University of Groningen Crossing the ultimate ecological ...

University of Groningen

Crossing the ultimate ecological barrierGill, Robert E.; Piersma, Theunis; Hufford, Garry; Servranckx, Rene; Riegen, Adrian

Published in:Condor

DOI:10.1650/7613

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2005

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Gill, R. E., Piersma, T., Hufford, G., Servranckx, R., & Riegen, A. (2005). Crossing the ultimate ecologicalbarrier: Evidence for an 11000-km-long nonstop flight from Alaska to New Zealand and eastern Australia byBar-tailed Godwits. Condor, 107(1), 1-20. https://doi.org/10.1650/7613

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-amendment.

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 19-10-2021

https://doi.org/10.1650/7613

https://research.rug.nl/en/publications/crossing-the-ultimate-ecological-barrier(531c931d-e4bd-427c-a6ad-1496c81d44c0).html

https://doi.org/10.1650/7613

[1]

FEATURE ARTICLES

The Condor 107:1–20q The Cooper Ornithological Society 2005

CROSSING THE ULTIMATE ECOLOGICAL BARRIER: EVIDENCEFOR AN 11 000-KM-LONG NONSTOP FLIGHT FROM ALASKA TO

NEW ZEALAND AND EASTERN AUSTRALIA BYBAR-TAILED GODWITS

ROBERT E. GILL JR.1,6, THEUNIS PIERSMA2, GARY HUFFORD3, RENE SERVRANCKX4 AND

ADRIAN RIEGEN5

1Alaska Science Center, U.S. Geological Survey, 1011 E. Tudor Road, Anchorage, AK 995032Animal Ecology Group, Center for Ecological and Evolutionary Studies, University of Groningen, P.O. Box14, 9750 AA Haren, The Netherlands, and Department of Marine Ecology and Evolution, Royal Netherlands

Institute for Sea Research (NIOZ), P.O. Box 59, 1790 AB Den Burg, Texel, The Netherlands3National Weather Service, National Oceanic and Atmospheric Administration, Alaska Region, 222 W. 7th Ave.

No. 23, Anchorage, AK 995134Canadian Meteorological Service, 2121 N. Service Road, Trans Canada Highway, Dorval, Quebec,

H9P 1J3, Canada5231 Forest Hill Road, Waiatarua, Auckland 8, New Zealand

Abstract. Populations of the Bar-tailed Godwit (Limosa lapponica; Scolopacidae) embarkon some of the longest migrations known among birds. The baueri race breeds in westernAlaska and spends the nonbreeding season a hemisphere away in New Zealand and easternAustralia; the menzbieri race breeds in Siberia and migrates to western and northern Aus-tralia. Although the Siberian birds are known to follow the coast of Asia during both mi-grations, the southern pathway followed by the Alaska breeders has remained unknown.Two questions have particular ecological importance: (1) do Alaska godwits migrate directlyacross the Pacific, a distance of 11 000 km? and (2) are they capable of doing this in asingle flight without stopping to rest or refuel? We explored six lines of evidence to answerthese questions. The distribution of resightings of marked birds of the baueri and menzbieriraces was significantly different between northward and southward flights with virtually nomarked baueri resighted along the Asian mainland during southward migration. The timingof southward migration of the two races further indicates the absence of a coastal Asia routeby baueri with peak passage of godwits in general occurring there a month prior to thedeparture of most birds from Alaska. The use of a direct route across the Pacific is alsosupported by significantly more records of godwits reported from within a direct migrationcorridor than elsewhere in Oceania, and during the September to November period than atother times of the year. The annual but rare occurrence of Hudsonian Godwits (L. haemas-tica) in New Zealand and the absence of their records along the Asian mainland also supporta direct flight and are best explained by Hudsonian Godwits accompanying Bar-tailed God-wits from known communal staging areas in Alaska. Flight simulation models, extreme fatloads, and the apparent evolution of a wind-selected migration from Alaska further supporta direct, nonstop flight.

Key words: Bar-tailed Godwit, energetics, flight mechanics, Limosa lapponica, migra-tion, Oceania, wind-selected migration.

Manuscript received 4 May 2004; accepted 22 November 2004.6 E-mail: [email protected]

2 ROBERT E. GILL JR. ET AL.

Atravesando la Barrera Ecologica Final: Evidencia de un Vuelo sin Escala de 11 000 km deLongitud desde Alaska a Nueva Zelanda y el Este de Australia por Limosa lapponica

Resumen. Las poblaciones de Limosa lapponica (Scolopacidae) se embarcan en una delas migraciones mas largas conocidas para aves. La raza baueri crıa en el oeste de Alaskay pasa la estacion no reproductiva a un hemisferio de distancia en Nueva Zelanda y el estede Australia; la raza menzbieri crıa en Siberia y migra hacia el oeste y el norte de Australia.Aunque se sabe que las aves de Siberia siguen la costa de Asia durante ambas migraciones,la ruta meridional que siguen las aves reproductivas de Alaska ha permanecido desconocida.Dos preguntas tienen particular importancia ecologica: (1) ¿las aves de Alaska migran di-rectamente a traves de Pacıfico, a lo largo de 11 000 km? y (2) ¿son capaces de hacerlo enun unico vuelo sin parar a descansar y reabastecerse? Exploramos seis lıneas de evidenciapara responder a estas preguntas. La distribucion de avistamientos de aves marcadas de lasrazas baueri y menzbieri fue significativamente diferente entre vuelos hacia el norte y elsur, sin que hubiera practicamente un solo avistamiento de individuos marcados de baueria lo largo del continente asiatico durante la migracion hacia el sur. El perıodo de la migracionhacia el sur de ambas razas indica la ausencia de una ruta costera asiatica para baueri, conun pico en el paso de las aves ocurriendo allı un mes antes de la partida de la mayorıa delas aves desde Alaska. El uso de una ruta directa a traves del Pacıfico tambien esta avaladopor un numero significativamente mayor de aves reportadas para un corredor migratoriodirecto que para cualquier otro lugar de Oceanıa, y para el perıodo entre septiembre ynoviembre que para otros momentos del ano. La presencia anual, aunque rara, de L. hae-mastica en Nueva Zelanda y la ausencia de registros a lo largo del continente asiaticotambien avalan la posibilidad de un vuelo directo y se explican mejor por el hecho de queL. haemastica acompana a L. lapponica desde areas de escala comunes en Alaska. Evidenciacomplementaria de un vuelo directo sin escalas esta dada por modelos de simulacion devuelo, la gran acumulacion de grasa en las aves y la aparente evolucion de una migracionseleccionada por el viento.

INTRODUCTION

The timing of human settlement of the Earth’sbiomes appears to be related not only to thephysical extent of ecological barriers encoun-tered but also to the inhospitable nature of thebarriers. In this sense, the Pacific Ocean argu-ably represents the most formidable ecologicalbarrier, with human expansion into the far reach-es of Oceania occurring only within the past3000–4000 years (Hurles et al. 2003). But doesthe Pacific present a similar ecological barrier tobirds? Obviously not to those forms adapted forexistence on and from the sea. And surprisinglynot for many terrestrial birds, as more such spe-cies have migrations crossing portions of the Pa-cific than across any other ocean (Williams andWilliams 1999). For example, several species ofshorebirds migrating from Alaska must cross aminimum of 3500 km of open ocean beforereaching Hawaii and even large portions of thesepopulations overfly the Hawaiian Archipelagoen route to the next available land 3000 km far-ther south (Thompson 1973, Williams and Wil-liams 1988, 1990, 1999, Marks and Redmond1994, Johnson 2003). The limits of such nonstopflights are pushed even further by Red Knots(Calidris canutus) and Bar-tailed Godwits (Li-mosa lapponica), which migrate northward from

southeastern Australia and New Zealand to stag-ing sites along the coast of the Yellow Sea, adistance of over 8000 km (Battley 1997, Battleyand Piersma 2005, J. Wilson and C. Minton, un-publ. data).

Two recognized subspecies of Bar-tailed God-wit occur in the central Pacific basin (Higginsand Davies 1996, Engelmoer and Roselaar 1998,McCaffery and Gill 2001). The L. l. menzbieripopulation breeds in central northern Siberiafrom the Yana River delta east to Chaun Gulfand spends the nonbreeding season in westernand northern Australia. Members of the L. l.baueri population nest in western and northernAlaska and spend the nonbreeding season inNew Zealand and eastern Australia. (In the Ana-dyr Basin area of Chukotka is a third, muchsmaller breeding population of unresolved tax-onomic affinity and nonbreeding area [Engel-moer and Roselaar 1998, McCaffery and Gill2001].) The menzbieri population, numberingabout 170 000 birds, appears to migrate bothnorth and south in a two-stage flight with themain northward leg from western Australia tothe Yellow Sea and Korean Peninsula entailinga 6000-km-long nonstop effort (Barter andWang 1990, Wilson and Barter 1998), and themain southward leg a 6000- to 8000-km-long

TRANSPACIFIC MIGRATION IN GODWITS 3

flight from the Sea of Okhotsk or Yellow Searegions to northwest Australia (M. Barter, pers.comm.). The baueri population is slightly small-er (Driscoll 1997, Gill and McCaffery 1999,McCaffery and Gill 2001) and during northwardmigration birds are thought to undertake a singleflight of between 5000 and 8000 km from NewZealand and eastern Australia to the KoreanPeninsula, Japan, and the north coast of the Yel-low Sea (Riegen 1999, Battley and Piersma2005, J. Wilson and C. Minton, unpubl. data).Intensive marking programs initiated in NewZealand and Australia in the late 1970s (Riegen1999, Minton 2004) have shown that birdsmarked within the nonbreeding range of bauerido not occur along the Asian coast during south-ward migration. This finding led Barter (1989)and others (Barter and Wang 1990, Riegen 1999,J. Wilson and C. Minton, unpubl. data) to spec-ulate that the southward flight is instead directacross the Pacific, a minimum distance of about9700 km to northeastern Australia and 10 800km to New Zealand.

