MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 30 (2004) 386–394 www.elsevier.com/locate/ympev

Nuclear and mitochondrial DNA evidence of polyphyly in the avian superfamily Muscicapoidea

Gary Voelker* and Garth M. Spellman

Barrick Museum of Natural History, Box 454012, University of Nevada Las Vegas, 4505 Maryland Parkway, Las Vegas, NV 89154, USA

Received 26 November 2002; revised 7 May 2003

Abstract

Nucleotide sequences of the nuclear c-mos gene and the mitochondrial cytochrome b and ND2 genes were used to assess the monophyly of Sibley and MonroeÕs [Distribution and Taxonomy of Birds of the World, Yale University Press, New Haven, 1990] Muscicapoidea superfamily. The relationships and monophyly of major lineages within the superfamily, as well as genera mem- bership in major lineages was also assessed. Analyses suggest that Bombycillidae is not a part of Muscicapoidea, and there is strongly supported evidence to suggest that Turdinae is not part of the Muscicapidae, but is instead sister to a Sturnidae + Cinclidae clade. This clade is in turn sister to Muscicapidae (Muscicapini + Saxicolini). Of the 49 Turdinae and Muscicapidae genera that we included in our analyses, 10 (20%) are shown to be misclassified to subfamily or tribe. Our results place one current Saxicolini genus in Turdinae, two Saxicolini genera in Muscicapini, and five Turdinae and two Muscicapini genera in Saxicolini; these relationships are supported with 100% Bayesian support. Our analyses suggest that c-mos was only marginally useful in resolving these ‘‘deep’’ phylogenetic relationships. Ó 2003 Published by Elsevier Science (USA).

Keywords: Bombycillidae; Cinclidae; Molecular systematics; Muscicapidae; Muscicapoidea; Sibley and Ahlquist; Sturnidae

1. Introduction been elevated to family status, leaving only the Old World groups Saxicolini and Muscicapini as elements of The avian family Muscicapidae has had a long hi- the Muscicapidae. story of revision. One generally consistent treatment Many of the proposed placements of groups in and throughout these revisions has been the identification of around the Muscicapidae relied on evidence of shared a core group, consisting of the Turdinae (thrushes), features such as jaw musculature, juvenal plumages, Muscicapinae (chats (Saxicolini) and muscicapine fly- tarsal scutellation, and the presence or absence of rictal catchers (Muscicapini)) and sylviine warblers (e.g., bristles (summarized in Sibley and Ahlquist, 1990). AOU, 1983; Hartert, 1910; Mayr and Amadon, 1951; However, the validity of using some of these features as Ripley, 1952). However, Muscicapidae has at various taxonomic characters has been questioned (e.g., Sibley times included, among others, such morphologically and Ahlquist, 1990) due, for example, to convergence on diverse groups as the Timaliinae (babblers), Polioptili- ecotypes or to characters being tied to foraging habits. nae (gnatcatchers), Malurinae (fairy wrens), Sylviinae Further, a number of these characters have a chaotic (Old World flycatchers), and Cinclidae (dippers) (e.g., distribution across taxa such that features attributed to Hartert, 1910; Mayr and Amadon, 1951; Storer, 1971). families generally offer no clear guide to relationships At the other extreme, under the current treatment by the (Sibley and Ahlquist, 1990). Either the characters are not American OrnithologistsÕ Union (1998), Turdinae has found in all members of a family, or the characters are not unique to single families (Sibley and Ahlquist, 1990). It was not until the DNA–DNA hybridization work * Corresponding author. Present address: Department of Biology, University of Memphis, 3700 Walker Avenue, Memphis, TN 38152, of Sibley and Ahlquist (1990), that a sweeping study USA. of passerine family relationships was conducted, E-mail address: [email protected] (G. Voelker). and an expansive phylogenetic hypothesis of ‘‘true’’

