Molecular phylogenetic analysis among species of Paridae, Remizidae and Aegithalos based on mtDNA sequences of COI and cyt b

Introduction

Tits and titmice are small, familiar cavity-nesting songbirds forming the family Paridae which contains 4 genera, 55 to 65 species, depending on different classifications (Harrap and Quinn, 1996; Salzburger et al., 2002; Dickinson, 2003). They are distributed in the entire Holarctic, Oriental and Afrotropical regions with a hot spot of diversity in the Himalayan and Chinese mountain ranges (Päckert et al., 2007). Because of their overall morphological similarity, all but three species have been classified into the genus Parus and the genus is divided into 10 subgenera (Harrap and Quinn, 1996) or 12 species groups (Eck, 1988). The three other genera, Melanochlora, Sylviparus and Pseudopodoces are monotypic. A well-supported, corroborated and presumably accurate phylogeny is essential for an understanding of ecological and behavioral variation within the group. Therefore, in recent years, a plethora of studies using molecular markers, such as allozyme comparisons (Gill et al., 1989), mitochondrial restriction fragment-length polymorphism (RFLP) data (Gill and Slikas, 1992; Gill et al., 1993), DNA-DNA hybridization distances (Sheldon et al., 1992; Slikas et al., 1996) and cytochrome-b sequences (Kvist et al., 1996; Gill et al., 2005; Martens et al., 2006) have been applied to the phylogeny of Paridae. However, the phylogenetic relationship among species tends to become increasingly controversial as different genetic markers emerge. For instance, based on nuclear DNA-DNA hybridization, Parus caeruleus and P. major formed a clade which is the sister group of all other Parids (Sheldon et al., 1992; Slikas et al., 1996), but the phylogeny inferred from cyt b sequences did not support that result (Gill et al., 2005).

Aegithalos, at one time, was classified as a genus in the family Paridae (Gadow, 1883; Hellmayr, 1903; Mayr and Amandon, 1951). However, Snow (1967) considered Aegithalos and a few related genera as a separate family, the Aegithalidae. Phylogenetic analyses based on DNA-DNA hybridization (Sibley and Ahlquist, 1990; Sheldon and Gill, 1996), protein allozyme comparisons (Ohta et al., 2000) and nuclear gene sequences (Barker et al., 2002; Spicer and Dunipace, 2004) suggested that Aegithalos was indeed not Parids and the closest sister group to Parids was the penduline tits, species of Remizidae. However, genera closely related to Aegithalos are still under dispute, as well as the phylogenetic relationships within Aegithalos (Päckert et al., 2010).

In this study, we sampled part of representative taxa of tits, long-tailed tits and penduline tits in an attempt to resolve the relationships among these three groups and also to infer the phylogenetic relationships within tits and long-tailed tits based on two mitochondrial gene sequences, COΙ and cyt b.

Materials and methods

Taxon selection and choice of outgroups

Altogether, twenty-nine individuals of ten species from Paridae, nine individuals of four species from Aegithalos and two Chinese penduline tits were included in this study. Six species had more than one subspecies included in the analysis (Table 1). According to Sibley and Monroe (1990), Sheldon and Gill (1996), Barker et al. (2002), Spicer and Dunipace (2004), an individual of Pica pica, Uragus sibiricus, Garrulax lunulatus and G. ocellatus were selected as outgroups.

Extraction, amplification and sequencing

Genomic DNA was extracted from blood, feathers or tissue specimens using the QIAamp™ DNA Mini Kit as per manufacturer's instructions. Nucleotide sequence data were obtained from the two mitochondrial, cytochrome c oxidase Ⅰ (COI) and cytochrome b (cyt b) genes. PCR amplification and sequencing of cytochrome c oxidase Ⅰ followed the method suggested by Sorenson et al. (1999), while Gill et al. (2005) described protocols for cytochrome b.

