Unravelling the migration and moult strategies of a long-distance migrant using stable isotopes: Red Knot Calidris canutus movements in the Americas

Ibis

(2005), doi: 10.1111/j.1474-919x.2005.00455.x

© 2005 British Ornithologists’ Union

Blackwell Publishing, Ltd.

Unravelling the migration and moult strategies of a long-distance migrant using stable isotopes: Red Knot

Calidris canutus

movements in the Americas

PHILIP W. ATKINSON,

1

* ALLAN J. BAKER,

2

RICHARD M. BEVAN,

3

NIGEL A. CLARK,

1

KIMBERLY B. COLE,

4

PATRICIA M. GONZALEZ,

5

JASON NEWTON,

6

LAWRENCE J. NILES

7

& ROBERT A. ROBINSON

1

1

British Trust for Ornithology, The Nunnery, Thetford, Norfolk IP24 2PU, UK

2

Centre for Biodiversity and Conservation Biology, Royal Ontario Museum, 100 Queen’s Park, Toronto, Ontario M5S 1C6, Canada and, Department of Zoology, University of Toronto, Queen’s Park, Ontario, Canada

3

University of Newcastle, School of Biology, King George VI Building, University of Newcastle, Newcastle upon Tyne NE1 7RU, UK

4

Delaware Coastal Programs, Division of Soil and Water Conservation, DNREC, 89 Kings Highway, Dover, DE 19901, USA

5

Fundación Inalafquen, Pedro Morón 385 (8520) San Antonio Oeste, Río Negro, Argentina

6

NERC Life Sciences Mass Spectrometry Facility, Scottish Universities Environmental Research Centre, Rankine Avenue, Scottish Enterprise Technology Park, East Kilbride G75 0QF, UK

7

Endangered and Nongame Species Program, New Jersey Division of Fish and Wildlife, PO Box 400, Trenton, NJ 08625, USA

For long-distance migrants, such as many of the shorebirds, understanding the demographicimplications of behavioural strategies adopted by individuals is key to understanding howenvironmental change will affect populations. Stable isotopes have been used in the terrestrialenvironment to infer migratory strategies of birds but rarely in marine or estuarine systems.Here, we show that the stable isotope ratios of carbon and nitrogen in flight feathers canbe used to identify at least three discrete wintering areas of the Red Knot

Calidris canutus

on the eastern seaboard of the Americas, ranging from southeastern USA to Patagonia andTierra del Fuego. In spring, birds migrate northwards via Delaware Bay, in the northeastern USA,the last stopping point before arrival in Arctic breeding areas, where they fatten up on eggsof spawning Horseshoe Crabs

Limulus polyphemus

. The isotope ratios of feather samplestaken from birds caught in the Bay during May 2003 were compared with feathers obtainedfrom known wintering areas in Florida (USA), Bahia Lomas (Chile) and Rio Grande (Argen-tina). In May 2003, 30% of birds passing through the Bay had Florida-type ‘signatures’, 58%were Bahia Lomas-type, 6% were Rio Grande-type and 7% were unclassified. Some of thesouthern wintering birds had started moulting flight feathers in northern areas, suspendedthis, and then finished their moult in the wintering areas, whereas others flew straight tothe wintering areas before commencing moult. This study shows that stable isotopes canbe used to infer migratory strategies of coastal-feeding shorebirds and provides the basis foridentifying the moult strategy and wintering areas of birds passing through Delaware Bay.Coupled with banding and marking birds as individuals, stable isotopes provide a powerfultool for estimating population-specific demographic parameters and, in this case, further ourunderstanding of the migration systems of the declining Nearctic populations of Red Knot.

Increasingly, population ecology is focusing at the levelof the individual. Understanding the demographicimplications of the behavioural decisions adopted by

individual animals and the variation between individ-uals is key to determining how changes in the animals’environment might have an impact at the populationlevel (Sutherland 1996, Grimm & Uchmanski 2002,Ydenberg

et al

. 2002, Stillman

et al

. 2003). Long-distance migrants, such as many shorebirds (Charadrii),

*Corresponding author.Email: [email protected]

2

P. W. Atkinson

et al.

© 2005 British Ornithologists’ Union,

Ibis

, doi: 10.1111/j.1474-919x.2005.00455.x

have evolved different migration strategies, operat-ing at different scales. Different populations maybreed and winter in geographically distinct regions,moving between them via common or different stag-ing sites. Within these regions, an individual’s choiceof wintering site can have important demographicconsequences (Gill

et al

. 2001). Wintering and stag-ing sites tend to be small in number, highly produc-tive and widely spaced (Butler

et al

. 2001). Thechoice of when to move between these sites, howlong to stay, at what weight to depart and when toreach the breeding grounds have all been shown todetermine individual fitness and affect demographicrates (Boyd 1992, Sandberg & Moore 1996, Marra

et al

. 1998, Ydenberg

et al

. 2002, Atkinson

et al

.2003, Bêty

et al

. 2003, Baker

et al

. 2004).Reliance on a limited number of staging sites

makes shorebirds extremely vulnerable to issuessuch as global climate change, habitat change or loss,human exploitation of their food resources and dis-turbance within these sites (Piersma & Baker 2000).To predict the effects of these issues on populationswe need to answer several key questions including:(a) how does the choice of a wintering site affectsubsequent survival, (b) what are the fitness conse-quences of variation in the timing of arrival at stagingand breeding sites and (c) to what extent do survivaland reproduction depend on the size of nutrientstores accumulated at the last stopover site in theannual spring migration (Alerstam & Hedenström1998, Gill

et al

. 2001, Madsen 2001, Drent

et al

.2003, Battley

et al

. 2004).Populations of mixed breeding or wintering origin

often occur in stopover sites in spring and autumn,e.g. Dunlin

Calidris alpina

, Bar-tailed Godwits

Limosalimosa

and Red Knot

Calidris canutus

in the WaddenSea in spring (Prokosch 1988; Goede

et al

. 1990)and Bar-tailed Godwits in the Wash estuary in autumn(Atkinson 1996). To have a fuller understanding ofthe changes in these different populations, it isimportant to calculate separate demographic para-meters for each group of birds. Unfortunately, in thecase of shorebirds, this is often difficult, as there israrely a clear feature that can identify the differentpopulations. In many migrant shorebirds, body sizevaries between breeding, and hence, wintering areas(Engelmoer & Roselaar 1998), but the differences arerarely sufficient to allocate birds to individual popu-lations. Although genetic markers can sometimespartly or wholly distinguish different subspecies ofshorebirds (Wenink

et al

. 1994), post-Pleistocenepopulations are generally too recently diverged to be

identified unequivocally with genetic tags (Baker 2002),and other intrinsic markers need to be explored.One such technique is to use the ratios of naturallyoccurring stable isotopes of various elements foundin feathers (Chamberlain

et al

. 1997, Hobson 1999,Webster

et al

. 2002). Feather proteins, formedduring moult, assimilate an isotopic signature thatis determined by the bird’s diet and consequentlyreflects that of the environment. Once grown, thefeather is metabolically inert and forms an isotopicrecord that reflects that of the environment in whichit was grown, until the next moult. If birds moult indifferent areas that span a stable isotope gradientor have different isotopic signatures, it is possibleto infer where the feather was moulted (Hobson1999). For example, in terrestrial systems hydrogenisotope ratios (

