Stable isotopes examined across a migratory dividein Scandinavian willow warblers (Phylloscopustrochilus trochilus and Phylloscopus trochilusacredula) re ect their African winter quartersC P Chamberlain1 S Bensch2 X Feng1 S ETHkesson2 and T Andersson2

1Department of Earth Sciences Dartmouth College Hanover NH 03755 USA2Department of Animal Ecology Lund University S-223 62 Lund Sweden

The C and N isotopes of feathers from two subspecies of willow warblers (Phylloscopus trochilus trochilus andPhylloscopus trochilus acredula) are isotopically distinct Our analysis of 138 adult males from 14 sites distributedacross Sweden shows that the mean 15N and 13C values of subspecies acredula (from latitudes above 638 N)were signicentcantly higher than the mean 15N and 13C values of subspecies trochilus (from latitudes below618 N) The analysed willow warbler feathers had been moulted in the winter quarters and the observedisotopic signatures should thus repoundect the isotopic pattern of food assimilated in Africa The isotopic dataobserved in Sweden match the cline in morphology both showing abrupt changes around 628 N This resultagrees with data from ringing recoveries indicating that the two subspecies occupy geographically andisotopically distinct wintering grounds in Africa Our isotopic data suggest that analysis of stable isotopes ofC and N is a promising method to track wintering quarters of European birds that migrate to Africa

Keywords stable isotopes nitrogen isotopes carbon isotopes migratory birds hybrid zonePhylloscopus trochilus

1 INTRODUCTION

Linking breeding and winter quarters of long-distancemigratory bird populations has been notoriously dicurrencultand relatively little is known about whether these popula-tions mix or remain separated on their winter groundsThe traditional method to identify migratory routes andwintering areas of migratory birds has until recentlyrelied exclusively upon analyses of recoveries of ringedbirds Despite a full century of ringing we only have afragmented picture of the winter ranges of most long-distance migratory bird species mainly because of thelow probability of obtaining intercontinental ringingrecoveries (one out of 103^105 for most small birds) Morerecent methods such as orientation experiments usingEmlen funnels (Emlen amp Emlen 1966) and satellitetracking (ARGOS 1994) will be of much help inrevealing migratory directions of local populations butthese techniques are time laborious and expensive whichseriously limits collection of data sets covering largerregions An alternative and novel approach is to usestable isotope compositions of animal tissues to trackmigratory patterns Such a method has recently beenapplied to neotropical migrant birds (Chamberlain et al1997 Hobson amp Wassenaar 1997 Marra et al 1998)monarch butterpoundies (Wassenaar amp Hobson 1998) Atlanticsalmon (Kennedy et al 1997 Harrington et al 1998) andAfrican elephants (Van der Merwe et al 1990 Vogel et al1990 Koch et al 1995)

In this study we apply this isotopic approach to twosubspecies of willow warblers (Phylloscopus trochilus trochilusand Phylloscopus trochilus acredula) that are known to havediiexclerent migratory routes and occupy geographically

distinct winter quarters A migratory divide is a regionin which two populations showing diiexclerent migratorydirection meet and ringing recoveries of willowwarblers in Europe have demonstrated a migratorydivide latitudinally in central Scandinavia (HedenstrIcircmamp Petterson 1984 1987) In autumn birds in southernScandinavia (trochilus) migrate south-west whereas birdsin northern Scandinavia (acredula) migrate south-eastboth heading towards Africa The available ringingrecoveries (centgure 1) suggest these birds occupy distinctareas of Africa in the winter Such diiexclerences in migra-tory direction and migratory distance most likely repoundectgenetic diiexclerences (Berthold 1996 Helbig 1996)Members of both subspecies start moulting all theirpoundight feathers on their African winter grounds inDecember or early January (Underhill et al 1992)Hence we test the hypothesis that the isotopic composi-tion of bird feathers collected in the breeding groundsshould show a change across the migratory dividebetween the two subspecies of willow warblers