1055-7903/$ - see front matter Ó 2003 Published by Elsevier Science (USA). doi:10.1016/S1055-7903(03)00191-X G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 387 membership in Muscicapidae existed. The result of this subunit 2 (ND2) and 998 bp of cytochrome b (2643 bp work was the formation of a Muscicapoidea superfam- total) from 23 taxa representing all Muscicapoidea ily. Included in this superfamily were Muscicapidae families, as well as from 7 Sylvioidea taxa and 7 Passe- (Turdinae + Muscicapinae), Bombycillidae (waxwings roidea taxa (see Appendix A). We also included for and allies), Cinclidae (dippers), and Sturnidae (starlings these taxa a 604 bp segment of the nuclear c-mos gene, to (Sturnini) and thrashers (Mimini)). The explicit set of add an independent (from the single linkage unit relationships within Muscicapoidea proposed by Sibley mtDNA genes) data partition. These data have been and Ahlquist (1990) suggests a sister relationship be- submitted to GenBank under the following accessions: tween Turdinae and Muscicapinae (Saxicolini + Musci- submission in progress. capini), with Sturnidae, Cinclidae, and Bombycillidae To further assess the monophyly of Muscicapidae being progressively more distant. and to assess the monophyly of the major subclades Of the families included in Muscicapoidea by Sibley Turdinae, Muscicapini, and Saxicolini we increased our and Ahlquist (1990), only the Bombycillidae had never, sampling to include 49 genera from those groups (see to our knowledge, previously been considered part of Appendix A). This sampling included 21 of 24 pur- ‘‘Muscicapidae’’ in taxonomic treatments. And in fact, ported Turdinae genera, 9 of 17 presumed Muscicapini due to conflicting evidence in their results Sibley and genera, and 19 of 31 genera currently assigned to Sax- Ahlquist were ultimately unconvinced of their decision icolini (see Appendix A). We also included as outgroups to place Bombycillidae in Muscicapoidea, rather than in the seven Sylvioidea and seven Passeroidea taxa dis- either Sylvioidea or Passeroidea (1990; p. 630). Recent cussed above. We analyzed sequence data from the ND2 evidence from nuclear genes suggests that Bombycillidae (1041 bp) and cytochrome b (1000 bp) genes for this set is not in fact part of Muscicapoidea (Barker et al., 2001). of taxa. These additional muscicapid data are part of Confusion over relationships in Muscicapoidea is not on-going studies of inter- and intra-generic studies of limited to these higher taxonomic categories. There has Muscicapoidea relationships, and will be submitted to been a long history of revisions dealing with the true GenBank at the conclusion of those studies. Requests membership of genera in either the Turdinae, Saxicolini, for these data should be addressed to the senior author. or Muscicapini, validity of genera within those groups, as well as revisions dealing with species membership in 2.2. Sequence data given genera (e.g., Dowsett and Dowsett-Lemaire, 1993; Orenstein, 1979; Ripley, 1952, 1962; Vaurie, 1953, We extracted total genomic DNA from all tissue and 1955a,b). For example, Monticola is usually placed near blood samples using a Qiaquick tissue extraction kit Turdus in the true thrushes (Turdinae), but some au- (Qiagen; see Appendix A for sample types). The cyto- thors place Monticola in Saxicolini due to the chat-like chrome b gene was amplified using the primers L14841 behavior of several species (Vaurie, 1955a). (Kocher et al., 1989) and H4A (Harshman, 1996) and In this paper, we use a combination of nuclear and sequenced using the primers L14841 and H15299 (Ko- mitochondrial genes to explore several issues relating to cher et al., 1989); B3, B4, and B5 (Lanyon, 1994); the superfamily Muscicapoidea. First, we seek to test the L15114, L15609, and H15547 (Edwards et al., 1991); monophyly of Muscicapoidea. Second, we seek to test and H4A. We amplified the ND2 gene using the primers the family relationships proposed by Sibley and Ahl- L5215 (Hackett, 1996) and H6313 (Johnson and quist (1990), with a main interest in determining whether Sorenson, 1998) and sequenced using L5215, L5758, Muscicapidae (Turdinae, Saxicolini, and Muscicapini) is H5776, and H5578 (Hackett, 1996; Johnson and monophyletic. Finally, we seek to determine whether Sorenson, 1998); L5758.2 (Voelker, 2002); and H6313. each subgrouping within Muscicapidae is monophyletic The nuclear c-mos gene was amplified and sequenced with respect to the taxonomy proposed by Sibley and using primers 944 and 1550 (Lovette and Birmingham, Monroe (1990). 