evolutionary status of icelandic redpolls carduelis flammea islandica (aves, passeriformes,...

14
ORIGINAL ARTICLE Evolutionary status of Icelandic Redpolls Carduelis flammea islandica (Aves, Passeriformes, Fringillidae) Julien Amouret 1 Katja Steinauer 1,2 Gunnar T. Hallgrimsson 1 Snæbjo ¨rn Pa ´lsson 1 Received: 12 September 2014 / Revised: 20 February 2015 / Accepted: 13 March 2015 Ó Dt. Ornithologen-Gesellschaft e.V. 2015 Abstract The Icelandic Redpoll Carduelis flammea is- landica is one of three subspecies of Carduelis flammea. The other two are C. f. rostrata, breeding in Greenland, and C. f. flammea, widely distributed at high latitudes in both North America and Eurasia. Recent studies on variation of the mtDNA control region and microsatellites among C. f. r. and C. f. f. and related species (Arctic Redpoll Carduelis hornemanni and Lesser Redpoll Carduelis cabaret) did not reveal clear genetic differentiation among the species. Here we include DNA sequences of mtDNA and nuclear markers of the Icelandic subspecies (C. f. islandica) and from additional samples of the other species and subspecies to evaluate further their taxonomic status within the com- plex, with special emphasis on C. f. islandica. Mitochon- drial and nuclear variation is large within species and does not provide support for the current subspecies and the species classification. Significant differences in haplotype frequencies of the combined genetic data are observed between the C. flammea subspecies, and C. cabaret. The slight genetic differentiation within the redpoll complex could result from introgression and/or incomplete lineage sorting following recent and rapid diversification in mor- phology, possibly driven by environmental factors. Keywords Holarctic Introgression Lineage sorting mtDNA Introns Subspecies Zusammenfassung Evolutiona ¨rer Status des isla ¨ndischen Birkenzeisigs Carduelis flammea islandica (Aves, Passeriformes, Fringillidae) Der isla ¨ndische Birkenzeisig Carduelis flammea islandica stellt eine der drei Unterarten des Birkenzeisig (Carduelis flammea) aus der Familie der Finken (Fringillidae) dar. Die anderen beiden Unterarten sind C. f. rostrata, der auf Gro ¨nland bru ¨tet, sowie C. f. flammea, welcher weit ver- breitet ist in den hohen Breitengraden sowohl Nor- damerikas als auch Eurasiens. Neuere Studien u ¨ber Variation innerhalb der Kontrollregion der mitochondrialen DNA (mtDNA) und Mikrosatelliten zwischen C. f. r. und C. f. f. sowie weiterer verwandter Arten (Arktischer Birkenzeisig Carduelis hornemanni und Kleiner Birken- zeisig Carduelis cabaret) konnten keine klare genetische interspezifische Differenzierung zeigen. Um den genauen taxonomischen Status der hier genannten Unterarten des Birkenzeisigs und insbesondere des isla ¨ndischen Birken- zeisigs C. f. islandica innerhalb des Artenkomplexes bewerten zu ko ¨nnen, beziehen wir in dieser Studie mito- chondriale DNA Sequenzen und nukleare Marker weiterer Proben von C. f. islandica sowie der anderen Unterarten ein. Sowohl auf mitochondrialer als auch nuklearer Ebene zeigt sich eine große Variation innerhalb der Arten, was der momentanen Klassifizierung der Arten und Unterarten widerspricht. Zwischen den Unterarten von C. flammea und Communicated by M. Wink. Electronic supplementary material The online version of this article (doi:10.1007/s10336-015-1208-3) contains supplementary material, which is available to authorized users. & Julien Amouret [email protected] 1 Department of Life and Environmental Sciences, University of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland 2 German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Deutscher Platz 5e, 04103 Leipzig, Germany 123 J Ornithol DOI 10.1007/s10336-015-1208-3

Upload: hi

Post on 24-Apr-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

ORIGINAL ARTICLE

Evolutionary status of Icelandic Redpolls Carduelis flammeaislandica (Aves, Passeriformes, Fringillidae)

Julien Amouret1• Katja Steinauer1,2

• Gunnar T. Hallgrimsson1• Snæbjorn Palsson1

Received: 12 September 2014 / Revised: 20 February 2015 / Accepted: 13 March 2015

� Dt. Ornithologen-Gesellschaft e.V. 2015

Abstract The Icelandic Redpoll Carduelis flammea is-

landica is one of three subspecies of Carduelis flammea.

The other two are C. f. rostrata, breeding in Greenland, and

C. f. flammea, widely distributed at high latitudes in both

North America and Eurasia. Recent studies on variation of

the mtDNA control region and microsatellites among C. f.

r. and C. f. f. and related species (Arctic Redpoll Carduelis

hornemanni and Lesser Redpoll Carduelis cabaret) did not

reveal clear genetic differentiation among the species. Here

we include DNA sequences of mtDNA and nuclear

markers of the Icelandic subspecies (C. f. islandica) and

from additional samples of the other species and subspecies

to evaluate further their taxonomic status within the com-

plex, with special emphasis on C. f. islandica. Mitochon-

drial and nuclear variation is large within species and does

not provide support for the current subspecies and the

species classification. Significant differences in haplotype

frequencies of the combined genetic data are observed

between the C. flammea subspecies, and C. cabaret. The

slight genetic differentiation within the redpoll complex

could result from introgression and/or incomplete lineage

sorting following recent and rapid diversification in mor-

phology, possibly driven by environmental factors.

Keywords Holarctic � Introgression � Lineage sorting �mtDNA � Introns � Subspecies

Zusammenfassung

Evolutionarer Status des islandischen Birkenzeisigs

Carduelis flammea islandica (Aves, Passeriformes,

Fringillidae)

Der islandische Birkenzeisig Carduelis flammea islandica

stellt eine der drei Unterarten des Birkenzeisig (Carduelis

flammea) aus der Familie der Finken (Fringillidae) dar. Die

anderen beiden Unterarten sind C. f. rostrata, der auf

Gronland brutet, sowie C. f. flammea, welcher weit ver-

breitet ist in den hohen Breitengraden sowohl Nor-

damerikas als auch Eurasiens. Neuere Studien uber

Variation innerhalb der Kontrollregion der mitochondrialen

DNA (mtDNA) und Mikrosatelliten zwischen C. f. r. und

C. f. f. sowie weiterer verwandter Arten (Arktischer

Birkenzeisig Carduelis hornemanni und Kleiner Birken-

zeisig Carduelis cabaret) konnten keine klare genetische

interspezifische Differenzierung zeigen. Um den genauen

taxonomischen Status der hier genannten Unterarten des

Birkenzeisigs und insbesondere des islandischen Birken-

zeisigs C. f. islandica innerhalb des Artenkomplexes

bewerten zu konnen, beziehen wir in dieser Studie mito-

chondriale DNA Sequenzen und nukleare Marker weiterer

Proben von C. f. islandica sowie der anderen Unterarten

ein. Sowohl auf mitochondrialer als auch nuklearer Ebene

zeigt sich eine große Variation innerhalb der Arten, was

der momentanen Klassifizierung der Arten und Unterarten

widerspricht. Zwischen den Unterarten von C. flammea und

Communicated by M. Wink.

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10336-015-1208-3) contains supplementarymaterial, which is available to authorized users.

& Julien Amouret

[email protected]

1 Department of Life and Environmental Sciences, University

of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland

2 German Centre for Integrative Biodiversity Research (iDiv)

Halle-Jena-Leipzig, Deutscher Platz 5e, 04103 Leipzig,

Germany

123

J Ornithol

DOI 10.1007/s10336-015-1208-3

dem Kleinen Birkenzeisig C. cabaret konnten signifikante

Unterschiede innerhalb der Haplotypfrequenz fur den

kombinierten genetischen Datensatz gefunden werden. Die

geringe genetische Differenzierung innerhalb des

Artenkomplexes der Birkenzeisige konnte auf Introgres-

sion und/oder unvollstandige Linientrennung zuruck-

zufuhren sein, welche ihrerseits aus der schnellen und

kurzfristigen morphologischen Diversifizierung resultiert,

die moglicherweise durch Umweltfaktoren verursacht ist.