δ

D) in feathers correlate with the

δ

Dof local precipitation patterns where the feather wasmoulted. This ratio varies in a gradient across conti-nental land masses and is useful for inferring geo-graphical origin in terrestrial passerines (Hobson &Wassenaar 1997, Rubenstein

et al

. 2002). Stableisotopes have rarely been used to determine loca-tions where shorebird populations have undergonemoult after arriving at their wintering areas (Farmer

et al

. 2003), and have not been used to distinguishbetween feathers moulted in different intertidal areas(Farmer

et al

. in press).In this study we attempt to use stable isotopes to

identify wintering areas of the populations of RedKnot that pass through Delaware Bay in spring, a majorspring staging area critical for at least six waderspecies (Fig. 1; Myers

et al

. 1987, Clark

et al

. 1993).Two races are thought to breed and winter in Northand South America,

C. c. rufa

(population decliningand now near 30 000, Wetlands International 2002,Morrison

et al

. 2004), which breeds in eastern arcticCanada, and

C. c. roselaari

(population 20 000,Wetlands International 2002), which is thought tobreed in Alaska and Wrangel Island (Tomkovich1992). In winter, a group of birds spend the winterin the southeastern USA states of South Carolina,Georgia and Florida and small numbers are foundaround the Gulf of Mexico to northern Mexico(Morrison & Harrington 1992). Another group win-ters in northern Brazil and Venezuela, but the largestgroup winters further south along the shores ofChile and Argentina. Small numbers spend the borealwinter at other sites along the South American coastand these probably refer to young birds. There issome debate as to the breeding origins of the differ-ent wintering populations. The wintering areas of

© 2005 British Ornithologists’ Union,

Ibis

, doi: 10.1111/j.1474-919x.2005.00455.x

Migration and moult strategies of the Red Knot

3

C. c. roselaari

are not definitively known but arethought to be either on the Pacific coasts of Califor-nia and Mexico or in areas around the Gulf of Mex-ico, particularly in the US states of Florida and Texas,and in Mexico (Morrison & Harrington 1992, Patten

et al

. 2003).

C. c. rufa

are known to winter in Tierradel Fuego and Patagonia and may also winter in thesoutheastern United States. A third race,

C. c. islandica

,breeds in Greenland and northwest Canada andwinters in northwestern Europe, and does not passthrough Delaware Bay.

In the boreal spring,

rufa

populations migratethrough a small number of staging sites and finallystop for 2–3 weeks in Delaware Bay, in the north-eastern United States (Harrington 1996). Here theyfeed almost exclusively on the eggs of spawningHorseshoe Crabs

Limulus polyphemus

before depart-ing to breeding areas in the Arctic (Castro & Myers1993, Tsipoura & Burger 1999, Harrington 2001).Red Knots approximately double their body massfrom 90–120 g to 180–240 g, at an average rate of4.7 g per day, making this the highest recorded rateof weight gain for Red Knot across all its subspeciesand other staging sites (Piersma

et al

. 2005). Duringthe past 10 years, there has been a decline in thenumber of crab eggs, associated with an increasedcommercial harvest of adult crabs for the bait indus-try (Walls

et al

. 2002), and an associated decline inRed Knots wintering in Tierra del Fuego. Numbers

declined from approximately 67 000 birds in 1982–85 to 30 000 in 2003 (Fig. 1; Morrison

et al

. 2004)and there is an urgent need to understand the mech-anisms behind this decline. Since 1995, Red Knotshave been banded as part of an international effort inthe Canadian breeding areas, South American stop-over and wintering areas and the final stopover sitein Delaware Bay. This provides an opportunity toestimate survival and recruitment rates but it has notbeen possible to separate out demographic parametersfor the different populations passing through the bay.In this study we investigate whether stable isotopesmay be used to determine an individual’s winteringarea and hence allow demographic rates to be cal-culated for specific parts of the Red Knot population.

MATERIALS AND METHODS

Choice of isotopes

Many studies have looked at movements of birdsbetween sites in terrestrial and marine environments,as there are predictable changes in

δ

13

C and

δ

15

N, butfew, if any, have attempted to distinguish movementsbetween sites located in intertidal areas (Hobson 1999).Marine environments are enriched in both C and Nand so

δ

13

C and

δ

15

N values increase with salinity.Within estuarine and marine environments bothisotopes are enriched the higher up the trophic levelthe organism feeds. In areas where they moult flightfeathers, Red Knots feed exclusively on benthicinvertebrates, predominantly bivalves (Baker 1996,Gonzalez

et al

. 1996, Rodrigues & Lopes 2000).Bivalves feed on benthic and suspended particulateorganic matter (POM), which are at the base of foodchains. Therefore, any differences found in these val-ues between sites is unlikely to be due to shifts in abird’s diet to higher trophic levels, instead probablybeing indicative of the underlying source of the Cand N in suspended and benthic organic matter.

Less well understood is how isotope ratios in inter-tidal benthic invertebrates vary within and betweensites, and whether any predictable relationships occur,i.e. can we predict where shorebirds that feed inintertidal areas moult? In the terrestrial environ-ment, carbon isotope values vary with the photosyn-thetic mechanism. C4 and CAM plants grow in aridenvironments and have higher

δ

13

C values than C3plants, and thus

δ

13

C values in arid areas tend to behigher than those from areas with higher rainfall. Ata larger scale, we hypothesized that

δ

13

C and

δ

15

Nwould be useful to differentiate between moulting

Figure 1. Numbers of wintering Red Knot on the easternseaboard of North and South America. Figures for Argentinaand Chile are taken from 2003 aerial surveys (Morrison et al.2004); data for Maranhão, Brazil, are from a 2003 field visit(A.J.B., P.M.G. unpubl. data); the most recent data for Florida arefrom Harrington (1996) and may be an overestimate of currentnumbers.

4

P. W. Atkinson

et al.

© 2005 British Ornithologists’ Union,

Ibis

, doi: 10.1111/j.1474-919x.2005.00455.x

areas in the southern United States and Tierra delFuego, as there is an abrupt increase in

δ

13

C and

δ

15

Nvalues in POM just north of 40

°

S in ocean water(Rau

et al

. 1982, Wada

et al

. 1987). Furthemore, inmarine systems, changes in

δ

15

N in are related to thesource, rate of input and degree of uptake of nitrate byalgae, and whether significant amounts of nitrogenfixation or denitrification occur (Altabet 1996).

