box 353010 Universityof Washington Seattle,Washington 98195 USA Abstract.--Molt-datafrom museumspecimens of Arctic Terns (Sternaparadisaea) were an- alyzedand consideredin relationto other life-historyaspects. Differences in shapeand color of outer tail featherswere reliable indicatorsof age class,allowing second-year birds to be distinguishedfrom third-yearbirds, and second-and third-yearbirds from after-third-year birds. No differencesin molt pattern were found betweensubadult age classes,or between subadultand adult age classes.The duration of primary molt wasestimated at 99 d for adults and 221 d for subadults.This differencein time allottedto molt probablyresults from con- straintsimposed on adultsby migrationand breeding.Adults replace all their flight feathers on the winteringgrounds, whereas subadults often completethis molt in tropicallatitudes where they summer.Arctic Terns appear to exhibit the secondinner primary molt featured by most other Sternaspecies. EL CICLO DE MUDA DE STERNA PARADISAEA Y COMENTARIOS SOBRE LOS CRITERIOS PARA ASIGNAR EDADES Sinopsis.--Seanalizaron los datosde muda de especimenesde museode Sternaparadisaea y se consideraronen relaci6n a otros aspectosde su historianatural. Las diferenciasen la forma y el color de las rectricesexternas fueron indicadoresconfiables de las edades,per- mitiendoque avesdel segundoafio se puedandistinguir de avesdel tercerafio, y que aves del segundoy tercer afio se puedan distinguirde avesmayores de tres aftos.No se hallaron diferenciasen los patronesde muda entre las clasesde avessubadultas, o entre las clasesde edad adultosy subadultos.La duraci6n de la muda primaria se estim6en 99 diaspara los adultosy de 221 dias para los subadultos.Esta diferenciaen tiempo asignadoa la muda probablementeresulta de limitacionesimpuestas en los adultospot la migraci6ny la re- producci6n.Los adultosremplazan todas sus plumas de vuelo en las1ocalidades invernales, mientrasque los subadultosa menudo completanesta muda en latitudestropicales donde pasanel verano. Sternaparadisaea parece exhibir la muda de la segundaprimaria interna caracteristicade la mayoriade las otrasespecias de Sterna. The yearlyenergy budgets of birds are taxed by three major costs:molt- ing, breeding,and migrating.Although a major energeticdemand in the annual cycle (Murphy and King 1992, Murphy and Taruscio1995, Tarus- cio and Murphy 1995, Walsberg1983), molt has been little studied.This is true despite the fact that museumspecimens constitute a huge and readily availabledata set for the study of molt (Rohwer and Manning 1990). Given the metabolic, thermoregulatory,and flight costsof molt, quantitativeknowledge of molt cyclesof individual species,rather than relianceon sweepinggeneralities of patternsamong groups, is criticalto understandingthe life-historystrategies of birds.A casein point is the Arctic Tern (Sternaparadisaea). The only quantitativedata availableon the molt cycleof adultsis a descriptionof feather conditionin 15 speci- mens,just five of which were activelygrowing primaries (Stresemann and Stresemann1966). This lack of data has led to conflictingspeculations as 400 Vol.68, No. 5 ArcticTern Molt [401 to when and how long Arctic Terns molt, as well as to conflictingideas over which feather tracts are involved in each molt and how often feather tracts are replaced (Cramp 1985, Ginn and Melville 1983, Stresemann and Stresemann 1966, Watson 1975). Furthermore, based on this limited data set, it was assumedthat this speciesdid not exhibit the secondre- placement of inner primaries (b' seriesof Stresemannand Stresemann 1966) characteristicof most other Sternaspecies. This studywas undertaken to attempt to clarifyand resolvethese issues. I present data that show that Arctic Terns appear to possessa second inner primary molt; they also replace at least some crown featherstwice in one nonbreeding season,a pattern similar to other speciesin the ge- nus,but a pattern which seemsto havebeen misinterpreted.Both of these molt featuresmay have important life-historyfunctions beyond the simple need to replace worn feathers (Voelker 1996). In this paper, I seekto (1) define reliable age charactersfor placing Arctic Terns in year classesand (2) describe the molts of the Arctic Tern. MATERIALS AND METHODS This studyis basedon 338 ArcticTern specimens:230 adult specimens and 108 sub-adultspecimens. Of these, 56 adult and 25 subadultspeci- menswere replacingflight feathers.Specimens were assembledfrom (1) major museumsfrom throughout the world, (2) museumsin countries with a history of Antarctic Research,and (3) larger museumsin countries within the range of the Arctic Tern (for museumsand museum abbrevi- ations, see Acknowledgments). Scoringmolt.