Maori folklore lends support to godwits cross-ing this large ecological barrier. When living ona small Pacific island north of New Zealand theynoticed that the kuaka (Bar-tailed Godwit) mi-grated every year in a southerly direction. Fromthis evidence they deduced that land was to befound to the south and canoes were outfitted fora voyage that eventually led to the discovery ofAotearoa (New Zealand), their new home (Gud-geon 1903, Te Paa 1912, Phillipps 1966, Riley2001).

Building upon these millennia-old observa-tions, our objective here was to answer two fun-damental and oft-pondered questions concerningthe southward migration of the baueri race ofBar-tailed Godwit: (1) do birds migrate acrossthe Pacific Ocean between Alaska and New Zea-land, a distance of 11 000 km? and (2) are theycapable of doing this in a single flight withoutstopping to rest or refuel? We addressed thesequestions by exploring six lines of evidence: (1)distributional records and chronology of occur-rence of godwits during migration periods, (2)differential resighting rates of leg-flagged birdsduring northward and southward migrations, (3)comparisons between departure and arrivalevents recorded at migration termini, (4) annualoccurrence of a congener, the Hudsonian Godwit(L. haemastica), in Oceania, (5) analyses ofmaximum flight ranges, and (6) synoptic weath-

er and wind-field analyses across the Pacific andatmospheric trajectory models at time of knowndeparture events.

METHODS

DISTRIBUTIONAL RECORDS ANDCHRONOLOGY THROUGHOUT OCEANIA

If godwits undertake a direct transpacific flightfrom Alaska to New Zealand and eastern Aus-tralia they would be expected to occur in centralOceania either as occasional fallouts from mi-grating flocks or at regularly used stopover sites.To assess this, we turned to various sources, in-cluding Byrne (1979), Heather and Sheehan(1990), references in serial publications such asAtoll Research Bulletin, Sea Swallow, and Birdsof North America (McCaffery and Gill 2001,Gill et al. 2002, Marks et al. 2002), and unpub-lished databases maintained by R. Pyle, P. Do-naldson, and D. Watling. We also referenced theextensive collections of birds and field notesfrom the Whitney South Seas Expedition and thePacific Ocean Biological Survey Program(maintained at the American Museum of NaturalHistory and the U. S. National Museum, respec-tively) and we searched ornithological collec-tions of major museums (Australia Museum,Bernice Bishop Museum, Honolulu; Field Mu-seum of Natural History, Chicago; and Museumof Vertebrate Zoology, Berkeley). Combined,the above sources represent over 300 field as-sessments (many with multiple records) collect-ed since the early 1920s. Admittedly, such dis-tributional records may present a somewhat bi-ased portrayal of the spatial and temporal oc-currence of birds because field efforts were notdistributed evenly across the region or season.Many studies reported when godwits were en-countered but did not provide the range of dateswhen investigators were present but no godwitswere observed. Because of these potential bias-es, we used these records to provide a generalpattern of monthly occurrence throughout the re-gion. Thus, the 568 ‘monthly records’ representthe presence of one or more investigators at aparticular site for one or more days in a givenmonth during which the presence or absence ofBar-tailed Godwits was noted.

We also assessed seasonal occurrence fromcensus data obtained at two sites (both in Fiji;but see Stinson et al. 1997 for the Mariana Is-lands) where counts have been conductedthroughout the annual cycle (Skinner 1983, D.


Watling, unpubl. data). For Skinner (1983) weused high monthly counts and for Watling,whose biweekly counts spanned a longer period,we present an average of all counts within a giv-en month.

Throughout this paper we define Oceania toinclude all insular bodies in the north-central tosouth Pacific Ocean, including those in Micro-nesia, Melanesia (but not the main island ofNew Guinea), Polynesia, and the Hawaii Archi-pelago, but excluding New Zealand proper, Aus-tralia, and Indonesia (after Bier 1995).

BAND RECOVERIES AND SIGHTINGS OF LEG-FLAGGED BIRDS

Since the early 1980s, about 10 000 godwitshave been marked with various colors of legflags specific to individual countries or re-gions—primarily in Australia and New Zealand,but also in China, Japan, and the Republic ofKorea (see Acknowledgments). Almost annuallysince the mid-1990s there have been efforts ded-icated to observing these marked birds, both onthe migration-staging grounds in Alaska (e.g.,Gill and McCaffery 1999) and at migration-stopover sites along the coast of Asia (Mintonet al. 2002). For this effort we relied heavily onprevious summaries of these recovery and re-sighting data (Riegen 1999, Minton et al. 2002,J. Wilson and C. Minton, unpubl. data).

TIMING OF ARRIVAL AND DEPARTURE

To assess levels of concordance between periodsof departure and arrival, we relied on availableseasonal census data from breeding, nonbreed-ing, and migratory stopover sites of both thebaueri and menzbieri subspecies. Most of thesestudies were conducted independently of eachother and focused on site-specific issues and notbroad geographic regions or range-wide assess-ments. Nevertheless, they are of sufficient num-ber and scope that comparisons can be made,especially within the past decade, when wemade concerted efforts to document departuresfrom Alaska and arrivals in New Zealand.

MAXIMUM FLIGHT RANGE PREDICTIONS

For an energy-based evaluation of the proposed11 000-km-long transpacific flight by godwitswe computed maximum flight range—i.e., thedistance flown until the fuel store was deplet-ed—under various conditions using ProgramFLIGHT 1.15 (Pennycuick 2004). This ad-

vanced program encompasses a family of flight-mechanic models that account for use of proteinstores during long-distance flights (Pennycuick1989, 1998). Use of this program to derive flightranges was greatly enhanced through recent val-idation of its performance by Pennycuick andBattley (2003), who looked at the 5400 km-long-flight of the Great Knot (Calidris tenuiros-tris), another long-distance migrant scolopacidwader. Using a set of realistic assumptions andempirical data on body mass and compositionobtained from birds collected at both the start(Northwest Australia) and end (China) of theflight (see Battley et al. 2000), they were able toestimate correctly fat used during flight andbody mass on arrival.

The variables we used in program FLIGHTand their settings are presented in Table 1. Thegeneral assumptions for the various parametersare justified in the explanatory notes that accom-pany the program (based on Pennycuick 1989,1998) or appear in Pennycuick and Battley(2003). However, we also make several specificassumptions. Primarily for comparison with oth-er studies, we have assumed that birds fly at sealevel, but we also ran simulations at 1500-m al-titude. Choice of flight height has only a veryminor effect on the predicted maximum flightrange because flying higher—with lower airdensities, higher flight speeds, and shorter flighttimes—comes at the expense of higher valuesfor specific work, specific power, and greater de-mands on the respiratory and circulatory sys-tems (Pennycuick and Battley 2003). We fol-lowed Pennycuick and Battley (2003) when con-sidering flight speed and, instead of using a cal-culated maximum range speed (Vmr), we set airspeed at 1.2 times the minimum power speed(Vmp) at the beginning of the flight and held itconstant throughout the flight. We did this be-cause as body mass declines, the (robust) esti-mate of Vmp also declines, leading to an increasein the ratio between V and Vmp (Pennycuick andBattley 2003). Although this leads to a smallreduction of flight range, flight at constant speedwould also lead to what Pennycuick and Battley(2003) called ‘‘an avoidance of unduly highpower requirements’’ when the birds are fullyloaded at the start of the migration. For bodydrag coefficients we considered two values, oneof 0.1 (from Pennycuick and Battley 2003) anda lower value (0.05) suggested by Pennycuick et


TABLE 1. Variables used in the simulation of maximum flight ranges (distance to depletion of fuel store) formale Bar-tailed Godwits departing Alaska on an 11 000-km-long flighta to New Zealand and eastern Australia(Program FLIGHT, version 1.15).

Variables (SI-units) Values

General assumptionsb

Basal metabolic rate equationc for non-passerinesInduced power factor 1.2Profile power ratio 0.903Acceleration due to gravity (m sec21) 9.81Fat energy density (J kg21) 3.90 3 107

Dry protein density (J kg21) 1.83 3 107

Protein hydration ratiod 2.2Conservation efficiency 0.23Circulation and respiration factor 1.1Density of muscle (kg m23) 1060Mitochondria inverse power density (m3 W21) 1.2 3 1026

Power density of mitochondria constant

Specific assumptionse

Altitude of flight (m) 0 or 1500Air density (kg m23) 1.23Starting ratio V:Vmp 1.2Flight speed during tripf constantSpecific work constantMinimum energy from protein (%) 5Body drag coefficient 0.1 or 0.05

Bird-related measurementsWing span (m)g 0.73Aspect ratioh 9.3Wing area (m2)i 0.0573Body mass at start (g)j 455, 485, or 515Fresh mass of pectoral muscle at start (g)j 67, 72, or 76Fat mass at start (g)j always 200Airframe mass at start (g)j 188, 213, 239

a The great circle distance between the most northerly Alaska staging site (Yukon Delta) and the northern tipof New Zealand is 10 700 km; that between the most southerly staging site (Nelson Lagoon) and northernQueensland, Australia, is 9700 km. We assume godwits follow a great circle route (orthodrome), though aconstant compass course (loxodrome route) would likely add little additional distance since the departure andarrival sites occur along a north-south axis.

b Based on standard settings in the program FLIGHT and as verified by Pennycuick and Battley (2003).c Changing it to the passerine equation in view of the high values of BMR in many shorebirds (Kersten and

Piersma 1987) has remarkably little effect on the model outcomes (see program FLIGHT).d This is the ratio of water released and lost through respiration as dry protein is combusted, assuming that

water makes up 69% of wet protein.e Specific to southward migrating baueri godwits, with the body drag coefficient and altitude being varied.f Flight speed (i.e., true air speed; see program FLIGHT) is a function of the starting body mass.g Based on a sample of 26 male baueri godwits from nonbreeding grounds in New Zealand (Battley and

Piersma 2005).h Based on a sample of wing tracings of three baueri godwits from Alaska (C. J. Pennycuick, pers. comm.).i Computed from wing span and aspect ratio.j Based on a variety of body mass and composition values.

al. (1996) as being more pertinent to birds withhighly streamlined bodies, such as shorebirds.