Introduction

The concept of subspecies has been used to assign in-

dividuals to geographically subdivided populations based

on morphological variation (Amadon 1949; Barrow-

clough 1982; Patten et al. 2002; Remsen et al. 2010).

Such categorizing has more commonly been done for

insular rather than continental areas (Phillimore et al.

2010). Individuals from a subspecies are also expected to

share a common ancestor, which has been described as

reciprocal monophyly (Avise 2000). The subspecies level

can be considered as a stepping stone towards allopatric

or incipient species (Mayr 1982), characterised by dif-

fuse genetic architecture (Ting et al. 2001). Thus studies

on subspecies can enhance our understanding of spe-

ciation, dispersal and geographic variation (Patten et al.

2002). In birds there is often conflicting support for

subspecies based on morphological and molecular data

(Price 2008).

Redpolls (Aves, Fringillidae), widely distributed at

high latitudes of the Holarctic, are currently classified as

three species (Knox 1988): the Lesser Redpoll Carduelis

cabaret, Common Redpoll Carduelis flammea, and the

Arctic Redpoll Carduelis hornemanni. The taxonomy is

geographically structured (Knox 1988), and is mainly

based on plumage variation (Troy 1985; Seutin et al.

1992). C. flammea includes three subspecies: C. f. flam-

mea, C. f. islandica and C. f. rostrata; C. hornemanni

includes two subspecies: C. h. hornemanni and C. h. ex-

ilipes. The species complex is known for a controversial

taxonomy due to different interpretations given to the

origin of these polytypic birds (Seutin et al. 1992) and its

morphology has been extensively studied since the 1980s

(Troy 1985; Knox 1988; Herremans 1990; Seutin et al.

1992, 1993). Recent studies on variation of the mtDNA

control region (CR) and microsatellites among C. f. ros-

trata and C. f. flammea, mainly from Scandinavia, and the

related species (C. hornemanni and C. cabaret) revealed

high genetic variation but no clear genetic differentiation

among the species (Seutin et al. 1993; Ottvall et al. 2002;

Marthinsen et al. 2008), possibly as a result of extensive

admixture or a lack of lineage sorting due to recent di-

versification. A more extensive sampling from the species

range and of the different subspecies such as from C. f.

islandica might help to resolve the evolutionary rela-

tionships within the complex.

C. f. islandica is known for high variation and its

taxonomic status has been uncertain (Seutin et al. 1992).

Three plumage forms have been described: ‘‘pale’’ which

shows whitish rump and underparts with moderately

streaked plumage, ‘‘dark’’ form presenting a boldly

streaked plumage and a lack of white rump and the ‘‘in-

determinate’’ form which has a whitish rump and a

relatively streaked plumage (Herremans 1990). Mainly on

the basis of the plumage, Herremans (1990) suggested that

the C. f. islandica is an exilipes-like pale form which has

evolved as a result of character release and more recently

by a secondary contact with C. f. rostrata. This apparent

introgression of the ‘‘dark’’ morph from C. f. rostrata into

the ‘‘pale’’ morph is supported by the irruptive and op-

portunistic behaviour of the redpolls (Troy 1983). Knox

(1988), in his description of the taxonomy of the redpolls

and based on observations made by Bird (1935) and

Timmermann (1938) of pale redpolls over the spring,

suggested that C. f. islandica needed to be examined from

different periods of the year.

Here we investigate the differentiation of the redpoll

complex (C. flammea, C. hornemanni and C. cabaret) with

a special emphasis on C. f. islandica and whether its sub-

species status is supported by genetic analysis of mito-

chondrial and nuclear variation. Distinct subspecies are

expected to be monophyletic or to share a common

ancestry within a lineage which may have diverged in al-

lopatry or in reproductive isolation over extended time

periods. In addition we analyse the geographical partition

of the variation among the taxonomic groups.

Materials and methods

Sampling

In total 199 redpoll individuals were sampled from Iceland

for this study: 57 during breeding season (May–August), 80

from early spring (March and April, referred to as

‘‘mixed’’) and 62 from the wintering period. From museum

collections we obtained eight C. f. rostrata and three C. h.

hornemanni, from northern Norway we got a sample of

feathers from 19 C. f. flammea individuals and from

Scotland we got a sample of feathers from 15 C. cabaret

individuals (see locations in Fig. 1; Table 1). From Gen-

bank we retrieved 148 mitochondrial sequences including

the CR and part of the cytochrome oxidase I (COI): 60 C. f.

flammea, 5 C. f. rostrata, 57 C. cabaret and 26 C. h.

J Ornithol

123

exilipes (see the accession numbers, information of the

origin of samples and sampling dates in Table 1), studied

previously by Ottvall et al. (2002), Hebert et al. (2004),

Kerr et al. (2007, 2009), Marthinsen et al. (2008) and

Johnsen et al. (2010).

DNA extraction

DNA was extracted from secondary feathers. The proximal

ends of the feathers were sliced in two pieces and dissolved

in 200 lL of 6 % Chelex 100 solution (Bio-Rad) and

2.5 lL of 1 % proteinase K.

Molecular sexing

Two primers sets were used in PCR to amplify introns of

the chromodomain helicase DNA binding protein gene

(CHD genes) in the W and Z sex chromosomes; 2987 F and

3112 R for W, and 3007 F and 3112 R for Z were used for

sexing, following the procedure in Fridolfsson and Ellegren

(1999). The PCR was performed in a volume of 10 lL

using 0.09 lL Taq polymerase (0.5 U lL-1), 0.75 lL

dNTP (2 mM), 1 lL Tween 20 (1 %), 1 lL Taq buffer

(Std Taq buffer), 1 lL BSA (5 mg mL-1) and 1 lL DNA

extract (about 10–100 ng lL-1). The primers 2550 F and

2718 R (Fridolfsson and Ellegren 1999), commonly used

for sexing, did not reveal any length variation. PCR frag-

ments were separated by gel electrophoresis on a 1.5 %

agarose gel, stained with ethidium bromide and viewed

under ultraviolet light.

PCR and DNA sequencing

Two mitochondrial fragments (CR and COI) and four nu-

clear markers were amplified using the same reagents as

listed above. The primers for the CR were from Tarr (1995)

and designed in this study, and those for COI were from

Lohman et al. (2009). The nuclear markers included two

sex-linked loci: GHR (primers in Tanaka et al. 1992) and

ACOI (primers in Kimball et al. 2009); and two autosomal

loci: TGFb2 (primers in Burt and Paton 1991) and FIB7

(primers in Prychitko and Moore 1997). The PCR products

were purified by applying a standard ExoSAP protocol and

sequenced using the ABI PRISM BigDye Terminator v3.1

Cycle Sequencing Kit in a reaction volume of 10 lL in-

cluding 1 lM primer and 0.8 lL of the digested PCR

products. The sequencing products were run on an ABI

PRISM 3100 Genetic Analyser (Applied Biosystems), read

in FinchTV 1.4.0 and aligned bye eye in MEGA 5 (Tamura

et al. 2011). Further details with annealing temperatures,

primer sequences and sequence lengths are given in Ap-

pendix 1 in Electronic Supplementary Material. All se-

quences are available in Genbank (Accession nos.

KM517973–KM518204 and KP823219–KP823394, Ap-

pendix 2 in ESM).