2000). All fragments were amplified in 50 ll reactions under the following conditions: denaturation at 94 °C, followed by 40 cycles of 94 °C for 30 s, 54 °C for 45 s, 2. Materials and methods and 72 °C for 2 min. This was followed by a 10 min extension at 72 and 4 °C soak. Products were purified 2.1. Specimens using a Qiagen PCR Purification Kit following manu- facturerÕs protocols. We performed 20 ll BigDye (ABI) Our sampling schemes were generally based on results sequencing reactions using 20–40 ng of purified and from the DNA–DNA hybridization studies of Sibley concentrated PCR product and following standard ABI and Ahlquist (1990) and the taxonomy of Sibley and protocols. Sequencing reactions were purified using Monroe (1990). To assess the monophyly of, and family Centrisep columns following manufacturerÕs protocols, and subfamily relationships within, Muscicapoidea, we dried in a centrivap concentrator, and run out on a used 1041 base pairs (bp) of NADH dehydrogenase Long Ranger (BMA) acrylamide gel with an ABI 377 388 G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 automated sequencer. Full complementary strands of analyses converged upon the same topology, and we each gene were unambiguously aligned using Sequen- therefore combined the 1750 saved trees from each cher 3.1 (GeneCodes). analysis to obtain posterior probabilities for nodes. In a ML bootstrap analysis, support for nodes was 2.3. Phylogenetic analysis obtained with searches for each of 100 pseudoreplicates using SPR branch swapping (which is computationally In our analysis of the monophyly of, and major clade faster than TBR branch swapping) on starting trees relationships within, Muscicapoidea, we conducted a obtained by NJ, with each replicate being limited to 250 successive approximations maximum likelihood (ML) rearrangements. In parsimony bootstrap analysis, sup- analysis in PAUP* (Swofford, 1999). We began by con- port for nodes was obtained using 100 pseudoreplicates, structing a neighbor-joining (NJ) tree with uncorrected each with 10 addition-sequence replicates. We used TBR distances and estimated values for the GTR + I + C branch swapping and weighted most codon positions by model from this tree. The GTR + I + C model was transition–transversion ratio values estimated from a NJ identified as the best fit model to our data using both the tree. The exceptions were cytochrome b and ND2 third LRT and AIC criteria in MODELTEST (Posada and positions, which due to extremely high values were in- Crandall, 1998). We then used these estimated values to stead weighted with tranversions 10 times transitions. conduct a maximum likelihood search, using the NJ tree For our sampling-intensive analysis of relationships as a starting tree. We ran this search until >8000 rear- within Muscicapidae, we used Bayesian inference to rangements (several days worth of rearrangements) had generate trees and measures of nodal support. Here been performed without a change in likelihood score (i.e., again we used the GTR + I + C model of evolution, ran after the score had reached a plateau), assuming that three separate analyses for 200,000 generations, sam- additional rearrangements would not result in further pling every 1000 generations, and combined 1750 trees change. We then re-estimated GTR + I + C values and from each analysis. began a second ML search using the first ML tree as a starting tree. This second search was also run until >8000 rearrangements had been performed without a change in 3. Results likelihood score. Because there was no difference in topology between the first and second ML trees, we 3.1. Sequence characteristics accepted this tree as our best estimate of relationships. Support for our estimate of relationships was con- C-mos accounted for 16% of the variable sites in our ducted using Bayesian inference (MRBAYES 2.01; reduced sampling scheme data set (11% parsimony in- Huelsenbeck and Ronquist, 2001), and ML and Parsi- formative), with 35% of sites within c-mos showing mony bootstrap analyses (PAUP*). In MRBAYES we variation (20% parsimony informative), whereas 61% used the GTR + I + C model and ran three separate (53%) and 49% (40%) of ND2 and cytochrome b sites analyses (with four chains per generation run simulta- were variable, respectively (Table 1). Within c-mos we neously) for 200,000 generations, sampling every 1000 found no evidence of synapomorphic indels that define generations. Each analysis began with a random tree. major clades. After each analysis, we plotted scores to determine the Both ND2 and cytochrome b show a deficiency of G point at which stationarity was reached, and sampled nucleotides that is common for these genes, whereas c-mos 1750 trees after this point. Each of our three separate shows a slight deficiency in A nucleotides (Table 1).