Herein we show that the N and C isotopes of feathersfrom the two subspecies of willow warblers changeabruptly at the presumed location of the migratorydivide ie the location for the change in morphologyand coloration (Bensch et al 1999) This result mostlikely repoundects the fact that trochilus and acredula occupydiiexclerent areas in Africa when replacing their feathersThe results from this study provide supporting evidencefor the use of stable isotopes for tracking migratorypatterns of birds in general In addition the signicentcantlydiiexclerent N and C isotope composition recorded inwillow warbler feathers moulted in west and east Africaopens new avenues of research aimed at disentanglingmigratory routes and wintering areas of Palaearctic^African migratory birds

Proc R Soc Lond B (2000) 267 43^48 43 copy 2000 The Royal SocietyReceived 2 September 1999 Accepted 27 September 1999

Author for correspondence (cpchamberlaindartmouthedu)

2 METHODS

(a) The study speciesIn addition to exhibiting a migratory divide we found willow

warblers suitable to study for four additional reasons Firstwillow warblers moult all their poundight feathers twice annually (inwinter and summer) (Underhill et al 1992) which means thatfeathers grown on winter grounds can be collected at breedingsites and vice versa Because it has been shown that the stableisotopic composition of feathers repoundects the food eaten duringtheir growth (Mizutani et al 1990 1992 Hobson amp Clark 1992)it should be possible to use feathers collected on northernbreeding sites to determine from approximately where in Africathe birds originated Second willow warblers have a largewinter range in tropical Africa (Cramp 1992) hence increasingthe possibility of detecting geographically consistent isotopepatterns Third willow warblers are very common with acontinuous distribution in Sweden (Cramp 1992) so enoughsamples can easily be obtained from any specicentc localityFourth the two subspecies are very similar with extensivemorphological overlap and existing data suggest no diiexclerencesin feeding behaviour or prey selection (Cramp 1992) whichotherwise could have biased the isotopic signals That the twosubspecies are recently derived is supported by the observationthat they lack diiexclerentiation at neutral genetic markers (mito-chondrial and microsatellites DNA) (Bensch et al 1999)

(b) Analyses of ringing recoveriesWe used the ringing recoveries reported to the Swedish

Museum of Natural History between 1960 and 1998 To identify

whether the ringing recoveries in Africa were representingPh t trochilus or Ph t acredula we used the following criteriaThose birds ringed in Sweden south of 608 N either before10 May (before the centrst Ph t acredula migrates through southernSweden (HedenstrIcircm amp Pettersson 1984)) or during June andJuly were identicented as Ph t trochilus (n ˆ 5) Those birds ringedeither north of 638 N or after 1 September in southern Sweden(when the majority of Ph t trochilus already have departed fromSweden (HedenstrIcircm amp Pettersson 1984)) were identicented asPh t acredula (n ˆ12) The subspecies identity of eight birds wasambiguous and therefore excluded from analyses

(c) Data collection and analyses of stable isotopesIn May andJune of 1996 and 1997 we sampled feathers of male

willow warblers captured on their breeding territories in mist-netswith aid of song playback at 32 sites in Sweden between 558 N and678 N (Bensch et al 1999) Diiexclerent primary feathers from thesame bird can have diiexclerent isotopic values repoundecting change indiet during growth of individual primaries (Thompson amp Furness1995 Bearhop et al 1999)To avoid these possible complications allsamples in this study were from the innermost primary feather onthe left wing In addition not all willow warblers moult theirpoundight feathers during the winter (Ginn amp Mellville 1983) In theseanalyses we therefore only included feathers from birds in freshplumage (495 of examined birds) which apparently hadundertaken a complete moult prior to spring migration Weanalysed feathers from 138 males representing 14 sites distributedacross Sweden for 15N and 13C values (table 1) It is also impor-tant to note that the N and C isotope ratios of feather keratinremain centxed after growth (Mizutani et al 1990 1992 Hobson ampClark 1992) In addition we routinely analyse a feather standardin our laboratory and there has been no change in the N and Cisotopic values of the standard since we began this line of researchin 1994 so any processes such as abrasion and deteriorationdo notaiexclect the isotopic ratio of feathers Because the N and C isotoperatios of feather keratin do not change after growth they provide anon-invasive method for the study of both recently collectedsamples and museum specimens (Thompson amp Furness1995)

