Laura Kvist, Department of Animal Ecology, Ecology Building, University of Lund, S-22362 Lund, Sweden. Tel.: +46-46-222-3613; fax: +46-46-222-4716; e-mail: firstname.lastname@example.org
We analysed variation of the mitochondrial control region from willow tits through its Palaearctic distribution range. Although we found high amount of genetic variation (π=1.114%), there was almost no differentiation between subspecies or geographical localities. This may be because of a combination of several ecological and genetic factors, including a relatively homogenic habitat through the distribution range, lack of geographical barriers, high gene flow and a large long-term effective population size. On the contrary, in the songar tit, which is sometimes considered to be conspecific with the willow tit, the mitochondrial lineages seem to correlate with the geographical locality and are clearly distinct from the willow tit. We concluded that the common ancestors of willow and songar tits existed some 1.5–2 Myr ago in the south-eastern Asia. After the last Ice Ages, the willow tit expanded all through the Palaearctic, whereas the songar tit remained in eastern Asia.
During the last decade, substantial knowledge has been accumulated on the phylogeographic patterns of various species at various geographical scales. These studies have mainly focused on European or North American species and only few studies have so far considered species having full Palaearctic distribution ranges. The geological histories of the Nearctic and the Palaearctic differ in that during the last glaciation North America was covered with continental ice to about 45–40°N, whereas the eastern Palaearctic was almost free of ice (Svendsen et al., 1999). Several different refuges affecting genetic differentiation of species have been proposed in the Nearctic (Pielou, 1991), but such small refuges have not been described from the eastern Palaearctic (except for in the Bering region). In Europe during the Weichselian glaciation the ice covered the continent to about 50°N, and at least three refuges have been proposed there (Bennet et al., 1991).
Although the continental ice has not formed geographical barriers to the eastern Palaearctic species, some of these species have genetically differentiated subpopulations. These species include invertebrates (Daphnia pulex, Weider & Hobæk, 1997), small mammals (Lemmus, Fedorov et al., 1999a; Dicrostonyx, Fedorov et al., 1999b) and migratory birds (Calidris alpina, Wenink et al., 1996; Anser erythropus; M. Ruokonen, personal communication). The main site forming a dispersal barrier seems to be the Taimyr-Yamal region and also other sites have been recognized along the Palearctic coast. The genetic structure of larger mammals (Rangifer tarandus, Gravlund et al., 1998; Ursus maritimus, Shields et al., 2000) has so far not been shown to be influenced by these distribution barriers.
The willow tit (Parus montanus) is a sedentary passerine bird with a distribution range covering a zone from north-eastern France and the British Isles through central and northern Europe and Russian taiga to the Pacific coast. According to Harrap & Quinn (1996), the willow tit belongs to the subgenus Poecile, superspecies atricapillus together with the songar tit (Parus songarus), the black-capped chickadee (P. atricapillus), the Carolina chickadee (P. carolinensis) and possibly also the Mexican chickadee (P. sclateri). The black-capped chickadee and the songar tit have sometimes been considered conspecifics with the willow tit (Eck, 1979). Altogether 11 subspecies of the willow tit have been described (Fig. 1). The willow tit is divided into three subspecies groups; salicarius, montanus and kamtschatkensis (Eck, 1980). The six subspecies of the salicarius group form an east–west cline from P. m. baicalensis at the Pacific coast and eastern Siberia, through P. m. uralensis in south-eastern Siberia, P. m. borealis in Fennoscandia, P. m. salicarius in Central Europe and P. m. rhenanus in continental Western Europe to P. m. kleinscmidti in the British Isles. The eastern birds are in general larger and paler than the western birds (Harrap & Quinn, 1996). Several subspecies have been proposed within the alpine montanus group (details in Eck, 1980) but they are not considered valid at present (Vaurie, 1959; Cramp & Perrins, 1993; Glutz von Blotzheim & Bauer, 1993). The kamtschatkensis group forms a north–south cline from P. m. kamtschatkensis of the Kamtschatka peninsula through P. m. sachalinensis of the Sachalin Island to P. m. restrictus of Japan. The southern birds are smaller and darker than the northern ones. Parus montanus anadyrensis, which is intermediate between P. m. kamtschatkensis and P. m. baicalensis, is also included in this group (Cramp & Perrins, 1993; Harrap & Quinn, 1996).
