At-Sea Distribution of Spectacled Eiders: a 120-Year-Old Mystery Resolved

The Auk 116(4):1009-1020, 1999

AT-SEA DISTRIBUTION OF SPECTACLED EIDERS: A 120-YEAR-OLD MYSTERY RESOLVED

MARGARET R. PETERSEN?3 WILLIAM W. EARNED,2 AND DAVID C. DOUGLAS• •U.S.Geological Survey, Biological Resources Division, Alaska Biological Science Center, 1011 EastTudor Road, Anchorage,Alaska 99503, USA;and 2U.S.Fish and Wildlife Service, Migratory Bird Management,Soldotna, Alaska 99669, USA

A13STRACT.--Theat-sea distribution of the threatenedSpectacled Eider (Sornateriafischeri) hasremained largely undocumented. We identified migration corridors, staging and molting areas,and winteringareas of adult SpectacledEiders using implanted satellite transmitters in birdsfrom each of thethree extant breeding grounds (North Slope and Yukon-Kuskokwim Delta in Alaska and arcticRussia). Based on transmitterlocations, we conductedaerial surveys to provide visual confirmationof eider flocksand to estimatenumbers of birds. We identifiedtwo principalmolting and stagingareas off coastalAlaska (Ledyard Bay and east- ernNorton Sound) and two off coastalRussia (Mechigmenskiy Bay on the eastern Chukotka Peninsula,and the areabetween the Indigirka and Kolymadeltas in the Republicof Sakha). We estimatedthat •10,000 birds molt and stagein monospecificflocks at Mechigmenskiy andLedyard bays, and several thousand molt and stage in easternNorton Sound. We further identifiedeastern Norton Sound as the principalmolting and stagingarea for femalesnesting on theYukon-Kuskokwim Delta, and Ledyard Bay and Mechigmenskiy Bay as the principalmolting and staging areas for femalesnesting on the NorthSlope. Males marked at all threebreeding grounds molt and stagein MechigmenskiyBay, Ledyard Bay, and the Indi- girka-Kolymadelta region. Males from the Yukon-KuskokwimDelta molt and stage mainly at MechigmenskiyBay. Equal numbers of malesfrom the North Slopemolt and stageat all threeareas, and mostmales from arcticRussia molt and stage at the Indigirka-Kolymadelta region.Postbreeding migration corridors were offshore in theBering, Chukchi, and Beaufort seas.In winter, eiderswere in the BeringSea south of St. LawrenceIsland. Our estimates from surveysin late winter and early springsuggest that at least333,000 birds winter in single-speciesflocks in the packice in the BeringSea. Received 14 August1998, accepted 10 February1999.

THE DISTRIBUTIONand ecologyof many ma- St. Matthew islands. These speculationsre- rine birds at sea are poorly known becauseof mainedunconfirmed because of high costsand the large distancesbirds traveland the great safetyissues associated with conductingsur- expenseof suchstudies. Seaducks (tribe Mer- veysin the BeringSea. gini) may spend one to four months on land The need for better information about the duringthe nestingperiod and the remainderof speciesbecame urgent when analysesof pop- theyear at sea.In generalthese marine locales ulationtrends suggested that the populationof are in northern latitudes where sea ice is com- SpectacledEiders nesting on the Yukon-Kus- mon.The at-seamolting and wintering areas of kokwim Delta (Y-K Delta) in Alaska declined the SpectacledEider (Somateriafischeri) have re- by 96%from the 1970sto the early 1990s(Stehn mained largely undocumentedsince the dis- et al. 1993,Ely et al. 1994).In addition,no data coveryof the species.One speculationwas that were available from other Alaskan and Russian SpectacledEiders wintered somewherein the breedinggrounds, but anecdotalinformation BeringSea, probably along the ice edge (Con- suggestedthat the specieshad declined in over 1926, Dement'ev and Gladkov 1952). Dau those areas as well (U.S. Fish and Wildlife Ser- and Kistchinski(1977) also speculated that this vice 1996). Consequently,all SpectacledEider speciesmolted in the Bering Strait and win- breedingpopulations were listed as threatened tered in polynyassouth of the ChukotkaPen- in 1993. Becauseit was thought that little insula and south of St. Lawrence,Nunivak, and changehad occurredon thenesting grounds in the SakhaRepublic and northernMaMadan Re- gion(arctic Russia), the NorthSlope of Alaska, E-mail:[email protected] and the Y-K Delta,environmental changes on