Isotope ratios for N and C are presented as -values whereˆ (RsampleRstandard)71 1000 where R ˆ15N14N and 13C12C

respectively The N and C isotope reference standards are air (N)and Peedee belemnite (PDB) (C) Reported uncertainties are thestandard error of the mean N and C isotopic analyses were madeon 03^07 mg of feather material All N and C isotope values weredetermined using an online Carlo Erba (Finnigan MAT BremenGermany) and a continuous poundow Finnigan MAT 252 mass spec-trometer located at Dartmouth College NH USA

N and C isotope values from the Carlo Erba mass spectro-meter were corrected by repeatedly analysing a standard(mesquite leaf tissue) during analysis and comparing the isotopicvalues of these standards with values for the same standard runoyenine These corrections typically are 72 for 15N and+06 for 13C In addition the precision of the isotopicanalyses for feather material was determined to be 03 and

02 for 15N and 13C (1sd n ˆ10) respectively

3 RESULTS

N and C isotopic values for individual birds are shown incentgure 2 and mean and1se values for individual populationsare given in table1Two results follow from this analysis

First the mean ( se) 15N and 13C values of sub-species acredula from latitudes above 638 N were

44 C P Chamberlain and others Stable isotopes repoundect winter quarters

Proc R Soc Lond B (2000)

trochilusacredula

Figure 1 Map of Europe and Africa showing the breedingrange of willow warblers of the two subspecies Ph t trochilus(dotted pattern) and Ph t acredula (stripped pattern) and theircontact zone (black) in Sweden Arrows show the averagemigratory direction through Europe of birds supposedlyoriginating from south of 608N (mean direction ˆ 2138) andnorth of 638 (mean direction ˆ 1748) (Bensch et al 1999)Also shown is the ringing recoveries of birds ringed as eitherPh t trochilus (circles) or Ph t acredula (squares)

+101 027 (n ˆ 48) and 7199 026 (n ˆ 48)respectively In contrast the mean 15N and 13C values ofsubspecies trochilus from latitudes below 618N were bothlower (+73 021 n ˆ 50 and iexcl215 029 n ˆ 50respectively) Diiexclerences between the mean values of

15N and 13C for the two subspecies were signicentcant(ANOVA p50001 for both 15N and 13C) Within thecontact zone between latitude 618N and 638 N the mean

15N value was +87 027 (n ˆ 40) and mean 13Cvalue was iexcl212 021 (n ˆ 40) (centgure 2) There was noeiexclect of year of sampling (table 1) for either C (F1134 ˆ 07p ˆ 04) or N (F1134 ˆ 08 p ˆ 04) in the ANOVA testingfor diiexclerences between regions Similarly there was noeiexclect of date of sampling (table 1) for either C(F1134 ˆ 004 p ˆ 08) or N (F1134 ˆ19 p ˆ 02)

Second the N and C isotopic data match the cline inmorphology for willow warblers in central Sweden(centgure 3) Recent work (Bensch et al 1999) has showndiiexclerences in size and plumage coloration in willowwarblers across the migratory divide in central Scandi-navia The relationship between mean ( se) andlatitude for two morphological traits (wing length n ˆ 32sites and body mass n ˆ 32) and C and N isotopecompositions of willow warblers captured in Sweden isshown in centgure 3 The lines in centgure 3 indicate theaverage values for the populations south and north of theidenticented transition zone (61^638 N) respectively Notethat the diiexclerences in 15N and 13C values occur at thesame latitudes as the changes in morphological traitsBecause the northern birds are larger than the southernbirds one could argue that the diiexclerences observed inisotopic signatures were a result of size-related foodpreferences However most of the size variation (480)is within and not between the subspecies and there wasno signicentcant eiexclect of size (wing length or body mass) on

15N or 13C in the ANOVAs testing for diiexclerencesbetween regions (all p4005)