Our earlier analyses of one of the subspecies (P. m. borealis, Kvist et al., 1998) have shown that at least this subspecies has not encountered any drastic changes in the population size during the last Ice Age, which suggested that the species has colonized the modern distribution area from the east instead of from some of the south European refuges.
In this paper, we have analysed variation of the mitochondrial control region from 13 populations of the willow tit from western Europe to the Pacific coast to address the following questions: (1) Is the willow tit genetically panmictic over the Palaearctic distribution range or is there genetic differentiation that might correlate either with the distribution range of the subspecies or with the geographical patterns found from some other Palaearctic species? (2) Is the mitochondrial variation correlated with a east–west cline in the salicarius group or a north–south cline in the kamtschatkensis group? (3) From where did the European birds originate, that is, was Europe colonized from the east or from the south European refuges? In addition we construct a gene tree from the willow tit and some of its close relatives to obtain more information of the phylogenetic status of especially the songar tit.
Materials and methods
DNA samples of the 94 willow tits and 11 songar tits were collected during the years 1993–2000 from the localities shown in Fig. 1 (also in Table 1). In addition, samples from four other Poecile species, and an outgroup, the great tit were used (Table 1). DNA was extracted from blood, pectoral muscle or heart and from embryonic plate as described by Kvist et al. (1998) and from feathers as described by Kvist et al. (1999a). Polymerase chain reaction (PCR) was performed with primers L16720 and H636 (Kvist et al., 1998), which cover the first and part of the second domain of the mitochondrial control region. For some feather samples this fragment was too large to amplify and therefore the region was amplified in two parts with primers L16720 (5′GACTCTCTCCAGGACC TGCGAC3′) and H411 (5′GGAGCTCCTGAAAGCTGTTG G3′) or L339 (5′GCATTCACGAAGT CCATAGC3′) and H636 (5′GAGATGAGGAGTATTCAACCGAC3′). PCR conditions were 94 °C for 10 min followed by 35 cycles of 94 °C for 1 min, 53 °C for 1 min and 72 °C for 1 min and a final extension of 72 °C for 5 min. The PCR products were purified from agarose gels by the method of Glenn & Glenn (1994). Sequencing reactions were performed with Big DyeTM Terminator Cycle Sequencing Kit (Applied Biosystems), with primers H636, H411 or H339 and run with ABI 377 automatic sequencer. Several regions were sequenced twice from independent PCR amplifications and the variable sites were checked for being consistent between the two sequences obtained.
Table 1. Summary of the species and populations sampled, sample sizes, and diversity estimates. Standard deviations are given in parentheses.
Table 1. (Cont'd)
The nucleotide sequences were aligned by eye using the Seqlab interface of the GCG program package. Pairwise Tajima–Nei distances were calculated with the GCG program. Nucleotide diversity (π;Nei, 1987, eqn 10.5), haplotype diversity (πNei, 1987 eqns 8.4 and 8.12), θ (=2Neμ, where Ne is effective population size and μ is mutation rate), estimated from the number of polymorphic sites per nucleotide (Tajima, 1996, eqn 10), Tajima’s D (Tajima, 1989, eqn 38), the mismatch distribution and raggedness statistics (Harpending, 1994) with coalescent simulations for the confidence intervals for raggedness index (r), were calculated with DnaSP version 3.5 (Rozas & Rozas, 1999). The long-term effective female population sizes were estimated from Ne=(108 × 0.5 × p)/g (Avise et al., 1988), where p is the mean pairwise genetic distance, g is the generation time of 2.26 years for the willow tit (Kvist et al., 1998) and a substitution rate of 2% per Myr between the lineages is assumed. This assumption is based on a calibration of great tit control region divergence with mitochondrial restriction fragment length polymorphism (RFLP) (Kvist et al., 1999b). The effective female population size under coalescent, describing the present day effective population size, was estimated from Ne=θ/(2μ), where μ is the mutation rate. Analysis of molecular variance (AMOVA) (Tajima– Nei distance and haplotype frequencies included) was performed with Arlequin version 2.0 (Excoffier et al., 1992), which was also used to construct the minimum spanning tree, analyse population differentiation and estimate numbers of migrants between the populations. Asymmetric migration between the populations was estimated using Migrate (version 1.1) with default parameters (Beerli, 1997–2000). The phylogenetic tree was constructed using the neighbour-joining method with maximum likelihood distance, transition–transversion ratio set to 26.73 (estimated from the data) and 1000 bootstraps using the Phylip program (Phylogeny Inference Package version 3.5c by Joseph Felsenstein, University of Washington 1993). To compare with, phylogenetic trees were also constructed using maximum parsimony, maximum likelihood and neighbour-joining (with Kimura 2-parameter distance) methods, each bootstrapped 100 times. Only two randomly chosen individuals per population of the willow tit were included (except from Fennoscandia only the Oulu population) in addition to all the individuals of the songar tit and representatives of the other Poecile species.