1009 1010 PETERSEN,LARNED, AND DOUGLAS [Auk, Vol. 116

TABLE 1. Year, location, and number of satellite three major nestingareas: arctic Russia(Indigirka transmittersdeployed on SpectacledEiders. River Delta, SakhaRepublic), North Slope(Prudhoe Bay region, Alaska), and Y-K Delta (Kashunukor Paired Failed Females with Manokinak rivers, Alaska). Males were considered Year males females broods pairedif capturedwhile flyingwith a female,flying Arctic Russia in a flock with an equal sex ratio, or attemptingto 1995 10 a 0 0 chaseanother pair from a wetland.Adult females North Slope, Alaska were capturedin spring when paired with adult 1995 0 0 10 males,during egg laying, within the first 10 daysof 1996 12 2 b 8 c incubation,or with a brood of ducklingsabout 30 days of age. Failed-breeding females abandoned Yukon-KuskokwimDelta, Alaska nestsafter surgery,failed to hatcheggs, or joined 1993 5 c 5 4 flocks of femalesduring the breeding season.Fe- 1994 11 11 • 0 maleswith broodsrejoined their brood after surgery 1995 0 2 8 c and probablyremained with their broodthrough the • Includesone bird shot on the nestingarea by a Iocal hunter (no remainderof the 45-to-55day brood-rearingseason. location data). Data from 82 of the 88 transmitters were received bIncludes one bird whosedeath on the nestinggrounds was com- plicatedby lead poisoning(no locationdata). throughthe Argos Data Collection and Location Sys- c Includes one bird with no location data. tem in Landover,Maryland. From a total of 6,595sat- ellite overpassevents that collectedone or morePTT transmissions, 3,888 locations were calculated for 82 the moltingand winteringgrounds were sus- individuals whose transmittersprovided location pectedin contributingto the decline(U.S. Fish data on a regularbasis (Table 1). Both standardand and Wildlife Service 1996). The locationsof auxiliary locationprocessing services (Service Argos 1996)were usedin this study.Only 1,165(30.0%) lo- principalmolting and winteringareas of Spec- cationswere acceptedfor analysis as independent tacled Eiders were unknown, so no assessment and biologicallyplausible using decisioncriteria de- could be made of habitat conditions and threats tailed in Petersenet al. (1995).Multiple locationsre- to the specieswhile at sea. cordedwithin a single6-h transmissionperiod were Here, we describethe distributionof Spec- not consideredindependent, so one locationwas se- tacled Eiders during the migration,staging, lectedfor eachindividual during each transmission molting, and wintering periodsderived from period to describethe locationof that individual. locationsof birdsthat were implanted with sat- Among multiple locations,the locationwith the ellite transmittersin eachof the three major highestArgos precision index was selected. If several breedingareas. We describemigration patterns locationswith similar precisionoccurred, the cen- tral-mostlocation was selected.Single locations dur- betweensexes and amongbreeding locations ing a transmissionperiod that were indexedwith as well as variationsin the use of principal poor precisionwere rejectedif not corroboratedby molting and wintering areas.In addition,we locationson previousor subsequentdays. For loca- provide estimatesof abundanceof birds in tions at sea, offshore distances were calculated to the principalmolting, staging, and wintering areas nearest land mass represented in a 1:l,000,000-scale basedon aerial surveys. digital chart of the world using ARC/INFO geo- graphicinformation system software. No transmitter METHODS providedacceptable location data everythree days throughoutthe lifetimeof its batteries.Thus, sample Satellitetracking.--All birds were markedwith sat- sizes vary for estimates of arrival and departure ellite platform transmittingterminal (PTT) transmit- times and numbersof birds at molting areas.Seven ters (Telemetry2000, Columbia, Maryland). Each transmittersmalfunctioned without providingloca- transmitterweighed 36 g and measured3.6 x 5.6 x tions beyond the breeding grounds. In addition, 0.8 cm exclusive of the external antenna. Transmit- threebirds on the Y-K Delta died 16 to 34 daysafter terswere programmed to transmitwith a 60-spulse surgery.This is consistentwith the levelof mortality rate for 6 h every 72 h (n = 86) or for 6 h every 120 from lead poisoningin the populationat that time h (n = 2) and were designedto last 6 to 8 months (Franson et al. 1995, Flint and Grand 1997, Flint et al. (Petersenet al. 1995). 1997, Fransonet al. 1998); however,the causeof Transmitters were deployed using an implant death of each individual was not determined. techniquedeveloped by Korschgenet al. (1996).Cap- Data were pooledamong years within locations. turing and handlingprocedures were describedin Departuredates among years for malesand females Petersen et al. (1995). We marked 88 adult breeders were consistentwith those observedduring other between 1993 and 1996 (Table 1) from each of the yearson the Y-K Delta (P. Flint pers. comm.)and October1999] SpectacledEiders at Sea 1011 with those from birds marked with conventional Methodsof estimatingabundance varied among transmitters during the same years birds were surveys.Abundance estimates from 11 of 15 (73.3%) markedwith PTTson the North Slope(D. Troy pers. surveyswere basedon uncorrectedocular estimates comm.). However, sample sizes often were insuffi- of the number of birds in flocks. Because ocular es- cientfor annual comparisonsof departure and arriv- timatesof denseflocks with large numbersof birds al dates (Table 1). may be low (Follestadet al. 1988),our abundancees- Aerial surveys.--Transectlines for aerial surveys timates should be considered minimal. All flocks lo- were established based on the most recent PTT lo- catedon the water during two late-wintersurveys cationsfor that region.The surveywas centeredon (1995 and 1997) and two surveys of eastern Norton thoselocations and then expandedin all directions Soundwere photographed. An estimatefor eachsur- in an attemptto find otherflocks. Survey design was vey was calculatedfrom countingthe number of similarto that describedby Pihl andFrikke (1992) for malesin all photographsand multiplyingby the open-water areas.Two observersrecorded species, mean ratio of total birds per male derived from a flock size, and sex ratio of flocks. Data were then in- subsampleof slides.To each of thoseproducts we tegrated with position coordinatesderived using added our ocular estimate of the number of birds in GPSTRAK(Anthony and Stehn1994) to providelo- flying flocks,which gaveus the annualpoint esti- cationinformation for eachflock. Aerial photographs matefor the survey. were taken of a nonrandomsample of flocksto de- Methods of estimating sex ratios also varied termine sex ratios after the surveywas completed. amongthe 15 surveys.No sex-ratiodata were col- During somesurveys, all flocksin the surveyarea lected on four surveys,ocular estimateswere made were photographedand the abundanceestimate de- on six, and countsof birds in photoson five. In the termined from a count of individuals in all flocks. five surveyswith photographsof flocks,a sexratio Movementdata.--A bird was consideredmigrating was calculated for each flock. Variance of this ratio when it remained in an area less than one week and was calculatedfollowing Cochran (1977) for cluster sequentiallocations indicated movement in a single samplingof proportions,with eachslide treated as a direction.Birds were identifiedas molting if theyre- samplingunit. Photographstaken for estimatingsex mained within a restricted area and moved less than ratios were a nonrandomsample of flocks,because 1.5 km/h for up to threeweeks. Staging areas were mostphotographs were taken of largeflocks that ap- identified from clusters of locations from several in- pearedto containthe highestconcentration of birds. dividuals. A bird was considered staging if it re- mainedin an area for at least10 daysor with move- RESULTS mentsexceeding 1.5 km/h duringthe periodfrom 1 Juneto 1 November.Wintering areaswere identified from clusters of locations from several individuals Timingof postbreeding dispersaL--Adult males during the period 1 Octoberto 20 November(early left the breedinggrounds in June,but depar- winter), 21 November to 31 January (mid-winter), ture datesvaried significantly (X 2 = 14.10,df = and 1 Februaryto 31 March (latewinter). 2, P = 0.001),with malesdeparting from arctic Because locations of individuals were calculated at Russiaand the North Slopealmost three weeks three-dayintervals, the precisedate that an individ- later than from the Y-K Delta (Table2). The de- ual left an area or arrived at an area could not be departure of femalesvaried accordingto nesting termined.An individual migratingto an areawas re- areaand the outcomeof the breedingattempt. corded as arriving by the first date that a location Femalesthat raised broods departed the breed- was determined; an individual leaving an area was consideredto have departedon the date of the last ing groundsabout the sametime (P > 0.05)in location from the area. lateAugust (Table 2). Femaleswhose breeding Statisticalanalysis.--Differences among arrival and attemptsfailed left the Y-K Delta almostseven departure datesand amongdistances birds were off- weeksbefore successfulfemales (X 2 = 10.73, df shorewere analyzedusing Kruskal-Wallistests on = 1, P = 0.001;Table 2). Furthermore,among ranked data. For the purposesof theseanalyses, all failed females on the Y-K Delta, departure locationsfrom individualsoffshore were pooled and datesdiffered (X2 = 27.0, df = 2, P < 0.001) be- treated as independentsamples. However, if all lo- tweenfemales whose eggs failed to hatch(me- cationsfor a particular analysiswere from a single dian = 10 July;interquartile range 3 to 17 July, individual, sampleswere consideredinsufficient for n = 10) and thosewho lost their broodssoon statisticalcomparisons with other birds. Data on datesand distancesoffshore are presentedas medi- afterhatching (median = 2 August,interquar- ans, interquartile ranges,and numbersof locations. tile range 28 July to 7 August, n = 5). We used Bonferroni tests of ranked data to determine Moltingareas.--We identified four mainmolt- similarities and differencesamong groups when ing areasfor SpectacledEiders (Fig. 1). Arrival Kruskal-Wallis testswere significant. datesto moltingareas varied significantly only 1012 PETERSEN,LARNED, AND DOUGLAS [Auk, Vol. 116