Table 1 Sequence characteristics of the c-mos, ND2, and cytochrome b (cyt b) genes for the 37 taxa included in the combined analysis of those gene regions Gene region c-mos ND2 cyt b Number of bases 604 1041 998 Number of variable bases (%) 209 (35) 638 (61) 489 (49) Number of parsimony informative bases 123 556 398 % A nucleotides 0.199 0.304 0.276 % C 0.318 0.346 0.345 % G 0.253 0.115 0.138 % T 0.238 0.235 0.241 % A nucleotide range across taxa 0.177–0.220 0.281–0.328 0.245–0.301 % C nucleotide range across taxa 0.306–0.327 0.283–0.391 0.319–0.379 % G nucleotide range across taxa 0.238–0.275 0.100–0.132 0.127–0.149 % T nucleotide range across taxa 0.215–0.242 0.205–0.279 0.226–0.264 G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 389

Overall, c-mos possesses a much higher GC content than Sibley and Ahlquist (1990, Fig. 1). Our maximum like- does either of the other genes (Table 1). Despite these lihood successive approximations analysis for the com- differences, the range of nucleotide content does not vary bined c-mos, ND2, and cytochrome b data yielded a widely across taxa, for any of the three gene regions, single best estimate of relationships (Fig. 2). This tree suggesting nucleotide bias may be unlikely to have an differs from the Muscicapoidea relationships proposed affect on phylogenetic analyses. Indeed, base composi- by Sibley and Ahlquist (1990) in three ways. First, tional heterogeneity across genes was not significant Muscicapoidea is polyphyletic in that Bombycillidae 2 across all taxa and all sites (v108 ¼ 91:029, P ¼ 0:880; and the remaining Muscicapoidea families were not each implemented in PAUP*, Swofford, 1999). However, base others closest relatives. Instead, the genus Polioptila compositional heterogeneity was significant when only (gnatcatchers, superfamily Sylvioidea) is placed as the 2 informative sites were considered (v108 ¼ 222:112, sister to Muscicapoidea (minus Bombycillidae). P 6 0:001), and thus may have an impact on our novel Second, Cinclidae is not sister to the remaining results with respect to some relationships within Musci- Muscicapoidea families Sturnidae and Muscicapidae, capoidea (see below). To test this idea, we excluded all nor is Sturnidae sister to Muscicapidae. Instead, Cin- outgroups except Polioptila, which is identified as the clidae and Sturnidae are sisters and are most closely closest outgroup to Muscicapoidea (which excludes related to Turdinae. Finally, the family Muscicapidae is Bombycillidae) in our analyses (see below), and again polyphyletic in that Turdinae and Muscicapidae are not analyzed base compositional heterogeneity. These results closest relatives. Turdinae is instead placed as sister to for heterogeneity in informative sites suggest that it may Sturnidae + Cinclidae, and together these three clades not be a factor for concern in either our limited sampling are sister to the major Muscicapidae lineages Muscica- 2 scheme (v63 ¼ 67:475, P ¼ 0:327) or in our increased pini + Saxicolini. The Shimodaira–Hasegawa test (as 2 sampling scheme (v282 ¼ 254:870, P ¼ 0:134). implemented in PAUP*) suggests that the topology we The range of sequence divergence (uncorrected) be- recovered (Figs. 1 and 2) is a significantly better estimate tween genera ranges from 2.8% between Thraupis and of relationships than is the Sibley and Ahlquist topology Passer to 9.9% between Zosterops and Bradornis for c- (P < 0:005), for this data set. mos. Sequence divergence in ND2 ranged from 11.9% Support for the relationships we recovered is mixed. between Melaenornis and Bradornis to 25.7% between Bayesian analysis, and maximum likelihood and parsi- Lamprotornis and Bombycilla, while divergences in cy- mony bootstrap analyses do not support Polioptila as tochrome b ranged from 7.8% between Turdus and sister to Muscicapoidea (minus Bombycillidae). Bayes- Cichlherminia to 19.0% between Sylvia and Bubalornis. ian analysis strongly supports (99%) the placement of these three groups as a polytomy, while ML bootstrap 3.2. Phylogenetic analyses analysis includes three other genera in this polytomy (Poecile, Trococercus, and Regulus) (not shown), all of Our results ultimately suggest a different set of which are, like Polioptila, members of the superfamily Muscicapoidea relationships than that proposed by Sylvioidea; this polytomy has ML bootstrap support of

Fig. 1. A comparison of the major Muscicapoidea clades and clade relationships suggested by Sibley and Ahlquist (left) and this study (right). Note that our results suggest that Bombycillidae is not part of Muscicapoidea, that Turdinae is not part of the Muscicapidae, and that Turdinae, Sax- icolini, and Muscicapini are not monophyletic as currently defined. 390 G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394

Fig. 2. Maximum likelihood tree of Muscicapoidea relationships, based on combined c-mos, cytochrome b, and ND2 sequence data. Numbers at the base of major clades reflect support for that clade based on, from top to bottom, Bayesian inference, maximum likelihood bootstrap analysis, and parsimony bootstrap analysis. See Section 2 for details of each analysis. The single asterisk denotes Bayesian support of 99% for a polytomy consisting of Bombycillidae, Polioptila, and Muscicapoidea. All nodes at and deeper in the phylogeny than the double asterisk are supported by >95% Bayesian support.

61%. Parsimony bootstrap analysis also places these Bayesian support for such a relationship approaches four Sylvioidea genera in a polytomy with Muscicapi- significance (93%; Fig. 2). Lack of support therefore dae, along with a clade consisting of the Sylvioidea suggests a polytomy consisting of Muscicapinae, Turdi- genera Sylvia, Turdoides, and Zosterops (not shown); nae, and Cinclidae + Sturnidae. All measures of support this polytomy has 68% bootstrap support. strongly uphold monophyly of the Turdinae, Saxicolini, All methods of assessing support for clades strongly and Muscicapini clades, with respect to the taxa that we suggest that Muscicapoidea (minus Bombycillidae) is a have included at this level of analysis (but see below). valid group (Fig. 2). A sister relationship between Cin- Results of our increased taxon sampling analysis, clidae and Sturnidae is strongly supported by Bayesian using only cytochrome b and ND2, confirm that inference and ML bootstrap support, but is only mar- Bombycillidae is not part of Muscicapoidea, as Poliop- ginally supported by parsimony bootstrap analysis. tila + Salpinctes are included in a polytomy along with Support for a sister relationship between Turdinae Bombycillidae and the remaining Muscicapoidea lin- and Cinclidae + Sturnidae is not convincing, although eages (Fig. 3). G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 391

Fig. 3. Phylogeny of Muscicapoidea relationships, based on cytochrome b and ND2 sequence data, and increased sampling of genera from Turdinae and Muscicapidae. Numbers at the base of major clades reflect support for that clade based on Bayesian inference.

