and monophyly of the group seems to be evident, since
both the bootstrap and the CP of the node are signifi-
cantly high (figs 2, 3). Therefore, Pinicola is not related
to Loxia or Carpodacus as previously thought [30] and
probably shares a common antecessor with bullfinches
(fig. 2). Pine grosbeak would be the one Pyrrhula finch,
found in North America, and showing a holoarctic distri-
bution.
Pine grosbeak has been grouped with the red-headed
rosefinch (Pinicola subhimachala) based on phenotypic
characters in the genus Pinicola (only these 2 species).
Both are larger (length ca 22 cm) than Pyrrhula and Car-
podacus finches. However, the genus Pinicola might be
split between Pyrrhula (pine grosbeak) and Carpodacus
(red-headed rosefinch), the latter sharing many pheno-
typic characteristics with Carpodacus (rosefinches) ex-
cept for the much bigger size (length 20 cm). Size should
not be a criterion for including the red-headed rosefinch
within the genus Carpodacus, since Haematospiza sipahi
(19 cm in length) is a sister species of Carpodacus ery-
thrinus roseatus (figs 2, 3). The phenotypically defined
subspecies Pyrrhula pyr. cinerea and griseiventris are
concordant with the molecular subspecies status sug-
gested in this work (see fig. 2). The time of divergence of
the Pyrrhula species (11 MYA) seems to be more recent
than that of grosbeaks and rosefinches but earlier than the
crossbill-redpoll divergence time.
3) Grosbeaks (genus Coccothraustes, 1 species; genus
Eophona, 2 species; genus Mycerobas, 4 species)
Coccothraustes coccothraustes (hawfinch) was phyloge-
netically unrelated to any of the tested Carduelini groups
[30]. To test hawfinch relatedness to the other groups
newly studied in the present work, a 306-nucleotide-long
cyt b sequence was retrieved from GenBank [26], aligned
with the BLAST program, and NJ, UPGMA, and maxi-
mum-parsimony trees were constructed. Hawfinch re-
mained as an isolated branch between the Pyrrhula and
6
A. Arnaiz-Villena et al.
Phylogeography of Carduelini birds
Figure 3. (a). Maximum unweighted parsimony tree. A Heuristic search was used (PAUP). Consistency and retention indexes were 0.42
and 0.53, respectively. The majority-rule bootstrap consensus tree based on 924 bases of mt cyt b genes is shown. The transition/transver-
sion ratio was 2:1; the observed 5:1 ratio made no differences. Parsimony bootstrap analysis was done with 1000 replications, and values
(in percent) are shown above branches. Parsimony was used unweighted (first, second, and third nucleotide) because weighting is only rec-
ommended for greater amounts of evolutionary divergence, which are not expected among the relatively closely related species analyzed
[21]. The number of events is shown underlined below branches. Bird species used are detailed in table 1. (b) Neighbor-joining bootstrap
tree (1000 replications) based on 924 bases of cyt b genes. Bootstrap values are shown above branches; branch lengths (¥ 1000) are shown
underlined below branches. The evolutionary model used was ‘the minimum evolution.’The transition /transversion ratio was 2:1. Distance
matrices were calculated based on maximum-likelihood analysis [14]. Bird species used are detailed in table 1.
A
B