TABLE2. Departure datesof SpectacledEiders from nestingareas. Values are median date and interquartile rangeof dates,with n in parentheses.

Females with Paired males Failed females broods

Arctic Russia 25 Jun, 19-30 Jun (8) No data No data North Slope, Alaska 22 Jun, 19-29 Jun (12) 21 Jul (1) 30 Aug, 26 Aug-4 Sep (15) Yukon-Kuskokwim Delta, Alaska 3 June,28 May-15 Jun (12) 12 Jul, 30 Jun-31 Jul (15) 29 Aug, 18 Aug-1 Sep (5) for femalesthat bred successfully(X 2 = 7.14,df Departuredates from moltingareas differed = 2, P = 0.03). Successfulfemales arrived later among males, successfulfemales, and unsuc- at MechigmenskiyBay (P = 0.02) than at east- cessfulfemales (X 2 = 11.65, df = 2, P = 0.003). ern NortonSound and LedyardBay, which had Males left molting areasthe earliest,followed similar arrival dates (P = 1.00; Table 3). One by failed femalesand then successfulfemales successful female that nested on the North (Table3). Arrival datesto the winteringarea Slope arrived off easternSt. LawrenceIsland alsodiffered in a similar pattern (X2 = 9.65, df on 21 September.Unsuccessful females from = 2, P = 0.008), with males and unsuccessful the Y-K Delta arrived in eastern Norton Sound femalesarriving first (P > 0.05), followedby in earlyAugust (Table 3). Arrival datesof males successfulfemales (P < 0.001). at moltingareas did not differsignificantly (Ta- Birds en route from breeding areas to molt- ble 3) amongbreeding areas (X 2 = 1.44,df = 2, ing areasand from moltingareas to wintering P > 0.05)or amongmolting areas (X 2 = 1.36,df areasremained primarily offshore(Fig. 1), ex- = 2, P > 0.05). ceptfor two adultmales from the North Slope

4 d•c

Bering S.e St.Mathew Is•

FIG. 1. At-seadistribution of SpectacledEiders, with all yearsand locationscombined. Molting areas: Indigirka/Kolyma river delta, LedyardBay, Mechigmenskiy Bay, and easternNorton Sound. October1999] SpectacledEiders at Sea 1013

TABLE3. Arrivaland departure dates of SpectacledEiders. Values are median date and interquartile range of dates,with n in parentheses.

