Mitochondrial phylogeographic structure of the white-browed piculet (Sasia ochracea): cryptic genetic differentiation and endemism in Indochina

Authors

  • Jérôme Fuchs,

    Corresponding author
    1. UMR5202, ‘Origine, Structure et Evolution de la Biodiversité’, Département Systématique et Evolution, Muséum National d’Histoire Naturelle, 55 rue Buffon, 75005 Paris, France
    2. Service Commun de Systématique Moléculaire, IFR CNRS 101, Muséum National d’Histoire Naturelle, 43 rue Cuvier, 75005 Paris, France
    3. DST-NRF Centre of Excellence at the Percy FitzPatrick Institute, University of Cape Town, Rondebosch 7701, Cape Town, South Africa
    4. Museum of Vertebrate Zoology and Department of Integrative Biology, 3101 Valley Life Science Building, University of California, Berkeley, CA 94720-3160, USA
    Search for more papers by this author
  • Per G.P. Ericson,

    1. Department of Vertebrate Zoology and Molecular Systematics Laboratory, Swedish Museum of Natural History, PO Box 50007, SE-104 05 Stockholm, Sweden
    Search for more papers by this author
  • Eric Pasquet

    1. UMR5202, ‘Origine, Structure et Evolution de la Biodiversité’, Département Systématique et Evolution, Muséum National d’Histoire Naturelle, 55 rue Buffon, 75005 Paris, France
    2. Service Commun de Systématique Moléculaire, IFR CNRS 101, Muséum National d’Histoire Naturelle, 43 rue Cuvier, 75005 Paris, France
    Search for more papers by this author

*Jérôme Fuchs, Museum of Vertebrate Zoology and Department of Integrative Biology, 3101 Valley Life Science Building, University of California, Berkeley, CA 94720-3160, USA. E-mail: fuchs@mnhn.fr

Abstract

Aim  Our understanding of the geographic patterns of gene flow between populations of birds in the Indo-Malayan faunal region is surprisingly poor compared with that in other parts of the world. A thorough knowledge of general patterns of phylogeographic structure is, however, of utmost importance for conservation purposes. Species with poor dispersal capabilities could serve as indicators of endemism and genetic isolation in the Indochinese subregion. From their morphology (tiny size, short tail, short and rounded wings), piculets of the genus Sasia are inferred to have poor dispersal capabilities, and thus form a suitable focal species. This study analysed the pattern of genetic variation within the White-browed Piculet (Sasia ochracea).

Location  Southeast Asia, north of the Isthmus of Kra.

Methods  We sampled 43 individuals throughout the breeding range of S. ochracea. DNA was extracted both from fresh tissues (= 15) and from toe pads from ancient museum skins (= 28). We amplified a 801-bp fragment of the mitochondrial ND2 gene to reconstruct the phylogeographic history of the White-browed Piculet. The sequence data were analysed using Bayesian inference, statistical parsimony, and population genetics methods (analysis of molecular variance, mismatch distributions). We estimated the amount of ongoing gene flow between populations using the coalescent-based method implemented in Mdiv.

Results  The analysis of molecular variance indicated that the current taxonomy does not adequately reflect the amount of genetic variation within S. ochracea, as the great majority of genetic variation was nested within the nominal subspecies, which is distributed from Nepal to southern Vietnam. Bayesian inference analyses and haplotype networks suggested the occurrence of five main lineages that are strongly correlated with geography. Our coalescent-based analyses indicated a very limited amount of ongoing gene flow between these five lineages. Our dating analyses suggested that the genetic structuring probably occurred during the last 400,000 years.

Main conclusions  Our analyses revealed that S. ochracea is composed of at least five lineages: south Vietnam (South Annam and ‘Cochinchina’), India and Nepal, Myanmar and India, the remainder of Indochina, and probably southern Myanmar (Tenasserim). We strongly recommend that studies aiming to understand the phylogeographic structure within Indo-Malayan species sample these areas.

Introduction

The Indo-Malayan faunal region is one of the most species-rich areas of the world. An impressive number of 1169 bird species (> 10% of all the species in the world) live there, of which 70% are endemic to the region (Newton, 2003). However, heavy exploitation of natural resources for wood and the pet trade has rendered the Indo-Malayan region among the most threatened on Earth (Sodhi et al., 2004, 2006). The severe threats have sparked intensive work to protect the fauna and flora, including the creation of conservation programs all over the region. Among many other things, such work requires a thorough understanding of the patterns of genetic diversity among and within species (Kahindo et al., 2007). Furthermore, widespread species may exhibit zones of molecular endemism, suggesting incomplete understanding of intraspecific taxonomy and the existence of morphologically cryptic species. In birds, such information is essentially lacking in all parts of the Indo-Malayan region, whereas it has received considerably more attention in other areas of the world, such as the Palaearctic (e.g. Pavlova et al., 2005; Zink et al., 2006), Afrotropics (e.g. Bowie et al., 2004, 2006), the Nearctic (e.g. Barrowclough et al., 2004; Burns & Barhoum, 2006), and the Neotropics (e.g. Marks et al., 2002; Bates et al., 2004). The only two studies from the Indo-Malayan region are of the White-crowned Forktail (Enicurus leschenaulti,Moyle et al., 2005) and Grey-cheeked Fulvetta (Alcippe morrisonia, Zou et al., 2007). These studies highlight regions of high genetic differentiation within Borneo (Moyle et al., 2005) and China (Zou et al., 2007). These two studies do not, however, involve a comprehensive geographic sampling across the breeding range of the study species, so a thorough understanding of the patterns of genetic diversity within an Indo-Malayan species has not yet been gained.