Muscicapoidea, minus Bombycillidae, is a strongly Results of our increased taxon sampling further show supported group, receiving 100% Bayesian support. The that the major muscicapid lineages Turdinae, Saxicolini, structure of relationships within Muscicapoidea sup- and Muscicapini are each paraphyletic as currently rec- ports the Muscicapinae lineage being sister to a strongly ognized. Turdinae includes the genus Cochoa, currently a supported clade containing Sturnidae, Cinclidae, and muscicapine, while supposed turdine genera Brachyp- Turdinae (Fig. 3). This result holds when using only teryx, Alethe, Myiophonus, Pseudocossyphus, and Mon- Polioptila as an outgroup (not shown), to control for ticola are here shown to be placed within Saxicolini, with potential effects of base compositional heterogeneity (see 100% Bayesian support (Fig. 3). Saxicolini further in- above). Within the Sturnidae + Cinclidae + Turdinae cludes Cyornis and Ficedula, each previously considered clade, a sister relationship between Sturnidae and Cin- Muscicapini genera, while Muscicapini includes Cerco- clidae is inferred, but not supported (just 73% Bayesian trichas and Copsychus, each previously placed in Saxico- support). Muscicapidae (Turdinae + Muscicapinae) is lini. The placement of these genera in alternative lineages clearly paraphyletic based on these results. are supported with 100% Bayesian support (Fig. 3). 392 G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394