Females with Location Paired males Failed females broods Arrival dates to molting areas Indigirka-Kolyma 17 Jul, 6-24 Jul (9) None None Ledyard Bay 6 Jul, 25 Jun-26 Jul (6) None 8 Sep,7-11 Sep (9) Norton Sound None 3 Aug, 20 Jul-10 Aug (13) 8 Sep,4-16 Sep (5) MechigmenskiyBay 25 Jul, 12 Jul-10 Aug (14) None 21 Sep, 17-21 Sep (4) Departure dates from molting areas Indigirka-Kolyma 5 Oct, 21 Sep-ll Oct (5) None None Ledyard Bay 19 Sep, 26 Aug-19 Oct (4) None 18 Oct, 6-25 Oct (9) Norton Sound None 28 Sep,17 Sep-10 Oct (5) 21 Oct, 2-29 Oct (5) MechigmenskiyBay 5 Oct, 25 Sep-12 Oct (8) None I Nov (2) Arrival dates to wintering area St. Lawrence I. 22 Oct, 28 Sep-2 Nov (8) 17 Oct, 12-22 Oct (4) 7 Nov, 1-11 Nov (9) that went onshoreat TeshekpukLake, Alaska. 1). No markedfemales from Y-K Delta or North When migrating through the BeaufortSea, Slopewent to the Indigirka/KolymaRiver del- males flew almost 10 km closer to shore than ta region (Fig. 2B). did females(X 2 = 8.94,df = 1, P = 0.003);how- The distances that females molted and ever,the tendencyto migrateoffshore did not stagedoffshore varied amonglocations (X 2 = differ betweenthe sexeswhen moving along 63.66, df = 1, P < 0.001). Femaleswere more the Chukchior BeringSea coasts (P > 0.05;Ta- than 10 km closer to shore at eastern Norton ble 4). Soundthan at LedyardBay (Table 4). Thenum- Malesand femalesused different molting ar- bers of locations for females at other sites were eas (X2 = 35.31, df = 3, P < 0.001).At leastone too small for meaningful statisticalcompari- male from eachbreeding area moltedand/or sons. stagedat eachof threelocations: Mechigmen- The distancesthat malesmolted and staged skiy Ba)• Indigirka/Kolyma river delta, and offshoreat their threemain moltingareas (Led- LedyardBay (Fig. 2A). Femalesnesting on the yard Bay,Mechigmenskiy Bay, and Indigirka/ Y-K Delta were found molting only in eastern Kolymariver deltas)also varied (X2 = 162.09, Norton Sound.Females nesting on the North df = 2, P < 0.001). Males were recordedfar- Slopemolted at LedyardBay (n = 10),Mechig- thestoffshore at LedyardBay (P < 0.001),clos- menskiyBay (n = 3), nearBukhoto Puoten (n er to shorein the Indigirka/Kolymariver delta = 1), and off easternSt. LawrenceIsland (n = (P < 0.001),and closest to shoreat Mechigmen-

TABLE4. Distances(km) SpectacledEiders were offshore when at sea.Values are median distance and in- terquartilerange of distances,with n in parentheses.

Location Males Females Both sexes Molt migration Beaufort Sea 6.6, 2.7-13.8 (24) 16.5, 8.4-24.8 (24) Chukchi Sea 34.7, 23.1-48.2 (69) 59.6, 21.2-93.4 (4) 34.9, 20.8-48.9 (73) Bering Sea 12.6, 9.3-38.8 (55) 26.2, 14.4-44.4 (31) 15.3, 0.1-30.6 (86) Molting areas Indigirka-Kolyma 16.9, 10.5-23.3 (103) LedyardBay 34.5, 30.9-38.1 (75) 37.9, 33.8-42.0 (93) Norton Sound 26.7, 21.7-30.9 (107) MechigmenskiyBay 6.9, 4.3-9.6 (122) 9.3, 5.8-12.7 (6) 7.1, 4.9-12.5 (128) Winter (St. Lawrence Island) Early winter 52.0, 32.1-71.9 (119) Mid-winter 108.2, 90.8-125.5 (19) 1014 PETERSEN,EARNED, AND DOUGLAS [Auk,Vol. 116

16 Observationsof birds during aerial surveys A -•, Y-KDelta 14 E22]North Slope at MechigmenskiyBay (Fig. 3A), easternNor- • Arctic Rus,sla •) 12 ton Sound(Fig. 3B), and Ledyard Bay (Fig. 3C) suggestedthat PTT locationsprovided accurate E 10 data on the locationand extentof moltingand •, 8 staging areasused by eiders. Winter locations.--All marked eiders win- I= 6 tered south and southwest of St. Lawrence Is- z 4 land (Figs. 4A-D). Areas used by eiders ap- 2 pearedto changefrom early to mid-winter,but useof theseareas was similar among years. An 18 E• •11 Y-KDe•a aerialsurvey conducted in March 1996corrob- •Z2] NorthSlope •) 16 oratedevidence from PTT locationssuggesting

• 14 that eiderswere absentfrom the portionof the E • 12 area that they used in early winter (Fig. 4C). '• 10 Aerial surveysconducted in late winter/early spring suggestedthat in spring some areas E ß'• 6 wereused repeatedly (Figs. 4C and4D), where- z asothers were used only periodically(Figs. 4B 4 and 4C). 2 The distances that birds were located off-

LB IlK EC ENS shorevaried betweenearly winter and mid- winter (X2 = 13.91,df = 1, P < 0.001).By mid- FIG.2. Molting areasused by markedadult Spec- winter, the median distance from shore was tacledEiders. LB = LedyardBay; I/K = Indigirka/ twice that of early winter (Table4). The maxi- Kolymacoast; EC = easternChukotka Peninsula (in- mum distance eiders were found offshore in cluding MechigmenskiyBay and BukhotoPuoten); winter was 206 km. ENS = eastern Norton Sotrod.(A) Molting areas usedby males;(B) molting areasused by females. Abundance.--Ourlargest estimate of Specta- cled Eider numbers(ca. 363,000birds; 95% CI 333,526to 392,532)came from aerialsurveys of skiy Bay (P < 0.001; Table 4). Both males and the BeringSea in late winter 1996-1997(Table femalesmarked with PTTsmolted in Ledyard 5). Estimatesof the number of birds in each Bayand Mechigmenskiy Bay. However, only at molting area varied among and within years. Ledyard Bay were samplesfor males and fe- We foundmales and femalesat all moltingand males sufficientfor meaningfulcomparisons. wintering areassurveyed (Table 5). Our survey The distance that birds were offshore at Led- data indicate that fewer eiders used eastern yard Bay differed between males and females Norton Soundthan LedyardBay and Mechig- (X2 = 6.23, df = 1, P = 0.01; Table 4). menskiyBay. Our bestestimates of the number