On the eastern seaboard of the Americas, RedKnot wintering sites span an area between approxi-mately 20

°

N and 53

°

S (Piersma & Davidson 1992),and so there is good justification for thinking that stableisotopes could be used successfully to distinguishthe moulting areas of this long-distance migrant.However, the gradients outlined above may be masked,as isotope ratios in the birds’ food will reflect a balancebetween the relative influence of terrestrial and marineinputs.

As there was uncertainty as to which elements woulddistinguish wintering sites, we ran a pilot study andanalysed 20 feathers for

δ

13

C,

δ

15

N,

δ

D and

δ

18

O.All these elements have been useful in determiningthe degree of terrestrial or freshwater influence. Thisanalysis indicated that

δ

13

C and

δ

15

N gave the bestseparation between reference sites and so the remainingfeathers were analysed for these two elements.

Site selection and feather collection

Feathers were obtained from three reference areaswhich were known wintering sites of Red Knot,two of which, Bahia Lomas and Rio Grande, were inTierra del Fuego and one in Florida, in the south-eastern USA (Table 1, Fig. 1). The choice of feathersto be sampled is an important one. Some Red Knotmay moult 1–3 of the smaller inner primaries, eitheron the breeding areas or at an intermediate locationen route to the main moulting area (Ginn & Melville1983), although this has been contested by Morrisonand Harrington (1992). They then suspend moult

and resume on arrival at their wintering areas. Toobtain an isotopic signature that was representativeof the main moulting area, the sixth primary or pri-mary covert measured descendantly, i.e. from theinside of the wing outwards, is an ideal feather. Thesefeathers were unlikely to experience suspendedmoult and were far enough away from the largeouter primaries that sampling should not adverselyaffect the flying ability of the bird. The sixth primarycovert was also easy to identify, being the outermostprimary covert to have a prominent white tip.

However at the time of the Rio Grande field visitin November 2002, the majority of birds caughtwere in active moult and in many cases the sixthprimary and its covert had yet to be moulted. Ratherthan have a mix of feathers from the current andprevious years, it was decided to sample the innerfirst primary. We were confident that these had beengrown during the current moult cycle as the featherswere new and not bleached or abraded, althoughsome of these feathers may have been moultedelsewhere. At Bahia Lomas, in February 2003, birdswere being sampled for other isotope studies, wherethe outermost secondary was being used, and sevenof these feathers were supplied to us.

A field visit to the southeastern US wintering areaswas not possible and therefore we used the sixthprimaries from 24 Red Knot specimens held at theRoyal Ontario Museum, Canada. These were collectedin Florida in January and February 1986 and 1987. InDelaware Bay, to minimize the impact on birds, sixthprimary coverts were clipped from a sample of 99birds caught in May 2003 on the coast in the Stateof Delaware. Birds selected for sampling were eitherchosen at random or because they were carryingbands that indicated they had been in South America.It was thought that primary coverts were moulted atthe same time as primaries but this may not be thecase. To test this, all the primaries and associatedprimary coverts were analysed from a wing of an

Table 1. Location, collection date and numbers of feathers obtained for isotopic analysis from (a) birds in known wintering areas and(b) birds of unknown origin collected on spring passage through Delaware Bay.

Country Location Collection dateNo. of

samples Feather collected

(a)USA Florida Jan./Feb. 1986 & 1987 24 Primary 6Chile Bahia Lomas Feb. 2003 7 Secondary 6Argentina Rio Grande Nov. 2002 28 Primary 1

(b)USA Delaware Bay May 2003 99 Primary covert 6

© 2005 British Ornithologists’ Union,

Ibis

, doi: 10.1111/j.1474-919x.2005.00455.x

Migration and moult strategies of the Red Knot

5

Argentinian-banded bird that had been found dead.There were no differences between the

δ

13

C and

δ

15

N signatures between the primaries and theirassociated primary coverts (paired two-sample

t

-test,

δ

13

C:

t

= 0.26,

P

= 0.8 ns;

δ

15

N:

t = 0.77, P = 0.46 ns,df = 9 in both cases). However there was a decreasein both δ13C and δ15N across the average values foreach primary and its associated covert. The range forthese was −11.6 to −8.8‰ for δ13C and 16.6–18.3‰for δ15N. These differences (2.8‰ for δ13C and1.8‰ for δ15N), although relatively small, will forma systematic bias if different feathers are collectedfrom different sites. It will therefore be important infuture studies either to collect the same feather orto correct for any bias. Without suspension, primarymoult in Red Knot in South Africa and the Wash,England takes approximately 95 days (Ginn & Melville1983, Underhill 2003) and in South Africa each pri-mary took between 13.7 and 23.5 days to grow. Mostof the difference in δ13C values was due to the ninthand tenth primaries, which are the last to grow, andalso take the longest. The difference between thefeathers used in this study, i.e. the first and sixth pri-mary and/or primary covert, was less than 1‰.

Feather isotope analyses

Each feather was taken from its sample bag, andsurface oils and waxes were removed by washingthe feather in 0.25 M sodium hydroxide solutionfollowed by two separate washes in pure water.The washed feathers were placed in clean, screw-topvials, labelled with sample number and reference.The clean feathers were then placed in a drying ovenat 50 °C overnight. After drying, the feathers wereclipped into fine sections (< 1 mm sections) in thesample vials using surgical scissors.

15N and 13C analysis

Approximately 1 mg of each feather was accuratelyweighed into tin capsules and loaded into an auto-matic sampler. Stable isotope measurements of 15Nand 13C were made using an EA-IRMS (ElementalAnalyser combustion continuous-flow Isotope RatioMass Spectrometry; 20/20 Mass Spectrometer PDZEuropa Ltd, Crewe, UK). All stable isotope valuesare reported in permil (‰) using the δ notation:

where δ isotope is the sample isotope ratio (13C or15N) relative to a standard, R is the ratio of heavy tolight isotopes (13C/12C or 15N/14N) in the sample orstandard. δ13C and δ15N are reported relative to theirinternational standards, namely Vienna Peedee Belemite(V-PDB) and atmospheric nitrogen, respectively.

The preweighed tin capsules containing the feathersamples were matched to %C and %N of the refer-ence. The reference material used in the analysis wasalbumin with a δ13C value of −18.4‰ vs. V-PDB anda δ15N value of 10.3‰ vs. air. This reference is trace-able to IAEA-NBS22 (oil) with an accepted δ13Cvalue of −29.7‰ vs. V-PDB, IAEA-N2 (ammoniumsulphate) with an accepted delta 15N value of 20.3‰vs. air and IAEA-NO-3 with an accepted value of+4.7‰. Routine measurements were precise towithin 0.3‰ for both δ13C and δ15N.