Lastly, we also made an important assumptionregarding types of fuel burned. If specific workof the muscles is constant, then, with the low-ering of body mass in the course of the flight,we assumed that excess muscle would be oxi-

dized as fuel, reducing the amounts that neededto be taken from the fat store. Given the evi-dence that birds actually need to consume pro-tein during long distance flights (Jenni and Jen-ni-Eiermann 1998, Battley et al. 2001, Jenni-Eiermann et al. 2002), and that they may takethis protein from organs other than the pectoral


muscles (Battley et al. 2000), we assumed thata minimum of 5% of the energy demand wasbased on the burning of protein. If the pectoralmuscles did not supply enough protein, the re-mainder was taken from the ‘airframe mass,’ the‘structural’ fat-free mass of the body.

For bird-related measurements required of theprogram we restricted our analysis to fully-grown adult males. (The performance values forfemales were very similar to those of males andare thus not presented.) The most critical lineardimension is wingspan, for which we use mea-surements of 26 adult male baueri godwits col-lected in New Zealand (Battley and Piersma2005). For aspect ratio we used a value of 9.3based on wing tracings of males and femalesfrom the same New Zealand sample (C. J.Pennycuick, pers. comm.); note that the predict-ed flight ranges are not very sensitive to wingspan values that do not deviate more than 0.05m. Adult male body mass at departure fromAlaska was based on the masses of two birds(Alaska Science Center specimen nos. 0051,0052) with maximum visual fat scores, eachweighing 485 g, collected on 20 September 1976at Nelson Lagoon, Alaska Peninsula (R. E. Gill,unpubl. data). These weights are at the high endof the range for adult males collected during fallstopover (McCaffery and Gill 2001) and closeto the average of 503 g in the adult males col-lected before northward departure from NewZealand (Battley and Piersma 2005). These NewZealand birds contained an average of 190.5 621.3 g of fat. Nine juveniles killed upon colli-sion with a radar dome on the Alaska Peninsulain mid-October 1987 (probably embarking on atranspacific flight) contained an average of 201.46 20.4 g of fat (Piersma and Gill 1998). Thesejuveniles were not fully grown with respect toskeleton and integument (P. Battley and T. Piers-ma, unpubl. data), thus their body size measure-ments cannot be used in the program FLIGHT(contra Pennycuick and Battley 2003). Based onthe fat levels in the New Zealand males and theAlaska Peninsula juveniles, we considered a fatload of 200 g to be a realistic value for adultmales starting their journey from Alaska. Lastly,given the high correlation between pectoral mus-cle mass and body mass in Bar-tailed Godwitsworldwide (Piersma and Gill 1998), we consid-ered pectoral mass 0.148 of body mass at thestart of the journey based on data for juvenilemales used in Piersma and Gill (1998). Subtrac-

tion of fat and pectoral muscle mass from bodymass led to the starting value for the airframemass.

Program FLIGHT uses a ‘time-marching’ rou-tine. Beginning with starting values it assumes thatall measurements stay constant over a 6-min in-terval and then it calculates how much fat andprotein are consumed in order to yield starting val-ues for body mass, wing-beat frequency, flightspeed, power, and a number of other variablesneeded to begin the subsequent 6-min interval. Itrepeats this procedure until all fat is used up andthen lists the total time spent in flight, body massand composition at arrival, distance flown, etc. Us-ing FLIGHT, we focused on the effects of a 60-grange in body mass centered on a starting mass of485 g while keeping fat mass constant at 200 g.We followed this procedure since we were partic-ularly interested in examining the question ofwhether the protein stores left before departure(see Piersma and Gill 1998, Piersma et al. 1999)might actually limit flight range. We also brieflyexplored the effects of (1) differences in pectoralmuscle mass on flight range while keeping bodymass at start constant, and (2) differences in fatmass while keeping lean mass constant. Of partic-ular relevance in this evaluation were the predictedarrival mass and the predicted size of the pectoralmuscles still remaining when the ‘virtual godwits’ran out of fat. The range in body masses found inmale Bar-tailed Godwits captured upon their arriv-al in October in New Zealand (in Higgins and Da-vies 1996, Miskelly et al. 2001) suggests that sim-ulations with assumptions that lead to predictedarrival masses lower than 210–225 g cannot beaccepted.

ENVIRONMENTAL DATA

In the previous simulations we assumed flightspeed was unaffected by winds, but a number ofwaterbird species staging in southwest Alaskahave been shown to have wind-aided southwardmigrations. To learn if departures of godwitsfrom Alaska were correlated with weather, welooked at synoptic weather and wind-field datafrom the September–November migration peri-od. From this we wanted to learn not only whatweather characteristics were associated withknown departure events but also the frequency,intensity, and tracks of storms that occurredthroughout the North Pacific during the stagingperiod. This in turn prompted us to look at enroute winds, both those associated with depar-


tures and those across the central and southernPacific Ocean. A thorough analysis of en routewinds is beyond the scope of this paper, but useof state-of-the-art global wind models in con-junction with documented departure eventscould reveal large-scale wind-field patternsthroughout the course of the documented migra-tions. To learn the extent of favorable winds pro-vided by storms during departure we used a La-grangian atmospheric trajectory model (CMC2001). The model uses three-dimensional windfields to estimate the origin or destination of anair ‘parcel’ from a specific location in space andtime. We used diagnostic wind fields at 6-hr in-tervals (from Global Data Assimilation Systemarchives of the Canadian Meteorological Centre[CMC]) and trajectories at various heights (fromnear surface to 2000 m) from sites with ob-served departures during model runs.

To assess winds over the Pacific Ocean oncebirds departed on migration we used two sourc-es. For the observed departure in 1987 we ob-tained data from the NOAA-CRIES Climate Di-agnostic Center (CDC 2004); for all other de-partures we used data from the CMC GlobalData Assimilation and Forecast System (CMC2004). For this effort both have been convertedto a 108 3 108 grid. Even though it is unlikelythat birds migrate at a constant altitude, we sim-plified our analysis by selecting winds at the 850mb (;1500 m) level, a general height at whichshorebirds in other studies have been shown tomigrate (citations in Green 2003). To developcomposite snapshots of winds en route for eachof the four observed departures we assumed theflight required about six days. We then lookedat winds at 24-hr intervals (beginning with de-parture) within three of the wind regimes acrossthe Pacific: (1) polar easterlies and north-tem-perate westerlies (;60–308N); (2) equatorialtrades and doldrums (308N–108S); and (3) south-ern hemisphere westerlies (10–408S). Lastly, werealize that godwits likely adjust air speed towind conditions en route and that transit timethrough wind regimes might vary by severalhours, especially at the start and end of the flightwhere winds are typically much more dynamiccompared to the middle, equatorial portionwhere winds are generally benign. Because ofthis, our graphic presentation of en route windsis based on differing time periods within threeregions of latitude that span the projected flightcorridor. Winds depicted during the initial stage

of the flight are at departure but also generallyrepresentative of conditions during the ensuing48-hr period; those through the middle latitudesdepict the subsequent 60-hr period; and those inthe south represent the conditions upon arrivaland over the preceding 36 hr.

STATISTICAL ANALYSES

To look specifically at the occurrence of godwitsin Oceania during southward migration, we firsthad to define the migration period and the likelypathway followed by birds. We considered themigration period to encompass the earliest de-parture from Alaska (early September) and thepeak arrival in New Zealand (into early Novem-ber). For a likely migration corridor we assumedbirds generally follow a great circle route be-tween the staging grounds in Alaska and thenonbreeding grounds in Australia and New Zea-land, with the outer bounds of the corridor in-scribed by great circle arcs that link the east andwest extents of the respective ranges. We thenplotted records of presence or absence of god-wits during the September–November periodand used a chi-square goodness-of-fit test tocompare the proportions occurring within andoutside of the transpacific migration corridor.We also used a chi-square test to compare theproportion of birds (maximum counts) occurringinside and outside the migration corridor. Birdswere considered to use a site if they had everbeen recorded there; these included sites sur-veyed during multiple years but for which birdswere not reported in some years. For numericcomparisons of birds inside and outside the pro-posed migration corridor, we took a conservativeapproach and assumed that some proportion ofbirds recorded at a particular site during south-ward migration remained there for extended pe-riods (see McCaffery and Gill 2001, D. Watling,pers. comm.) and may have been included incounts spanning two or more consecutivemonths. Thus, to mitigate possible issues ofpseudoreplication, we used the single highest ofthe monthly maximum counts from sites atwhich two or more months were reported. Wethen compared the average maximum count in-side and outside the migration corridor using aMann-Whitney U-test. Differences in seasonaland regional resighting rates of leg-flagged birdswere assessed using chi-square tests. The distri-bution of resightings in Asia, however, is biasedgeographically in that most effort to date has


FIGURE 1. (A) Percentage of total records of occur-rence (n 5 254) of Bar-tailed Godwits throughout Oce-ania each month. Numbers above bars show numberof records. (B) Number of godwits recorded duringmonthly censuses at Suva Point, Fiji; solid bars fromSkinner (1983) and open bars from D. Watling (un-publ. data).

occurred in Japan and Republic of Korea andlarger areas like the Yellow Sea and Sea ofOkhotsk have received comparatively little cov-erage, especially during the southward migrationperiod. Despite this shortcoming, coverage hasbeen adequate to establish seasonal patterns overbroad regions. Data are reported as means 6 SD.

RESULTS

SEASONAL OCCURRENCE, DISTRIBUTION,AND NUMBERS OF BIRDS IN OCEANIA

We found records of godwits in Oceania duringevery month of the year (Fig. 1A), but most fre-quently during the southward migration period(September–November), which comprised 49%of all monthly records (n 5 254). No othermonth accounted for more than 8% of the total.We found a similar temporal pattern at Suva Pt.,Fiji, the only site in Oceania at which systematiccounts of godwits have been conducted for ex-tended periods (Fig. 1B).

The geographic occurrence of godwits in Oce-ania was widespread, with birds noted frommost (77%) of the 30 major archipelagoes andfrom over 350 different atolls and islands within.Only from central and eastern Polynesia (e.g.,Southern Cook, Marquises, Austral, Gambier,

Line, most of the Tuamotu, and Pitcairn islands)have birds not been recorded.