For the sequencing reactions, the same primers were used. Both strands of each PCR product were sequenced. For each gene and sample, multiple sequence fragments were obtained by sequencing with different primers. Complete sequences were assembled using Seqman Ⅱ (DNASTAR^®). Sequences were compared visually to the original chromatograms to avoid reading errors. Assembled sequences were aligned by eye. All sequences were deposited in GenBank.

Phylogenetic analysis

Based on a priori assumption and partition homogeneity test (p = 0.44), the two mitochondrial genes were analyzed as one data set with a total nucleotide length of 2149 base pairs (bp). The data were analyzed using maximum likelihood (ML, Felsenstein, 1981) and Bayesian inference methods (BI, Rannala and Yang, 1996; Yang and Rannala, 1997; Larget and Simon, 1999). Statistics for nucleotide variation and genetic distance were computed with MEGA 4 (Tamura et al., 2007). A Jukes-Cantor estimate of the number of nucleotide substitutions per site was computed for the cyt b gene (Jukes and Cantor, 1969). Following the suggestion of Nei and Kumar (2000), if the Jukes-Cantor distance is less than 0.05, the p-distance would be used. Otherwise, a more complicated distance model would be employed.

Following alignment, we partitioned the data by genes in order to allow different rates for the various partitions for ML and BI analyses. Nucleotide substitution models were selected separately by genes and then used for different data partitions in reconstruction. However, the TVM + I + G model (−lnL = 15397.6758, K = 9, AIC = 30813.3156) was identified as the best fit for these two genes, using both the likelihood-ratio test (LRT) and Akaike Information Criterion (AIC) implemented in MODELTEST 3.7 (Posada and Crandall, 1998). The parameter values for the model include a symmetric rate matrix specifying relative probabilities for all possible nucleotide changes (Rmatrix = 0.6396 [A-C], 8.0280 [A-G], 1.1611 [A-T], 0.0498 [C-G], 8.0280 [C-T], 1.0000 [G-T]), the proportion of invariant sites (pinvar = 0.6183) and the shape parameter for the gamma distribution of rate variation (shape = 1.2128). The base frequencies were set as follows: A = 0.3263, C = 0.4100, G = 0.1009, T = 0.1628.

ML reconstruction (1000 replicates) was performed in TREEFINDER (Jobb, 2007) and BI performed in MRBAYES 3.1.2 (Ronquist and Huelsenbeck, 2003). In the BI analysis, we ran two analyses of two million generations and trees sampled every 100 generations. One cold and three heated Markov chains were used in our analysis. The trees saved during the "burn-in" phase (the first 200000 generations in our analysis) were discarded. The remaining trees from two runs were used to create a 50% majority rule consensus tree.

Results

Sequence characteristics

We obtained a total of 88 sequences, with their GenBank accession numbers listed in Table 1. No stop codons were identified in a contiguous 1201 bp stretch of the COI gene and 948 bp of cyt b. The overlapping sequences from different PCR products and a single peak in the electropherograms suggest that these sequences do not come from "numts" (nuclear sequences of mitochondrial origin). The average base composition of sequence was skewed, which is similar to that found in previous avian studies (Barhoum and Burns, 2002; Webb and Moore, 2005). The characteristics of these sequences are summarized in Table 2.