4. Discussion such that Bombycillidae is removed, and placed as a family incertae sedis within, or near, Sylvioidea. Note 4.1. Did c-mos help in resolving deep relationships? that Sibley and Ahlquist (1990) were not convinced of their decision to place Bombycillidae in Muscicapoidea; Many studies seeking to address ‘‘deep’’ relationships their rather equivocal results also supported placement among clades often include a nuclear gene in analyses. of Bombycillidae in Sylvioidea. The commonly held notion is that because nuclear genes While our results, and those of Barker et al. clearly tend to evolve slowly, as compared to mtDNA, they do support removing Bombycillidae from Muscicapoidea, not experience the same level of saturation and thus will neither study included sufficient sampling of Sylvioidea not confound analyses of deeper splits in the same taxa to better establish true bombycillid relationships. manner in which high levels of sequence divergence can Furthermore, given the ladder-like placement of the affect mtDNA-based analyses. Therefore, nuclear genes outgroups in our nuclear plus mitochondrial DNA should be much more effective at resolving relationships analyses, it is likely that Sylvioidea and Passeroidea may among older divergences. The lower percentage of var- also prove to be polyphyletic (see also Barker et al., iable sites and the lower range of sequence divergence 2001; Groth, 1998; Sheldon and Gill, 1996), making a found in our c-mos data set, as compared to percentages more conclusive statement of bombycillid affinities im- and ranges from the mtDNA genes, does suggest that possible at this time. c-mos is indeed evolving more slowly. However, our analyses suggest that the nuclear c-mos 4.3. Relationships among, and membership in, major gene contributed little if anything to the overall analysis Muscicapoidea clades of relationships in Muscicapoidea. Indeed c-mos when analyzed alone (using Bayesian inference; not shown) Our results strongly support a different view of rela- provided no support for the sister relationship between tionships among major Muscicapoidea clades than that Turdinae and Cinclidae + Sturnidae that was reflected in proposed by Sibley and Ahlquist (1990). Our nuclear analyses of combined nuclear and mtDNA data, and in plus mitochondrial analysis clearly indicates a clade analyses of mtDNA alone. C-mos failed to recover sig- consisting of Turdinae, Sturnidae, and Cinclidae, al- nificant support for the Muscicapini clade, and also though the relationships among the three groups is un- failed to support the strong Muscicapoidea + Bomby- clear. The Turdinae + Sturnidae + Cinclidae clade is cillidae + Polioptila relationship that we show here. Only clearly sister to Muscicapinae. Our expanded mito- in supporting a sister relationship between Cinclidae and chondrial analysis unequivocally supports a sister rela- Sturnidae did the inclusion of c-mos provide resolution tionship between Sturnidae and Cinclidae, and also over using mitochondrial genes alone. Furthermore, we supports a Turdinae + Sturnidae + Cinclidae clade being found no evidence of synapomorphic insertions or de- sister to Muscicapinae. A close Turdinae–Cinclidae re- letions in c-mos which would serve to identify major lationship was proposed by Stejneger (1905), and is clades; such taxonomically informative indels have been supported by other molecular evidence (Voelker, 2002). found in other studies (Lovette and Birmingham, 2000). This set of relationship obviously results in Muscicapi- These results suggest that nuclear genes, or at least dae being polyphyletic as currently recognized. c-mos, may not represent a panacea for resolving deeper Generic membership within Turdinae, Saxicolini, and relationships. Muscicapini is also clearly in need to reassessment. Our results unambiguously support the shift of 10 genera 4.2. Defining a monophyletic Muscicapoidea from their current subfamily or tribal placement ac- cording to Sibley and Monroe (1990) to a different Our results clearly support polyphyly of the super- subfamily or tribe. For several of these genera, such family Muscicapoidea, as it is currently defined (Sibley shifts have been argued previously. For example, most and Ahlquist, 1990; Sibley and Monroe, 1990). In our taxonomic schemes place Monticola in Turdinae (e.g., combined nuclear and mitochondrial DNA analysis, the Ripley, 1952). However several treatments have argued Sylvioidea genus Polioptila forms a polytomy with that due to behavioral similarities, Monticola should be Bombycillidae and the remaining Muscicapoidea clades. moved to a position near the chats (Vaurie, 1955a, and In our expanded mitochondrial DNA analysis, this same references therein); our results support this, showing relationship holds, but with Salpinctes placed as sister to that Monticola is firmly embedded within Saxicolini. Polioptila. Our maximum likelihood and parsimony The difficulty in determining relationships among bootstrap analyses further suggested that additional genera is clearly a function of there being relatively few Sylvioidea taxa may be closer to Muscicapoidea than is good morphological and/or behavioral characters sepa- Bombycillidae. Based on these results, and those of rating those genera (reviewed in Sibley and Ahlquist, Barker et al. (2001) who