Based on morphological features Bensch et al (1999)estimated the transition zone between northern andsouthern willow warblers to be less than 350 km Ourisotopic data show a geographical change at least asdistinct as for the morphological traits These results

suggest that the isotopic pattern observed in Scandinaviarepoundects distinctly diiexclerent African winter quarters of thetwo subspecies of Scandinavian willow warblers

4 DISCUSSION

The willow warbler has a winter range covering mostof sub-Saharan Africa and is found mainly in savannahand forest clearings avoiding closed evergreen forests(Cramp 1992) The few ringing recoveries suggest thattrochilus migrates to western Africa and acredula migratesto central-eastern Africa (centgure 1) However despite theringing of 715 000 willow warblers in Sweden up to 1995(Stolt et al 1997) the few ringing recoveries accumulatedso far have not enabled us to locate the position andwidth of the migratory divide

Stable isotopes repoundect winter quarters C P Chamberlain and others 45

Proc R Soc Lond B (2000)

Table 1 N and C isotope compositions of willow warbler feathers from Sweden

sample site 15N 13N

latitude longitude sample date mean se mean se n

55842rsquo N 13827rsquo E 7^10 May 1996 +725 062 iexcl2043 091 1056815rsquo N 14800rsquo E 15^17 May 1997 +794 030 iexcl2100 058 1058819rsquo N 14849rsquo E 28^30 May 1997 +720 055 iexcl2100 062 1059810rsquo N 15825rsquo E 12^17 May 1996 +660 040 iexcl2174 051 1060824rsquo N 17858rsquo E 21^22 May 1996 +757 041 iexcl2167 044 1061843rsquo N 17825rsquo E 26^27 May 1996 +876 048 iexcl2159 038 1061853rsquo N 12844rsquo E 16^17 June 1996 +907 068 iexcl2170 052 1062802rsquo N 14803rsquo E 8^10 June 1997 +811 044 iexcl2138 022 1062828rsquo N 17829rsquo E 28^30 May 1996 +872 053 iexcl2008 030 1063821rsquo N 12833rsquo E 10^11 June 1996 +1016 084 iexcl2021 054 1063829rsquo N 19842rsquo E 3^4 June 1996 +939 038 iexcl1999 018 1065858rsquo N 16807rsquo E 6^8 June 1996 +978 047 iexcl2017 023 1065858rsquo N 24800rsquo E 22^24 June 1997 +1061 066 iexcl2036 031 1067813rsquo N 20848rsquo E 26^27 June 1997 +1045 052 iexcl1844 130 8

trochilusacredulacontact zone

ndash10

ndash14

ndash18

ndash22

ndash26

d13

Cfe

athe

r

d 15Nfeather

4 6 8 10 12 14 16

Figure 2 13C and 15N mean values (see table 1) in feathersof willow warblers collected in Sweden south of 618 N (circles)between 618 N and 638 N (open triangles) and north of 638 N(squares) The lines represent 95 concentdence limits using alldata for contact-zone willow warblers subspecies acredula andsubspecies trochilus Dashed line birds south of 618 N lightsolid line birds between 618 N and 638 N and heavy solid linebirds north of 638 N

We found that the C and N isotopes of two subspeciesof willow warblers (acredula and trochilus) changedabruptly at the same latitude (628 N) as the change inmorphological traits It has been shown that (i) the Nand C isotopes of feathers repoundect the diet and localenvironmental conditions were the tissues were grown(Mizutani et al 1990 1992 Hobson amp Clark 1992) and(ii) with the exception of hydrogen isotopes (Chamberlainet al 1997) the isotopic composition of feather keratin iscentxed after growth Thus the diiexclerences in the 15N and

13C values for acredula and trochilus most likely result fromthese two subspecies of willow warblers wintering inisotopically distinct areas in Africa a result which isconsistent with rare ringing recoveries mentioned above