Table  1.  List of taxonomic samples and sequences used in the study

Family	Genus	Species and subspecies	Museum No.	Collection locality	GenBank Accession Nos.
					cyt b	COΙ
Paridae	Parus	Parus major commixtus	IOZ1254	Guangxi	HM185345	HM185334
		Parus major subtibetanus	IOZ5236	Yunnan	HM185346	HM185333
		Parus major subtibetanus	IOZ3918	Sichuan	HM185347	HM185332
		Parus major artatus	IOZ5661	Jilin	HM185348	HM185331
		Parus monticolus yunnanensis	IOZ3633	Sichuan	HM185349	HM185330
		Parus monticolus yunnanensis	IOZ2465	Shaanxi	HM185350	HM185329
		Parus spilonotus rex	IOZ3026	Fujian	HM185351	HM185328
		Parus cyanus	IOZ 781	Xinjiang	HM185352	HM185327
		Parus venustulus	IOZ2806	Hubei	HM185353	HM185326
		Parus venustulus	IOZ1945	Shaanxi	HM185354	HM185325
		Parus ater aemodius	IOZ1094	Gansu	HM185355	HM185324
		Parus ater	IOZ2331	Zvolen	HM185356	HM185323
		Parus ater ater	IOZ9029	Heilongjiang	HM185357	HM185322
		Parus palustris hypermelas	IOZ1239	Gansu	HM185358	HM185321
		Parus palustris hypermelas	IOZ7991	Hubei	HM185359	HM185320
		Parus palustris brevirostris	IOZ5675	Jilin	HM185360	HM185319
		Parus palustris hellmayri	IOZ8680	Shanxi	HM185361	HM185318
		Parus palustris hellmayri	IOZ2458	Shaanxi	HM185362	HM185343
		Parus montanus baicalensis	IOZ5944	Heilongjiang	HM185363	HM185317
		Parus montanus baicalensis	IOZ5945	Heilongjiang	HM185364	HM185316
		Parus montanus	IOZ2237	Zvolen	HM185365	HM185315
		Parus montanus affinis	IOZ2178	Shaanxi	HM185366	HM185342
		Parus dichrous dichrous	IOZ2489	Shaanxi	HM185367	HM185314
		Parus dichrous dichrous	IOZ6117	Shaanxi	HM185368	HM185313
		Parus cristatus	IOZ2333	Zvolen	HM185369	HM185312
	Sylviparus	Sylviparus modestus modestus	KIZglgs1240	Yunnan	HM185370	HM185311
		Sylviparus modestus modestus	KIZglgs1241	Yunnan	HM185371	HM185310
	Pseudopodoces	Pseudopodoces humilis	IOZ4783	Qinghai	HM185372	HM185309
		Pseudopodoces humilis	IOZ4785	Qinghai	HM185373	HM185308
Remizidae	Remiz	Remiz consobrinus	IOZ10640	Liaoning	HM185374	HM185298
		Remiz consobrinus	IOZ10652	Liaoning	HM185375	HM185297
Aegithalidae	Aegithalos	Aegithalos caudatus caudatus	IOZ5677	Jilin	HM185376	HM185307
		Aegithalos caudatus glaucogularis	IOZ2769	Hubei	HM185377	HM185306
		Aegithalos caudatus glaucogularis	IOZ3322	Shaanxi	HM185378	HM185305
		Aegithalos concinuus concinnus	IOZ1152	Gansu	HM185379	HM185304
		Aegithalos concinuus concinnus	IOZ1499	Hunan	HM185380	HM185303
		Aegithalos concinuus concinnus	IOZ3661	Sichuan	HM185381	HM185302
		Aegithalos concinuus talifuensis	IOZ5344	Yunnan	HM185382	HM185301
		Aegithalos bonvaloti	IOZ3766	Sichuan	HM185383	HM185300
		Aegithalos fuliginosus	IOZ2476	Shaanxi	HM185384	HM185337
Timaliidae	Garrulax	Garrulax lunulatus	IOZ2594	Shaanxi	HM185385	HM185341
		Garrulax ocellatus	IOZ5083	Yunnan	HM185386	HM185340
Fringillidae	Uragus	Uragus sibiricus	IOZ5870	Heilongjiang	HM185387	HM185344
Corvidae	Pica	Pica pica	IOZ3931	Xinjiang	HM185388	HM185339
Note: IOZ, Institute of Zoology, Chinese Academy of Sciences; KIZ, Kunming Institute of Zoology, Chinese Academy of Sciences.