Haplotype estimation and descriptive statistics

The haplotypes of each nuclear marker were inferred by

using PHASE 2.0: 1000 burn-in and 1000 iterations

(Stephens and Donnelly 2003). Only sequences with

Fig. 1 Breeding distribution

range of the redpoll complex

(Knox 1988) and sampling

localities (details are given in

Table 1). The geographical

groupings represent the main

localities used in the analyses of

molecular variance (AMOVAs)

J Ornithol

123

Ta

ble

1In

form

atio

no

nsa

mp

les

fro

mth

esi

xre

dp

oll

tax

a

Tax

on

Lo

cali

tyC

ou

ntr

y# (F

ig.

1)

Co

ord

inat

esC

oll

ecti

on

dat

en

CR

nC

OI

Acc

essi

on

no

.C

rA

cces

sio

nn

o.

CO

I

Ref

eren

ces

C.

f.is

lan

dic

aB

reed

ing

Icel

and

16

4�N

18

�W2

00

9,

20

12

31

38

KM

517

97

3–

KM

51

80

03

KM

51

80

74–

KM

518

11

1

a

‘Mix

ed’

Icel

and

16

4�N

18

�W2

00

7–

201

31

62

3K

M5

18

00

4–

KM

51

80

21

KM

51

81

67–

KM

518

18

8

a

Win

teri

ng

Icel

and

16

4�N

18

�W2

00

7–

201

31

83

4K

M5

18

02

2–

KM

51

80

37

KM

51

81

52–

KM

518

16

6

KM

51

81

89–

KM

518

20

4

a

C.

f.fl

am

mea

Div

idal

No

rway

26

8.9

3�N

19

.52�E

Sep

tem

ber

20

12

18

19

KM

518

04

3–

KM

51

80

60

KM

51

81

12–

KM

518

13

0

a

Hei

md

alen

No

rway

46

1.4

2�N

8.8

7�E

May

–Ju

ly1

99

8–

20

03

22

EU

400

48

1–

EU

40

05

02

Mar

thin

sen

etal

.(2

00

8)

San

dfj

ord

enN

orw

ay3

70

.50

�N3

0.5

3�E

July

20

04

3E

U4

00

50

3–

EU

40

05

05

Mar

thin

sen

etal

.(2

00

8)

Fal

ster

bo

Sw

eden

15

55

.23

�N1

2.5

0�E

Au

tum

n1

99

7,

19

99

13

Ott

val

let

al.

(20

02)

Eid

sber

gN

orw

ay4

59�3

10 N

11�1

40 E

19

93

–1

99

44

Ott

val

let

al.

(20

02)

Hei

md

alen

No

rway

46

1�2

50 N

8�5

20 E

19

93

–1

99

42

Ott

val

let

al.

(20

02)

(Onta

rio

)C

anad

a1

41

DQ

43

342

7K

err

etal

.(2

00

7)

Sam

eia

No

rway

45

9.5

2�N

11

.32�E

Sep

tem

ber

20

00

2G

U5

71

31

5,

GU

57

131

3

Joh

nse

net

al.

(20

10)

Hei

md

alen

No

rway

46

1.4

2�N

8.8

7�E

July

20

03

1G

U5

71

31

4Jo

hnse

net

al.

(20

10)

Rek

a

Ag

ayak

an

Russ

ia5

63

.33

�N1

41

.73�E

July

19

90

1G

Q4

81

47

3K

err

etal

.(2

00

9)

Kh

asy

nsk

iyR

uss

ia5

60

.08

�N1

50

.78�E

Sep

tem

ber

19

88

1G

Q4

81

47

0K

err

etal

.(2

00

9)

Mez

ensk

iyR

uss

ia8

65

.72

�N4

4.3

7�E

Jun

e1

99

81

GQ

48

146

9K

err

etal

.(2

00

9)

Ter

iber

ka

Russ

ia8

68

.98

�N3

4.8

9�E

July

19

90

1G

Q4

81

46

8K

err

etal

.(2

00

9)

An

ady

rR

uss

ia6

64

.18

�N1

78

.25�E

July

19

88

1G

Q4

81

46

7K

err

etal

.(2

00

9)

Ust

-ku

tR

uss

ia9

56

.67

�N1

05

.77�E

July

19

98

1G

Q4

81

47

5K

err

etal

.(2

00

9)

An

cast

erC

anad

a1

44

3.1

9�N

80

.02�W

1A

Y6

66

47

4H

eber

tet

al.

(20

04)

Isak

ov

tsy

Russ

ia7

58

.43

�N5

0.4

5�E

Oct

ober

19

97

1G

Q4

81

46

6K

err

etal

.(2

00

9)

J Ornithol

123

Ta

ble

1co

nti

nu

ed

Tax

on

Lo

cali

tyC

ou

ntr

y# (F

ig.

1)

Coord

inat

esC

oll

ecti

on

dat

en

CR

nC

OI

Acc

essi

on

no.

Cr

Acc

essi

on

no.

CO

I

Ref

eren

ces

Lab

ytn

ang

iR

uss

ia7

58

.02

�N6

8.6

0�E

Jun

e1

99

31

GQ

48

147

1K

err

etal

.(2

00

9)

Mar

ko

vo

Russ

ia6

64

.69

�N1

70

.42�E

Jun

e1

99

91

GQ

48

147

2K

err

etal

.(2

00

9)

Ves

tpo

llen

No

rway

26

8.3

2�N

14

.67�E

July

20

03

1G

U5

71

31

6Jo

hnse

net

al.

(20

10)

No

yab

r’sk

Russ

ia6

3.4

8�N

74

.87�E

Jun

e1

99

21

GQ

48

147

4K

err

etal

.(2

00

9)

C.

f.ro

stra

taIs

un

gu

aG

reen

lan

d1

06

7.1

7�N

50

.25�W

July

20

09

24

KM

518

03

9–

KM

51

80

40

KM

51

81

35–

KM

518

13

7

a

Go

dh

avn

Gre

enla

nd

10

69

.25

�N5

3.6

3�W

Au

gu

st1

99

01

1K

M5

18

13

1M

arth

inse

net

al.

(20

08)

Kv

and

alG

reen

lan

d1

17

4.5

9�N

56

.89�W

July

19

94

33

EU

400

50

6K

M5

18

13

2–

KM

518

13

4

Mar

thin

sen

etal

.(2

00

8)

Gre

enla

nd

16

75

.0�N

20

.00�W

Au

gu

st–

Oct

ob

er1

90

6,

19

14,

19

18,

19

30

4E

U4

00

50

7–

EU

40

05

09

Mar

thin

sen

etal

.(2

00

8)

Uts

ira

No

rway

45

9.1

9�N

4.5

3�E

Sep

tem

ber

20

04

1E

U4

00

51

0–

EU

40

05

12

,

EU

40

05

14

EU

400

51

5

Mar

thin

sen

etal

.(2

00

8)

C.

cabare

tG

edse

rD

enm

ark

15

54

.57

�N1

1.9

3�E

Au

gu

st2

01

12

KM

518

04

1–

KM

51

80

42

a

Bla

avan

ds

Hu

k

Den

mar

k1

55

5.5

6�N

8.1

2�E

Jun

e1

99

0a

Kaa

sD

enm

ark

15

57

.20

�N9

.67�E

Au

gu

st2

00

4a

Uts

ira

No

rway

45

9.1

9�N

4.5

3�E

May

–S

epte

mb

er2

00

3,

20

04

14

EU

400

45

1–

EU

40

04

64

Mar

thin

sen

etal

.(2

00

8)

Kv

ines

dal

No

rway

45

8.1

7�N

6.5

4�E

Sep

tem

ber

20

03

11

EU

400

46

5–

EU

40

04

75

Mar

thin

sen

etal

.(2

00

8)

Jom

fru

lan

dN

orw

ay4

58

.52

�N9

.36�E

May

20

03

5E

U4

00

47

6–

EU

40

04

80

Mar

thin

sen

etal

.(2

00

8)

Sco

tlan

d1

35

7�N

4�W

13

10

KM

518

06

1–

KM

51

80

73

KM

51

81

42–

KM

518

15

1

a

Fal

ster

bo

Sw

eden

15

55�2

30 N

12�5

00 E

Au

tum

n1

99

7,

19

99

17

Ott

val

let

al.