We also found a significant difference be-tween the southward migration period and therest of the year (x2

1 5 32.4, P , 0.001) whenwe looked at geographic distribution of recordsin Oceania by season. Most sites where godwitswere recorded during the September–Novemberperiod occurred throughout a corridor linkingAlaska and the nonbreeding grounds in easternAustralia and New Zealand (Fig. 2). The samepattern was found when total maximum countsper site were compared inside and outside themigration corridor. When adjusted for sites withmulti-year records, 93% (n 5 868) of all godwitsnoted during the southward migration periodcame from sites within the likely migration cor-ridor. The proportion increased to 97% when re-cords just outside but east of the corridor (Ha-waiian and Cook islands) were also considered(Fig. 2). Most (87%) of the 868 birds were re-corded from four sites in the southern half of themigration corridor: Mankin Atoll (120 birds) inthe Tungaru Islands, Rewa River (200 birds) andSuva Point, Fiji (121 birds), and Chatham Island(314 birds) east of New Zealand (Fig. 2). Whenthese four sites are not considered, the averagemaximum number of godwits recorded at siteselsewhere in Oceania during southward migra-tion was similar both outside (4.6 6 7.1, range1–21) and inside (2.8 6 3.9, range 1–20) thecorridor (Mann-Whitney U-test: z 5 0.9, n 513, 37, P 5 0.19).

RESIGHTINGS OF MARKED BIRDS

The proportions of color-flagged godwits of thetwo subspecies that were resighted along thecoast of Asia during northward and southwardmigrations differed markedly (Table 2; x2

1 536.8, P , 0.001). On northward migration bothbaueri and menzbieri regularly used intermedi-ate stopover sites; during southward migration,however, menzbieri were still commonly sightedalong the coast of Asia whereas baueri—withbut three exceptions—have gone unreported.Sightings of marked baueri (n 5 136; R. Gilland B. McCaffery, unpubl. data)—but not ofmarked menzbieri—on the Alaska staginggrounds from late August through September(1999–2004) further indicate the extent of sep-aration of the two subspecies during southwardmigration.


FIGURE 2. Distribution of records of Bar-tailed Godwits throughout Oceania during the southward migrationperiod (September–November). Filled circles 5 sites reporting godwits; unfilled circles 5 sites at which nogodwits were noted during the period. Map projection 5 Orthographic (central meridian 5 180; reference latitude5 210). Lateral bounds of stippled region 5 plotted great circle routes.

TABLE 2. Seasonal distribution of resightings and recoveries of baueri and menzbieri subspecies of Bar-tailedGodwits along the coast of East Asia during northward and southward migration. All putative baueri weremarked on nonbreeding grounds in New Zealand and eastern Australia; menzbieri were marked in westernAustralia (but see footnotes for few exceptions). See methods section for details on the sources of data.

Location observed

Northward

baueri menzbieri

Southward

baueri menzbieri

Russia 2 3 2 5Japan 84a 1 2North Yellow Sea 18 38 1 2Republic of Korea 62 54 35Hong Kong, Taiwan, SE Chinab 2 52 1Total 168 148 3 45

a Includes one bird flagged in Japan in August and seen in New South Wales, Australia, the following February.b Includes 40 birds (1 baueri and 39 menzbieri) shot by hunters; remainder are resightings of flagged birds

including one juvenile flagged in Hong Kong and recovered in northwest Australia.

TIMING OF DEPARTURE AND ARRIVAL

The average peak departure of godwits fromAlaska and peak arrival in Fiji, New Zealand,and eastern Australia occur within a 2- to 3-

week period in late September through mid-Oc-tober (Fig. 3). Both departure from Alaska andarrival in New Zealand can be earlier, however,as recorded in 2003 when birds were seen leav-


FIGURE 3. Timing of southern migration of Bar-tailed Godwit races baueri (unfilled bars) and menz-bieri (cross-hatched bars) at different locations alongtheir migration routes. Filled portions of bars indicateperiods of peak passage; dashed lines indicate periodsof movement. Information for Alaska from Gill andMcCaffery (1999), McCaffery and Gill (2001), R. Gill,unpubl data; for Asia, Moores (1999), Satoshi (2003);for Fiji, Smart (1971), Skinner (1983), D. Watling, un-publ. data; for NW Australia, Higgins and Davies(1996), J. Wilson and C. Minton, unpubl. data, D. Rog-ers and M. Barter, in litt.; for SE Australia, Higginsand Davies (1996), Keates (2003), M. Barter, pers.comm.; for New Zealand, Hawkins (1980), Sagar etal. (1999), Riegen (1999), A. Riegen, unpubl. data.

TA

BL

E3.

Con

diti

ons

duri

ngde

part

ures

ofB

ar-t

aile

dG

odw

its

onso

uthw

ard

mig

rati

onfr

omsi

tes

onth

eA

lask

aP

enin

sula

.

Dep

artu

repe

riod

aN

umbe

rof

bird

s(n

umbe

rof

floc

ks)b

Sto

rmch

arac

teri

stic

s

Low

cent

erpo

siti

onP

ress

ure

(mb)

Dis

tanc

e&

dire

ctio

ncfr

omde

part

ure

poin

t

Dep

artu

resi

tew

inds

Spe

ed(m

sec2

1 )D

irec

tion

Tai

lwin

dco

mpo

nent

d

Spe

ed(m

sec2

1 )D

irec

tion

Fet

ch(k

m)e

20O

ct19

87.

9(?

)47

8N,

1748

W97

611

00km

;22

5810

N10

–21

NN

W-N

W12

00(0

9:30

)24

Sep

1996

492

(1)

438N

,17

58W

974

1650

km;

2378

4–8

NN

E8–

23N

-NW

1300

(03:

00–0

7:00

)10

–11

Oct

2000

.40

00(?

)55

8N,

1608

W98

665

0km

;18

485–

10N

NE

13–2

3N

-NW

900

(04:

00–2

0:00

)3–

5S

ep20

0350

90(1

5)47

8N,

1628

W99

890

0km

;18

380

–5N

8–15

N-N

NW

1100

(09:

00–1

0:00

)

a19

87de

part

ure

poin

t:C

old

Bay

(558

209N

,16

2850

9W);

1996

and

2000

:N

elso

nL

agoo

n(5

6800

9N,

1618

009W

);20

03:

Ege

gik

Bay

(588

109N

,157

8309

W).

All

tim

esC

oord

inat

edU

nive

rsal

Tim

e(a

dd10

hrfo

rlo

cal,

Ala

ska

Day

ligh

tT

ime)

.b

For

deta

ils

of19

87de

part

ure

see

Pie

rsm

aan

dG

ill

(199

8);

1996

obse

rvat

ion

byR

.G

ill

and

M.

Ow

ens;

2000

obse

rvat

ion

byR

.G

ill;

2003

obse

rvat

ion

byR

.G

ill

and

D.

Rut

hrau

ff.

cD

irec

tion

rela

tive

totr

ueno

rth

from

depa

rtur

esi

te.

dW

inds

alon

gli

kely

init

ial

mig

rati

onro

ute

from

depa

rtur

esi

te.

eF

etch

isth

elo

nges

tdi

stan

cew

ith

sust

aine

dw

inds

from

tail

orqu

arte

ring

tail

dire

ctio

n.ing during the first week of September (Table 3)and first arrivals were noted in New Zealand 6–10 days later (A. Riegen and P. Battley, unpubl.data). In contrast, southward passage of menz-bieri godwits along the coast of Asia and arrivalin Western Australia are a month earlier and es-sentially over before most baueri godwits departAlaska (Fig. 3).

MAXIMUM FLIGHT RANGE PREDICTIONS

With a starting body mass of 485 g, a fat massof 200 g (41% body fat), and pectoral musclesadjusted to body mass, male Bar-tailed Godwitswould be able to cover 11 000 km under still airconditions only if their body drag coefficientwas as low as 0.05 (Table 4, Fig. 4). Under ei-ther assumption for body drag, arrival bodymass (213–214 g, or 75% of the lean mass atstart) would be reasonable. (Note that Battley etal. [2000] found lean mass of Great Knots thatarrived after a 5400 km long flight to be ca. 80%of lean mass at departure.) Pectoral muscle mas-ses of godwits at arrival (21–33 g) were small,but not unrealistically so (Landys-Ciannelli et al.2003). A body drag coefficient of 0.05 appearsto be realistic for godwits since it produces amore consistent prediction of air speed (18.6 msec21, or 67 km hr21), i.e., a value that is muchcloser to empirical values obtained by radar forgodwits of the L. l. taymyrensis subspecies dur-


TABLE 4. Predicted performance (according to program FLIGHT) of male Bar-tailed Godwits initiating flightwith given fuel stores and flight parameters (see Table 1) and flown until fat stores are depleted. ProgramFLIGHT assumes that a small part of the energy used comes from burning protein, primarily from pectoralmuscle but also from other components of lean mass as well.

Body dragcoefficient

At start of flight

Bodymass (g)

Leanmass (g)

Fatmass (g)

Pectoralmuscle mass

(g)

At fat depletion

Bodymass (g)

Pectoralmuscle mass

(g)

Distancecovered(km)a

Days inthe air

Air speed(m sec21)

0.10 455 255 200 67 183 28 9303 7.0 15.30.10 485 285 200 72 213 32 8154 6.0 15.60.10 515 315 200 76 243 36 7240 5.3 15.90.05 455 255 200 67 188 30 12 883 8.2 18.20.05 485 285 200 72 214 33 11 308 7.0 18.60.05 515 315 200 76 244 37 10 049 6.1 19.00.05 485 285 200 54 213 24 11 308 7.0 18.60.05 515 285 230 76 202 29 12 928 7.9 19.0

a Distances covered based on flight at sea level.