δδδδD analysis

We weighed 0.80 ± 0.02 mg of each feather into3.5 × 5-mm pressed silver capsules, which were thenpelletized and loaded into a Costech zero-blank electricautosampler. Samples were reduced over glassy carbonat > 1500 °C in a Costech ECS 4010 ElementalAnalyser fitted with a high-temperature inductionfurnace (model HTG-02) and passed via He into aPDZ Europa 20/20 Mass Spectrometer. Seven silvercapsules of isotopic standard mineral oil IA R002(Iso-Analytical, Cheshire, UK) of different masseswere used to calculate and correct for the productionof H3+ in the mass spectrometer source. As approxi-mately one-fifth (Wassenaar & Hobson 2000) of thetotal hydrogen content of feather keratin is readilyexchangeable with ambient water vapour, it is nec-essary to correct the data further so that only the δDof the non-exchangeable hydrogen is measured andtherefore any local, climate-related effects are negated.For this purpose we used the keratin isotope standardsCFS and BWB-II (Wassenaar & Hobson 2003) andan internal standard ISB. All δD results are reportedwith respect to standard mean ocean water.

Statistical analysis and interpretation

The aim of the analysis was to determine the simi-larity of the feather isotopic signatures of the birdspassing through Delaware Bay on spring migration tothe three sets of reference samples. A two-stage processwas used. First, δ13C and δ15N values of the samplesfrom the three reference sites were divided intothree groups using Ward’s method in the CLUSTER

δ isotopeRR

sample

standard=

−

×1 1000

6 P. W. Atkinson et al.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

and TREE procedures in SAS (SAS Institute 2001).As cluster analysis is sensitive to outliers, 5% (equivalentto three samples) of the points, identified by the pro-cedure, were removed. The proportion of samplesfrom each of the reference sites in each cluster, andthe error in sample classification, was calculated.

The reference sample data, classified into thethree clusters identified above, were entered, withthe birds of unknown origin from Delaware Bay, intoa canonical discriminant analysis using the DIS-CRIM procedure in SAS. This allocated each samplefrom Delaware Bay into one of the groups, based onthe highest probability of group membership andcalculated canonical coefficients for each sample.Although the analysis allocated samples to groups, itdid not take into account the likelihood that individ-ual samples belonged to each group (i.e. the distanceeach sample was from the group mean). The canon-ical variables were therefore used to identify samplesthat fell well outside the group means. Canonicalvariables have a mean of zero and variance of one. Toremove samples that had less than 1% chance ofbelonging to a particular group, any sample that hada value of either canonical variable of greater than2.58 from the group mean was deemed unclassified.This critical value, based on the t distribution, is thenumber of standard deviations away from the groupmean over which group membership is predicted tobe < 1% (Zar 1998).

RESULTS

Isotopic signature from reference sites

The three points dropped from the initial clusteringincluded two outlying samples from Rio Grande andone from Florida (Fig. 2a). There was no overlapbetween Florida (all in Group A) and Bahia Lomas(all in Group B) samples, but samples from Rio Grandewere found in all three clusters (Tables 2a, 3).Group C contained only samples from Rio Grande.

Rio Grande is the most southerly wintering site.These results provide support that some of the birdscaught at this site (nine out of 28) had either moultedin, or started to moult in, other more northerly areas,and had either flown south in active moult, or hadsuspended and resumed once in Rio Grande. One ofthese birds was placed in Group A, six were placedin Group B and two were removed as outliers.Where moult information was available, four (out offour) of the birds placed in Group B, the one GroupA bird and one of the two Rio Grande outliers were

in active moult, indicating that these birds hadmoulted their first primary away from Rio Grande.

Signatures from Delaware Bay

The samples taken from migrating birds in DelawareBay corresponded well to the three clusters, withonly six samples lying outside the 1% confidencelimit of group membership (Fig. 2b). Four of thesesamples lay well outside the range and represented adistinct group of samples that had high δ13C and lowδ15N values.

Samples taken in the Bay were made up of birdsthat had been banded previously in wintering areas

Figure 2. (a) Plot of δ13C vs. δ15N for reference feather samples.Large ellipses indicate the groups in which each sample wasplaced using Ward’s cluster analysis (A, B or C). R = Rio Grande,B = Bahia Lomas, F = Florida. Samples surrounded by smallcircles identify those that were excluded as outliers by the clusteranalysis. (b) Plot of δ13C vs. δ15N for feather samples taken fromstaging birds of unknown origin in Delaware Bay in spring 2003.Each individual is indicated by a letter according to the group inwhich it was placed by the discriminant analysis. Points that felloutside the 1% criterion of group membership were classified asU. Large dots and error bars represent the means and 2standard deviations of groups A, B and C.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

Migration and moult strategies of the Red Knot 7

and also some unbanded birds, effectively chosen atrandom (Table 2b). Of the random sample, 27 wereplaced in Group A (Florida type), 52 in Group B(Bahia Lomas type), five in Group C (Rio Grandetype) and six were unclassified. Of the nine birdsthat had been banded originally in Tierra del Fuego,four were placed in Group B and five in Group C. Allthe Group C birds were banded in Rio Grandebetween 2000 and 2002. Two of the four Group Bbirds had been banded originally during the australsummer at Rio Grande (November 2001) and BahiaLomas (February 2002). The remaining two wereoriginally caught on northwards passage in theaustral autumn, one at San Antonio Oeste (March2003, 300 km north of Peninsula Valdes – Fig. 1) andthe other in southern Brazil (exact date unknown).

DISCUSSION

Inferring the moulting areas of Red Knot using stable isotopes

Our analysis has shown that stable isotopes can beused successfully in intertidal areas to distinguish

between areas in which Red Knot underwent wingmoult, even for two sites that were relatively closeto one another. Our initial prediction that δ13C andδ15N values would be higher in sites at latitudeshigher than 40°S proved wrong, although thesetwo elements did provide a suitably robust way ofseparating out the northern and southern moultingsites. The mechanisms determining δ15N in intertidalareas over large spatial scales are poorly understood,and will depend in some part on large-scale environ-mental gradients operating in the marine environ-ment, the degree of influence that terrestrial andfreshwater habitats have in the intertidal area, andthe amount of anthropogenic inputs. Given the verydifferent δ15N values, it seems as though the sourceof N in Florida and southern Tierra del Fuego sites isvery different.