 | Show Table

DownLoad: CSV

Table  2.  Molecular characterization of the sequences

Genes	No. of sequences	Total sites	Variable sites (%)	Avg R (si/sv)	Nucleotide frequencies
					A (%)	T (%)	G (%)	C (%)
COΙ	40	1201	370	1.7	25.8	23.8	17.3	33.1
cyt b	40	948	326	1.5	26.6	23.8	13.2	36.3

 | Show Table

DownLoad: CSV

Pairwise genetic distances

Because the mean Jukes-Cantor distance was 0.112 ± 0.006 SE (> 0.05) for the cyt b sequence data set, the Tamura-Nei model was chosen to compute the genetic distance (Tamura and Nei, 1993). Sequence divergences within species range from 0% (Parus dichrous) to 2.8% (Aegithalos concinnus), with an average divergence of 1.15%. The mean cyt b sequence distance of Aegithalos was 0.081 ± 0.007 SE, with the lowest 0.002 between A. bonvaloti and A. fuliginosus and the highest 0.118 between A. concinnus and A. caudatus. The mean distance of genus Parus was 0.077 ± 0.005 SE, with the lowest 0.053 between P. major and P. monticolus. Between taxa in Parus and the other two species in the Paridae, average pairwise divergences are as follows: 0.115 ± 0.009 SE to Sylviparus modestus and 0.097 ± 0.007 SE to Pseudopodoces humilis. The distances were 0.16 ± 0.01 SE between Aegithalos and Paridae, 0.18 ± 0.013 SE between Aegithalos and Remizidae, and 0.126 ± 0.01 SE between Paridae and Remizidae.

Phylogenetic relationships

The consensus tree from the Bayesian analysis is identical to the ML tree, except for a few weakly supported nodes. A few nodes that are resolved in the ML tree are polytomies in the Bayesian consensus tree (Fig. 1). In all optimal trees, no species which includes several subspecies were found to be paraphyletic in our study. Paridae, Remizidae, Aegithalos and Garrulax species formed a monophyletic group with high support (bootstrap support = 70%, Bayesian posterior probability = 97%) and branched off to two monophyletic clades. The clade, representing individuals from species of Paridae and Remidae, was highly supported (bootstrap = 98%, BPP = 100%). Remiz consobrinus positions as a sister species to the Paridae, including Sylviparus, Pseudopodoces and the species in Parus. Aegithalos is monophyletic, grouped with two Garrulax species and, with weak support, formed the other clade.

Figure  1.  Trees obtained from the analysis of the combined COΙ and cyt b data: (a) Bayesian and (b) maximum likelihood. The optimal model as estimated by Modeltest following the AIC and LRT was estimated to be TVM + I + G. Support values are indicated to the left of the nodes.

DownLoad: Full-Size Img PowerPoint

Within Aegithalos, four species are monophyletic and grouped in all optimal trees (Bayesian posterior probability = 100%, bootstrap = 100%). Within the clade, Aegithalos bonvaloti is sister to A. fuliginosus and that pair is sister to A. caudatus. Aegithalos concinnus is sister to these three species. Phylogenetic relationships have strong support (Bayesian posterior probability = 100%, bootstrap > 95%) and are identical in the two phylogenetic trees.

Within the family Paridae, Sylviparus modestus positions as a sister species to the remaining species of Paridae from the two trees. These remaining species form two high-support monophyletic clades in the BI tree (BPP = 99%), but the bootstrap value is rather low (bootstrap = 44%) in the ML tree. One clade includes Parus cyanus, P. monticolus, P. major, P. spilonotus and Ps. humilis. The relationships among these five species are congruent in the two trees except for the position of Ps. humilis. In the two trees, Parus cyanus is a sister species to the remaining species, although the nodal support value is low. Parus major and P. monticolus pair as sister taxa with high support. In the BI tree, P. spilonotus and Ps. humilis pair as sister taxa but in the ML tree, Pseudopodoces humilis is a sister species to P. spilonotus, P. major and P. monticolus. However, the nodal supports are low. Another clade includes P. ater, P. venustulus, P. palustris, P. montanus and two Eurasian crested tits, P. dichous and P. cristatus. Parus palustris and P. montanus, P. ater and P. venustulus, P. dichous and P. cristatus pair as sister taxa with high support. Relationships among these three pairs are unclear because of polytomies in the BI tree and low nodal support value in the ML tree.