(20

02)

Bai

nh

amN

ew Zea

land

40

.45

�S1

72

.33�E

No

vem

ber

19

90

4R

atn

asin

gh

aman

d

Heb

ert

(20

07)

Ey

rew

ell

Fo

rest

New Z

eala

nd

43

.24

�S1

72

.24�E

No

vem

ber

19

93

1R

atn

asin

gh

aman

d

Heb

ert

(20

07)

Eid

sber

gN

orw

ay1

55

9.3

1�N

11

.14�E

19

93

–1

99

45

Ott

val

let

al.

(20

02)

J Ornithol

123

Ta

ble

1co

nti

nu

ed

Tax

on

Lo

cali

tyC

ou

ntr

y# (F

ig.

1)

Coord

inat

esC

oll

ecti

on

dat

en

CR

nC

OI

Acc

essi

on

no.

Cr

Acc

essi

on

no.

CO

I

Ref

eren

ces

C.

h.

ho

rnem

an

ni

Th

ule

Gre

enla

nd

12

77

.48

�N6

9.3

4�W

Au

gu

st2

00

81

1K

M5

18

03

8K

M5

18

14

1a

Ell

esm

ere

Isla

nd

Can

ada

12

78�N

84

�WJu

ne–

July

19

79

2K

M5

18

13

9–

KM

518

14

0

a

C.

h.

exil

ipes

Kir

un

aS

wed

en2

67

.87

�N2

0.2

5�E

July

19

94

1E

U4

00

52

0M

arth

inse

net

al.

(20

08)

Kan

in

Pen

insu

la

Russ

ia8

68

.43

�N4

5.4

2�E

Au

gu

st1

99

41

EU

400

52

1M

arth

inse

net

al.

(20

08)

Chu

ko

tka

Russ

ia6

64

.58

�N1

77

.33�E

July

20

05

1E

U4

00

52

3M

arth

inse

net

al.

(20

08)

r-V

aran

ger

No

rway

36

9.0

8�N

29

.00�E

Jun

e1

91

11

EU

400

53

0M

arth

inse

net

al.

(20

08)

San

dfj

ord

enN

orw

ay3

70

.50

�N3

0.5

3�E

July

20

04

2E

U4

00

53

1–

EU

40

05

32

Mar

thin

sen

etal

.(2

00

8)

Kra

sno

jars

kR

uss

ia9

56

.13

�N9

3�E

Feb

ruar

y1

91

21

EU

400

52

9M

arth

inse

net

al.

(20

08)

Ko

lym

aR

uss

ia6

69

.5�N

16

1�E

Jan

uar

y1

91

0,

19

16

2E

U4

00

52

7–

EU

40

05

28

Mar

thin

sen

etal

.(2

00

8)

No

rway

Feb

ruar

y1

98

2,

Oct

ober

20

07

3E

U4

00

52

5,

EU

400

53

3–

EU

40

05

34

Mar

thin

sen

etal

.(2

00

8)

Sw

eden

Feb

ruar

y1

99

8,

Mar

ch2

00

52

EU

400

52

2,

EU

400

52

4M

arth

inse

net

al.

(20

08)

Fal

ster

bo

Sw

eden

15

55

.38

�N1

2.8

2�E

Oct

ober

,N

ov

emb

er1

99

04

EU

400

51

6–

EU

40

05

19

Mar

thin

sen

etal

.(2

00

8)

San

dfj

ord

enN

orw

ay3

70

.5�N

30

.53�E

July

20

04

1G

U5

71

31

9Jo

hnse

net

al.

(20

10)

Cher

skiy

Russ

ia5

69

.40

�N1

58

.50�E

July

19

88

1G

Q4

81

48

0K

err

etal

.(2

00

9)

Av

ton

om

nay

aR

uss

ia5

69

.4�N

15

8.5

�EJu

ly1

99

21

DQ

43

342

8K

err

etal

.(2

00

7)

Ok

rug

Russ

ia6

64

.12

�N1

78

.25�E

July

19

92

2D

Q4

33

42

9,

DQ

43

343

1

Ker

ret

al.

(20

07)

Ob

last

Russ

ia3

68

.98

�N3

4.8

9�E

July

19

94

2D

Q4

33

43

0,

DQ

43

343

2

Ker

ret

al.

(20

07)

San

dfj

ord

enN

orw

ay3

70

.50

�N3

0.5

3�E

July

20

04

1G

U5

71

32

0Jo

hnse

net

al.

(20

10)

To

tal

22

41

64

aS

amp

lin

gfr

om

the

pre

sen

tst

ud

y

J Ornithol

123

phasing posterior probability more than 0.6 were consid-

ered resolved and thereby selected for further analysis

(Harrigan et al. 2008). Standard diversity indices were

summarized using ARLEQUIN 3.1 (Excoffier et al. 2005),

including the number of haplotypes (#h), haplotype di-

versity (H), nucleotide diversity (p) and number of seg-

regating sites (S). Haplotypic richness (AR), controlling

for differences in sample size, was calculated using the

allelic richness function in the hierfstat package in R

(Goudet 2005). To visualize the relationship among the

haplotypes, networks for each marker were drawn

separately (and also for the two mtDNA regions, as the

records in Genbank are not from the same individuals) by

using the median-joining network method (Bandelt et al.

1999) implemented in the program NETWORK 4.6.0.0

(http://www.fluxus-engineering.com). A combined net-

work based on a subset of the data where the same indi-

vidual was sequenced for both mitochondrial regions was

also constructed. As a result of high variation in the CR

and observed homoplasy, resulting in loops in the network

and thus in incorrect genetic relationships, we identified

sites where multiple mutations or hits had occurred. The

method is based on computation of what we refer to as the

background nucleotide diversity pij for each site i and

nucleotide j, segregating at that particular site. By com-

paring pij to the overall nucleotide diversity (ptotal) for the

whole data set we can see whether the variation of the

nucleotide at that particular site is in clear conflict with

rest of the sequence variation as may result from multiple

hits. When pij was 10 % larger than ptotal, we omitted the

site i when drawing the network (see the ratios for the

combined loci in Appendix 3.1 and 3.2 in ESM).

Population divergence

Genetic variation was partitioned by AMOVA with respect

to taxonomic status of the sampled populations, consider-

ing both pairwise distances between sequences (U) and

solely the haplotype frequencies (F), and tested, with 1000

permutations using Arlequin 3.1 (Excoffier et al. 2005).

Averages of differentiation among species (FCT), sub-

species (FST) and subspecies within species (FSC) for all

nuclear markers and combined with one of the two mtDNA

markers at each time were then calculated by weighting

each variance component with samples size and the asso-

ciated P value with the Fisher’s method for combining

P values (e.g. Sokal and Rohlf 2012). Differentiation be-

tween the breeding and non-breeding Icelandic samples

was tested for the mtDNA variation. The pairwise genetic

differentiation between samples was further analysed by

calculating pairwise UST and FST distances for each

marker, averaged and summarised with a multidimensional

scale plot. Kruskal’s stress (Venables and Ripley 2002)

was calculated to assess how well the multidimensional

scale plots matched the initial distances.