Red Knot are dependent on benthic invertebrates atboth Bahia Lomas and Rio Grande and yet the isotopicsignatures were very different. The two groups wereseparated based on δ13C values, but δ15N overlapped.In marine systems, δ13C and δ15N values tend to behigher than in terrestrial habitats and are also enrichedat higher trophic levels. Red Knot at Bahia Lomasmainly eat the bivalve Darina, whereas at Rio Grande,since 2001, Darina has declined in the sandflats(P.M.G. unpubl. data) and birds have been eatingsnails as well. Marine snails also tend to eat organicmatter and, as Red Knots are feeding on similar foodat the two sites, any difference in trophic level isunlikely. The habitats at Rio Grande and Bahia Lomasare very different and it is probably more useful tothink of the different δ15N and δ13C values as broadhabitat signatures. Bahia Lomas is a large, open,sandy bay, whereas the habitat at Rio Grande is pri-marily ‘restinga’, a hard tabular platform formed byred dust being blown off the land by strong offshorewinds. In this habitat, numerous pools are exposed atlow tide and birds feed on mussels clinging to therough restinga surface. With sites close togetherand experiencing broadly similar climatic conditions,the δ13C values are more likely to be affected by theamount of terrestrial, and particularly freshwater, input.The lower values at Bahia Lomas would perhapsindicate a higher degree of freshwater influence.This hypothesis was supported by δD measure-ments obtained during the pilot analyses (mean δDfor samples from Group C (Rio Grande type):−60.44‰ ± 1.28 sd, n = 6; Group B (Bahia Lomastype): −84.78‰ ± 5.04 sd, n = 4), the more negativevalues indicating freshwater, and from δ15N whichalso was slightly higher in Rio Grande (Table 3).

Table 2. (a) Group membership for each of the referencesamples as allocated by Ward’s cluster analysis; (b) allocation offeathers collected from birds caught on spring passage inDelaware Bay to the three different groups, divided into a newlycaught random sample and those which had been previouslybanded in South America.

Table 3. Mean δ13C and δ15N (‰, ± se) of the three groupsidentified by Ward’s cluster analysis. A = Rio Grande, B = BahiaLomas, C = Florida.

Group A B C Unclassified

(a)Florida, USA 23 1Bahia Lomas, Chile 7Rio Grande, Argentina 3 4 14 2

(b)Random sample 27 52 5 6Previously banded birdsRio Grande 1 5Bahia Lomas 1Brazil (spring) 1San Antonio Oeste (spring) 1

Group A B C

δ13C −9.9 ± 0.30 −14.1 ± 0.13 −15.2 ± 0.29δ15N 18.1 ± 0.13 16.4 ± 0.22 11.2 ± 0.23

8 P. W. Atkinson et al.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

In the pilot sample δD values for Florida birds, at−83.67‰ ± 3.78 sd, n = 5, were similar to those forGroup B birds, despite very different δD in rainfallfrom the two areas (IAEA 2001). Although we donot know the exact mechanisms involved, it is pos-sible that a large flat bay such as Bahia Lomas wouldhave a salinity gradient.

Anthropogenic inputs may also be important. RíoGrande is a city of about 50 000 people on the shoreof the Río Grande River. Red Knots feed in areasclose to and in the river mouth. In the past, organicwaste from the city, including that from a chickenfarm, has been released at high tide over the flatswhere the Red Knots feed. Bahia Lomas is in aremote area and organic waste from human settle-ments is likely to be minimal.

Despite applying caution to assigning birds to spe-cific sites, based on isotopic signatures, aerial surveysof Patagonia and Tierra del Fuego during February2003 indicated that 95% of all birds counted wererecorded on these two sites (Morrison et al. 2004).We can therefore assign birds to these two areas withsome degree of confidence. Aerial counts put theratio of Rio Grande to Bahia Lomas birds at 1 : 7.3(3500 : 25 500, Morrison et al. 2004), and in Dela-ware Bay the cluster analyses indicated that in therandom samples the ratio was slightly higher at1 : 10.4 (5 : 52). We can, however, be less sure of theorigins of the four birds plotted in Figure 1(b),which have very different signatures. They may befrom northern Brazil, where approximately 8000birds are known to have wintered in the past (Mor-rison & Ross 2004).

We can use the ratios of northern and southernbirds in Delaware Bay to estimate the size of thepopulation wintering in northern areas. The totalcounted in Argentina and Chile in 2003 was 30 475birds (Morrison et al. 2004), which included un-known proportions of adults, second-year birdsand juveniles, and these make up Groups B and C.Between March and May, first-year birds (birds bornthe previous boreal summer) and second-year birds(birds born two boreal summers ago) migrate northin the boreal spring. Using the random samples fromDelaware, we estimate from the ratios recorded there,assuming a similar age-structure to Tierra del Fuegoand that similar proportions pass through DelawareBay in spring, that Group A (Florida-type birds)represents approximately 14 500 birds and thatthe unclassified group represents 3000–4000 birds.This also assumes that birds from different winter-ing areas distribute themselves randomly across

the Bay. Further work is needed to test theseassumptions.

Breeding areas of the northern wintering populations of Red Knot

These results open up some interesting questionsabout the breeding origins of the birds wintering inthe Gulf of Mexico. It is very likely that the GroupA birds found on spring passage in Delaware corre-spond to the northern hemisphere wintering popu-lation of rufa referred to by Morrison and Harrington(1992). Although up to 10 000 birds have beenknown to winter in the Gulf of Mexico and theAtlantic coasts of the southeastern USA, the breed-ing, and racial, origins are debatable. Tomkovich(1992) thought that these might be C. c. roselaari,which breed in Alaska and Wrangel Island (Fig. 1)and are thought to migrate along the Pacific coast ofNorth America. However, using plumage and men-sural characteristics of specimens collected at theSalton Sea, Patten et al. (2003) concluded thatPacific coast Red Knots are C. c. roselaari, whichappear to winter on the Pacific coast of North Amer-ica and western Mexico, and that Gulf of Mexicobirds are rufa. Morrison and Harrington (1992)acknowledge that the breeding areas of the Florida/Gulf of Mexico/Caribbean wintering group are notprecisely known. Birds banded in this wintering pop-ulation have been seen on spring passage in both theinterior (Manitoba/Saskatchewan) and east Atlantic(Delaware Bay) flyways. They suggest that birdstaking the interior route occupy the western end ofthe current known range of rufa, but also accept thata proportion may also breed in Alaska.

The results from this study indicate that a sub-stantial proportion of the birds passing throughDelaware in spring are from these northerly winteringareas. If they were roselaari, and therefore breedingin Alaska and Wrangel Island, a migration route thattakes them through Delaware Bay in spring seemsunlikely although not impossible. Shorter alterna-tives would be either to take a northwesterly routethrough the centre of the United States, stopping offat lakes in Saskatchewan along with many othershorebird species (Skagen et al. 1999) or to fly upthe west coast of America (as suggested by Pattenet al. 2003). Interestingly, genetic analyses indicatedthat Florida wintering birds were more similarto C. c. rogersi that breed in eastern Russia (Baker1992) than to rufa. The present study does notanswer this conundrum and further analyses of

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

Migration and moult strategies of the Red Knot 9

samples taken from the roselaari breeding areas,Pacific coast wintering areas, Saskatchewan stagingareas and western edge of the rufa breeding rangewould be needed to do so.

Feather selection and moult strategy

As some individuals moult and winter in differentsites, the choice of feather is crucial. In Rio Grande,birds had just commenced moult during thesampling period and, to ensure we obtained a signa-ture from the current year, the inner primary wassampled because it is small and would be replacedquickly. If it was in active moult it provided a signa-ture from that site, as turnover time in blood is amatter of days (Hobson & Clark 1992). Birdscommence primary moult at different times and wedecided to take the innermost primary, rather thantaking whichever feather was in active moult, toobtain the range of variation in delta values for thatfeather in that site. As values of isotopes are likely tochange over time, taking whichever feather was inactive moult would only have given a signature thatwas representative of that site on the days the featherhad been grown.