FIGURE 4. Predicted maximum flight range (the dis-tance to fuel-depletion) in male Bar-tailed Godwits.Plot is a function of initial body mass (515, 485, or455 g), variation in the percent fat of body mass at thestart of the flight from Alaska (39–44%; with corre-lated variation in body mass, fat-free mass, and pec-toral muscle mass, but with a constant fat load of 200g), and two values for the body drag coefficient(BDC). Solid circles indicate simulations using a BDCvalue of 0.1, unfilled circles indicate simulations usinga value of 0.05. The solid horizontal lines representgreat circle distances transited by godwits: upper lineat 10 800 km is distance between the northernmoststaging site in Alaska (Yukon Delta) and the northerntip of the North Island, New Zealand; lower line at9700 km is distance between southernmost staging sitein Alaska (Nelson Lagoon) and Townsville, Queens-land, Australia, the suspected northern portion of thenonbreeding range of baueri in Australia. For a birdwith a starting body mass of 485 g, a BDC of 0.1, anaverage constant flight speed of 15.4 m sec21, but withan average tailwind of 4.5 m sec21 for the entire dis-tance, the flight range would be increased by 2000 km,as indicated by the arrow and broken horizontal line.Similar proportional increase in flight range would oc-cur in birds having a BDC of 0.05 (unfilled circles).

ing northward migration (18.4 m sec21; M.Green and T. Piersma, unpubl. data), than for airspeeds (15.6 m sec21) obtained with a body dragcoefficient of 0.1 (Table 4).

Reducing body mass by 30 g and leaving fatmass at 200 g (44% fat) enhanced the predictedmaximum flight range (Fig. 4), but led to inap-propriately low arrival masses and very smallpectoral muscle masses (Table 4). Increasingbody mass by 30 g lean tissue (39% fat) led tolower maximum flight ranges but also to reason-able values for remaining body and pectoralmuscle masses at the point of fat depletion (Ta-ble 4). When we decreased pectoral muscle massat departure to 54 g based on the fraction ofbody mass measured in the sample of bauerifrom New Zealand (0.111; Battley and Piersma2005), final body mass was of the right orderbut pectoral muscle mass (24 g) remaining afterthe flight was certainly too low (Table 4). Whenwe gave birds with a lean mass of 285 g an extra30 g of fat (thus increasing pectoral muscle massfrom 72 g to 76 g), they increased their maxi-mum flight range, but reduced final body massand pectoral muscle mass to quite low values.Lastly, when we ‘‘flew’’ birds at 1500 m altitudeinstead of at sea level they increased their rangeby only 60–100 km but increased their optimalflight speed about 7%, resulting in a concomitantreduction in travel time (;9–12 hr).

DEPARTURES IN RELATION TO WEATHER

Actual departures of birds on southward migra-tion from Alaska were observed on four occa-sions between 1987 and 2003 (Table 3). The


FIGURE 6. Air flow over a 5-day period for ob-jects entering the air column at Nelson Lagoon,Alaska, at the time of the observed departure of god-wits on 23 September 1996. The objects, enteredinto the air column at various altitudes, are thentracked at successive 6-hr intervals with filled star5 surface, unfilled circle 5 750 m, unfilled square5 1500 m, and unfilled triangle 5 2500 m elevationabove sea level (ASL). Their respective filled sym-bols denote 24-hr periods. During the initial 60 hrof the model run between Nelson Lagoon and about408N latitude, winds at the 750-m and 500-m alti-tude varied between 10–15 m sec21. The arrow toNew Zealand approximates direction but not nec-essarily route taken.

←

FIGURE 5. En route winds (850 mb altitude 5 1500 m) associated with the four recorded departures ofgodwits from Alaska. See Table 3 for location of departure sites. The upper panels represent winds at time ofdeparture (63 hr, CUT 5 Coordinated Universal Time), middle panels show typical conditions between the 49th

and 110th hr of the flight, and lower panels show the conditions at the end of the projected 6-day transit time(but are also representative of the preceding 24–36 hr). Long, unflagged portion of the axis of each wind barpoints towards direction wind is blowing; number and type of short flags perpendicular to long bar indicatewind speed. Original wind vectors were depicted in knots and converted here to m sec21. Legend indicates rangeof directions that would be tailwinds, headwinds, or side winds along proposed transpacific migratory corridorbetween Alaska and eastern Australia and New Zealand.

1987 event entailed recovery of birds that diedafter colliding at night with a large radar domenear the distal end of the Alaska Peninsula(Piersma and Gill 1998); the three other depar-tures were documented from on-ground obser-vations (1996 and 2003) or from aerial surveys(2000). Observed departures spanned almost a7-week period between early September andmid-October (Table 3).

All four events occurred in association withmoderate troughs with imbedded storms havingcentral pressures between 976 mb and 998 mb(average 984 mb) and centered between 650 kmand 1400 km south of the departure sites (Table3, Fig. 5). Such storms are propagated in theAleutian low pressure system along a track that,beginning in September, passes south along theAleutian Islands and then northward into theGulf of Alaska. During the period 1976–2000,storms with central pressure of between 975 mband 1000 mb occurred annually along this trackon average twice in September, between two andthree times in October, and just over three timesin November. Local winds during the departureevents varied in both direction and intensity,ranging between 0–10 m sec21 from north tonortheast. The positions of the storm centers attime of departure (Fig. 5) suggest birds wouldhave to have flown on a slight west-southwestheading (200–2408) before obtaining maximumbenefit from tailwinds, but once positioned with-in the ‘upstream’ side of the systems birds flyingin a southerly direction would have encounteredstrong direct to quartering tailwinds averaging15 m sec21 (midpoint of ranges, Table 3). Windsof this approximate speed and direction wouldhave been maintained on average over a distanceof about 1000 km, with conditions associatedwith the 1996 departure, for example, extendingalmost 1500 km south (Table 3, Fig. 6).

Winds along the mid-portion (latitudes 208Nto 208S) of each of the four suspected flight


paths (Fig. 5) were very similar and character-ized by light (2–8 m sec21) crosswinds or quar-tering winds. Once into the southern realm ofthe southeast trades and austral westerlies atabout days five and six of the flight, godwitsagain experienced strong direct or quarteringtailwinds over the last 1000 km of the flight,especially if New Zealand was the destination.Any birds attempting to go to eastern Australiaduring the 2003 event would have experiencedmoderate to strong headwinds from centralQueensland south to Victoria, but mostly calmwinds if landfall were in northern Queensland(Fig. 5).

DISCUSSION

The evidence we present supporting a directnonstop flight by baueri godwits between Alas-ka and New Zealand is straightforward and com-pelling: (1) baueri godwits are extremely rarealong the central East Asian mainland on south-ward migration, (2) peak southward departurefrom Alaska and peak arrival in New Zealandoccur within the same relatively short period,and both are a month later than for godwits (L.l. menzbieri) that do follow a continental Asianroute, (3) too few godwits have been noted inOceania to suggest any regularly used interme-diate stopover site(s), but the birds that are re-corded there peak in occurrence and number inOctober and within a direct corridor linkingAlaska and New Zealand/eastern Australia, es-pecially near the terminus, where fallout of tran-sients would be expected, (4) the annual occur-rence of Hudsonian Godwits in New Zealandand eastern Australia—but their absence frommainland Asia—can best be explained by theiraccompanying Alaskan Bar-tailed Godwits on atranspacific flight, (5) birds appear energeticallyand mechanically capable of such a flight basedon current knowledge of aerodynamics and mea-sured fuel sources, and (6) known departuresfrom Alaska coincide with favorable winds fora southward flight but are in opposition to amore southwesterly continental route. Aspects ofseveral of these lines of evidence warrant addi-tional discussion.

FACTORS CONSTRAINING FLIGHT RANGE

The simulations with program FLIGHT suggestthat even under still air conditions Bar-tailedGodwits leaving Alaska staging sites with real-istic body and fat mass values should be able to

reach New Zealand in a nonstop flight of be-tween 9800 and 10 700 km. If the godwits areable to routinely use tailwinds en route, we canrelax the assumption of a body drag coefficientof 0.05 (but see Elliott et al. 2004) and accept avalue closer to the more often used 0.1 (Kvistet al. 2001, Pennycuick and Battley 2003). Inaddition to fat as fuel, protein availability andwater (dehydration) can limit flight range(Klaassen 1995, Jenni and Jenni-Eiermann1999). For the L. l. taymyrensis subspecies dur-ing a 4300-km-long northward flight from west-ern Africa to Europe, Landys et al. (2000) con-cluded that flights ranging in altitude from sealevel to 3000 m would avoid dehydration, andin fact found no evidence for dehydration in ar-riving godwits. Interestingly, Landys et al.(2000) also had to assume a body drag coeffi-cient as low as 0.05 for the virtual godwits tocomplete their flight.

Our simulations have made clear that the nec-essary minimum protein use during nonstopflights does limit maximum flight range. Birdsleaving Alaska with a lean mass lower than 275g are predicted to have exhausted their fat whentheir lean mass is as low as 200 g and their pec-toral muscles have become tiny (Table 4). Thesmall-sized juvenile Bar-tailed Godwits with alean mass of only 166 g and a fat store of 200g (Piersma and Gill 1998) are predicted to beable to cover more than 11 000 km nonstop. Notsurprisingly, they are also predicted to arrivewith perhaps unrealistically low lean and pec-toral muscle masses (for a body drag coefficientof 0.05, lean mass after 11 000 km of flight un-der still air conditions would be 170 g and pec-toral muscle mass 25 g; with a coefficient of 0.1the predicted final mass values are 130 g and 20g, respectively). Given such values upon arrivalin New Zealand, it seems unwarranted to expectthese birds to have been capable of reaching theSouth Pole, an additional distance of 6000 km,as predicted by Pennycuick and Battley (2003),who accommodated unrealistic lean mass val-ues. Lowering the minimum energy obtainedfrom protein to 2% does not resolve the prob-lem. In view of the absence of hard body com-position data for adult godwits, and the problem-atic departure condition of the juveniles from1987, detailed studies examining departure bodycondition in relation to performance during theensuing flight and pinpointing exact arrival timein New Zealand are clearly needed.


A DIRECT ROUTE OR ONE WITH STOPOVERS?