There were altogether 56 polymorphic sites of which 25 were phylogenetically informative in the 592 bp region sequenced from the willow tit (Supplementary Material). There were no alignment gaps. The overall haplotype diversity was high (0.976) and 62 different haplotypes were found from the 94 analysed individuals.
Phylogeography of the willow tit
The minimum spanning tree of the willow tits is shown in Fig. 2. All the possible connections between the haplotypes are not shown in the figure, because the overall structure remains the same. The most common haplotype is shared between 11 individuals originating from Oulu (individuals 1, 11, 23), Tovetorp (68, 69, 70, 73), the Urals (189) and the Kamtschatka peninsula (151, 152, 154), representing the distribution areas of three subspecies. Eight individuals (17, 153, 161, 196, 206, 209, 211, 214), originating from five populations, share the next common haplotype, which differs from the most common one by only one substitution. The rest of the shared haplotypes represent individuals from three to one populations. All the individual haplotypes are scattered around the few common haplotypes with up to eight substitutions to the common haplotypes (excluding the samples from the songar tit). The minimum spanning tree clearly shows that there is no phylogeographic structure that would correlate either with the populations or with the subspecies distribution ranges. This same result is obtained also from the AMOVA, where 86.44% of the total variance occurred within the populations and 89.99% within the subspecies.
The nucleotide diversities, π, within the populations varied from 0.387% in Latvian population to 0.806% in Magadan. θ, estimated from the number of polymorphic sites per nucleotide, varied from 0.414% in Kamtschatka to 1.044% in the Alps (Table 1). There was a trend to increasing nucleotide diversity, haplotype diversity and number of polymorphic sites within the salicarius group from east to west according to the described morphological cline (P. m. baicalensis through P. m. uralensis, P. m. borealis to P. m. salicarius). Within the kamtschatkensis group, no evidence for such a trend could be seen (Fig. 3). The mismatch distribution of the willow tits fits the distribution expected under population expansion (P [r ≤ 0.0202]=0.515, 95% confidence intervals for r: 0.00658, lower limit and 0.6473, upper limit; Rogers & Harpending, 1992; Harpending, 1994), with parameters θ0=0.44922, θ∞=infinite and τ=3.513 (Fig. 4). Both estimated neutrality tests were significantly negative: Tajima’s D=–2.05571, P < 0.05 and Fu and Li’s D=–3.96756, P < 0.02, which also indicates past population expansion. The long-term effective female population sizes were 135 000, 159 000 and 148 000 or when using the coalescent 69 000, 559 000 and 926 000 for the western (salicarius and montanus groups), eastern (kamtschatkensis group) and all populations, respectively.
The pairwise Fst values between the populations are shown in Table 2a. The highest values are concentrated in the Kraslava and Nagano populations, which accordingly show the lowest amount of migrants (Table 2b). The migration from east to west, estimated by mean values of emigrants from the eastern and western populations, was almost 100 times as high than vice versa (Nm=2 × 108 from east and 3 × 106 from west).
Table 2. a Pairwise Fst values between the populations of the willow tit. Significant values (P < 0.05) are shown in bold.