Discussion

Phylogenetic relationships among Paridae, Remizidae and Aegithalos

Our data support the classification for the family Paridae proposed by Sibley and Ahlquist (1990). Based on DNA-DNA hybridization, they combined family Remizidae with Paridae, hence the family Paridae includes two subfamilies, Remizinae and Parinae. Sheldon and Gill (1996) studied the phylogeny of song birds using DNA-DNA hybridization and concluded that the sister group of Paridae is the Remizidae. Based on combined myoglobin intron Ⅱ and cytochrome b sequences, Alström et al. (2006) revealed the close relationships between P. major and R. pendulnus. In our study, the Chinese penduline tit (Remizidae) positioned as sister group to the Parids (Paridae) with high support. This may indicate very close relationships between Paridae and Remizidae. Although, in the present study, only one species of Remizidae was tested, the high nodal support (> 95%) signifies that the designated branch is most possibly unaffected by sampling among variation existing in the data (Slikas et al., 1996).

Long-tailed tits, species of Aegithalos, because of their morphological similarity with tits, were earlier believed to be Parids by several authors (Gadow, 1883; Hellmayr, 1903; Mayr and Amandon, 1951). However, Stresemann (1923) described several characteristics of Aegithalos that differ from those of the Parids, such as the presence of a complete juvenile-molt, nest structure and naked hatching. Since then, Aegithalos and two other monotypic or small genera are usually placed in the family Aegithalidae (e.g., Paynter, 1967; Morony et al., 1975; Sibley and Ahlquist, 1990; Sibley and Monroe, 1990). Sturmbauer et al. (1998) suggested a close relationship between Leptopoecile and Aegithalos based on the mitochondrial 16S sequence. Alström et al. (2006) confirmed the close relationship between Aegithalos and Leptopoecile based on the combined myoglobin intron Ⅱ and cytochrome b sequences. Using protein allozyme comparison, Ohta et al. (2000) found the genetic distance between Aegithalos and Parus appears to be at a familial level. In our study, two distinct clades and Aegithalos grouped with two outgroup Garrulax species, in spite of weak nodal support, show distant relationships between Aegithalos and Paridae. In addition, the genetic divergence of cyt b between Aegithalos and Paridae is larger than that between Remizidae and Paridae. Hence, the morphological similarity between tits and long-tailed tits may have resulted from evolutionary convergence.

Phylogenetic relationships within Paridae

Based on plumage pattern, Hellmayr (1903) had recognized eleven subgenera for the genus Parus. However, the delimitation of these subgenera is in dispute. For instance, Wolters (1982) classified the species of Pardaliparus into Pariparus and this classification was supported by molecular analyses (Slikas et al., 1996; Gill et al., 2005). In our study, Parus ater and P. venustulus also formed a closely related group, but the divergence of cyt b is ~7% (uncorrected p-distance). According to Hellmayr's classification, our samples belong to seven subgenera: Cyanistes (P. cyanus), Baeolophus (P. dichous, P. cristatus), Poecile (P. montanus, P. palustrs), Periparus (P. ater), Pardaliparus (P. venustulus), Machlolophus (P. spilonotus) and Parus (P. major, P. monticolus). Similar to previous studies, our data did not reveal phylogenetic relationships among these subgenera. Therefore, the Parus phylogeny needs to be studied further. We intend to add more species and subspecies and new markers, including nuclear loci to classify this challenge in the future.

Although our results came to quite similar conclusions as previous studies, different phylogenetic relationships among species were detected as well. The question of greatest interest is: what is the phylogentically most closely related species to Ps. humilis in Parids? This aberrant and enigmatic Tibetan species, earlier thought to be a corvid, turned out to be a Parid (James et al., 2003). Past phylogenetic analyses based on cyt b gene sequences positioned Ps. humilis as sister species to P. major (James et al., 2003; Gill et al., 2005). However, in optimal trees based on the cyt b and COΙ gene sequences in our study, Ps. humilis is sister to P. spilonotus, although the nodal support is weak. This result may indicate that more data of the species and different markers are needed to determine the phylogenetic status of Parids.