FrG.3. PTT locationsand surveyareas of SpectacledEiders on stagingand moltingareas at (A) Mechig- menskiyBay, (B) easternNorton Sound, and (C) LedyardBay. Squares = locationsof flocksrecorded during aerialsurveys and triangles= PTT locations.Aerial surveyareas are delineatedby dashedlines. October1999] SpectacledEiders at Sea 1015

' I?•øW 17•"

A)WIN•ER 1993-94• • 4-• 63'N- ß63'N q- 4- •

10-13 NOV 1994 [ I • AerialSurvey• • •. 62'N- + q- -62'N +

0, KILOMETERS,100] • Aerial6APR Survey1995

174•'W 172•'W 170t'W Sate?elocation: 07FEB •1995 • C)WINTER 1995-96 D)WINTER 1996-97

AerialSurvey • / ,_ ,' • ß _ •,a[•"/ '"-d23 MAR1996 7-8 MAR 1997 ' -• AerialSu•ey AerialSurvey + + +

,,• • 22 MAR1996 • AerialSu•ey

FzG.4. PTT locationsand surveyareas of SpectacledEiders in the BeringSea during (A) winter 1993- 1994,(B) winter 1994-1995, (C) winter1995-1996, and (D) winter1996-1997. Triangles = locationsof marked birds from 1 Octoberto 20 November,circles = locationsof marked birds from 21 November to 31 December, and diamondis the locationof a markedbird on 7 February1995. Survey areas are delineated by solidlines. Reconnaissanceareas (no birds located)are markedby dashedlines. See Table 5 for surveytotals corre- spondingto survey dates.

of birds in each area, however, varied substan- ment'ev and Gladkov (1967), and Dau and tially amongyears. Survey conditionsvaried Kistchinski(1977), was partly substantiatedby amongyears, and timing of surveysoften was our satellitetelemetry data. The molting areas, suboptimalowing to extendedbad weather; however, were more restricted in locationthan thus, the numberof birds likely was underes- envisionedby Dau and Kistchinski(1977) and timated during somesurveys. Because no sur- were in near-shore waters. It is understandable veyswere conductedin the Indigirka/Kolyma that the relativelysmall and localizedmolting delta area, near Bukhoto Puoten, or in eastern areaswere previouslyundocumented because St. LawrenceIsland, the relativeimportance of birdsgenerally were too far offshoreto be seen these areasfor molting and stagingbirds re- during incidentalflights or aircraft surveys mains undetermined. along the coast.In addition, most birds mi- gratedfar enoughoffshore that ground-based DISCUSSION observers would have been unable to detect large numbersof birds movingfrom breeding The at-seadistribution of SpectacledEiders areas to molting areas. in winter, as proposedby Conover(1926), De- Distribution.--Eastern Norton Sound was the 1016 PETERSEN,LARNED, AND DOUGLAS [Auk, Vol. 116

TABLE5. Estimatednumbers of SpectacledEiders at moltingand winter locations.

