Molecular phylogeny of Carduelinae (Aves, Passeriformes, Fringillidae) proves polyphyletic origin of the genera Serinus and Carduelis and suggests redefined generic limits

https://doi.org/10.1016/j.ympev.2008.10.022Get rights and content

Abstract

Relationships of the 133 species of the subfamily Carduelinae (Fringillidae) are poorly resolved. For a more robust phylogenetic resolution, we sequenced two mitochondrial protein-coding genes (ATPase 6 and ND3), two nuclear introns (myoglobin intron 2 and transforming growth factor-β2 intron 5) and one nuclear protein-coding gene (c-mos) from 50 cardueline taxa representing especially the large genera Serinus and Carduelis. A total of 2934 bp obtained was subjected to maximum likelihood and Bayesian inference. Three of the five loci, as well as the combined dataset recovered the monophyly of the basal placement of Fringilla in the monophyletic Fringillidae, and the monophyly of the Carduelinae. While relationships within this group are moderately resolved by some individual gene trees (myoglobin and c-mos loci), high nodal support is provided in other individual gene trees and the combined tree. Among the well resolved terminal cardueline groups, Linurgus, Loxia and Pyrrhula are found to be monophyletic while genera Carpodacus, Carduelis and Serinus appear para- or polyphyletic. Within Serinus and Carduelis, the obtained phylogenetic structure corresponds well with the subdivisions suggested by H.E. Wolters, based on traditional methods. Thus, we support his generic subdivision (Ochrospiza, Dendrospiza and Crithagra for Serinus, and Chloris, Spinus, Sporagra, Pseudomitris, Acanthis and Linaria for Carduelis). Otherwise, we notice several cases of significant genetic divergence within traditional species suggesting incipient speciation in Linurgus olivaceus, Loxia curvirostra, Serinus mozambicus and Serinus burtoni. Some of these cases need a further phylogeographical study with a denser geographical sampling but for the case the most noteworthy, that of Serinus burtoni, we suggest a taxonomic change in this study.

Introduction

The finches (Aves: Passeriformes, Fringillidae) represent a diverse group of “seed-eating” passerine birds (Dickinson, 2003). While the highest generic diversity is in the Highlands of Asia, some genera are more widespread and represented in Africa and North and even South America. There is also one marked island radiation (the Drepanidini) in the Hawaii Islands (Dickinson, 2003). This group has recently been broadened with two genera of colorful birds previously referred to the Neotropic family Thraupidae, Euphonia and Chlorophonia, which are now recognized as representing deep branches in the Fringillidae family (Yuri and Mindell, 2002).

The Old World Finches are characterized primarily by their stout and conical bills, which have caused the taxonomic confusion with other finch-like, nine-primaried oscine birds (see Sibley and Ahlquist, 1990). Recent DNA-based studies have defined the Fringillidae in different ways, either as a very broad family encompassing ca. 1000 species of true finches (Fringillinae) as well as New World emberizine finches and their allies (Sibley and Monroe, 1990, Yuri and Mindell, 2002), or more narrowly as a family of 168 species, which can be divided into three subfamilies the Fringillinae, Carduelinae and Drepanidinae (Groth, 1998, Van der Meij et al., 2005). The last definition of the finches, which includes 168 “seed-eating” species, appears more structured and we prefer to follow it.

Within this group, the Carduelinae subfamily, which is the focal group of this study, comprises 133 species of serins, canaries, Old World seedeaters, goldfinches, siskins, rosefinches and others, representing a morphologically, behaviorally and ecologically quite complex group (Hall and Moreau, 1970, Dickinson, 2003, Fry and Keith, 2004). Two genera are particularly complex: Serinus with 38 species of which 31 are endemic to Africa, and Carduelis with 32 species broadly distributed across the Holartic biogeographic region and in South America (Sibley and Monroe, 1990, Dickinson, 2003). Relationships of the Carduelinae have been studied using morphological and behavioral characteristics (Beecher, 1953, Tordoff, 1954, Elzen and Nemeschkal, 1991), and Hall and Moreau (1970) tried to group African forms in superspecies and species groups. Several recent studies using DNA sequences data tried to resolve phylogenetic relationships between them. However, most of these studies included few species (Groth, 1998, Yuri and Mindell, 2002, Van der Meij et al., 2005), or focused on one genus, assuming a priori, that this represented one natural group (Arnaiz-Villena et al., 1998, Arnaiz-Villena et al., 1999, Ryan et al., 2004, Zamora et al., 2006). The molecular study by Arnaiz-Villena et al. (2001) used a fairly broad taxon sampling, but included only one Serinus species. All these data were combined in a comprehensive supertree by Jønsson and Fjeldså (2006), and most recently, Arnaiz-Villena et al. (2007a) combined data in a Bayesian analysis. However, this is based on a single molecular marker (cytochrome b), which has already been shown to give poor nodal support in this rapidly radiating group (Ryan et al., 2004). Some terminal branches in this analysis gain high support values but it is very difficult to evaluate whether this single gene tree reflects the species tree (Ericson et al., 2003). A more careful scrutiny is therefore needed.