A direct flight by a congener. Hudsonian God-wits breed in subarctic and temperate NorthAmerica and migrate to southern South America(Elphick and Klima 2002), yet are rare annualvisitors to New Zealand (Higgins and Davies1996, Elphick and Klima 2002) and occasionallyelsewhere in Oceania (Watling 2001), with up tonine different individuals seen in a single year.It is highly unlikely that Hudsonian Godwits,though also long-distance migrants (McCafferyand Harwood 2000), reach southern Oceania byfollowing a continental route via the East Asianmainland, a distance of over 16 000 km. Indeed,we could only find a single record of the speciesfrom Asia and that from Chukotka almost 150years ago (in Kessel and Gibson 1978). It is alsohighly unlikely that birds reach New Zealand viaa 9000-km-long flight across the SouthernOcean after an initial flight of 8000 to 11 000km from eastern Canada to southern SouthAmerica. The most logical explanation for theiroccurrence in New Zealand is that they accom-pany Bar-tailed Godwits on the godwits’ south-ward flight across the Pacific (see also Kesseland Gibson 1978). Recent observations of smallnumbers of Hudsonian Godwits (all juveniles todate) among large flocks of Bar-tailed Godwitsat principal Alaskan staging sites (R. Gill and B.McCaffery, unpubl. data) support this idea.

Stopovers. As it is energetically more favor-able to cover a certain migration distance inmany small steps than in one long hop (Piersma1987), the assembled evidence that some150 000 Bar-tailed Godwits annually make a11 000-km-long nonstop flight from Alaska toNew Zealand/eastern Australia begs the questionof why do they not make stopovers, either alongthe East Asian mainland if such a route is fol-lowed or during a transpacific crossing. First, theevidence we have assembled fails to support useof a continental route, with or without use ofstopover sites. Given the paucity of records ofbaueri along the East Asian mainland duringsouthward migration, if they were migratingalong the Asian coast it would entail a nonstopflight of almost 16 000 km, 40% longer than aflight directly across the Pacific. Such a longflight is improbable given the fuel loads of de-parting godwits and predicted arrival mass uponfuel depletion (contra Pennycuick and Battley2003). In addition, during the recorded departureevents, birds would have initially encountered

moderate to strong headwinds and then a longfetch of strong southerly winds if they had fol-lowed a more southwesterly route along theAsian mainland. Such a flight would have forcedbirds to either fly into opposing winds or detouraround the systems, both of which would haveadded substantially to energetic costs.

We likewise found no evidence suggestinguse of intermediate stopover sites if birds fol-lowed a direct route across the Pacific. Argu-ably, the Pacific Ocean is vast and it could har-bor yet undiscovered stopover site(s), but the re-gion has received considerable attention fromornithologists, and its indigenous peoples are in-timately in tune with their natural resources. Wefind it beyond reason to expect the annual useby 150 000 godwits at stopover site(s) in Oce-ania to have gone undetected by either group ofpeople. Indeed, the 80-year span of records wesearched accounted for a total of only about4000 godwits having been recorded throughoutall of Oceania—this from a projected total ofsome 12 million godwits that could havestopped somewhere en route during this period.And though we cannot rule out that birds alighton open ocean waters during transit, it likelywould be for relatively short periods and thenrelated to adverse conditions (Piersma et al.2002) and not for rest.

Thus, a single flight over the Pacific is notonly likely but in several ways advantageous: itmay be safer (there are rarely aerial predators incentral Oceania; cf. Ydenberg et al. 2002),healthier (as encounters with pathogens will beavoided; Piersma 1997), and faster and more di-rect (as the time to settle at new stopover areasis avoided; Alerstam and Lindstrom 1990). Itcould also indicate the high quality of the west-ern Alaskan staging sites relative to potentialstaging/stopover areas along the East Asiancoastline or throughout Oceania (Gill and Han-del 1990, Gudmundsson et al. 1991). Indeed,soft substrate intertidal habitat, the preferredfeeding substrate for nonbreeding godwits, is ex-tremely limited throughout Oceania, occurringmostly on Fiji, the one site in Oceania that reg-ularly hosts godwits (Watling 2001).

Our story has its perplexities. There are threereports of putative baueri from the East Asianmainland during the southward migration peri-od—one of a Victoria, Australia-flagged birdseen in the North China Sea region in late Au-gust, and two of New Zealand-banded birds


from Kamchatka, Russia; one seen in mid-Au-gust and another shot in early October (Riegen1999, Gosbell 2004, J. Wilson and C. Minton,unpubl. data). One or more scenarios likely ex-plain these three exceptions to an otherwiseclear pattern of subspecies segregation duringsouthward migration. First, though the two sub-species are largely segregated on the nonbreed-ing grounds, there are records of each occurringin the other’s range (J. Wilson and C. Minton,unpubl. data) and thus potentially occurringelsewhere outside normal distributions. A con-founding factor is the uncertain taxonomic status(Engelmoer and Roselaar 1998, Tomkovich andSerra 1999) and unknown nonbreeding range ofthe small population of godwits that nests in theAnadyr Lowlands and southern Chukotks Pen-insula, Russia, intermediate between the rangesof menzbieri in Siberia and baueri in Alaska(McCaffery and Gill 2001). Phenotypic differ-ences are slight, with Anadyr birds more closelyresembling baueri in both size and plumage thanthey do menzbieri and thus more likely to beidentified as baueri during banding. The recov-ery of the New Zealand-banded bird in southernKamchatka, Russia, in October is a bit more dif-ficult to explain, especially since most godwitshave departed estuaries along the west coast ofKamchatka by early September (Gerasimov andGerasimov 1998). If the Anadyr populationproves distinct, it is likely the source of birdsreported in Japan in September and in Micro-nesia where, though few in number, specimensof putative baueri have been recorded duringsouthward migration (Stickney 1943, Baker1951, Stinson et al. 1997), considerably west ofa direct migration corridor between Alaska andNew Zealand or eastern Australia. The occur-rence of godwits in Micronesia is likely ex-plained by fallout of migrants coming from theSea of Okhotsk and Japan to northern Australiaalong a direct route that passes over the North-ern Mariana archipelago. A radar study of shore-bird migration over Guam during mid-August tolate October showed just such a pattern (Wil-liams and Williams 1988, 1999).

THE ROLE OF WIND SYSTEMS OVERTHE PACIFIC

The baueri subspecies of Bar-tailed Godwit canbe added to a growing list of birds that haveevolved wind-sensitive migration strategies—es-pecially during southward migrations—within

the subpolar marine low pressure belt that circlesthe northern hemisphere (Richardson 1979,1990, Akesson and Hedenstrom 2000, Green2003, M. Green and T. Piersma, unpubl. data).This phenomenon is especially evident in theNorth Pacific, where the Aleutian low pressuresystem shapes and dominates weather and windpatterns throughout the year (Christoforou andHameed 1997, Overland et al. 1997). In partic-ular, two taxa of large-bodied, medium-distancemigrant geese annually initiate transoceanicflights from the Alaska Peninsula to the PacificCoast of North America in conjunction with thepassage of moderate to strong low pressure sys-tems (Dau 1992, Gill et al. 1997). Not surpris-ingly, even small-bodied birds such as Dunlin(C. alpina) having similar nonbreeding ranges,often depart on the same weather systems asused by geese (Warnock and Gill 1996, R. Gill,unpubl. data). The emerging pattern is that god-wits, geese, and Dunlin can use the same storms,only varying their departures in accordance withthe position of the storm center and the birds’final destination.

Evolving a migration system in conjunctionwith winds at the departure site is one thing but,in the case of godwits that are transiting the en-tire Pacific, they must pass through at least fiveother latitudinal zones of defined winds andpressure. It is beyond the scope of this paper topresent a thorough analysis of wind conditionsalong the entire projected flight path (cf. Piersmaand Jukema 1990, Piersma and van de Sant1992, Akesson and Hedenstrom 2000). But whatemerges from the four cases we studied suggeststhat winds were generally favorable throughoutthe migration corridor during the calculated 6-day transit time and certainly in no instance wasthere strong opposing wind for any appreciabledistance. The most obvious question relating tothis is to what extent local departure cues arerelated to favorable ‘downstream’ winds? Isweather across the Pacific teleconnected suchthat certain departure cues at northern latitudesassure relatively favorable conditions alongmost of the route (McCaffery and Gill 2001)?The Aleutian low pressure center is a large-scaledominating feature of the North Pacific (Chris-toforou and Hameed 1997, Overland et al. 1997)that has obviously shaped the evolution ofequally large-scale geographical migration pat-terns, similar to systems described elsewhere inthe northern hemisphere (citations in Green


2003). But the godwits’ migration strategy, in-volving flights that span two hemispheres, waslikely not selected for based solely on factorsoccurring over just a portion of the range andindependent of those elsewhere along the migra-tion corridor.

This raises obvious questions about global cli-mate change and its effects on wind regimes andthus on wind-selected avian migrants. The eco-logical effects of climate fluctuations are manyand projected to be most profound in regionswith large-scale patterns of climate variabilitysuch as the North Atlantic and North Pacific(e.g., Stenseth et al. 2002). Global climatechange models suggest an intensification ofpropagating weather systems moving across theNorth Pacific. Such would result in a shift of theAleutian low center eastward that would in turnincrease the number and intensity of storms andbring stronger northerly winds over a longerfetch on the backside of individual low pressurecenters. Godwits may, however, be able to adaptto this as the Aleutian Low and adjacent Ha-waiian High have been found to shift positionand intensity (seesaw pattern) on a decadal scaleduring most of the twentieth century (Christo-forou and Hameed 1997, Overland et al. 1997).The phenomenon, however, also needs to be as-sessed in the southern hemisphere in terms ofteleconnection between the two hemispheres.

Development of suitable remote sensing sat-ellite technology would greatly enhance our un-derstanding of the complexity of the godwits’migration system and flight behaviors of long-distance transoceanic migrants in general. Withsuch technology, answers would be forthcomingto questions about (1) mechanisms of orienta-tion, (2) how birds select winds (vertical andlateral) at all stages of the flight, (3) whether andhow they compensate for wind drift, (4) whetherthey adjust airspeed during the course of theflight, and (5) the extent to which they can as-sess and react to changes in downstream flightconditions.

THE PACIFIC AS AN ECOLOGICAL BARRIER?