Location Surveydate Estimate Proportionmale a Molting areas Norton Sound 31 Aug 1993 551b 10.7 _+3.2 (8 flocksc) 23 Aug 1994 1,460b No data 6 Sep1994 4,030b 16.0 _+1.3 (23 flecksc) 18 Aug 1996 3,303a No data 28 Sep1996 2,793a 14.4 _+4.3 (6 photos) LedyardBay 18 Aug 1994 202e 0 (2 flocks,100 E 1 F) 21 Sep1995 33,192b 85.3 _+2.5 (28 photos) 1 Oct 1996 14,116e 62.0 _+2.7 (212 flecksc) MechigmenskiyBay 21 Aug 1994 37,397b No data 22 Aug 1994 41,209b 49.8 _+4.0 (8 fleckse) 5 Sep 1995 55,731b 28.4 _+9.1 (11 photos) Bering Sea Early winter 10-13 Nov 1994 32,698•f 56.7 _+1.7 (70 flecksc) Late winter 26 Mar 1996 226,810 b No data 7-8 Mar 1997 363,030g 54.4 + 2.3 (13 photos) Spring 6 Apr 1995 148,059g 56.7 _+3.6 (15 photos) "Females"may includehatching year maIesand subadultfemales of all agesand unknownproportions; values are .f _+SE. Ocular estimateof aII birds in surveyarea. Ocular estimateof percentin male plumage. Photo estimates. Ocular estimateextrapolated to includearea betweensurvey lines. Incompletesurvey. Ocular and photo estimatescombined. primary molting area for femalesthat nested asa moltingarea because the areacan be half- on the Y-K Delta. Our maximum estimate of fe- coveredto completelycovered with iceby Oc- males in eastern Norton Sound (3,385) corre- tober (Gloersenet al. 1992). spondsrelatively well with the recentestimate Differences in the distances birds were off- of 3,000 or fewer nesting femaleson the Y-K shoreamong the differentmolting areas per- Delta (U.S. Fish and Wildlife Service1996). If haps reflectsthe bioticand abioticvariability femalesfrom other breeding areascurrently amongthese areas. For instance,the widthsof molt in easternNorton Sound, they are proba- the relativelyshallow shelves of the Ledyard bly few in number. Bay,Mechigrnenskiy Bay, and Indigirka-Koly- Basedon the large numbersof eiders esti- ma delta areas appear to vary, and distance mated during aerial surveysat Ledyard and from shoreand depthsare probably highly cor- Mechigmenskiybays, we concludethat these related.Thus, if molting birds are limited by are important molting areasfor femalesthat factorscorrelated with water depth (such as breedon the North Slopeand for malesfrom preferredfoods), the distancefrom shorethey all areas.Our data suggestthat the area off- are foundwill differ amongmolting areas.Sim- shorefrom the Indigirka/Kolymadeltas is an ilarly, the distanceice is from shorealong the important molting area for males, becausea Beaufort Sea during molt migration may be high proportionof malesfrom the Indigirka correlated with difference distances that males Delta molted there, as did some males from the and successful females were found offshore in Y-K Delta and the North Slopebreeding areas. the Beaufort Sea. However, the numbers and sex ratios of birds PostbreedingdispersaL--The differences in molting offshorefrom the Indigirka/Kolyma timing of molt migrationbetween male and fe- river deltas were not determined because aerial male SpectacledEiders in our study are consis- surveysin the areacould not be conducted.Fe- tent with thosefound in other speciesof sea males from arcticRussia may molt there.It is ducks(Weller 1964) and with previousobser- unlikely,however, that the areaoff the Indigir- vationsof SpectacledEiders (Johnsgard 1964, ka-Kolymariver deltasis consistentlyused by Kistchinski and Flint 1974, Dau and Kistchinski largenumbers of successfullybreeding females 1977). Similarly, it was not surprising that October1999] SpectacledEiders at Sea 1017 malesin the two northernmostnesting grounds Bent 1925) that successfulfemales molt in sum- (North Slope and arctic Russia)departed the mer on the breeding groundswhile raising nesting grounds later than males from the their broods. moresouthern Y-K Delta, because nesting chro- Exceptfor the two males locatedat Teshek- nology is later at the more northern areas puk Lake, Alaska,molt migrationand fall mi- (North Slopemedian nest initiation 15 June[D. gration occurredin offshorewaters. This is Troy pers. comm.];arctic Russia nest initiation consistentwith incidentalobservations report- 17 June to 2 July [Kondratevand Zadorina ed in Palmer (1976), Dau and Kistchinski 1992]) than at the Y-K Delta (median nest ini- (1977), and Johnsonand Herter (1989). We tiation 27 May [Grand and Flint 1997]).This found no evidencethat eiders nesting on the conclusion differs from Dau and Kistchinski North Slopemigrated over the coastalplain or (1977), who suggestedthat departure dates nearthe BrooksRange, as suggested in Johnson were similar from Y-K Delta and Indigirka Del- and Herter (1989). ta (arctic Russia),but is consistentwith John- Our observation that males often took several son and Herter (1989) for the Beaufort Sea weeksto migratethrough the Beaufort,Chuk- coast.The tendencyfor malesfrom all nesting chi, and Beringseas suggests that theseareas areas to arrive at molting areas at about the are important staging and/or feeding areas sametime suggeststhat molt is synchronous during molt migration.However, the physio- amongmales. The'date a male leavesthe breed- logicalcondition of birds when they departthe ing area may be a consequenceof timing and breeding grounds and arrive at molting areas nestingsuccess of the femalewith which he is is unknown,as are the type,quality, and quan- paired.However, migration may be eitherpro- tity of foodsavailable to eidersduring molt mi- longedor relativelyrapid such that most males grationand staging.The importanceof these arrive at molting areas and begin to molt at areasrelative to the timing of molt, survival about the same time. during the molting period,and conditionafter Failed and nonbreedingfemales have been moltingis unknown.However, the availability reportedto leavethe IndigirkaDelta by 30 July and qualityof keyresources in thoseareas dur- (Kistchinskiand Flint 1974), which is within ing the prolongedmigration period ultimately the range of datesthat unsuccessfulfemales may influencethe survival of SpectacledEi- left the Y-K Delta in our study (26 Juneto 9 Au- ders. gust). Our data for females on the Y-K Delta Winterrange.--The wintering area south of suggestthat the timing of reproductivefailure St. LawrenceIsland that we identifiedfor Spec- further influencestiming of molt migration. tacledHiders includes some of the waterspro- This is likely true for femalesnesting in arctic posedby Dau andKistchinski (1977) as a pos- Russiaand the North Slope,but too few unsuc- siblewintering area. We found,however, that cessful females at the North Slope were SpectacledHiders were not restrictedto polyn- marked. yas,but insteadused areasthat may, in some Our observationsof eiders migrating off- years,have •60% ice coverage,especially in shorewhen leavingthe nestinggrounds are late winter and early spring (Gloersenet al. similar to previous observations(Dau and 1992). Becausewe surveyedonly the areas Kistchinski 1977, Johnson and Herter 1989). where marked birds were located, we did not Males began postbreedingmigration in sum- considerother potential wintering areas.No mer (Juneto July) and stagedor migrated in surveyswere conductedin areasidentified by offshore waters. Subadult and unsuccessful fe- Dau and Kistchinski(1977), such as south of St. malesleft their nestingareas later in the sum- Matthew and Nunivak islands and Chukotka mer. Our observations, as well as those sum- Peninsula,or in areasidentified by AlexeiMik- marized by Dau and Kistchinski (1977), hailovichTrukhin (E H. Fay pers. comm.)in showedthat successfulfemales left the nesting the northern Kurils and southeast Kamchatka. grounds in late August to September and Observers conducting aerial surveys for moved to offshore waters. None of the females large marine mammals in the Bering Sea in we capturedwith broodswas flightless or had spring have reported "eiders" and "large recentlyundergone molt of body or wing feath- ducks"among the seaice (J.Brueggeman and ers.This is contraryto Nelson'ssuggestion (in G. Garnerpers. comm.),but the identity of the 1018 PETERSEN,LARNED, AND DOUGLAS [Auk, Vol. 116 specieswas unknown. At 62ø00'N,176ø26'W in but no trend data can be inferred with so few the Bering Sea aboard the U.S. Coast Guard years of data. cutter PolarStar, E H. Fay (pers. comm.)noted Conclusions.--Here,we provide a basisto be- flocksof SpectacledEiders flying eastwardon gin more comprehensivestudies of eider dis- 17 May 1980over the ice "at which time they tribution relative to physicaland biological were just about on a straight-linecourse be- characteristicsand potentialpopulation effects tweenCape Navarin and the YukonDelta." due to changesin humanuse of the Beaufort, Abundance.