Gill et al. (2005) recommended upgrading six subgenra of Parus into genera for facilitating future evolutionary analyses among Parids. The retained genus Parus only included P. major, P. monticolus, P. xanthogenys, P. spilonotus, P. holsti and the African tits (P. leucomelas, P. niger, P. carpi, P. albiventris, P. leuconotus, P. rufiventris, P. funereus, P. fringillinus, P. fasciiventer, P. thruppi, P. griseiventris, P. cinerascens and P. afer). However, in our present study, we also believe that the genus Machlolophus should be recognized in consideration of the phylogenetic relationships between P. spilonotus and Ps. humilis and distinct plumage pattern between P. major and P. spilonotus (Ohta et al., 2000). Furthermore, in our data, the divergence of the cyt b gene between P. major–P. monticolus group and P. spilonotus (8.4%, uncorrected p-distance) is no less than the divergence among recommended genera (ranging from 7.3% between Pariparus and Baeolophus to 9.7% between Poecile and Machlolophus, uncorrected p-distance).

Intra- and interspecies relationships and history of Aegithalos taxa

On the one hand, intraspecific divergence is significant. The Black-throated Tit (Aegithalos concinnus) includes six highly distinct subspecies. These subspecies are remarkably different in their plumage pattern. Divergence differences of the cyt b gene among these subspecies are also significant. In our data, the pairwise p-distance between specimens from Bengal (India, ssp. iredalei, sequence obtained from GenBank) and Gaoligong (Yunnan, ssp. talifuensis) was 4.5%. p-distance between specimens from Gaoligong and Pingjiang (Hunan, ssp. concinnus) was 6.1%. p-distance between specimens from Bengal and Pingjiang was 5.1%. Eck and Martens (2006) sequenced three specimens, including subspecies ssp. iredalei, ssp. talifuensis and ssp. manipurensis, and they found the pairwise p-distance was 5.1% between ssp. iredalei and ssp. talifuensis, 5.3% between ssp. talifuensis and ssp. manipurensis and 6% between ssp. iredalei and ssp. manipurensis. These four subspecies are reddish-crowned. The other two subspecies A. c. pulchellus and A. c. annamensis are grey-headed. They are obviously different in presence of reddish breast band, the latter of which is absent. Hence, it seems possible that A. concinnus represents an unresolved species swarm (Eck and Martens, 2006).

As for the Long-tailed Tit (Aegithalos caudatus), based on comparative morphology, up to 19 subspecies are currently recognized, divided by Harrap and Quinn (1996) into four groups. Zink et al. (2005) studied the mitochondrial phylogeography, including five subspecies or representatives of A. caudatus group and A. alpinus group of the long-tailed tit. Within their samples, no correspondence between five subspecies and mitochondrial subdivision has been discovered. However, two geographically unsorted lineages were displayed and differentiation seems to exist between the two main groups. In this study, the cyt b divergence of two specimens belonging to the subspecies A. c. caudatus and A. c. glaucogularis, is 1.3%. Thus, phylogeographic structures would be developed if additional subspecies were taken into consideration.

On the other hand, interspecies molecular divergence is small. The morphologically distinct sister species A. fuliginosus and A. bonvaloti are separated by unexpectedly small cyt b divergences (0.2%, uncorrected p-distance) and associated very short branch lengths in the cyt b tree. The pairwise cyt b divergence between the species is comparable to that within populations of the same species in other passerine birds. This could indicate recent separation and divergence from their common ancestor. However, surprisingly, the more slowly evolving nuclear locus β-fibrinogen intron 7 shows relatively greater divergence (2.5%, uncorrected p-distance) than the faster-evolving cyt b (unpublished data). This might be the result of amplification of nuclear pseudogenes instead of mitochondrial DNA (Zhang and Hewitt, 1996; Sorensen and Quinn, 1998). However, our sequences show no evidence of being of nuclear origin. Introgression of mitochondrial DNA seems to be a more likely explanation. A. fuliginosus and A. bonvaloti are not known to hybridize, but their current distributions partly overlap. Hybridization or past hybridization leading to introgression is nevertheless a possibility. Weckstein et al. (2001) argued that introgressive hybridization is the cause of discordant patterns of mitochondrial and allozyme data in the North American sparrows Zonotrichia leucophrys and Z. Atricapilla.