Despite remarks to the contrary by Morrison andHarrington (1992), some individuals, in commonwith other waders, appeared to have started andfinished primary moult in different areas. In the RioGrande sample, one of the birds fell into Group B,corresponding to birds collected in Florida, and twowere outliers. In fact, all three of these were at theextreme range of δ13C values for the Florida refer-ence samples and were closer to the cluster of‘unknown’ samples from Delaware Bay (Fig. 2a,2b).From the known distribution of birds, this cluster islikely to refer to areas in northern Brazil (Morrison& Ross 1989). Two of these three birds were in activemoult when caught in Rio Grande. Six other activelymoulting birds, caught at Rio Grande, had a firstprimary isotope signature similar to that of BahiaLomas, a site some 200 km away. Overall, in thesample of 27 actively moulting birds from RioGrande where moult was recorded, the birds withthe seven highest moult scores (Ginn & Melville1983) had isotope signatures that indicated that thefirst primary had been grown away from Rio Grande.

In this case, it may be expected that some birdsshow signatures that are mixed Bahia Lomas and RioGrande as the two areas are relatively close, butthese results indicate that a significant proportion ofthe southern-wintering birds had started to moult in

areas far from the main wintering area. We cannot beentirely sure that the birds that had started moultfurther north were in fact adults. Few first-year birdsreach South America in their first year of life, as theproportion of juveniles in catches there is low(P.M.G. and A.J.B. unpubl. data), and they are likelyto spend their first boreal winter and following sum-mer in areas to the north of Patagonia and Tierra delFuego, before migrating south for their second borealwinter. Birds that had grown their first primary inareas other than Rio Grande were more advanced inmoult than the other birds, suggesting they wereimmature birds (i.e. 16–18-month-old birds born inthe northern summer of the previous year) that hadstarted moult earlier in areas to the north.

Stable isotopes can therefore provide a way ofquantifying the proportion of birds undergoingsuspended moult, which is otherwise difficult to estim-ate. In northern hemisphere shorebird populations,the phenomenon of moulting a few inner primaries onthe breeding or staging areas is well known and hasbeen observed in many species (Ginn & Melville 1983).It is most extreme in the Grey Plover Pluvialis squa-tarola, in which up to 30% of birds may be found tobe in suspended moult on the Wash, England, in mid-winter (Underhill 2003). Starting moulting either onthe breeding grounds or at an intermediate site isseen as a way of being able to complete the energet-ically costly task of moult before the onset of harshwinter conditions, and Grey Plovers may be poorlyadapted to wintering conditions in western Europe(Underhill 2003).

Application of stable isotopes in future shorebird studies

Using different tissues, a bird in the hand providesthe opportunity to infer information on the diet overperiods ranging from days to perhaps several years(Hobson & Clark 1992). In studies such as this,which aim to elucidate different migration strategiesof long-distance migrants, it is desirable to gatherinformation about several important locations,which include breeding, wintering, spring andautumn staging areas. Ideally this should be in asnon-invasive a way as possible. Moult of primaryfeathers is understood for many northern-breedingspecies that winter in coastal areas, and can provideinformation about diet over several months, but isgenerally less well understood in waders that areresident in the tropics. Secondary, median and bodymoult is less well studied in shorebirds but, at least

10 P. W. Atkinson et al.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

in plovers, frequently shows complex patterns, withseveral different generations of feathers to be foundin the wing (Clark & Clark 1986, Jukema et al.2001). Birds begin to grow summer plumage in thewintering and spring staging sites but body feathersmay be renewed throughout the year. Blood, parti-cularly plasma, shows a high turnover and thus pro-vides information on diet during the previous fewdays. Toenails provide a continuous record of dietover the past few months. For a bird caught on springmigration in Delaware we could obtain informationon where the bird was in autumn, winter and spring.If some tracts, e.g. secondaries, contain different gen-erations of feathers, information can be gleaned evenfurther back, although this is rare in Calidris waders(N.A.C. unpubl. data).

Stable isotopes give us the opportunity to investi-gate the diet of shorebirds at different times of year,infer geographical origin of wintering populations,calculate population-specific demographic para-meters and, in multispecies comparisons, determinewhere individual shorebird species and their preyoccur in food webs. However, much of this dependson birds growing feathers or other tissues in areaswith predictably different isotope ratios and insimilar habitats, and this may not be the case in otherstudy systems. For example, in a study of AmericanPluvialis dominica and Pacific Golden Plovers P.fulva, feathers grown in the wintering groundsshowed no differences in δD, δ13C and δ15N valuesbetween species, although they wintered on differ-ent continents (Rocque 2003).

In this study we were fortunate that differences inisotope ratios occurred between the small numberof Red Knot wintering areas, but recognize that insome cases differences may be small. Isotopes maynot allow us to pinpoint a bird to an exact locationbut they can provide information on the strategiesadopted by individuals in the population of this,and potentially other, long-distance migrant shore-birds. Coupled with banding data, stable isotopes infeathers can provide a powerful tool to understandthe ecological and demographic consequences of anindividual’s life-history decisions.

We would like to thank all the volunteers who helped outwith the shorebird banding programmes in Argentina,Chile and Delaware Bay. We are grateful to the US Fishand Wildlife Service for funding this study and would liketo acknowledge Greg Breese for his help and support ofthe project. In particular we would like to thank MarkPeck and Monica Abril for feather collection in BahiaLomas and Rio Grande, and Adrian Farmer for advice.

Armada Argentina and Rubén Pissaco recovered the deadRed Knot at San Antonio Oeste. We also thank PabloHavelka and Juan Cejas from Sub Secretaría de Rec. Nat-urales de Tierra del Fuego, Miguel Isla from Dirección deCiencia y Técnica de Tierra del Fuego, Nora Loekemeyer,Prefectura Naval Argentina, Museo de la Ciudad, ConcejoDeliberante de Río Grande and Jorge Amena for theirassistance in our work in Argentina. Stable isotope ana-lyses were carried out at the NERC Life Sciences MassSpectrometry Facility, East Kilbride (δD) and by GillianTaylor at the University of Newcastle (δ13C and δ15N).

REFERENCES

Alerstam, T. & Hedenström, A. 1998. The development of birdmigration theory. J. Avian Biol. 29: 343–369.

Altabet, M.A. 1996. Nitrogen and carbon isotopic tracers of thesource and transformation of particles in the deep sea. InIttekkott, V., Schafer, P., Honjo, S. & Depetris, P.J. (eds)Particle flux in the Ocean: 155–184. Scientific Committee OnProblems of the Environment (SCOPE) Series no. 57. UK:John Wiley & Sons Ltd.