--Aerialsurvey data provided Chukchi,and Beringseas. With the identifica- minimum estimates of birds that used three of tion of eastern Norton Sound, Ledyard Bay, the four main moltingareas. Breeding males MechimeginskiyBay, and the Indigirka-Koly- and females arrived at and molted in these ar- ma deltasas important molting areas,studies easat differenttimes of the year Malesarrived canbe designedthat focuson the ecologyof at molting areas almost two months earlier SpectacledEiders during molt. The delineation than successfulfemales and beganleaving the of the winteringarea south of St. LawrenceIs- molting areas as successfulfemales arrived. land providesa basisto initiate more detailed Thus, a comprehensiveestimate of the number studiesof populationdynamics and ecologyof of males and females molting in each area winteringbirds. Bioticand abioticfactors that wouldrequire at leasttwo surveys,one each for probablyinfluence the distributionof Specta- males and females.We do not know what pro- cled Eiderswithin and amongwinters include portion of subadultSpectacled Eiders molts in sea ice conditions, benthic invertebrate com- theseareas, or whetherthey molt primarily in munities,ocean depth, and behavioralcharac- other areas.Thus, a worldwide populationes- teristicsof eiders.Once the wintering area is timate from surveysof the molting grounds defined more clearly, hypothesesregarding wouldbe difficultto obtain.Except for perhaps factorsthat occurin winter that may limit re- Y-K Delta birds in eastern Norton Sound, it is coveryof the populationcan be addressed. unlikelythat estimatesof moltingbirds in oth- The tendencyfor SpectacledEiders to clump er areaswould provide a breeding-population into discreteareas during molting (and per- estimatefor arcticRussia or the North Slopeof haps in winter) leavesthem vulnerableto nat- Alaska. ural and human-causedenvironmental pertur- Our minimum populationestimate of about bations.During molt, eiders are particularly 333,000birds in 1997suggests that the number vulnerableto contactwith petroleumproducts of SpectacledEiders has increasedsince 1977, and to becomingentangled in fishinggear or when Dau and Kistchinski(1977) proposed a otherfloating objects. Because they cannotfly world population of 200,000 breeding birds. during molt, hazing them away from suchar- Dau and Kistchinski's(1977) estimatewas ex- easwould be difficult.There is strongevidence trapolatedfrom provisionalestimates based on that mostfemale Spectacled Eiders nesting on small portionsof the breedingarea (C. Dau the Yukon-Kuskokwim Delta use eastern Nor- pers. comm.). In addition, their estimate did ton Sound as their principal molting area. not include nonbreedingor subadult birds. Thus,perturbations such as new trawl fisher- Thus,the totalpopulation size can only be con- ies, fuel spills, or other disruptionsto birds siderednot to havechanged dramatically dur- and/or their habitat could have a major effect ing theperiod of rapid decline(a 96%decrease; on thisbreeding population, which has already Stehn et al. 1993) of the Y-K Delta breeding experienceda major decline.Similarly, winter- population.Based on data from aerial surveys, ing SpectacledEiders appear to usea relatively a substantialpopulation increase appears to restrictedarea in theBering Sea. Major disrup- haveoccurred between 1995 and 1997(Table 5). tionsof food resources,other habitat changes, Aerial surveysat seain winter were limited by or direct threatsto birds could have major in- shortdaylight periods, chronically bad weath- fluenceson the worldwide population. er, and fuel restrictionsof surveyaircraft. Sur- Postbreedingmigration routes of Spectacled veysrarely were continued beyond areas where Eiders tend to be offshore where birds normal- birds were no longer seen.Thus, surveyesti- ly wouldnot be at risk to deathor injury.How- matesrepresent the best minimum estimates ever,because petroleum development occurs in for the conditionsduring a particularsurvey, theseoffshore areas, eiders may have an in- October1999] SpectacledEiders at Sea 1019 creasedrisk of flying into fixed structuresor COCHRAN,W. G. 1977. Sampling techniques.John supplyvessels and being exposed to petroleum Wiley and Sons,New York. productsthat may spill or accumulatein areas CONOVER,H. g. 1926.Game birds of the HooperBay of open water Region,Alaska. Auk 42:162-180,303-318. DAU, C. P., AND S. A. KISTCHINSKI.1977. Seasonal A more comprehensiveanalysis to statisti- movementsand distributionof the Spectacled cally definehigh-use areas and habitatcharac- Eider Wildfowl 28:65-75. teristicsthat may help identify criticalhabitats DEMENT'EV, G. P., AND N. A. GLADKOV. 1952. Birds for molting, migrating, and wintering eidersis of the SovietUnion. Vol. 4. Israel Programfor needed. This analysis will be important for Scientific Translations, Jerusalem. evaluatingproposed activities such as a ship- ELY, C. g., C. P. DAU, AND C. A. BABCOCK.1994. De- ping port for petroleumproducts in Ledyard cline in a populationof SpectacledEiders nest- Bay;exploration, development, and shipping of ing on the Yukon-KuskokwimDelta, Alaska. petroleumproducts in the BeaufortSea; and Northwestern Naturalist 75:81-87. expandedfisheries in easternNorton Sound. FLINT,P. L., ANDJ. g. GRAND.1997. Survival of Spec- tacledEider adult femalesand ducklingsduring brood rearing. Journalof Wildlife Management ACKNOWLEDGMENTS 61:217-221. FLINT, P. L., M. R. PETERSEN,AND J. B. GRAND. 1997. Thisproject would not havebeen possible without Exposureof SpectacledEiders and otherdiving the surgicalexpertise of Dan Mulcahy,Kathy Burek, ducks to lead in western Alaska. Canadian Jour- and PamTuomi. Many people assisted with thesur- nal of Zoology75:439-443. gery, particularly Mark Cunningham,Kim Mills, FOLLESTAD,A., B. H. LARSEN,t. NYG•.RD, AND N. Todd O'Hare, Abby Parks,and JohnPearce. Kevin Roy. 1988. Estimatingnumbers of moulting ei- Kenowand Carl Korschgentaught us the surgical ders Sotnateria tnollissirna with different flock size technique.Declan Troy, Paul Flint, BarryGrand, Joel and flock structure.Fauna Norvegica SeriesC Schmutz,John Pearce,and Andre Degteyereval- Cinclus 11:97-99. lowedus to markbirds on theirstudy areas, and they FRANSON,J. C., M. R. PETERSEN,L. H. CREEKMORE,P. and their crews were instrumentalin capturing L. FLINT, AND M. R. SMITH. 1998. Blood lead con- birds.Paul Howey was particularly responsive to our centrationsof SpectacledEiders near the Kashu- phone calls from the field to modify the implant nuk River, Yukon Delta National Wildlife Ref- transmitter Gregory Balogh, Bill Eldridge, Jack uge, Alaska. Ecotoxicology7:175-181. Hodges, Karen Laing, Brian McCaffery, Robert FRANSON,J. C., M. R. PETERSEN,C. U. METEYER,AND Platte,and Tim Tipladyassisted with aerialsurveys. M. R. SMITH.1995. Lead poisoning of Spectacled J. Brueggeman,E H. Fay,Paul Flint, Gerald Gamer, Eiders (Somateriafischeri) and of a Common Ei- and DeclanTroy allowedus to usetheir unpublished der (Somateriatnollissirna) in Alaska. Journalof data.In-kind supportwas provided by BritishPetro- Wildlife Diseases 31:268-271. leum-Exploration(AK) Inc. on behalfof PrudhoeBay GLOERSEN,P., W. I. CAMPBELL,D. J. CAVALIERI,J. C. Unit owners,North SlopeBorough, Yakutsk Institute COMISO, C. L. PARKINSON,AND J. J. ZWALLY. of Biology,and Yukon Delta National Wildlife Ref- 1992. Arctic and antarctic sea ice, 1978-1987: uge.This project was funded by theU.S. National Bi- Satellite passive-microwaveobservations and ological Service,Alaska ScienceCenter (now U.S. analysis. National Aeronauticsand Space Ad- Geological Survey, Biological ResourcesDivision, ministration,Washington, D.C. Alaska BiologicalScience Center); U.S. Fish and GRAND,J. B., AND P. L. FLINT.1997. Productivity of Wildlife Service,Region 7, Migratory Bird Manage- nestingSpectacled Eiders on the lowerKashu- ment;and U.S. Fish and Wildlife Service,Region 7, nuk River, Alaska. Condor 99:926-932. EcologicalServices. We thank T. W. Arnold, D. V. JOHNSGARD,P. A. 1964.Observations on the biology Derksen, D. Esler, J. B. Grand, R. E. Gill, Jr., R. Oates, of the SpectacledEider. Wildfowl 15:104-107. C. Smith,and two anonymousreviewers for com- JOHNSON,S. R., AND D. R. HERTER.1989. The birds of mentson variousdrafts of the manuscript. the BeaufortSea. BP Exploration(Alaska) Inc., Anchorage,Alaska. LITERATURE CITED KISTCHINSKI,A. A., AND V. E. FLINT. 1974. On the bi- ologyof the SpectacledEider. Wildfowl 25:5-15. ANTHONY,R. M., ANDR. A. STEFIN.1994. Navigating KONDRATEV,A. V., AND L. V. ZADORINA. 1992. Com- aerialtransects with a laptopcomputer Wildlife parative ecologyof the King Eider (Sotnateria SocietyBulletin 22:125-129. spectabilis)and the SpectacledEider (S.rischer 0 BENT, A. C. 1925. Life histories of North American on the ChaunTundra. Zoological Journal 71:99- wild fowl. Part 2. Smithsonian Institution. Unit- 108. ed States National Museum Bulletin No.126. KORSCHGEN,C. E., K. P. KENOW, A. GENDRON-FITZ- 1020 PETERSEN,EARNED, AND DOVCLAS [Auk, Vol. 116