Table 2. b Number of migrants/generation between the willow tit populations based on Slatkins linearized Fsts.
When two subspecies of the songar tit (hereafter assigned as P. m. songarus and P. m. affinis), and the nine subspecies of the willow tit were included in AMOVA, 76.7% of the variation was distributed among the subspecies. Instead, when only the subspecies of the willow tit were included, only 10.0% the variation was explainable by the subspecies. In the neighbour-joining tree (Fig. 5), the subspecies of the songar tit are monophyletic (with bootstrap support of 99.5% for P. m. songarus and 99.5% for P. m. affinis) and group together with the willow tits with bootstrap support of 92.3%. The haplotypes from the different subspecies of the willow tit are intermingled. The black-capped chickadee P. atricapillus, which has earlier been considered conspecific with the willow tit, is clearly as distant from it as the other Poecile species included. The trees constructed using maximum likelihood and maximum parsimony resulted in similar topology, as well as the usage of different distance estimates (which did not even change the bootstrap values). The pairwise Tajima–Nei distances are shown in Table 3.
Table 3. Tajima–Nei distances (%) between some species of the Poecile group.
Phylogeography of the willow tit in the Palaearctic
The results clearly show that there is no phylogeographic structure in the mitochondrial DNA (mtDNA) of the sampled populations of the willow tits through the Palaearctic and the same holds true when the populations are pooled according to the distribution ranges of the subspecies. The reason for this genetic panmixia may be a result of a combination of several ecological and genetic factors, including a relatively homogeneous habitat available throughout the distribution range, lack of obvious geographical barriers, high gene flow and a large long-term effective population size. These are discussed in the following sections in more detail.
The willow tit is found in coniferous forests (Fennoscandia, Siberia) as well as in mixed and broadleaved forests (western and central Europe). These habitats are abundant as a more or less continuous belt throughout the distribution range of the species, from the Atlantic coast of British Isles to the Pacific coast. In Europe the edge of the distribution range is east of the Pyrenees and north of Greece and includes the whole Alps. In Asia, the several mountainous regions do not seem to have had any marked effects on the distribution and actually the willow tits have been recorded to breed at high elevations (up to 2400 m in Japan, Harrap & Quinn, 1996 and even up to at least 2900 m in China, J. Martens, personal observation). This certainly has affected the dispersal capacity and enabled high gene flow even over the Urals and the Altai mountains. However, whether the present gene flow is still high can be questioned, because the borders of the song types today are sharp (Martens et al., 1995) and the differences in morphological traits like coloration and size of the subspecies contradict high present gene flow.
Both estimates of the effective female population size, the long-term and the coalescent estimate were large. Especially the large long-term Nef indicates that there have not been any strong population bottlenecks in the recent history of the willow tit, although suitable forest habitats during the Weichselian glaciation were restricted in the eastern Palaearctic and were almost absent from the western Palaearctic. However, south of Mongolia, there were regions covered with forest vegetation (Adams, 1997), which the species could have inhabited.
It takes approximately 4 N generations to reach reciprocal monophyly of mitochondrial haplotypes in a stable-sized population and even longer if the population is expanding (Avise et al., 1984). Given the large long-term female effective population size of 148 000 in willow tits, it would take about 1 300 000 years for the monophyly to occur in a stable population (assuming a generation time of 2.26 years and a substitution rate of 2% between lineages/Myr) and the coalescent time (estimated from 0.5 × 108 × p, where p is the mean pairwise genetic distance; Avise et al., 1988) is only 570 000 years. Evidently the willow tit population has expanded in size and range after the Ice Ages. This means that the time to monophyly would be even longer. If a population is less than 0.5 N generations old, it is highly likely that there are many mitochondrial lineages present which predate the age of the population (Avise et al., 1984). In other words, it is possible that although dispersal would be restricted and differentiation of subspecies would be in progress, not enough time has passed for monophyletic groups of subspecies to occur in a mitochondrial gene tree. The problems and approaches of distinguishing between the coalescence time of the haplotypes and the population divergence time are discussed also by Edwards & Beerli (2000), where the authors point out that multilocus and multispecies studies should be accomplished in order to estimate population divergence times.