Atkinson, P.W. 1996. The origins, moult, movements andchanges in numbers of Bar-tailed Godwits Limosa lapponicaon the Wash, England. Bird Study 43: 60–72.

Atkinson, P.W., Appleton, G.F., Clark, J.A., Clark, N.A.,Gillings, S., Henderson, I.G., Robinson, R.A. & Stillman, R.A.2003. Red Knots Calidris canutus in Delaware Bay 2002.Survival, Foraging and Marking Strategy. BTO ResearchReport no. 308. Thetford: British Trust for Ornithology.

Baker, A.J. 1992. Molecular genetics of the genus Calidris, withspecial reference to Knots. Wader Study Group Bull. 64 (Suppl.):29–35.

Baker, A.J. 2002. The deep roots of bird migration: inferencesfrom the historical record preserved in DNA. Ardea 90: 503–513.

Baker, A.J., González, P.M., Piersma, T., Niles, L.J., Inês deLima Serrano do Nascimento, Atkinson, P.W., Clark, N.A.,Minton, C.D.T., Peck, M. & Aarts, G. 2004. Rapid populationdecline in Red Knots: fitness consequences of decreasedrefuelling rates and late arrival in Delaware Bay. Proceedingsof the Royal Society B 271: 875–882.

Baker, A.J., Manriquez, R.E., Benegas, L.G., Blanco, D.,Borowik, O., Ferrando, E., de Goeij, P., González, P.M.,González, J., Minton, C.D.T., Peck, M., Piersma, T. &Ramírez, M.S. 1996. Red Knots at their farthest south: aninternational expedition to Tierra del Fuego, Argentina, inFebruary 1995. Wader Study Group Bull. 79: 103–108.

Battley, P.F., Piersma, T., Rogers, D.I., Dekinga, A., Spaans, B.& Van Gils, J.A. 2004. Do body condition and plumageduring fuelling predict northwards departure dates of GreatKnots Calidris tenuirostris from northwest Australia? Ibis 146:46–60.

Bêty, J., Gauthier, G. & Giroux, J.-F. 2003. Body condition,migration, and timing of reproduction in Snow Geese: a testof the condition-dependent model of optimal clutch size.Am. Nat. 162: 110–121.

Boyd, H. 1992. Arctic summer conditions and British Knotnumbers: an exploratory analysis. Wader Study Group Bull.64 (Suppl.): 144–152.

Butler, R.W., Davidson, N.C. & Morrison, R.I.G. 2001.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

Migration and moult strategies of the Red Knot 11

Global-scale shorebird distribution in relation to productivityof near-shore ocean waters. Waterbirds 24: 224–232.

Castro, G. & Myers, J.P. 1993. Shorebird predation on eggs ofHorseshoe Crabs during spring stopover on Delaware Bay.Auk 110: 927–930.

Chamberlain, C.P., Blum, J.D., Holmes, R.T., Feng, X.,Sherry, T.W. & Graves, G.R. 1997. The use of isotopetracers for identifying populations of migratory birds. Oecologia109: 132–141.

Clark, N.A. & Clark, J.A. 1986. Secondary moult of GoldenPlover populations wintering on the Firth of Forth, Scotland.Wader Study Group Bull. 48: 6.

Clark, K.E., Niles, L.J. & Burger, J. 1993. Abundance anddistribution of migrant shorebirds in Delaware Bay. Condor95: 694–705.

Drent, R., Both, C., Green, M., Madsen, J. & Piersma, T. 2003.Payoffs and penalties of competing migratory schedules.Oikos 103: 274–292.

Engelmoer, M. & Roselaar, C. 1998. Geographical Variation inWaders. Dordrecht: Kluwer Academic Publishers.

Farmer, A., Abril, M., Fernández, M., Torres, J., Kester, C. &Bern, C. in press. Using stable isotopes to associate migra-tory shorebirds with their wintering locations in Argentina.Ornitol. Neotrop. Suppl. in press.

Farmer, A., Rye, R., Landis, G., Bern, C., Kester, C. & Ridley, I.2003. Tracing the pathways of Neotropical migratory shore-birds using stable isotopes: a pilot study. Isotopes Environ.Health Studies 39: 169–177.

Gill, J.A., Norris, K., Potts, P.M., Gunnarsson, T.G., Atkinson,P.W. & Sutherland, W.J. 2001. The buffer effect and large-scale population regulation in migratory birds. Nature 412:436–438.

Ginn, H.B. & Melville, D.S. 1983. Moult in Birds. BTO Guide 19.Tring: British Trust for Ornithology.

Goede, A.A., Nieboer, E. & Zegers, P.M. 1990. Body massincrease, migration pattern and breeding grounds of Dunlins,Calidris a. alpina, staging in the Dutch Wadden Sea in spring.Ardea 78: 135–144.

Gonzalez, P.M., Piersma, T. & Verkuil, Y. 1996. Food, feedingand refuelling of Red Knots during northward migration atSan Antonio Oeste, Rio Negro, Argentina. J. Field Ornithol.67: 575–591.

Grimm, V. & Uchmanski, J. 2002. Individual variability andpopulation regulation: a model of the significance of within-generation density dependence. Oecologia 131: 196–202.

Harrington, B.A. 1996. The Flight of the Red Knot. New York:W.W. Norton Co.

Harrington, B.A. 2001. Red Knot (Calidris canutus). In Poole, A.& Gill, F. (eds) The Birds of North America, no. 563. Philadel-phia, PA: Academy of Natur Sciences; Washington, DC:American Ornithologists’ Union.

Hobson, K.A. 1999. Tracing origins and migration of wildlifeusing stable isotopes: a review. Oecologia 120: 314–326.

Hobson, K.A. & Clark, R.G. 1992. Assessing avian diets usingstable isotopes. I: Turnover of 13C in tissues. Condor 94: 181–188.

Hobson, K.A. & Wassenaar, L.I. 1997. Linking breeding andwintering grounds of Neotropical migrant songbirds usingstable hydrogen isotopic analysis of feathers. Oecologia 109:142–148.

IAEA. 2001. GNIP Maps and Animations, International AtomicEnergy Agency, Vienna. Accessible at http://isohis.iaea.org.

Jukema, J., Piersma, T., Hulscher, J.B., Bunskoeke, E.J.,Koolhaas, A. & Veenstra, A. 2001. Goudplevieren enwilsterflappers: eeuwenoude fascinatie voor trekvogels.Utrecht: Fryske Akademy.

Madsen, J. 2001. Spring migration strategies in Pink-footedGoose Anser brachyrhynchus and consequences for springfattening and fecundity. Ardea 89: 43–55.

Marra, P.P., Hobson, K.A. & Holmes, R.T. 1998. Linking winterand summer events in a migratory bird by using stable-carbon isotopes. Science 282: 1884–1886.