The C and N isotope values of animals are stronglyinpounduenced by diet and trophic structure (DeNiro ampEpstein 1978 1981 Minagawa amp Wada 1984 Wada et al1987 Mizutani et al 1990 1992 Hobson amp Clark 1992)During the winter season willow warblers are primarilyinsectivorous but the exact trophic structure of the twosubspecies is unknown It is therefore possible thattrochilus and acredula occupy the same geographical region(although this is not supported by ringing recoveries) andthe diiexclerent isotopic signatures repoundect intrinsic diiexcler-ences in diet between the two subspecies We consider thisexplanation unlikely for the following reasons First bodysize did not correlate with isotopic signatures Such acorrelation would be expected if larger birds were feedingon larger prey types containing diiexclerent isotope ratiosThe lack of a correlation between body size and isotoperatios and the fact that most of the size variation occurswithin rather than between subspecies suggest that anydiiexclerence in trophic structure is not the cause of the

isotopic diiexclerence between trochilus and acredula Secondwe only included males in the analyses so possible diiexcler-ences in prey selection between the sexes (NystrIcircm 1991)cannot have biased the result Third we analysed thesame feather the innermost primary in all birds This isthe centrst feather the birds are replacing in their moultand because of the small wing gap at this stage of moult(Bensch amp Grahn 1993) they probably can keep theirpoundight performance and usual feeding technique whengrowing this feather (HedenstrIcircm 1998) In additionstart of moult varies only moderately in Africa earlyDecember in Guinnea-Bissau late December in SouthAfrica and early January in Uganda (Underhill et al1992) Hence most of the analysed feathers should havebeen replaced within the same month-long period so thatit is unlikely that the isotopic diiexclerences between trochilusand acredula repoundect diiexclerences in diet at diiexclerentmoulting times

We therefore suggest that the diiexclerences betweenisotopic ratios of trochilus and acredula primarily resultfrom the two subspecies wintering in isotopically distinctareas However without regional isotopic maps offeathers from willow warblers from Africa it is impossibleto know exactly where in Africa the two subspecieswinter Since local environmental conditions stronglyinpounduence N and C isotopic values it is possible to placeconstraints on where the two subspecies of willowwarblers winter As mentioned above willow warblers inAfrica are found mainly in savannah and forest clearings(Cramp 1992) The C isotope values of willow warblerfeathers are indicative of an ecosystem dominated by C3African trees shrubs and grasses (approximately 27(Smith amp Epstein 1971 Koch et al 1995)) In addition the

46 C P Chamberlain and others Stable isotopes repoundect winter quarters

Proc R Soc Lond B (2000)

(a)73

72

71

70

69

68

win

g le

ngth

(m

m)

(c)105

10

95

9

85

body

mas

s (g

)

55 57 59 61latitude

63 65 67 69

(b)12

11

10

9

8

7

6

15N

d13

Cd

(d)ndash17

ndash18

ndash19

ndash20

ndash21

ndash22

ndash2355 57 59 61

latitude63 65 67 69

trochilusacredulacontact zone

Figure 3 The relationship between mean ( se) and latitude for two morphological traits and C and N isotope compositionsof willow warblers captured in Sweden (a) Wing length (n ˆ 32 sites) (b) body mass (n ˆ 32) (c) 15N (n ˆ 14) and (d) 13C(n ˆ 14) Mean values for wing length and body size are based on measurements of ten to 20 males per site The lines indicate theaverage values for the populations south and north of the identicented transition zone (61^638 N) respectively Contact-zone willowwarblers are presented by open diamonds subspecies acredula by squares and subspecies trochilus by circles

relatively high 15N values of willow warblers are consis-tent with a trophic structure originating at base levelAfrican plants ( 15N of approximately 5 to 3) in areaswith relatively low mean annual precipitation less than500 mm per year (Heaton 1987 Johnson et al 1997) Ourisotopic data on willow warblers are therefore consistentwith the observations that the dominant habitat of willowwarblers in Africa is savannah and forest clearings