Although the Pacific may be an ecological bar-rier for humans and for numerous Palearctic andNearctic migratory landbirds (Baker 1951, Prattet al. 1987), it does not appear to be a barrierfor the baueri race of the Bar-tailed Godwit, atleast under the observed ecological conditionsassociated with departures from Alaska and ar-

rivals in New Zealand and eastern Australia. ThePacific as a barrier is further mollified consid-ering that southward migration by godwits ap-pears to be wind-selected, at least at departure,and that such winds appear to confer significantenergy savings. With inferred high rates of an-nual survival (A. Riegen in McCaffery and Gill2001, R. Gill, unpubl. data), experienced adultsmay actually encounter few losses during theirsouthward journeys, but it is quite possible thatthe situation is different for juveniles migratingtheir first time, especially if they fly in flockscomposed only of inexperienced juveniles. Andsuch appears to be the case, at least concerninglate-departing birds from Alaska and late arriv-als in New Zealand (Piersma and Gill 1999, Rie-gen 1999, R. Gill, unpubl. data). The age struc-ture of the population occurring during the non-breeding season in Fiji is poorly known (D. Wat-ling, pers. comm.), but one could expect a highproportion of inexperienced juveniles to occurthere since Fiji is the last suitable land before a2000-km-long open-ocean crossing to New Zea-land.

Future studies should focus primarily on ques-tions related to flight behaviors throughout themigration corridor, but equally rewarding wouldbe to learn both local and regional patterns ofarrival and subsequent movements of birds. Forexample, is the movement of godwits fromQueensland, Australia, to New Zealand (Riegen1999) an annual event and are these mostly ju-veniles that are incapable of the longer directflight to New Zealand?

ACKNOWLEDGMENTS

This work benefited from numerous discussions withcolleagues throughout the Pacific basin, including M.Barter, D. Rogers, D. Milton, and C. Minton in Aus-tralia; P. Battley in New Zealand; G. McCormack inthe Cook Islands; P. Tomkovich and Y. Gerasimov inRussia; and C. Dau, C. Handel, B. McCaffery, and D.Ruthrauff in Alaska. Unpublished information on theoccurrence of godwits throughout Oceania was provid-ed by P. Bruner, P. Donaldson, W. Johnson, J. Marks,G. McCormack, R. Pyle, F. Ward, and D. Watling. M.Kashiwage and Y. Shigeta provided information fromJapan. A cadre of New Zealanders helped assess thearrival of godwits following observed departures inAlaska, especially P. Battley, H. and Z. Clifford, T.Habraken, D. Lawrie, D. Melville, R. Schuckard, andB. and B. Woolley. Banding records were extractedfrom the extensive databases maintained by the Aus-tralian Bird and Bat Banding Scheme, the New Zea-land Bird Banding Scheme, and New Zealand’s Mi-randa Shorebird Center. M. Whalen helped prepare


graphics. Earlier versions of the manuscript benefitedby comments and suggestions from P. Battley, M. Bar-ter, D. Dobkin, R. Drent, C. Handel, M. Klaassen, C.Minton, L. Tibbitts, and two anonymous reviewers.

LITERATURE CITED

AKESSON, S., AND A. HEDENSTROM. 2000. Wind selec-tivity of migratory flight departures in birds. Be-havioral Ecology and Sociobiology 47:140–144.

ALERSTAM, T., AND A. LINDSTROM. 1990. Optimal birdmigration: the relative importance of time, energy,and safety, p. 331–359. In E. Gwinner [ED.], Birdmigration: physiology and ecophysiology. Spring-er, Berlin.

BAKER, R. H. 1951. The avifauna of Micronesia, itsorigin, evolution, and distribution. University ofKansas Publications of Museum Natural History3:1–359.

BARTER, M. 1989. Bar-tailed Godwit Limosa lapponicain Australia, Part 1: races, breeding areas and mi-gration routes. Stilt 14:43–48.

BARTER, M., AND T. H. WANG. 1990. Can waders flynon-stop from Australia to China? Stilt 17:36–39.

BATTLEY, P. F. 1997. The northward migration of Arc-tic waders in New Zealand: departure behaviour,timing and possible migration routes of Red Knotsand Bar-tailed Godwits from Farewell Spit, North-West Nelson. Emu 97:108–120.

BATTLEY, P. F., M. W. DIETZ, T. PIERSMA, A. DEKINGA,S. TANG, AND K. HULSMAN. 2001. Is long-distancebird flight equivalent to a high-energy fast? Bodycomposition changes in freely migrating and cap-tive fasting Great Knots. Physiological and Bio-chemical Zoology 74:435–449.

BATTLEY, P. F., AND T. PIERSMA. 2005. Body composi-tion and flight ranges of Bar-tailed Godwits (Li-mosa lapponica baueri) from New Zealand. Auk,in press.

BATTLEY, P. F., T. PIERSMA, M. W. DIETZ, S. TANG, A.DEKINGA, AND K. HULSMAN. 2000. Empirical evi-dence for differential organ reductions duringtrans-oceanic bird flight. Proceedings of the RoyalSociety of London Series B 267:191–195.

BIER, J. A. 1995. Reference map of Oceania: the Pa-cific islands of Micronesia, Polynesia, and Mela-nesia. University of Hawaii Press, Honolulu.

BYRNE, J. E. [ED.]. 1979. Literature review and syn-thesis of information on Pacific Island ecosystems.U.S. Fish and Wildlife Service, Office of Biolog-ical Services, Washington, DC. FWS/OBS-79/35.

CHRISTOFOROU, P., AND S. HAMEED. 1997. Solar cyclesand the Pacific ‘centers of action.’ GeophysicalResearch Letters 24:293–296.

CDC [ONLINE]. 2004. Canadian Diagnostic Centre.NCEP/NCAR reanalysis 1. Retrieved from theNOAA-CRIES Climate Diagnostic Center WorldWide Web. ,http://www.cdc.noaa.gov/cdc/data.ncep.reanalysis.html. (23 November 2004).

CMC [ONLINE]. 2001. Canadian Meteorological Cen-tre, Long range transport models. ,http://www.cmc.ec.gc.ca/cmc/CMOE/modele/A-modele.html. (23 November 2004).

CMC [ONLINE]. 2004. Canadian Meteorological Cen-tre, Global Forecast System. Retrieved from the

Canadian Meteorological Centre World Wide WebLibrary. ,http://www.smc-msc.ec.gc.ca/cmc/oppsystems/globalpforecastpe.html. (23 Novem-ber 2004).

DAU, C. P. 1992. The fall migration of Pacific FlywayBrent Branta bernicla in relation to climatic con-ditions. Wildfowl 43:80–95.

DRISCOLL, P. 1997. The distribution of waders alongthe Queensland coastline, p. 80–122. In P. Straw[ED.], Shorebird conservation in the Asia-PacificRegion. Australasia Wader Studies Group Report,Canberra, Australia.

ELLIOTT, K. H., M. HEWETT, G. KAISER, AND R. W.BLAKE. 2004. Flight energetics of the MarbledMurrelet, Brachyramphus marmoratus. CanadianJournal of Zoology 82:644–652.

ELPHICK, C. S., AND J. KLIMA. 2002. Hudsonian Godwit(Limosa fedoa). In A. Poole and F. Gill [EDS.], Thebirds of North America, No. 629. The Birds ofNorth America, Inc., Philadelphia, PA.

ENGELMOER, M., AND C. ROSELAAR. 1998. Geographi-cal variation in waders. Kluwer Academic Pub-lishers, Dordrecht, The Netherlands.

GERASIMOV, N. N., AND Y. N. GERASIMOV. 1998. Theinternational significance of wetland habitats inthe lower Moroshechnaya River (West Kamchat-ka, Russia) for waders. International Wader Stud-ies 10:237–242.

GILL, R. E., JR., C. A. BABCOCK, C. M. HANDEL, W. R.BUTLER JR., AND D. G. RAVELING. 1997. Migra-tion, fidelity, and use of autumn staging groundsin Alaska by Cackling Canada Geese Brantacanadensis minima. Wildfowl 47:42–61.

GILL, R. E., JR., AND C. M. HANDEL. 1990. The im-portance of subarctic intertidal habitats to shore-birds: a study of the central Yukon-KuskokwimDelta, Alaska. Condor 92:702–725.

GILL, R. E., JR., AND B. J. MCCAFFERY. 1999. Bar-tailed Godwits Limosa lapponica in Alaska: apopulation estimate from the staging grounds.Wader Study Group Bulletin 88:49–54.

GILL, R. E., JR., B. J. MCCAFFERY, AND P. S. TOMKO-VICH. 2002. Wandering Tattler (Heteroscelus in-canus). In A. Poole and F. Gill [EDS.], The birdsof North America, No. 642. The Birds of NorthAmerica, Inc., Philadelphia, PA.

GOSBELL, K. 2004. Self-funding volunteers help inKamchatka. Tattler 41:3–5.

GREEN, M. 2003. Flight strategies in migrating birds:when and how to fly. Ph.D. dissertation, Depart-ment of Ecology, Lund University, Sweden.

GUDGEON, W. E. 1903. The whence of the Maori, Part3. Journal of the Polynesian Society. Vol. 12.

GUDMUNDSSON, G. A., A. LINDSTROM, AND T. ALERS-TAM. 1991. Optimal fat loads and long-distanceflights by migrating Knots Calidris canutus, Sand-erlings C. alba and Turnstones Arenaria interpres.Ibis 133:140–152.

HAWKINS, J. M. 1980. Seasonal fluctuations in numbersof Bar-tailed Godwits on Nelson Haven. Notornis27:88–90.

HEATHER, B. D., AND P. M. SHEEHAN. 1990. Fifty yearsof birds study in New Zealand: an index to No-


tornis 1939–1989. Ornithological Society of NewZealand.

HIGGINS, P. J., AND S. J. J. F. DAVIES [EDS.]. 1996.Handbook of Australian, New Zealand and Ant-arctic birds. Vol. 3: Snipe to Pigeons. Oxford Uni-versity Press, Melbourne.

HURLES, M. E., E. MATISOO-SMITH, R. D. GRAY, AND

D. PENNY. 2003. Untangling oceanic settlement:the edge of the knowable. Trends in Ecology andEvolution 18:531–540.

JENNI, L., AND S. JENNI-EIERMANN. 1998. Fuel supplyand metabolic constraints in migrating birds. Jour-nal of Avian Biology 29:521–528.

JENNI, L., AND S. JENNI-EIERMANN. 1999. Fat and pro-tein utilisation during migratory flight. Interna-tional Ornithological Congress 22:1437–1449.