also found Bombycillidae to be 1990). This has obviously made the placement of genera misplaced, we suggest that Muscicapoidea be revised into higher taxonomic categories difficult and rather G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 393 subjective. For example, Ripley (1952) placed both Appendix A Myadestes (solitaires) and Sialia (bluebirds) with the chat thrushes (largely Saxicolini), while Sibley and Taxa and specimens included in this study, following Ahlquist (1990) and our data clearly place them in the taxonomy of Sibley and Monroe (1990). Museum Turdinae proper (true thrushes). Brachypteryx is placed voucher number (or other identifying information) fol- in the chat thrushes by Ripley (1952) and by our data, low each species listed. Museum or other contributor but is considered part of Turdinae by Sibley and Mon- abbreviations are as follows: ANSP, Academy of Natural roe (1990). Sciences, Philadelphia; AMNH, American Museum of We have no doubt that as more genera are added to Natural History; BMNH, Bell Museum of Natural Hi- this analysis, additional genera will prove to be incor- story, University of Minnesota; COP, Copenhagen Mu- rectly placed. Furthermore, preliminary data strongly seum (no voucher); FMNH, Field Museum of Natural indicate that several turdine and muscicapine genera are History; LSUMNH, Louisiana State University Mu- paraphyletic, and that still other genera are invalid seum of Natural History; MBM, Marjorie Barrick Mu- (Klicka and Voelker, unpubl. data; Voelker, unpubl. seum of Natural History UNLV; PF, Percy FitzPatrick data). Indications of this is evidence shown herein of a Institute, Cape Town (no voucher); STRI, Smithsonian potentially paraphyletic Muscicapa, and a recent study Tropical Research Institute (no vouchers); UWBM, (Goodman and Weigt, 2002) which supports merging University of Washington Burke Museum. Material the genus Pseudocossyphus into Monticola. To again from COP and PF consisted of blood samples, STRI highlight the historical confusion in determining thrush material consisted of purified DNA, and material from relationships, Ripley (1952) suggested that Pseudocoss- all other institutions consisted of frozen tissues. An as- yphus should be merged into Cossypha, basing his de- terisk identifies those specimens used in the nuclear plus cision solely on the shared feature of a broad culmen mitochondrial and mitochondrial only DNA analyses. with prominent bristles, and apparently ignoring the Specimens used: Muscicapoidea:Bombycillidae: striking overall similarity in morphology between Bombycilla cedrorum* MBM 8769, Bombycilla garrulus* Pseudocossyphus and Monticola. UWBM 53989, Phainopepla nitens* MBM 6124; Cincli- In conclusion, our results support the recognition of dae: Cinclus schultzii MBM 6912, Cinclus mexicanus* the true thrushes as a distinct family Turdidae, which MBM 5779; Sturnidae: Creatophora cinerea* MBM has been recognized by some recent revisions (e.g., 5948, Lamprotornis nitens* MBM 8225, Toxostoma lec- AOU, 1998), a reduced Muscicapidae comprising the ontei* MBM 6000; Muscicapidae:Turdinae: Neocossy- tribes Saxicolini and Muscicapini, and a rearrangement phus rufus* FMNH 389378, Pseudocossyphus sharpei of those genera shown here to be allied with different FMNH 363837, Monticola saxatilis UWBM 46533, subfamiliies or tribes than currently recognized. We Myiophoneus caeruleus AMNH (JGG 1202), Zoothera recommend that the strongly supported relationships cameronensis FMNH 391736, Ridgwayia pinicola presented here be considered in future taxonomic BMNH 25591, Ixoreus naevius BMNH 42283, Neso- revisions. cichla eremita PF, Cichlherminia lherminieri* STRI DO- CLH1, Sialia currucoides MBM 5654, Myadestes townsendi* MBM 5645, Cichlopsis leucogenys STRI EC- Acknowledgments CLE11769, Entomodestes coracinus ANSP 766, Catharus aurantiirostris* MBM 6639, Hylocichla mustelina MBM We thank the curators and staff of the Academy of 6227, Platycichla leucops COP NK4-110291, Pso- Natural Sciences, Philadelphia, the American Museum phocichla litsipsirupa MBM 5853, Turdus chiguanco* of Natural History, the Bell Museum of Natural History, MBM 5431, Chlamydochaera jefferyi LSU B-36481, the Field Museum of Natural History, the Louisiana Brachypteryx montana FMNH 396295, Alethe poliophrys State Museum of Natural Science, the Smithsonian FMNH 355627; Muscicapini: Bradornis infuscatus* Tropical Research Institute, and the University of MBM 7880, Dioptrornis fischeri FMNH 385188, Mela- Washington Burke Museum for providing us with sam- enornis mariquensis* MBM 5903, Sigelus silens MBM ples of specimens in their care. Peter Ryan of the Percy 5855, Rhinomyias goodfellowi FMNH 357501, Musci- FitzPatrick Institute graciously provided the Nesocichla capa striata* MBM 7473, Muscicapa adusta* MBM sample. J. Klicka provided helpful comments on the 7455, Myioparus griseigularis FMNH 391779, Ficedula manuscript. Collection of some of the specimens used in parva* UWBM 44005, Cyornis caerulata LSU B-36414; this study and laboratory work was supported by Na- Saxicolini: Pogonocichla stellata MBM 7440, Swynner- tional Science Foundation Grant DEB-9903544 to G.V.; tonia swynnertoni LSU B-21159, Stiphrornis erythrotho- additional funding was received from the Ornithological rax FMNH 391713, Sheppardia aequatorialis FMNH Collections Fund and the Genetics Laboratory Fund, 384998, Erithacus rubecula UWBM 57174, Tarsiger both administered by the Barrick Museum of Natural cyanurus* UWBM 51572, Cossypha dichroa* MBM History. 