The relationship of the average genetic and geographic

distance matrices among samples was evaluated with a

Mantel test (Mantel 1967). The genetic divergence of the

Icelandic sample from the other populations was espe-

cially analysed along with how its divergence was de-

pendent on geographic distance. The geographic distance

of each population to the Icelandic population was cal-

culated as the average distance of different coordinates of

different samples by using the geosphere package in R

(Hijmans et al. 2014) and geographic grouping described

in Fig. 1.

The genetic structure, considering all the markers, was

further analysed by performing the clustering methods of a

discriminant analysis of principal components (DAPC),

implemented in the adegenet package (Jombart et al. 2010)

in R (R Development Core Team 2014), both using the

taxonomical affiliation as a prior and without any priors to

testing the genetic composition of each taxonomic entity.

Demographic analysis

Demographic changes of the different species and sub-

species were estimated by comparison to the expectation of

the sudden expansion model (Rogers and Harpending

1992). The deviation was tested with a parametric boot-

strap procedure implemented in ARLEQUIN 3.1 (Excoffier

et al. 2005). A ragged and erratic mismatch distribution can

result from a population which has been at equilibrium for

a long time, or alternatively by admixture of two separate

populations. A generalized non-linear least-squares ap-

proach in ARLEQUIN 3.1 (Excoffier et al. 2005) was used

to estimate the median (s parameter) of the distribution to

date the onset of the expansion. We used also ARLEQUIN

3.1 to perform Fu’s Fs (Fu 1997) and Tajima’s D (Tajima

1989) tests. Negative values of these statistics are expected

when populations have recently expanded from a recent

bottleneck (or from the effect of directional selection),

whereas positive values may reflect an admixture of two

lineages.

Results

Genetic variation

All markers were highly variable and similar variation was

in general observed within subspecies as within species

(Table 2). The nucleotide diversity (p) was lower in both

mitochondrial markers than in the four nuclear markers.

The CR p value was particularly high for the wintering

Icelandic population. AR was consistent across loci. The

J Ornithol

123

Table 2 Genetic variability indexes obtained from two mitochondrial and four nuclear loci of different taxa of redpoll

Loci CR COI ACO1 FIB7 GHR TGFb2 Average (nuDNA)

Fragment size (bp) 608 572 828 502 256 567

N

Taxon

C. flammea 140 138 80 90 142 130 110.5

C. f. islandica

Breeding 31 38 37 68 119 66 72.5

Wintering 34 57 – – – – –

C. f. rostrata 11 8 2 10 9 12 8.3

C. f. flammea 64 35 41 12 14 52 29.8

C. hornemanni 19 11 7 2 2 6 4.3

C. h. hornemanni 1 3 4 – 2 2 2.7

C. h. exilipes 18 8 3 2 – 4 3.0

C. cabaret 65 10 19 4 34 4 15.3

Overall 224 159 106 96 178 140 130

#h (S)

C. flammea 46 (32) 43 (35) 30 (52) 44 (44) 30 (33) 53 (49) 39.3 (44.5)

C. f. islandica

Breeding 11 (9) 15 (13) 13 (27) 34 (38) 30 (33) 29 (33) 26.5 (32.8)

Wintering 10 (10) 18 (18) – – – – –

C. f. rostrata 5 (10) 5 (3) 1 (0) 4 (12) 2 (2) 6 (12) 3.3 (6.5)

C. f. flammea 32 (27) 14 (13) 19 (45) 7 (10) 3 (2) 22 (22) 12.8 (19.8)

C. hornemanni 10 (11) 6 (6) 4 (11) 1 (0) 2 (1) 4 (7) 2.8 (4.8)

C. h. hornemanni 1 (1) 3 (2) 2 (1) – 2 (1) 1 (0) 1.7 (0.7)

C. h. exilipes 9 (10) 4 (5) 2 (10) 1 (0) – 3 (6) 1.5 (5.3)

C. cabaret 22 (21) 5 (5) 6 (18) 1 (0) 10 (11) 3 (3) 5 (8)

Overall 62 (37) 48 (38) 34 (55) 46 (44) 36 (38) 59 (57) 43.8 (48.5)

p (9100)

C. flammea 0.298 ± 0.002 0.202 ± 0.001 0.684 ± 0.004 1.307 ± 0.007 0.416 ± 0.003 0.621 ± 0.004 0.757

C. f. islandica

Breeding 0.200 ± 0.020 0.220 ± 0.020 0.583 ± 0.003 1.280 ± 0.007 0.464 ± 0.003 0.642 ± 0.004 0.742

Wintering 0.271 ± 0.002 0.214 ± 0.002 – – – – –

C. f. rostrata 0.190 ± 0.010 0.130 ± 0.020 0.000 1.560 ± 0.011 0.444 ± 0.002 0.917 ± 0.006 0.730

C. f. flammea 0.380 ± 0.020 0.176 ± 0.013 0.797 ± 0.004 0.854 ± 0.005 0.159 ± 0.002 0.488 ± 0.003 0.575

C. hornemanni 0.311 ± 0.002 0.216 ± 0.002 0.574 ± 0.004 – 0.391 ± 0.006 0.623 ± 0.004 0.529

C. h. hornemanni – 0.233 ± 0.002 1.000 ± 1.000 – 0.391 ± 0.006 0.000 0.464

C. h. exilipes 0.310 ± 0.030 0.219 ± 0.002 0.797 ± 0.006 – – 0.676 ± 0.005 0.737

C. cabaret 0.390 ± 0.030 0.202 ± 0.002 0.968 ± 0.006 – 0.441 ± 0.003 0.323 ± 0.003 0.577

Overall 0.344 ± 0.002 0.203 ± 0.001 0.703 ± 0.004 1.272 ± 0.007 0.420 ± 0.003 0.611 ± 0.003 0.752

H ± SD

C. flammea 0.879 ± 0.023 0.731 ± 0.042 0.805 ± 0.051 0.985 ± 0.008 0.555 ± 0.051 0.933 ± 0.021 0.820

C. f. islandica

Breeding 0.787 ± 0.071 0.787 ± 0.064 0.794 ± 0.073 0.984 ± 0.008 0.595 ± 0.054 0.937 ± 0.026 0.828

Wintering 0.825 ± 0.052 0.702 ± 0.068 – – – – –

C. f. rostrata 0.618 ± 0.164 0.786 ± 0.151 0.000 1.000 ± 0.177 0.222 ± 0.166 1.000 ± 0.096 0.555

C. f. flammea 0.947 ± 0.016 0.708 ± 0.086 0.797 ± 0.073 0.964 ± 0.077 0.385 ± 0.149 0.905 ± 0.039 0.763

C. hornemanni 0.854 ± 0.069 0.727 ± 0.144 0.900 ± 0.161 – 1.000 ± 0.500 0.867 ± 0.129 0.922

C. h. hornemanni 1.000 1.000 ± 0.272 1.000 ± 0.500 – 1.000 ± 0.500 0.000 0.667

C. h. exilipes 0.837 ± 0.075 0.643 ± 0.184 0.667 ± 0.314 – – 0.833 ± 0.222 0.750

J Ornithol

123

haplotype networks reflect the high genetic variation and

they were characterized by large admixture with no distinct

reciprocally monophyletic lineages, neither for the sub-

species nor the species (Fig. 2, Appendix 4 in ESM). Many

unique haplotypes were diverging from the main common

haplotype in all markers. Interestingly all species shared

the most frequent mitochondrial haplotypes, C1 (CR) and

D1 (COI). Two COI haplotypes (C5 and C8) were shared

by non-breeding C. f. islandica and C. cabaret from

Scotland. Similar admixture was observed in the network

for CR. A shared CR haplotype was observed in C. cabaret

and C. f. flammea from southern Scandinavia (D12). Aside

from the singletons, two haplotypes were observed only in

the mainland populations from Scandinavia and Russia

(D34 and D41).