We suggest that the higher 15N and 13C values ofsubspecies acredula compared with subspecies trochilusrepoundect the fact that acredula winters in an area (EastAfrica) with a more arid climate and a higher ratio ofC4^C3 plants This argument is based on the followingtwo reasons First studies have shown that elevated 15Nvalues of plant (Heaton 1987) and animal tissues (Van derMerwe et al 1990 Heaton et al 1986 Sealy et al 1987Ambrose 1991 Ambrose amp DeNiro 1989 Johnson et al1997) in Africa occur in areas with relatively low rainfallThis increase in 15N values is presumably the result of15N enrichment of soil N in dry soils due to higher rates ofN loss (Shearer et al 1978) and propagation of this Nisotope signal through the food web The elevated 15Nacredula is consistent with these N isotope studies Secondthe relatively high 13C in animal tissues repoundects inpart the increased abundance of C4 relative to C3 plantsin the ecosystem because C4 plants have higher 13Cvalues (mean approximately iexcl13) than C3 plants(mean approximately iexcl27) (Smith amp Epstein 1971) Inaddition C3 plants have higher 13C values in water-limited habitats (Ehleringer amp Cooper 1988) which alsowould result in elevated 13C of animal tissues in aridareas We therefore speculate that the elevated 13Cvalues observed in acredula repoundect an increase in the ratioof C4^C3 plants in a water-limited environment

Our interpretation is consistent with regional isotopicstudies of elephant bone collagen in Africa (Van derMerwe et al 1990 Vogel et al 1990 Koch et al 1995) Inthese isotopic studies it was shown that individual popula-tions of African elephants had distinct 15N and 13Cvalues (Van der Merwe et al 1990 Vogel et al 1990)although there was considerable variability within a givenpopulation due to local environmental eiexclects (Koch et al1995) In general east and south African elephants hadhigher 13C values and similar or higher 15N values thanwest African elephants (Koch et al 1995 centg 4) Theseauthors (Van der Merwe et al 1990 Vogel et al 1990) alsoattributed the high 15N and 13C values to increasedaridity and relative abundance of C4 plants

Our isotopic data agree with the results from ringingrecoveries that trochilus and acredula winter in distinctareas in Africa However for birds within the contactzone (those birds between latitudes 618N and 638 N) incentral Sweden it is unclear from the isotopic datawhether these birds represent a mix of the two subspeciesor whether these birds winter in a diiexclerent area inAfrica The C isotopic ratios (centgures 2 and 3 table 1)suggest that local sites consist either of pure trochilus(three sites) or pure acredula birds (one site) In contrastmean N isotopic ratios within these local sites were inter-mediate between those of trochilus and acredula with themean 15N value diiexclering signicentcantly from the mean ofeither subspecies (p5005) Hence the N isotopic ratiossuggest that the sites in the contact zone either consist of

a mix of birds from the two subspecies or that the birdsin the contact zone winter in an area with a distinct Nisotopic signature The former explanation is consistentwith data on morphology which indicate the presence oflinkage disequilibrium between size and coloration withinthe contact zone (Bensch et al 1999) This explanation isalso consistent with the hypothesis that there is a contin-uous inpoundux into the contact zone of birds from the north(which are larger and have grey-brown plumage) andbirds from the south (which are smaller and have green-yellow plumage) (Bensch et al 1999)

Thus from our isotopic data alone it remains unclearwhich migratory route is followed by contact-zone willowwarblers Further research using stable isotope signaturesfrom willow warblers breeding in this contact zone willallow us to test whether birds mate randomly or assorta-tively with respect to migratory direction and winterlocation (as measured by isotopic signatures) Forexample contact-zone birds that show intermediateisotope signatures (supposed hybrids) can be used inorientation experiments directly testing whether theytake up an intermediate course of migration in autumn

5 CONCLUSIONS

In conclusion the strong correlation between morpho-logical features and stable isotopic ratios of two subspeciesof willow warblers which migrate to geographicallydistinct areas in Africa supports the use of stable isotopesas a method for tracking migratory patterns of birds(Chamberlain et al 1997 Hobson amp Wassenaar 1997) Wetherefore suggest that the signicentcantly diiexclerent N and Cisotope composition recorded in willow warbler feathersmoulted in West and East Africa respectively opens newavenues of research aimed at disentangling migratoryroutes and wintering areas of Palaearctic^African migra-tory birds in general The addition of other isotopic ratiossuch as D and 87Sr which have been used elsewhere(Chamberlain et al 1997 Hobson amp Wassenaar 1997)when combined with N and C will allow further discrimi-nation of wintering ranges of bird populations and reducethe potential problem of high isotopic variability due tolocal ecological shifts (Koch et al 1995) An important taskwill be to collect data enabling the construction of regionalisotopic maps of feather keratin for migratory species intheir African wintering grounds like those underconstruction for neotropical birds in North America(Chamberlain et al 1997 Hobson amp Wassenaar1997)