In the present study, we focus on the terminal parts of the Carduelinae phylogeny, as we include representatives of most species groups of the large traditional genera Carduelis and Serinus, altogether 18 currently recognized species of the first genus and 20 of the second. Thus, the main aim of our study was to scrutinize the monophyly of these two genera and obtain a more robust phylogenetic hypothesis based on points of congruence between different gene trees.

For this, we use a total of 2934 bp including two mitochondrial protein-coding genes (ATPase 6 and ND3), two nuclear introns (myoglobin intron 2 and one transforming growth factor-β2 intron 5) and one nuclear protein-coding gene (c-mos). It is documented that the combination of several molecular markers brings a more robust phylogenetic resolution (Ericson et al., 2003, Nguembock et al., 2008a).

Section snippets

Taxon sampling

The sampling for this study focused primarily on the species-rich genera Carduelis and Serinus and putative related genera within the Carduelinae. We used tissue from fresh specimens as well as a few Genbank sequences. Fresh tissues came from tissue collection of the ZMUC and MNHN, and others were collected in the Cameroon Volcanic Line. For the species taxonomy, we followed Dickinson (2003), which uses a moderate level of generic splitting. We included eight of the 24 potential genera listed

Individual and concatenated data sets

We obtained between 443 (Serinus mozambicus ZMUC 132570) and 684 bp (majority of samples) for the ATPase 6 gene. MrMODELTEST selected the GTR + Г + I as the best-fit model. The parameters estimates are detailed in the Table 3. Both methods supported the same topology which taken as a whole received high bootstrap and posterior probability except a few nodes for which the support was not strong (Fig. 1). For the ND3 gene, sequence lengths ranged from 211 (Fringilla montifringilla MNHN2000-1646) to 351

Monophyly of Carduelinae and the deeper branching within the group

Several molecular studies (Groth, 1998, Yuri and Mindell, 2002, Van der Meij et al., 2005, Arnaiz-Villena et al., 2007a) already recovered the monophyly of the subfamily Carduelinae but these studies included very few carduelines taxa, or used only one genetic marker. In the study by Arnaiz-villena et al. (2007a), most of the deeper branches received low nodal support, except for the group including all members of Carduelis and Serinus, and some small genera, which gained 99% posterior

Acknowledgements

We are grateful to the various researchers and institutions that caught and provided several samples for our study, LSUMNS, UMMZ, ZMUC and all staff of WWF Cameroon (Nyassosso-Cameroon). We thank Eric Pasquet of the National Museum of the Natural History of Paris who provided several cardueline samples. In addition, we wish to thank J. Lambourdière and C. Bonillo for their help during laboratory work and several anonymous referees for their comments on an earlier version of this manuscript. We

References (47)

  • A. Arnaiz-Villena et al.

    Phylogeography of crossbills, bullfinches, grosbeaks and rosefinches

    Cellular and Molecular Life Sciences

    (2001)
  • A. Arnaiz-Villena et al.

    Bayesian phylogeny of Fringillinae birds: status of the singular African oriole finch Linurgus olivaceus and evolution and heterogeneity of the Genus Carpodagus

    Acta Zoologica Sinica

    (2007)
  • A. Arnaiz-Villena et al.

    Evolution of the major histocompatibility complex class I genes in Serinus canaria from the Canary Islands is different from that of Asia and African continental Serinus species

    Journal of Ornithology

    (2007)
  • W.J. Beecher

    A phylogeny of the oscines

    Auk

    (1953)
  • R.T. Chesser

    Molecular systematics of the rhinocryptic genus Pteroptochos

    Condor

    (1999)
  • A. Cooper et al.

    Mass survival of birds across the cretaceous-tertiary boundary: molecular evidence

    Science

    (1997)
  • E.C. Dickinson

    The Howard and Moore Complete Checklist of the Birds of the World

    (2003)
  • J.R. Eberhard et al.

    Phylogeny and biogeography of the Amazona ochrocephala (Aves: Ptsittacidae) complex

    Auk

    (2004)
  • R. van den Elzen et al.

    Radiation in african canaries (Carduelinae): a comparaison of different classificatory approaches

    Acta XX, Congressus Internationalus Ornithologici

    (1991)
  • P.G.P. Ericson et al.

    Evolution, biogeography, and patterns of diversification in passerine birds

    Journal of Avian Biology

    (2003)
  • J. Felsenstein

    Confidence limits on phylogenies: an approach using the bootstrap

    Evolution

    (1985)
  • C.H. Fry et al.
    (2004)
  • S. Guindon et al.

    A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood

    Systematic Biology

    (2003)
  • Cited by (32)

    • The phylogenetic relationships and generic limits of finches (Fringillidae)

      2012, Molecular Phylogenetics and Evolution
      Citation Excerpt :

      The relationships among these clades are not fully resolved, with conflicting evidence provided by the nuclear and mitochondrial genomes (Fig. 2). Incongruent signals from different genes were observed also by Nguembock et al. (2009), but even using the same gene, cytochrome b, changes in taxa included in the analysis resulted in quite different topologies (Arnaiz-Villena et al., 1998, 1999, 2001; van den Elzen et al., 2001; Ryan et al., 2004; Yang et al., 2006). Despite these differences, some common patterns emerge.

    View all citing articles on Scopus
    View full text