PATRICK,W. L. GREEN,AND F. J. DEIN. 1996. Im- Komdeur, J. Bartelsen, and G. Cracknell, Eds.). planting intra-abdominal radiotransmitters IWRB SpecialPublication No. 19. with externalwhip antennasin ducks.Journal of SERVICEARGOS. 1996. User'smanual. ServiceArgos, Wildlife Management 60:132-137. Inc., Landover,Maryland. PALMER,R. S. (Ed.). 1976.Handbook of North Amer- STEHN, R. A., C. P. DAU, B. CONANT, AND W. I. BUT- icanbirds. Vol. 3. YaleUniversity Press, New Ha- LER,JR. 1993. Decline of SpectacledEiders nest- ven, Connecticut. ing in WesternAlaska. Arctic 46:264-277. PETERSEN,M. R., D. C. DOUGLAS,AND D. M. MUL- UNITED STATES FISH AND WILDLIFE SERVICE. 1996. CAHY.1995. Use of implantedsatellite transmit- SpectacledEider recovery plan. Anchorage, ters to locate SpectacledEiders at-sea. Condor Alaska. 97:276-278. WELLER,M. W. 1964. Generalhabits. Pages15-34 in PIHL,S., AND J. FRIKKE.1992. Countingbirds from The waterfowl of the world. Vol. 4 (J. Delacour, aeroplane.Pages 8-23 in Manual for aeroplane Ed.). Country Life Limited, London. and ship surveysof waterfowland seabirds(J. Associate Editor: T. W. Arnold