Furthermore, species limits within the A. niveogularis are not, as yet, reliably defined. Harrap and Quinn (1996) proposed the species to be split in two species while Dickinson (2003) proposed a three way split. Therefore, currently, five (Martens and Eck, 1995) or six (Harrap and Quinn, 1996) or seven (Dickinson, 2003) species have been recognized within the genus Aegithalos. Our study includes only a few species and subspecies from Aegithalos because of insufficient or unavailable samples. Recently, Päckert et al. (2010) studied the phylogeny of long-tailed tits and allies based on mitochondrial and nuclear markers and resolved the status of some species on the basis of phylogenetic relationships. However, by considering the long evolutionary time of species and differences within species, both morphological and molecular characters, a study which includes more species and subspecies is necessary to define the phylogeny of Aegithalos.

We suggest that the southwestern mountain ranges of China might be the center of origin of Aegithalos species and that the taxa of this genus colonized new habitats from west to east across China. Supporting this hypothesis are the patterns of taxonomic diversity and endemism. For example, members of all long-tailed tits are found in the areas of southwestern China except A. niveogularis. The oldest linage in the genus in our study is the A. concinnus talifuensis which is distributed in Yunnan Province of China. The historical biogeography can be deduced by the relationships among species and the areas of species distribution (Gill et al., 2005). For instance, the closely related phylogenetic relationships between Aegithalos fuliginosus and A. bonvaloti is accompanied by a partly overlapped distribution. Aegithalos caudatus caudatus is older than A. c. glaucogularis in our phylogenetic analysis and is thus distributed farther from the original areas than A. c. glaucogularis. Given a rough calibration of 2% divergence per million years, we hypothesize that the separation of A. concinnus may occur ~5.5 Mya, on the basis of the estimate of 11% sequence divergence between A. concinnus and other three species. Then, about 4.5 Mya ago, A. caudatus was divided and colonized the palearctic region. More recently, separation between Aegithalos fuliginosus and A. bonvaloti is hypothesized to have taken place in the late Pleistocene, i.e., ~100000 years ago. However, that break may be earlier than late Pleistocene, for the possible introgression between two species may obscure the real separation time.

Congruent genetic differentiation among co-distributed species

Comparative phylogeography helped elucidate the relative effect of shared historical earth events on current patterns of biodiversity by comparing historical patterns of gene flow and divergence among species that overlap in time and space (Hickerson et al., 2010). In other words, co-distribution species may have similar population genetic differentiation or congruent phylogeographical patterns as a result of sharing common environmental and geological changes. In our study, the Great Tit (Parus major) and the Black-throated Tit (A. concinnus) overlap in southern and southwestern China, the Coat Tit (P. ater), the Willow Tit (P. montanus), the Marsh Tit (P. palustris) and the Long-tailed Tit are distributed in the northeast, center and west of China. Although our sample size was small, we found congruent genetic differentiation among these sympatric distribution taxa. The great tit and the black-throated tit have similar phylogeographical patterns. They differentiated into two clades, one including the samples of Yunnan Province and the other one belonging to the other sample locations in China. Similarly, the coat tit, the marsh tit, the willow tit and the long-tailed tit also divided into two branches, one covering the areas of northeastern China and Europe and the other including the species distribution in central and western China. These phylogeographical patterns or population structures of sympatric species may indicate that the historical earth events they experienced, including climatic and geological, had almost the same effect on these closely related taxa.

Acknowledgements

We are grateful to Tao Li for laboratory assistance and anonymous reviewers who provided helpful comments to the improvement of the manuscript. This work was financially supported by the National Natural Science Foundation of China (Grant No. 30870270) and the National Science Funds for Distinguished Young Scientists (No. 30925008) to Fumin Lei.