Morrison, R.I.G. & Harrington, B.A. 1992. The migration sys-tem of the Red Knot Calidris canutus rufa in the New World.Wader Study Group Bull. 64 (Suppl.): 71–84.

Morrison, R.I.G. & Ross, R.K. 1989. Atlas of Nearctic Shore-birds on the coast of South America. 2 volumes. CanadianWildlife Service, Ottawa, Canada.

Morrison, R.I.G., Ross, R.K. & Niles, L.J. 2004. Declines inwintering populations of Red Knots in southern South Amer-ica. Condor 106: 60–70.

Myers, J.P., Morrison, R.I.G., Antas, P.Z., Harrington, B.A.,Lovejoy, T.E., Sallaberry, M., Senner, S.E. & Tarak, A.1987. Conservation strategy for migratory species. Am. Sci.75: 19–26.

Patten, P.A., McCaskie, G. & Unitt, P. 2003. Birds of the SaltonSea. Status, Biogeography, and Ecology. Berkeley, CA:University of California Press.

Piersma, T. & Baker, A.J. 2000. Life history characteristics andthe conservation of migratory shorebirds. In Gosling, L.M. &Sutherland, W.J. (eds) Behaviour and Conservation: 105–124. Cambridge, UK: Cambridge University Press.

Piersma, T. & Davidson, N. 1992. The migrations and annualcycles of five subspecies of Knots in perspective. WaderStudy Group Bull. 64 (Suppl.): 187–197.

Piersma, T., Rogers, D.I., González, P.M., Zwarts, L., Niles, L.J.,de Lima do Nascimento, I., Minton, C.D.T. & Baker, A.J.2005. Fuel storage rates before northward flights in red knotsworldwide: facing the severest constraint in tropical intertidalenvironments? In Greenberg, R. & Marra, P.P. (eds) Birds ofTwo Worlds: 262–273. Washington DC: Smithsonian Institu-tion Press.

Prokosch, P. 1988. Das Schleswig-Holsteinische Watten-meer als Frühjahrs-Aufenthaltsgebeit arktische Watvogel-Populationen am Bespiel von Kiebitzregenpfeifer (Pluvialissquatarola, L 1758), Knutt (Calidris canutus, L 1758) undPfuhlschnepfe (Limosa lapponica, L 1758). Corax 12: 274–442.

Rau, G.H., Sweeney, R.E. & Kaplan, I.R. 1982. Plankton 13C/12C ratio changes with latitude: differences between northernand southern oceans. Deep-Sea Res. 29: 1035–1039.

Rocque, D.A. 2003. Intrinsic markers in avian populations:explorations in stable isotopes, contaminants and genetics.PhD thesis, University of Alaska, Fairbanks.

Rodrigues, A.A.F. & Lopes, A.T.L. 2000. The occurrence ofRed Knots Calidris canutus on the north-central coast ofBrazil. Bull. Br. Ornithol. Club 120: 251–259.

Rubenstein, D.R., Chamberlain, C.P., Holmes, R.T., Ayers, M.P.,Waldbauer, J.R., Graves, G.R. & Tuross, N.C. 2002. Link-ing breeding and wintering ranges of a migratory songbirdusing stable isotopes. Science 295: 1062–1065.

Sandberg, R. & Moore, F.R. 1996. Fat stores and arrival on thebreeding grounds: reproductive consequences for passerinemigrants. Oikos 77: 577–581.

12 P. W. Atkinson et al.

© 2005 British Ornithologists’ Union, Ibis, doi: 10.1111/j.1474-919x.2005.00455.x

SAS Institute. 2001. The SAS System for Windows V. 8.02.Cary, NC: SAS Institute Inc.

Skagen, S.K., Sharpe, P.B., Waltermire, R.G. & Dillon, M.B.1999. Biogeographical Profiles of Shorebird Migration inMidcontinental North America. Biological Science ReportUSGS/BRD/BSR – 2000–0003.

Stillman, R.A., West, A.D., Goss-Custard, J.D., Caldow, R.W.G.,McGrorty, S., Durell, S.E.A., Le V., dit, Yates, M.G.,Atkinson, P.W., Clark, N.A., Bell, M.C., Dare, P.J. &Mander, M. 2003. An individual behaviour-based model canpredict shorebird mortality using routinely collected shellfisherydata. J. Appl. Ecol. 40: 1090–1101.

Sutherland, W.J. 1996. From Individual Behaviour to PopulationEcology. Oxford: Oxford University Press.

Tomkovich, P.S. 1992. An analysis of geographical variability inKnot Calidris canutus based on museum skins. Wader StudyGroup Bull. 64 (Suppl.): 17–23.

Tsipoura, N. & Burger, J. 1999. Shorebird diet during springmigration stopover on Delaware Bay. Condor 101: 635–644.

Underhill, L. 2003. Within ten feathers: primary moult strategiesof migratory waders (Charadrii). In Berthold, P., Gwinner, E.& Sonnenschein, E. (eds) Avian Migration: 187–197. Berlin:Springer-Verlag.

Wada, E., Terazaki, M., Kabaya, Y. & Nemoto, T. 1987. 15N and13C abundances in the Antarctic Ocean with emphasis on thebiogeochemical structure of the food web. Deep-Sea Res.34: 829–841.

Walls, E.A., Berkson, J. & Smith, S.A. 2002. The horseshoe

crab, Limulus polyphemus: 200 million years of existence,100 years of study. Rev. Fish. Sci. 10: 39–73.

Wassenaar, L.I. & Hobson, K.A. 2000. Improved methodfor determining the stable-hydrogen isotope composition δDof complex organic materials of environmental interest.Environ. Sci. Technol. 34: 2345–2360.

Wassenaar, L.I. & Hobson, K.A. 2003. Virtual equilibrationand online technique for rapid determinations of non-exchangeable δD of feathers for avian migration studies.Environ. Sci. Technol. 39: 211–217.

Webster, M.S., Marra, P.P., Haig, S.M., Bensch, S. & Holmes, R.T.2002. Links between worlds: Unraveling migratory connectiv-ity. Trends Ecol. Evol. 17: 76–82.

Wenink, P.W., Baker, A.J. & Tilanus, M.G.J. 1994. Mitochon-drial control-region sequences in two shorebird species, theturnstone and the dunlin, and their utility in population geneticstudies. Mol. Biol. Evol. 11: 22–31.

Wetlands International. 2002. Waterbird Population Estimates,3rd edn. Wetlands International Global Series, no. 12. Wage-ningen: Wetlands International.

Ydenberg, R.C., Butler, R.W., Lank, D.B., Guglielmo, C.G.,Lemon, M. & Wolf, N. 2002. Trade-offs, condition depend-ence and stopover site selection by migrating sandpipers.J. Avian Biol. 33: 47–55.

Zar, J.H. 1998. Biostatistical Analysis, 3rd edn. Upper SaddleRiver, NJ: Prentice Hall.

Received 9 April 2004; revision accepted 21 April 2005.