It has been suggested that colonization of new areas from ‘a leading edge’ (Hewitt, 1996) would reduce genetic variation and might even lead to a fairly homogenous population structure because of sequential bottlenecks. This indicates that the areas with lower diversity are more distant from the refuges than the areas with higher diversity (Fry & Zink, 1998). Decrease of nucleotide diversity with increasing latitude has been documented from some widespread bird species in the Nearctic (red-winged blackbird Agelaius phoeniceus, Ball et al., 1988 and Fig. 3a in Fry & Zink, 1998) and Palaearctic (greenfinch Carduelis chloris, Meriläet al., 1997). In willow tits there was no correlation between latitude and nucleotide diversity, instead we found a trend of increasing nucleotide diversity towards east. This combined with the large long-term effective population size (indicating lack of recent bottlenecks) and the palaeovegetation maps reconstructed to present vegetation cover about 18 000 years ago (Adams, 1997) suggests that during the last Ice Ages the willow tit could have occupied fairly large regions in the south-eastern Palaearctic. Thus they may have colonized the present western distribution area from the east. This scenario is supported by a study by Martens et al. (1995), who concluded that different territorial songs of the willow tit may have developed and radiated in the eastern Palaearctic region. To some extent the nucleotide diversity values followed the cline described for the subspecies in the salicarius group, but because of the lack of morphological measures of any kind from the individuals sampled this cannot be tested nor confidently stated.
This study describes the population genetic structure of a sedentary bird species having a Palaearctic distribution range. A plethora of sedentary birds have been studied in America (e.g. chickadees, Gill et al., 1993; fox sparrow Passerella iliaca, Zink, 1994; song sparrow Melospiza melodia, Fry & Zink, 1998; rock ptarmigan Lagopus mutus, Holder et al., 1999, 2000) and some in western Europe (greenfinch Carduelis chloris, Meriläet al., 1997; blue tit Parus caeruleus, Taberlet et al., 1992 and Kvist et al., 1999a; great tit P. majorKvist et al., 1999b), but very little is known so far from the eastern Palaearctic. Existence of divergent intraspecific lineages has been found in many of the sedentary Nearctic bird species although presently no barriers seem to prohibit gene flow. At least some of these intraspecific lineages have been related either to a specific refuges (e.g. rock ptarmigan; Holder et al., 1999) or morphological subspecies (e.g. fox sparrow, Zink, 1996). However, many studies have revealed also species with no obvious phylogeographic structuring, many of those exhibiting shallow mitochondrial trees and indications of recent population expansion (e.g. Ball & Avise, 1992; Seutin et al., 1995; Zink, 1996). In western Europe, the lack of divergent lineages in the greenfinch and the great tit could also be attributed to population growth after a bottleneck during the Weichselian glaciation (Meriläet al., 1997; Kvist et al., 1999b). In southern Europe divergent lineages found from the blue tits, could instead be related to two subspecies and two refuges (Kvist et al., 1999a). In the willow tit no divergent lineages were found, but the high amount of genetic variation found implies that the species has not encountered a similar history than the relative species in Europe, instead the species survived in the east when the conditions became harsh during the Ice Ages.
Phylogenetic status of the songar tit
The phylogenetic status of the songar tit has been controversial, as it is sometimes considered to be a subspecies of the willow tit and sometimes an own species. Voous (1977) placed the songar tit within the willow tit, but stated that the conspecificity has not been proven. Eck (1979, 1980) divided the willow tit into three ‘sectors’, the first including P. m. montanus and the salicarius group, the second including the kamtchatkensis group and the third including the songar tit. He also considered the North American black-capped chickadee to be conspecific with the willow tit. Cramp & Perrins (1993) and Harrap & Quinn (1996) treated the songar tit and the willow tit as separate species, but noted the problematic status. Martens et al. (1995) analysed the territorial songs of the willow and songar tits in the Palaearctic and concluded them to be conspecifics.