7428, Cercotrichas coryphaeus MBM 5878, Copsychus 394 G. Voelker, G.M. Spellman / Molecular Phylogenetics and Evolution 30 (2004) 386–394 albospecularis* FMNH 352903, Phoenicurus auroreus Hartert, E., 1910. Die vogel€ der Palaarktischen€ fauna. R. Friedlander, UWBM 47088, Chaimarrornis leucocephalus AMNH Berlin. (JGG 1023), Rhyacornis fuliginosus* FMNH 347850, Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. Enicurus scouleri* FMNH 347851, Cochoa viridis Johnson, K.P., Sorenson, M.D., 1998. Comparing molecular evolution AMNH (RTC 580), Saxicola torquata MBM 7453, in two mitochondrial protein coding genes (Cytochrome b and Oenanthe monticola MBM 7456, Cercomela sinuata ND2) in the dabbling ducks (Tribe: Anatini). Mol. Phylogenet. MBM 5829, Myrmecocichla formicivora MBM 5851, Evol. 10, 82–94. €€ Thamnolaea cinnamomeiventris * MBM 11043; Sylvioi- Kocher, T.D., Thomas, W.K., Meyer, A., Edwards, S.V., Paabo, S., Villablanca, F.X., Wilson, A.C., 1989. Dynamics of mitochondrial dea: Poecile gambeli* MBM GMS 280, Regulus calen- DNA evolution in animals: amplification and sequencing with dula* MBM 9325, Polioptila caerulea* MBM 9190, conserved primers. Proc. Natl. Acad. Sci. USA 86, 6196–6200. Zosterops pallidus* MBM 7430, Trococercus albonota- Lanyon, S.M., 1994. Polyphyly of the black bird genus Agelaius and tus* MBM 11345, Sylvia borin* MBM 11430, Turdoides the importance of assumptions of monophyly in comparative bicolor* MBM 8230; Passeroidea: Passer melanurus* studies. Evolution 48, 679–693. Lovette, I.J., Bermingham, E., 2000. c-mos variation in songbirds: MBM 7479, Prunella atrogularis UWBM 46573*, Anthus molecular evolution, phylogenetic implications, and comparisons cinnamomeus* MBM 5860, Motacilla capensis* MBM with mitochondrial differentiation. Mol. Biol. Evol. 17, 1569–1577. 5842, Bubalornis niger* MBM 5748, Thraupis episcopis* Mayr, E., Amadon, D., 1951. A classification of recent birds. Am. MBM 7057, Emberiza rustica* UWBM SVD 141. Mus. Nov. 1496, 1–42. Orenstein, R.I., 1979. The systematic position of the water redstarts, Chaimarrornis and Rhyacornis. Ibis 121, 220–224. Posada, D., Crandall, K.A., 1998. MODELTEST: testing the model of References DNA substitution. Bioinformatics 14, 817–818. Ripley, S.D., 1952. The thrushes. Postilla 13, 1–48. American OrnithologistsÕ Union, 1983. Check-list of North American Ripley, S.D., 1962. Brief comments on the thrushes. Postilla 63, 1–5. birds, sixth ed. American OrnithologistsÕ Union, Washington, DC. Sheldon, F.H., Gill, F.B., 1996. A reconsideration of songbird American OrnithologistsÕ Union, 1998. Check-list of North American phylogeny, with emphasis on the evolution of titmice and their birds, seventh ed. American OrnithologistsÕ Union, Washington, sylvioid relatives. Syst. Biol. 45, 473–495. DC. Sibley, C.G., Ahlquist, J.E., 1990. Phylogeny and Classification of Barker, F.K., Barrowclough, G.F., Groth, J.G., 2001. A phylogenetic Birds: a Study in Molecular Evolution. Yale University Press, New hypothesis for passerine birds: taxonomic and biogeographic Haven. implications of an analysis of nuclear DNA sequence data. Proc. Sibley, C.G., Monroe Jr., B.L., 1990. Distribution and Taxonomy of R. Soc. Lond. B 269, 295–308. Birds of the World. Yale University Press, New Haven. Dowsett, R.J., Dowsett-Lemaire, F., 1993. Comments on the taxon- Stejneger, L., 1905. Birds of the genus Cinclus and their geographical omy of some Afrotropical bird species. Tauraco Res. Rep. 5, 323– distribution. Smithsonian Misc. Coll. 47, 421–430. 389. Storer, R.W., 1971. Classification of birds. In: Farner, D.S., King, J.R. Edwards, S.V., Arctander, P., Wilson, A.C., 1991. Mitochondrial (Eds.), Avian Biology, vol. 1. Academic Press, New York, pp. 1–18. resolution of a deep branch in the genealogical tree for perching Swofford, D.L., 1999. Phylogenetic Analysis Using Parsimony (* and birds. Proc. R. Soc. Lond. B 243, 99–107. other Methods), Version 4. Sinauer Associates, Sunderland, MA. Goodman, S.M., Weigt, L.A., 2002. The generic and species relation- Vaurie, C., 1953. A generic revision of the flycatchers of the tribe ships of the reputed endemic Malagasy genus Pseudocossyphus Muscicapini. Bull. Am. Mus. Nat. Hist. 100, 459–538. (family Turdidae). Ostrich 73, 26–35. Vaurie, C., 1955a. Systematic notes on Palearctic birds. No. 14. Groth, J.G., 1998. Molecular phylogenetics of finches and sparrows: Turdinae: the genera Erithacus, Luscinia, Tarsiger, Phoenicurus, consequences of character state removal in cytochrome b se- Monticola, Erythropygia, and Oenanthe. Am. Mus. Nov. 1731, quences. Mol. Phylogenet. Evol. 10, 377–390. 1–30. Hackett, S.J., 1996. Molecular phylogenetics and biogeography of Vaurie, C., 1955b. Systematic notes on Palearctic birds. No. 15. Tanagers in the genus Ramphocelus (Aves). Mol. Phylogenet. Evol. Turdinae: the genera Turdus, Grandala, and Enicurus. Am. Mus. 5, 368–382. Nov. 1733, 1–18. Harshman, J., 1996. Phylogeny, evolutionary rates, and ducks. Ph.D. Voelker, G., 2002. Molecular phylogenetics and the historical bioge- dissertation, Univ. of Chicago, Chicago. ography of dippers (Cinclus). Ibis 144, 577–584.