This research was supported by Fredrick Hall Professor funds(to CPC) Uddenberg Nordingska Stiftelsen and the SwedishNatural Science Research Council (to SB and SETH) We thankPaul Koch and Richard Holmes for their thoughtful reviews

REFERENCES

Ambrose S H 1991 Eiexclects of diet climate and physiology onnitrogen isotope abundance in terrestrial food webs J ArchSci 18 293^317

Ambrose S H amp DeNiro M J 1989 Climate and habitatreconstruction using stable carbon and nitrogen isotope ratiosof collagen in prehistoric herbivore teeth from Kenya QuatRes 31 407^422

Stable isotopes repoundect winter quarters C P Chamberlain and others 47

Proc R Soc Lond B (2000)

ARGOS 1994 Animal tracking Newsletter no 48 Tolouse CedexFrance CLS Service Argos

Bearhop S Thompson D R Waldron S Russel I CAlexander G amp Furness RW1999 Stable isotopes indicate theextent of freshwater feeding by cormorants Phalacrocorax carboshot at inland centsheries in England J Appl Ecol 3675^84

Bensch S amp Grahn M 1993 A new method for estimatingindividual speed of molt example from the willow warblerCondor 95 305^315

Bensch S Andersson T amp ETHkesson S 1999 Morphologicaland molecular variation across a migratory divide in willowwarblers Phylloscopus trochilus Evolution (In the press)

Berthold P 1996 Control of bird migration London Chapman ampHall

Chamberlain C P Blum J D Holmes R T Feng XSherry T W amp Graves G R 1997 The use of isotope tracersfor identifying populations of migratory birds Oecologia 109132^141

Cramp S 1992 The birds of the western Palearctic vol 6 OxfordUniversity Press

DeNiro M J amp Epstein S 1978 Inpounduence of diet on the distri-bution of carbon isotopes in animals Geochim Cosmochim Acta42 495^506

DeNiro M J amp Epstein S 1981 Inpounduence of diet on the distri-bution of nitrogen isotopes in animals Geochim CosmochimActa 45 341^351

Ehleringer J R amp Cooper T A 1988 Correlations betweencarbon isotope ratio and microhabitat in desert plantsOecologia 76 562^566

Emlen S T amp Emlen J T 1966 A technique for recordingmigratory orientation of captive birds Auk 83 361^367

Ginn H B amp Melville D S 1983 Moult in birds BTO guide 19Tring UK Maund amp Irvine Ltd

Harrington R R Kennedy B P Chamberlain C P BlumJ D amp Folt C L 1998 15N enrichment in agricultural catch-ments centeld patterns and applications to tracking Atlanticsalmon (Salmo salar) Chem Geol 147 281^294

Heaton T H E 1987 The 15N14N ratios in plants in SouthAfrica and Namibia relationship to climate and coastalsaline environments Oecologia 74 236^246

Heaton T H E Vogel J C Von la Chevallerie G amp GolletG 1986 Climatic inpounduence on the isotopic composition ofbone nitrogen Nature 322 822^823

HedenstrIcircm A 1998 The relationship between wing area andraggedness during molt in the willow warbler and otherpasserines J Field Ornithol 69 103^108

HedenstrIcircm A amp Pettersson J 1984 The migration of willowwarbler Phylloscopustrochilus at OttenbyVOcircr FOcircgelvIgraverld 43 217^228

HedenstrIcircm A amp Pettersson J 1987 Migration routes andwintering areas of willow warblers Phylloscopus trochilus (L)ringed in Fennoscandia Ornis Fennica 45 1^7