Of the four subspecies described for the songar tit (P. m. songarus, P. m. weigoldicus, P. m. affinis and P. m. stoetzneri) we were able to include two in our study. Both, P. m. songarus and P. m. affinis were monophyletic. In addition, there were two mitochondrial lineages in P. m. affinis, one from Gansu and other from Qinghai, China. Also the P. m. songarus individual from Kirghizia differed a lot from the two individuals from Kazakhstan. The songarus–montanus group was monophyletic with a bootstrap support of 97%. The Tajima–Nei distances between P. m. songarus, P. m. affinis and the rest of P. montanus ranged from 3.15 to 4.13% (Table 3), the largest being between P. m. affinis and P. m. songarus. The distances to and between the other species of the Poecile group were roughly twofold. When compared with the distances of the North American chickadee species, estimated from mtDNA (RFLP), the observed distances between subspecies were about half of those estimated between recognized species within the Poecile group (Gill et al., 1993). Genetic distance, estimated from the mitochondrial control region sequences, between two subspecies of the blue tit has been found to be 1.3% (Kvist et al., 1999a), smaller than the distance between the songar and the willow tits.
According to the mitochondrial gene tree and genetic distances both subspecies of songar tit are as distinct from each other as the willow tit from them. On the basis of the phylogenetic species concept (PSC), defined as the smallest diagnosable cluster of individual organisms within which there is a parental pattern of ancestry and descent (e.g. Vogler & DeSalle, 1994), the songar tit cannot be considered as one species. Whether the two subspecies should be treated as subspecies of the willow tit or as two separate species cannot be answered only on the grounds of mtDNA. Martens et al. (1995) proposed on the basis of acoustic characters of the songar tit, that the territorial song of the songarus group points to an origin from the borealis group of the willow tit, and that according to the song characters, the songar tit cannot be considered as an independent species. To conclude, the phylogenetic status of the songar tit cannot be definitely determined, but there are however, indications that it should be considered conspecific with the willow tit.
To summarize, the common ancestors of the willow and songar tits could have existed somewhere in the south-eastern Asia, some 1.5–2 Myr ago (assuming the 2% divergence rate). The split of the ancestral species, into the ‘subspecies of songar tit’ and into the willow tit happened almost simultaneously. As the climate began to warm after the last Ice Ages, willow tit was able to expand all through the Palaearctic, whereas P. m. songarus and P. m. affinis were restricted to eastern Asia, more or less within their presumed areas of origin. The rapid range expansion, population growth and large effective population size of the willow tit are mainly responsible for the observed lack of phylogeograhic patterns in the mtDNA. It is possible that there are genetic differences, other than those studied by us between the willow tit populations, but they may have evolved so recently, simultaneously with the expansion, that they cannot yet be revealed by the mtDNA.
We are grateful to the following for providing us with samples: C. Bronson, L. Brotons, S. Ernst, V. Fedoroff, J. Hering, H. Higuchi, K. Koivula, I. Krams, K. Lahti, T. Mokwa, U. Olsson, B. Petri, V. Pravosudov, S. Rytkönen, A. Thessing, P. Welling and L. Zelenskaya. We also like to thank our helpers in the field: in the Russian Far East O. Valchuk and S. I. Golovatch, in Kasakhstan, and Kyrgysztan S. I. Golovatch and W. Schawaller and many colleagues from the local Kasakh and Kyrgys Academies; in China A. Gebauer, M. Kaiser, Y.-H. Sun and Y. Fang. Valuable comments by Staffan Bensch and Darren Irwin are gratefully acknowledged. This work was funded by the Research Council for Biosciences and Environment of the Academy of Finland, Thule Institute at the University of Oulu, University of Oulu (LK and MO), Deutschen Ornithologen-Gesellschaft, Forschungskommission, Gesellschaft für Tropenornithologie and Feldbausch Foundation, Fachbereich Biologie, Universität Mainz (JM).
Table S1 Variable sites in the first 592 bp of willow tit control region. Cons denotes for the consensus sequence, origin of the individual is assigned in the beginning of each row (songarKa=P. m. songarus from Kazahstan, Russia, songarKi=P. m. songarus from Kirghizia, Russia, affinisQ=P. m. affinis from Qinghai, China and affinisG=P. m. affinis from Gansu, China), and the numbers refer to the individual numbers presented in Table 1.