--Arctic Terns have 11 primariesand 18 secondaries(Ginn and Melville 1983) per wing, and 12 rectrices.Primaries were numbered P1-P10, P1 being most proximal (closest)to the body. The eleventh pri- mary is vestigial(Stresemann 1963, Stresemannand Stephan 1968), and I did not include it in this study.Secondaries were numbered S1-S18, S18 being most proximal to the body. Rectriceswere numbered R1 (in- nermost) to R6 (outermost) on each side. I determined the presenceof molting flight feathers by lifting covert featherswith forcepsand a dis- sectingprobe. A molt scorewas assignedto each flight feather (remiges and rectrices)as follows:(0) old; (1) missingor in pin; (2) opened pin to one-third grown; (3) one-third to two-thirdsgrown; (4) two-thirdsto full-grown, with sheathing; (5) full-grown, with no sheathing (Ashmole 1962, 1963, Ginn and Melville 1983). Possiblescores range from zero (all old feathers) to 340 (all new feathers). I consideredmissing or partially grown feathersto be molting only if molt wasoccurring, or had recently been finished, in the correspondingtracts on both sidesof the body. Becausebirds tend to begin molt with primary feathers, I considered growing tail feathers that preceded initiation of primary molt to have been lost adventitiouslyunless replacement was symmetrical.I also ex- amined the head and body of each specimenfor molt, using a dissecting probe to lift feathers so I could examine feather basesand tracts,assign- ing a plus (molting) or minus (not molting) to the following regions: 402] G. Voelker J.Field Ornithol. Summer 1997 head, upperbody, and underbody. Molt terminology followsthat of Hum- phrey and Parkes (1959). I estimatedthe rate and duration of molt using Pimm's (1976) regressionmethod employingdate as the dependentvari- able and molt score as the independent variable, which providesan esti- mate of molt duration for individuals.For the regressionanalysis, collec- tion dateswere convertedto Julian dates (e.g., 1 Jan. = JD 1, 31 Dec. -- JD 365). Becausethis speciesmolts acrosstwo calendar years, I made Julian dates additive in the secondyear. Few specimenswere collected awayfrom the breeding range. To add birds from as many different datesas possible I included sevenmolt scores that are based either on birds I did not examine, or on birds with uncer- tain collection dates. I examined one (AMNH 448054) of the six speci- mens collected by the British, Australian, and New Zealand Expedition; the other five are unaccountedfor (B. Gill, pers. comm.). Falla (1937) gavea descriptionof the stageof molt for all six of these specimens,and basedon my examinationof AMNH 448054 and the descriptionsgiven by Falla, I have given a molt scoreto the five missingspecimens. Because body, primary, and rectrix condition were accurately described for the specimenI examined, I scored these feathers on the unseen specimens accordingto the given descriptions.Secondaries were not described,but becauseAMNH 448054 had almost completed secondarymolt, and be- causemost molting specimensin the studyhad finished growing second- aries before primaries, I assumedthat the secondariesfor the other five subsequentlycollected specimenswere new and fully grown. Friedmann (1945) described the body and primaries of a specimen collectedby H. M. Bryant on 25 Feb. 1940; I assumedcompleted growth of both the secondaries and rectrices. The seventhscore requiring explanation was assignedto AM 22090, a specimen with no collection date and of dubious collection locality, known to have been collected by Mawson'sExpeditions to Antarctica, 1911-1914. I examined both the specimenand Mawson's(1915) accounts of his expeditions,and found that he noted the collectionof Arctic Terns on two occasions,the first being on Macquarie Island. As the collection occurred on 6 Aug. 1913 and the bird is heavily molting, I have ruled out Macquarie Island as the collecting site. The other bird was collected on 16 Jan. 1914, a date that, by comparison to the molt stage of other Antarctic specimensobserved in January (Zink 1981a) would fit this spec- imen. The exact collectionlocality is not listed,but on 17 Januarythe position of the expedition was 62ø21'S,95ø9'E. I have used this as its collection locality. Molt and migration.--Becausethe Arctic Tern has a circumpolarbreed- ing distribution (Cramp 1985, Harrison 1983), I consider any specimen collected at greater than 50ø North latitude to be on the breeding grounds (n = 142). Becausemost Arctic Terns are generally associated with pack ice on their wintering grounds (Cline et al. 1969, Erickson et al. 1972, Zink 1981b, but see Morant et al. 1983), and the limit of pack ice varies around Antarctica between