Helbig A J 1996 Genetic basis mode of inheritance and evolu-tionary changes of migratory directions in Palearctic warblers(Aves Sylvidae) J Exp Biol 199 49^55

Hobson K A amp Clark R G 1992 Assessing avian diet usingstable isotopes ITurnover of 13C in tissues Condor 94 181^188

Hobson K A amp Wassenaar L I 1997 Linking breeding andwintering grounds of neotropical migrant songbirds usingstable hydrogen isotopic analysis of feathers Oecologia 109142^148

Johnson B J Miller G H Fogel M L amp Beaumont P B1997 The determination of late Quaternary paleoenviron-ments at Equus Cave South Africa using stable isotopes andamino acid racemization in ostrich eggshell PalaeogeogPalaeoclimat Palaeoecol 136 121^137

Kennedy B P Folt C L Blum J D amp Chamberlain C P1997 Natural isotope markers in salmon Nature 387 766

Koch P L Heisinger J Moss C Carlson R Fogel M Lamp Behrensmeyer A K 1995 Isotopic tracking of changein diet and habitat use in African elephants Science 2671340^1343

Marra P P Hobson K A amp Holmes R T 1998 Linkingwinter and summer events in a migratory bird by usingstable-carbon isotopes Science 282 1884^1886

Minagawa M amp Wada E 1984 Stepwise enrichment of 15Nalong food chains further evidence and the relation between

15N and animals age Geochim Cosmochim Acta 48 1135^1140Mizutani H Fukuda M amp Wada E 1990 Carbon isotope

ratio of feathers reveal feeding behaviour of cormorants Auk107 400^437

Mizutani H Fukuda M amp Kabaya Y 1992 15N and 13Cenrichment factors of feathers of 11 species of adult birdsEcology 73 1391^1395

NystrIcircm K G K 1991 On sex-specicentc foraging behavior in thewillow warbler Phylloscopus trochilus Can J Zool 69 462^470

Sealy J C Van der Merwe N J Lee-Thorp J A ampLanhamm J L 1987 Nitrogen isotopic ecology in southernAfrica implications for environmental and dietary tracingGeochim Cosmochim Acta 51 2707^2717

Shearer G Kohl D H amp Chien S-H 1978 The nitrogen-15abundance in a wide variety of soils Soil Sci Soc Am J 42899^902

Smith B N amp Epstein S 1971 Two categories of 13C12C ratiosin higher plants Plant Physiol 47 380^384

Stolt B-O EkstrIcircm L Fransson T Kroon C Staav RSIgravellstrIcircm B amp SIgravellstrIcircm U B 1997 Report on Swedish birdringing for 1995 Stockholm Swedish Museum of NaturalHistory (Bird Ringing Centre)

Thompson D R amp Furness R W 1995 Stable-isotope ratios ofcarbon and nitrogen in feathers indicate seasonal dietaryshifts in northern fulmars Auk 112 493^498

Underhill L G Prys-Jones R P Dowsett R J Herroelen PJohnson D N Lawn M R Norman S C Pearson D J ampTree A J 1992 The biannual primary moult of willowwarblers Phylloscopus trochilus in Europe and Africa Ibis 134286^297

Van der Merwe N J Lee-Throp J A Thackeray J F Hall-Martin A Kruger F J Coetzee H Bell R H ampLindeque M 1990 Source-area determination of elephantivory by isotopic analysis Nature 346 744^746

Vogel J C Eglinton B amp Auret J M 1990 Isotope centnger-prints in elephant bone and ivory Nature 346 747^749

Wada E Terazki M Kabaya Y amp Nemoto T 1987 15N and13C abundances in the Antarctic Ocean and emphasis on thebiogeochemical structure of the food web Deep-Sea Res 34829^841

Wassenaar L I amp Hobson K A 1998 Natal origins ofmigratory monarch butterpoundies at wintering colonies inMexico new isotopic evidence Proc Natl Acad Sci USA 9515 436^15 439

48 C P Chamberlain and others Stable isotopes repoundect winter quarters

Proc R Soc Lond B (2000)