Population divergence

The overall AMOVA revealed highly significant differ-

ences among subspecies, either based on pairwise or

Table 2 continued

Loci CR COI ACO1 FIB7 GHR TGFb2 Average (nuDNA)

Fragment size (bp) 608 572 828 502 256 567

C. cabaret 0.904 ± 0.018 0.667 ± 0.163 0.952 ± 0.096 – 0.748 ± 0.088 0.833 ± 0.222 0.844

Overall 0.899 ± 0.015 0.724 ± 0.040 0.837 ± 0.042 0.986 ± 0.007 0.590 ± 0.046 0.936 ± 0.019 0.837

AR

C. flammea 5.45 4.27 3.40 4.86 5.64 4.33 4.56

C. f. islandica

Breeding 4.14 4.66 3.33 5.28 3.69 4.44 4.19

Wintering 4.32 3.88 – – – – –

C. f. rostrata 4.52 4.52 – – 2.50 3.71 3.11

C. f. flammea 6.29 3.95 3.45 4.48 2.87 4.16 3.74

C. hornemanni 5.96 4.04 3.75 – – 4.29 4.02

C. h. hornemanni 5.96 4.04 – – – – 4.02

C. h. exilipes – – – –

C. cabaret 5.54 3.83 3.38 – 4.21 – 3.80

s

C. f. islandica 1.377 1.273 14.225 6.301 0.680 3.668 6.219

C. f. rostrata 2.004 0.957 – 11.758 – 2.945 7.352

C. f. flammea 2.273** 1.092 0 5.498 1.711 3.240 2.612

C. h. hornemanni – 1.629 – – – – –

C. h. exilipes 1.766 2.109 12.180 – – 6.412 9.296*

C. cabaret 2.625 1.461 1.289 – 0.969 2.475 1.578

Tajima’s D

C. f. islandica -1.430 -1.886* -0.430 -0.943 -2.434 -1.642* -1.362

C. f. rostrata -1.220 -1.448 0 1.984 0 -1.012 0.243

C. f. flammea -1.874* -2.178** -1.485* 0.512 -1.095 -1.761* -0.957*

C. h. hornemanni 0 0 – – 0 0 0

C. h. exilipes -1.278 -1.595* 0 0 – 1.662 0.554

C. cabaret -1.414 -1.388 -1.053 0 -2.002** 1.090 -0.491

Fu’s F

C. f. islandica -7.195 -13.105 -0.449 -24.446 -31.053 -25.075 -20.256

C. f. rostrata -1.684* -3.576 0 0.043 0 -2.299* -0.564

C. f. flammea -27.007 -13.692 -5.224* -4.086* -0.080 -26.249 -8.910**

C. h. hornemanni 0 -1.216 – – 0 0 0

C. h. exilipes -3.978** -0.785 3.635 0 – 1.099 1.578

C. cabaret -13.657 -1.896** -2.695* 0 -10.214 0.006 -3.226

N number of sequences, #h number of different haplotypes, S number of polymorphic sites, p nucleotide diversity, H haplotype diversity, AR

allelic richness (corrected for rarefaction), s median of the mismatch distribution

* 0.05 [ P [ 0.01; ** 0.01 [ P [ 0.001

J Ornithol

123

haplotype differences (FST range 0.057–0.074; P \ 0.001).

Similar but less differentiation was observed among sub-

species within species (FSC range 0.030–0.046;

P \ 0.002), which is mainly due to the differences within

C. flammea (Table 3). No significant differentiation was

found among the species. The output from the AMOVA on

COI was all non-significant. Based on the haplotype dif-

ferences, the other markers showed significant differences

among subspecies (range 0.006–0.225). Only the autoso-

mal marker TGFb2 presented a significant difference based

on haplotype frequencies among the species (FCT = 0.067,

P \ 0.01). Slightly larger differentiation was generally

obtained with FST than UST suggesting that the differen-

tiation is recent, with a lack of phylogenetic signal. The

main patterns of differentiation between the samples, based

on all markers, can be seen in the multidimensional scale

plot in Fig. 3 and Table 4. Despite lack of monophyly there

is some indication that the C. flammea subspecies group

together and are intermediate between the other two spe-

cies. C. cabaret is clearly different from the other samples,

all pairwise comparisons were large and significant with

the C. flammea subspecies (FST = 0.07–0.13,

UST = 0.04–0.10). C. f. islandica differed from the other

C. flammea subspecies when considering only the UST

values (C. f. flammea: 0.02, P \ 0.01 and C. f. rostrata:

0.05, P \ 0.05), the FST remained not significant. These

significant differentiations among the C. flammea sub-

species are consistent with the output of the AMOVAs

Table 3 Analyses of molecular variances of redpoll populations for all loci

Source of variation Loci Weighted

average

all nuclear

markers ? COI

Weighted

average

all nuclear

markers ? CR

ACOI GHR FIB7 TGFb2 COI CR

Among species (UCT) -0.0061 0.0028 -0.0147 0.0929 0.0082 0.0888 0.020 0.038

Among subspecies (UST) -0.0145 0.0125 0.0936** 0.1387*** 0.0169 0.0892*** 0.057*** 0.067***

Among subspecies

within species (USC)

-0.0084 0.0097 0.1067** 0.0505* 0.0087 0.0004 0.037** 0.030**

Among species (FCT) 0.0812 -0.0685 0.2204 0.0670** -0.0149 -0.0107 0.030 0.039

Among subspecies (FST) 0.1304*** 0.0065* 0.2251*** 0.0812** -0.0166 0.0414*** 0.072*** 0.074***

Among subspecies

within species (FSC)

0.0535* 0.0702** 0.0061** 0.0152* -0.0017 0.0515*** 0.035*** 0.046***

The variance is partitioned by the taxonomic classification to species and subspecies. Fixation indexes are given for pairwise distances (U) and

based on haplotypes frequencies (F). P values were obtained by 1000 permutations

# Negative variance components result from estimation error and can be interpreted as 0

* 0.05 [ P [ 0.01; ** 0.01 [ P [ 0.001; *** 0.001 [ P

C8

C5C1

C. cabaretC. hornemanniC. f. flammeaC. f. rostrataC. f. islandica

D12

D41

D1

D8

D9

D34

a b

Fig. 2 Unrooted median joining network of the mitochondrial

haplotypes within the redpoll complex, a cytochrome oxidase I

(COI) and b the control region (CR). The size of the pie charts refers

to the observed frequencies of haplotypes, the shadings to different

species and subspecies. The length of the branches are drawn in

proportion to the number of mutations between haplotypes. Homo-

plasic sites were omitted to reduce the multiple hits in the network

(see ‘‘Materials and methods’’). Haplotype names presented in the

figure are referred to in the text, see also Appendix 3 in ESM

J Ornithol

123

(Tables 3, 4). It should though be noted that the number of

specimens behind C. hornemanni and C. f. rostrata are low,

and thus the P values are only suggestive.

The overall differentiation based on pairwise differences

(UST) and haplotype differences (FST) did not show any

relation to geographic distance between the localities de-

fined in Fig. 1, tested by Mantel test with 1000 permuta-

tions. However the genetic deviation of C. f. islandica from

the other samples was associated with geographic distances

(nuclear markers ? COI; FST; t = 7.14; P \ 0.01 and

UST; t = 3.34, P = 0.04).

Three genetic clusters were observed with the DAPC

analysis of the multiple markers without any priors

(Fig. 4a), all clusters were found in each taxon except for

C. hornemanni for which the sample size was extremely

low. The proportion of each cluster did not differ between

the taxa, but C. f. islandica was about three to four times

more common in cluster 1 than C. f. flammea and C.

cabaret (Fig. 4b). A high proportion of the variance

(79.3 %) was though conserved according to the putative

taxonomic affiliation. C. hornemanni, although based on a

small sample size, is very differentiated from the other on

the first axis, and C. cabaret is positioned between the three

C. flammea subspecies on the first two axes but differs

clearly on the third axis; this pattern is consistent with the

results given by the pairwise genetic distances across all

markers (Fig. 4c; Table 4). The C. flammea subspecies

show small differentiation along the first two axes, C. f.

rostrata is between the two other C. flammea subspecies

and C. hornemanni on the first axis and a split between C. f.

islandica–C. f. rostrata versus C. f. flammea is observed on

the second axis.