JENNI-EIERMANN, S., L. JENNI, A. KVIST, A. LINDSTROM,T. PIERSMA, AND G. H. VISSER. 2002. Fuel use andmetabolic response to endurance exercise: a windtunnel study of a long-distance migrant shorebird.Journal of Experimental Biology 205:2453–2460.

JOHNSON, O. W. 2003. Pacific and American Golden-Plovers: reflections on conservation needs. WaderStudy Group Bulletin 100:10–13.

KEATES, A. 2003. A survey of Bar-tailed Godwits insouth-east Queensland. Stilt 44:50–51.

KERSTEN, M., AND T. PIERSMA. 1987. High levels ofenergy expenditure in shorebirds: metabolic ad-aptations to an energetically expensive way oflife. Ardea 75:175–187.

KESSEL, B., AND D. GIBSON. 1978. Status and distribu-tion of Alaska birds. Studies in Avian Biology 1.

KLAASSEN, M. 1995. Water and energy limitations onflight range. Auk 112:260–262.

KVIST, A., A. LINDSTROM, M. GREEN, T. PIERSMA, AND

G. H. VISSER. 2001. Carrying large fuel loads dur-ing sustained bird flight is cheaper than expected.Nature 413:730–732.

LANDYS, M. M., T. PIERSMA, G. H. VISSER, J. JUKEMA,AND A. WIJKER. 2000. Water balance during realand simulated long-distance migratory flight inthe Bar-tailed Godwit. Condor 102:645–652.

LANDYS-CIANNELLI, M. M., T. PIERSMA, AND J. JUKEMA.2003. Strategic size changes of internal organsand muscle tissue in the Bar-tailed Godwit duringfat storage on a spring stopover site. FunctionalEcology 17:151–159.

MARKS, J. S., AND R. L. REDMOND. 1994. Migration ofBristle-thighed Curlews on Laysan Island: timing,behavior and estimated flight range. Condor 96:316–330.

MARKS, J. S., T. L. TIBBITTS, R. E. GILL JR., AND B. J.MCCAFFERY. 2002. Bristle-thighed Curlew (Nu-menius tahitiensis). In A. Poole and F. Gill [EDS.],The birds of North America, No. 705. The Birdsof North America, Inc., Philadelphia, PA.

MCCAFFERY, B. J., AND R. E. GILL JR. 2001. Bar-tailedGodwit (Limosa lapponica). In A. Poole and F.Gill [EDS.], The birds of North America, No. 581.The Birds of North America, Inc., Philadelphia,PA.

MCCAFFERY, B. J., AND C. M. HARWOOD. 2000. Statusof Hudsonian Godwits on the Yukon-KuskokwimDelta, Alaska. Western Birds 31:165–177.

MINTON, C. 2004. Summary of activities. VictorianWader Study Group Bulletin 27:1–11.

MINTON, C., R. JESSOP, P. COLLINS, J. DELEYEV, AND L.BEASLEY. 2002. Sightings of waders leg-flagged inVictoria: report number 9. Stilt 42:56–72.

MISKELLY, C. M., P. M. SAGAR, A. J. D. TENNYSON,AND R. P. SCOFIELD. 2001. Birds of the Snares Is-lands, New Zealand. Notornis 48:1–40.

MOORES, N. 1999. A survey of the distribution andabundance of shorebirds in South Korea during1998–1999: interim summary. Stilt 34:18–29.

OVERLAND, J. E., J. M. ADAMS, AND N. A. BOND. 1997.Decadal variability in the Aleutian low and its re-lation to high latitude circulation. Journal of Cli-mate 12:1542–1548.

PENNYCUICK, C. J. 1989. Bird flight performance: apractical calculation manual. Oxford UniversityPress, Oxford, UK.

PENNYCUICK, C. J. 1998. Computer simulation of fatand muscle burn in long-distance bird migration.Journal of Theoretical Biology 191:47–61.

PENNYCUICK, C. J. [ONLINE]. 2004. FLIGHT. Version1.15. ,http://www.bio.bris.ac.uk/people/staff.cfm?key595. (12 November 2004).

PENNYCUICK, C. J., AND P. F. BATTLEY. 2003. Burningthe engine: a time-marching computation of fatand protein consumption in a 5420-km non-stopflight by Great Knots Calidris tenuirostris. Oikos103:323–332.

PENNYCUICK, C. J., M. KLAASSEN, A. KVIST, AND A.LINDSTROM. 1996. Wingbeat frequency and thebody drag anomaly: wind-tunnel observations ona Thrush Nightingale (Luscinia luscinia) and aTeal (Anas crecca). Journal of Experimental Bi-ology 199:2757–2765.

PHILLIPPS, W. J. 1966. Maori life and custom. A. H.and A. W. Reed, Wellington, New Zealand.

PIERSMA, T. 1987. Hop, skip or jump? Constraints onmigration of Arctic waders by feeding, fattening,and flight speed. Limosa 60:185–194.

PIERSMA, T. 1997. Do global patterns of habitat use andmigration strategies co-evolve with relative in-vestments in immunocompetence due to spatialvariation in parasite pressure? Oikos 80:623–631.

PIERSMA, T., AND R. E. GILL JR. 1998. Guts don’t fly:small digestive organs in obese Bar-tailed God-wits. Auk 115:196–203.

PIERSMA, T., G. A. GUDMUNDSSON, AND K. LILLIEN-DAHL. 1999. Rapid changes in the size of differentfunctional organ and muscle groups during refu-eling in a long-distance migrating shorebird. Phys-iology and Biochemical Zoology 72:405–415.

PIERSMA, T., AND J. JUKEMA. 1990. Budgeting the flightof a long-distance migrant: changes in nutrient re-serve levels of Bar-tailed Godwits at successivestaging sites. Ardea 78:315–337.

PIERSMA, T., B. SPAANS, AND A. DEKINGA. 2002. Areshorebirds sometimes forced to roost on water inthick fog? Wader Study Group Bulletin 97:42–44.

PIERSMA, T., AND S. VAN DE SANT. 1992. Pattern andpredictability of potential wind assistance for wad-ers and geese migrating from West Africa and theWadden Sea to Siberia. Ornis Svecica 2:55–66.


PRATT, H. D., P. L. BRUNER, AND D. G. BERRETT. 1987.A field guide to the birds of Hawaii and the trop-ical Pacific. Princeton University Press, Princeton,NJ.

RICHARDSON, W. J. 1979. South-eastward shorebird mi-gration over Nova Scotia and New Brunswick inautumn: a radar study. Canadian Journal of Zo-ology 57:107–124.

RICHARDSON, W. J. 1990. Timing of bird migration inrelation to weather: updated review, p. 78–101. InE. Gwinner [ED.], Bird migration: physiology andecophysiology. Springer, Berlin.

RIEGEN, A. C. 1999. Movements of banded Arcticwaders to and from New Zealand. Notornis 46:123–142.

RILEY, M. 2001. Maori bird lore. Viking Seven SeasNZ, Ltd., Paraparaumu, New Zealand.

SAGAR, P. M., U. SHANKAR, AND S. BROWN. 1999. Dis-tribution and numbers of waders in New Zealand,1983–1994. Notornis 46:1–43.

SATOSHI, M. [ED.]. 2003. Report of the shorebird pop-ulation change monitoring census in Japan, 2002.WWF Japan and Ministry of Environment, Japan.

SKINNER, N. J. 1983. The occurrence of waders at SuvaPoint, Fiji. Notornis 30:227–232.

SMART, J. B. 1971. Notes on the occurrence of wadersin Fiji. Notornis 18:267–279.

STENSETH, N., A. MYSTERUD, G. OTTERSEN, J. HUR-RELL, K.-S. CHAN, AND M. LIMA. 2002. Ecologicaleffects of climate fluctuations. Science 297:1292–1296.

STICKNEY, E. H. 1943. Birds collected during the Whit-ney South Sea expedition. American Museum No-vitates 1248:1–9.

STINSON, D. W., G. J. WILES, AND J. D. REICHEL. 1997.Occurrence of migrant shorebirds in the MarianaIslands. Journal of Field Ornithology 68:42–55.

TE PASS, W. 1912. The story of the Kuaka’s flight.Journal of the Polynesian Society 21.

THOMPSON, M. 1973. Migratory patterns of RuddyTurnstones in the central Pacific region. LivingBird 12:5–23.

TOMKOVICH, P. S., AND L. SERRA. 1999. Morphometricsand prediction of breeding origin in some Holarc-tic waders. Ardea 87:289–300.

WARNOCK, N., AND R. E. GILL JR. 1996. Dunlin (Ca-lidris alpina). In A. Poole and F. Gill [EDS.], Thebirds of North America, No. 203. The Academyof Natural Sciences, Philadelphia, PA, and TheAmerican Ornithologists’ Union, Washington,DC.

WATLING, D. 2001. A guide to the birds of Fiji andwestern Polynesia including American Samoa,Niue, Samoa, Tokelau, Tonga, Tuvalu and Wallisand Futuna. Environmental Consultants (Fiji) Ltd.

WILLIAMS, T. C., AND J. M. WILLIAMS. 1988. Radar andvisual observations of autumnal (southward)shorebird migration on Guam. Auk 105:460–466.

WILLIAMS, T. C., AND J. M. WILLIAMS. 1990. The ori-entation of transoceanic migrants, p. 9–21. In E.Gwinner [ED.], Bird migration: physiology andecophysiology. Springer, Berlin.

WILLIAMS, T. C., AND J. M. WILLIAMS. 1999. The mi-gration of land birds over the Pacific Ocean. In-ternational Ornithological Congress 22:1948–1957.

WILSON, J. R., AND M. A. BARTER. 1998. Identificationof potentially important staging areas for ‘‘longjump’’ migrant waders in the East-AustralasianFlyway during northward migration. Stilt 32:16–27.

YDENBERG, R. C., R. W. BUTLER, D. B. LANK, C. G.GUGLIELMO, M. LEMON, AND N. WOLF. 2002.Trade-offs, condition dependence and stopoversite selection by migrating sandpipers. Journal ofAvian Biology 33:47–55.

University of Groningen Crossing the ultimate ecological ...

Documents

Transcript of University of Groningen Crossing the ultimate ecological ...