Two individuals sampled in Iceland were assigned to C.

f. flammea and another to C. f. rostrata. Two individuals C.

f. rostrata from Greenland were assigned to C. f. islandica

and two C. f. flammea from Scandinavia were assigned to

C. f. islandica. Four C. cabaret were wrongly assigned; two

to C. f. islandica and two to C. f. flammea (Fig. 4d).

Demographic patterns

The medians of the mismatch distribution (s values) for the

mitochondrial markers varied from 1.38 to 2.63 for CR and

0.96 to 2.11 for COI and were generally lower than the

average medians across nuclear markers (range 1.58–9.30)

(Table 2). The variation in all markers and subspecies

followed the expectation of the sudden expansion model

and negative Tajima’s D and Fu’s F were also observed for

almost all markers in different species and subspecies but

were only significant over all markers for C. f. flammea

(Table 2).

Discussion

The variation in mitochondrial DNA and in nuclear

markers within the redpoll species is characterized by high

variation and lack of monophyletic support for the different

taxonomic units. Neither the different species (C. flammea,

C. hornemanni and C. cabaret) nor the different subspecies

within the species harbour monophyletic lineages. Never-

theless the partition of the genetic variation among sub-

species showed slight and significant differentiation

(FST = 0.07, UST = 0.06). C. f. islandica differed sig-

nificantly from the two other C. flammea subspecies; C. f.

rostrata (UST = 0.05) and C. f. flammea (UST = 0.02). The

largest deviation was found between C. cabaret and the

other subspecies of the other species. The multivariate

DAPC ordination provided further support to the genetic

divergence between the three species.

Although a significant differentiation was detected be-

tween the taxonomic units, it is though more like patterns

often observed among populations within species than

between species, and the divergence does thus clearly not

Table 4 Pairwise genetic distances between subspecies and species

within the redpoll complex

C. f. i. C. f. r. C. f. f. C. h. h. C. h. e. C. c.

C. f. i. 0.052 0.021 0.071 0.078 0.044

C. f. r. 0.076 0.076 0.242 0.061 0.060

C. f. f. 0.013 0.087 0.082 0.095 0.100

C. h. h. 0.151 0.361 0.144 0.121 0.262

C. h. e. 0.031 0.027 0.040 0.200 0.130

C. c. 0.065 0.125 0.066 0.276 0.071

The distances are based on averages of the two mtDNA markers and

four nuclear markers combined (FST, below the diagonal; UST, above

the diagonal). The significant results (P \ 0.05) are presented in bold

Fig. 3 Multidimensional scaling plot of the redpoll species and

subspecies, based on average pairwise genetic distances (UST) of two

mitochondrial markers (CR and COI) and four nuclear markers

(ACOI, GHR, FIB7 and TGFb2), Kruskal’s stress = 0.10

J Ornithol

123

Fig. 4 Discriminant analysis of principal components based on two

mitochondrial markers (control region and COI) and four nuclear

markers (ACOI, GHR, FIB7 and TGFb2). a Bayesian information

criterion according to different numbers of cluster. b A bar plot

representing the proportion of individuals from each subspecies

assigned to each of these clusters without any priors, samples size are

given in brackets. c A scatter plot of the three first PCs of the DAPC

with putative taxonomic affiliation as priors. d A bar plot representing

the assignment probability of individuals to the different subspecies

(the morphological attribution is defined on the y-axis); asterisk

represents assignment of individuals to another species or subspecies

J Ornithol

123

meet the different species criteria which have been sup-

posed on the basis of COI mtDNA (Hebert et al. 2004; Witt

et al. 2006). The main difference was observed between C.

cabaret and the other species, supporting the hypothesis of

Ottvall et al. (2002) that this species might have survived in

a different refugium during the last glaciation, than the

other redpoll species. Although a larger genetic differen-

tiation of C. cabaret from other species was observed when

taking the absolute genetic distances into account for the

mtDNA CR marker (average UST = 0.103, P \ 0.001 and

average FST = 0.073, P \ 0.001), indicating a historical

divergence, the averages across all markers were almost

identical (UST = 0.119, P \ 0.001, FST = 0.121,

P \ 0.001). No differentiation was observed for the mito-

chondrial marker COI among a small sample of Scandi-

navian and North American redpolls (Johnsen et al. 2010),

and between C. flammea and C. hornemanni from North

America (Kerr et al. 2007). Most of the genetic variation

across all markers was found within the subspecies and a

similar proportion (FSC = 0.05, P \ 0.001) of the genetic

variation in CR was explained by the subspecies within

species as described by Marthinsen et al. (2008) in analysis

of the three species (FST = 0.05, P \ 0.01). C. f. islandica

is genetically differentiated from C. f. rostrata

(UST = 0.052, P = 0.024), which is also morphologically

intermediate in size and colour between the two extremes,

C. cabaret and C. hornemanni.

The lack of differentiation based on the molecular var-

iation can reflect either introgression and incomplete lin-

eage sorting of the mtDNA which can be difficult to

distinguish (e.g. Holder et al. 2001) but all the evidence

points to a recent morphological diversification within the

redpoll complex, the genealogies for the different markers

are shallow and if introgression or admixture has occurred

it has been among closely related lineages. Our extension

of the previous mtDNA data analysis, by adding both se-

quences from C. f. islandica and new data on nuclear

markers, suggests that the redpoll complex is a single ge-

netic evolutionary unit which has experienced a sudden

expansion, supporting the previous conclusion by

Marthinsen et al. (2008). Further analysis of C. f. rostrata

and C. h. hornemanni, which were underrepresented in the

present study as in the previous study by Marthinsen et al.

(2008), may though add further insight into the evolution of

this species complex.

Origin of the subspecies C. f. islandica

Two possible hypotheses can explain the high molecular

variation within C. f. islandica: (1) introgression following

hybridisation between related species, or (2) incomplete

lineage sorting due to recent morphological divergence and

ancestral polymorphism. Williamson (1961) claimed that

the Arctic Redpolls C. hornemanni were breeding in Iceland

during an earlier and colder epoch, but as a result of an

amelioration of the climate a sub-Arctic form colonised

Iceland from southwest Iceland and C. f. islandica might

thus be a result of an admixture of Arctic and sub-Arctic

forms of redpoll. The explanation by Williamson (1961)

might also apply to the situation in the Palearctic distribution

of the redpoll regarding the lack of genetic differentiation.

During winter and spring, redpolls in Iceland might be

mixed with C. f. rostrata [FST = 0.022, n.s. (CR)] as shown

by increased genetic variation (CR: p = 0.271 ± 0.002)

compared to the summer (CR: p = 0.200 ± 0.020).

On the basis of the comparison of the mitochondrial and

nuclear variation, we can conclude that the redpoll com-

plex represents a single genetic evolutionary unit and has

undergone recent diversification and consequently shows

incomplete lineage sorting. There is no evidence of distinct

lineages which may have diverged in allopatric refugia

during the Pleistocene, and thus no phylogenetic evidence

for the different species nor the subspecies status of C. f.

islandica.

Acknowledgments We are grateful to the University of Iceland

research fund, the University of Iceland doctoral fund and the Palmi

Jonsson’s Fund for Nature Conservation for financial support. We are

also grateful to Yann Kolbeinsson for his help in the field. We thank

Ron Summers and Karl-Birger Strann for providing samples from

Scotland and Norway, respectively. We thank also employers at the

Natural History Museum of Oslo, Natural History Museum of Den-

mark and the Canadian Museum of Nature for sending samples from

their collections. We are also thankful to the editor and an anonymous

reviewer who provided helpful comments which improved the

manuscript. All the experiments and sampling procedures comply the

current laws in Iceland. We declare no conflict of interest related to

this study.

References

Amadon D (1949) The seventy-five per cent rule for subspecies.

Condor 51:250–258

Avise JC (2000) Phylogeography: the history and formation of

species. Harvard University Press, MA

Bandelt H-J, Forster P, Rohl A (1999) Median-joining networks for

inferring intraspecific phylogenies. Mol Biol Evol 16:37–48

Barrowclough GF (1982) Geographic variation, predictiveness, and

subspecies. Auk 99:601–603

Bird C (1935) The birds of Jan Mayen Island. Ibis 77:837–855

Burt DW, Paton IR (1991) Molecular cloning and primary structure of

the chicken transforming growth factor-b2 gene. DNA Cell Biol

10:723–734

Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0): an

integrated software package for population genetics data

analysis. Evol Bioinform Online 1:47–50

Fridolfsson A-K, Ellegren H (1999) A simple and universal method for

molecular sexing of non-ratite birds. J Avian Biol 30:116–121

Fu Y-X (1997) Statistical tests of neutrality of mutations against

population growth, hitchhiking and background selection.

Genetics 147:915–925

J Ornithol

123

Goudet J (2005) Hierfstat, a package for R to compute and test

hierarchical F-statistics. Mol Ecol Notes 5:184–186

Harrigan RJ, Mazza ME, Sorenson MD (2008) Computation vs.

cloning: evaluation of two methods for haplotype determination.

Mol Ecol Resour 8:1239–1248

Hebert PD, Stoeckle MY, Zemlak TS, Francis CM (2004) Identifi-

cation of birds through DNA barcodes. PLoS Biol 2:e312

Herremans M (1990) Taxonomy and evolution in redpolls Carduelis

flammea–hornemanni; a multivariate study of their biometry.

Ardea 78:441–458

Hijmans RJ, Williams E, Vennes C (2014) Package ‘‘geosphere’’:

spherical trigonometry. http://cran.r-project.org/web/packages/

geosphere/

Holder MT, Anderson JA, Holloway AK (2001) Difficulties in

detecting hybridization. Syst Biol 50:978–982

Johnsen A, Rindal E, Ericson PG, Zuccon D, Kerr KC, Stoeckle MY,

Lifjeld JT (2010) DNA barcoding of Scandinavian birds reveals

divergent lineages in trans-Atlantic species. J Ornithol

151:565–578

Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of

principal components: a new method for the analysis of

genetically structured populations. BMC Genet 11:94–109

Kerr KC, Stoeckle MY, Dove CJ, Weigt LA, Francis CM, Hebert PD

(2007) Comprehensive DNA barcode coverage of North

American birds. Mol Ecol Notes 7:535–543

Kerr KC, Birks SM, Kalyakin MV, Red’kin YA, Koblik EA, Hebert P

(2009) Filling the gap-COI barcode resolution in eastern

Palearctic birds. Front Zool 6:29–42

Kimball RT, Braun EL, Barker FK, Bowie RC, Braun MJ,

Chojnowski JL, Hackett SJ, Han K-L, Harshman J, Heimer-

Torres V (2009) A well-tested set of primers to amplify regions

spread across the avian genome. Mol Phylogenet Evol

50:654–660

Knox A (1988) The taxonomy of redpolls. Ardea 76:1–26

Lohman DJ, Prawiradilaga DM, Meier R (2009) Improved COI

barcoding primers for Southeast Asian perching birds (Aves:

Passeriformes). Mol Ecol Resour 9:37–40

Mantel N (1967) The detection of disease clustering and a generalized

regression approach. Cancer Res 27:209–220

Marthinsen G, Wennerberg L, Lifjeld JT (2008) Low support for

separate species within the redpoll complex (Carduelis flammea–

hornemanni–cabaret) from analyses of mtDNA and microsatel-

lite markers. Mol Phylogenet Evol 47:1005–1017

Mayr E (1982) Of what use are subspecies? Auk 99:593–595

Ottvall R, Bensch S, Walinder G, Lifjeld JT (2002) No evidence of

genetic differentiation between lesser redpolls Carduelis flam-

mea cabaret and common redpolls Carduelis f. flammea. Avian

Sci 2:237–244

Patten MA, Unitt P, Sheldon F (2002) Diagnosability versus mean

differences of Sage Sparrow subspecies. Auk 119:26–35

Phillimore AB, Winker K, Haig S (2010) Subspecies origination and

extinction in birds. Ornithol Monogr 67:42–53

Price T (2008) Speciation in birds. Roberts, Greenwood Village

Prychitko TM, Moore WS (1997) The utility of DNA sequences of an

intron from the b-fibrinogen gene in phylogenetic analysis of

woodpeckers (Aves: Picidae). Mol Phylogenet Evol 8:193–204

R Core Team (2014) R: a language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna.

http://www.R-project.org/

Ratnasingham S, Hebert PDN (2007) BOLD: the barcode of life data

system (http://www.barcodinglife.org). Mol Ecol Notes 7:355–364

Remsen J Jr, Winker K, Haig S (2010) Subspecies as a meaningful

taxonomic rank in avian classification. Ornithol Monogr

67:62–78

Rogers AR, Harpending H (1992) Population growth makes waves in

the distribution of pairwise genetic differences. Mol Biol Evol

9:552–569

Seutin G, Boag PT, Ratcliffe LM (1992) Plumage variability in

redpolls from Churchill, Manitoba. Auk 109:771–785

Seutin G, Boag PT, Ratcliffe LM (1993) Morphometric variability in

redpolls from Churchill, Manitoba. Auk 110:832–843

Sokal R, Rohlf F (2012) Biometry: the principles and practice of

statistics in biological research, 4th edn. Freeman, New York

Stephens M, Donnelly P (2003) A comparison of Bayesian methods

for haplotype reconstruction from population genotype data. Am

J Hum Genet 73:1162–1169

Tajima F (1989) Statistical method for testing the neutral mutation

hypothesis by DNA polymorphism. Genetics 123:585–595

Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S

(2011) MEGA5: molecular evolutionary genetics analysis using

maximum likelihood, evolutionary distance, and maximum

parsimony methods. Mol Biol Evol 28:2731–2739

Tanaka M, Maeda K, Okubo T, Nakashima K (1992) Double antenna

structure of chicken prolactin receptor deduced from the cDNA

sequence. Biochem Biophys Res Commun 188:490–496

Tarr C (1995) Primers for amplification and determination of

mitochondrial control-region sequences in oscine passerines.

Mol Ecol 4:527–530

Timmermann G (1938) Die Vogel Islands. Isafoldarprentsmidja,

Reykjavik

Ting C-T, Takahashi A, Wu C-I (2001) Incipient speciation by sexual

isolation in Drosophila: concurrent evolution at multiple loci.

Proc Natl Acad Sci 98:6709–6713

Troy DM (1983) Recaptures of redpolls: movements of an irruptive

species. J Field Ornithol 54:146–151

Troy DM (1985) A phenetic analysis of the redpolls Carduelis

flammea flammea and C. hornemanni exilipes. Auk 102:82–96

Venables WN, Ripley BD (2002) Modern applied statistics with S.

Springer, New York

Williamson K (1961) The taxonomy of the redpolls. Br Birds

54:238–241

Witt JD, Threloff DL, Hebert PD (2006) DNA barcoding reveals

extraordinary cryptic diversity in an amphipod genus: implica-

tions for desert spring conservation. Mol Ecol 15:3073–3082

J Ornithol

123