In our previous survey of the phylogenetic relationships and biogeographical history of the piculets (Fuchs et al., 2006), we sampled four white-browed piculets (Sasia ochracea Hodgson, 1836) and discovered substantial genetic differentiation among the four individuals. Here, we aim to describe and understand, using a more thorough sampling of the breeding range, the geographic structure of the genetic variation within the White-browed Piculet, a woodpecker species endemic to Southeast Asia. Sasia ochracea has a widespread distribution throughout the Indochinese subregion, and three subspecies are currently recognized (Winkler & Christie, 2002): S. o. ochracea Hodgson, 1836 is present from northern India to southern and central Vietnam; S. o. kinneari Stresemann, 1929 can be found in northern Vietnam (‘Tonkin’) and South China (Yunnan and Guangxi); and S. o. reichenowi Hesse, 1911 is restricted to south Myanmar (Tenasserim) and south-west Thailand south to the Isthmus of Kra, where the species is replaced by its closest relative, the Rufous Piculet (Sasia abnormis) (Fuchs et al., 2006). Sasia ochracea inhabits dense, low vegetation in broadleaved evergreen and mixed deciduous forest but is also tolerant of more degraded habitats such as secondary forest. It can be found at various altitudes between 250 m and 1850 m in Southeast Asia, and even up to 2600 m in the Indian subcontinent. From their morphology (tiny size, short tail, short and rounded wings), piculets of the genus Sasia are inferred to have poor dispersal capabilities. Indeed, one individual of Sasia abnormis was captured only 800 m from its initial capture location several years earlier (Winkler & Christie, 2002). Species with poor dispersal capabilities, such as S. ochracea, could serve as models with which to detect areas of local endemism as a result of genetic isolation. Here, we aim to address the phylogeographic structure within S. ochracea in order to: (1) propose new hypotheses about patterns of genetic variation within widespread Southeast Asian species, and (2) highlight areas where high genetic distinctiveness occurs in order to inform conservation practices.

Materials and methods

Laboratory procedures

We sampled 43 individuals of S. ochracea, and used two individuals of S. abnormis as outgroups. Sampling localities encompassed the distribution of S. ochracea (Fig. 1, Appendix S1). Several ancient collecting localities were retrieved using Hennache & Dickinson (2000) and Lozupone et al. (2004). DNA was extracted from fresh tissues (blood, liver, muscle) (= 15) and from toe-pads from museum skins collected during the period 1910–1970 (= 28) using a Cethyl Trimethyl Ammonium Bromide (CTAB)-based protocol (Winnepenninckx et al., 1993). Two further sequences (one White-browed and one Rufous Piculet) were retrieved from GenBank (http://www.ncbi.nlm.nih.gov). We amplified an 801-bp fragment of the ND2 gene using primer pairs L5219–H6313, L5219–SaH650 and SaL450–H6313 for fresh samples, and L5219–Sa200H, SaL150–SaH350, SaL300–SaH500, SaL450–SaH650, SaL600–SaH800, SaH750–H6313, 750F–950R for museum samples (Appendix S2). The amplification protocol was standard (2 min at 94°C, followed by 36 cycles of 94°C for 40 s, 54°C for 45 s, 72°C for 40 s, and a final extension at 72°C for 5 min). Three-microlitre samples of the amplification products were electrophoresed on a 1.5% agarose gel and examined under UV light with ethidium bromide to check for the correct fragment size, control for the specificity of the amplifications, and rule out contaminations (positive blank). The polymerase chain reaction (PCR) products were purified using a QiaQuick PCR purification kit (Qiagen, Holden, Germany). Cycle-sequencing reactions were performed using a CEQ Dye terminator cycle sequencing kit (Beckman Coulter, Inc., Fullerton, CA, USA) or a Big Dye (Applied Biosystems, Inc., Foster City, CA, USA) terminator chemistries kit using the same primers as for PCR amplifications. DNA strands were sequenced on an automated CEQ2000 DNA analysis system or ABI3100 sequencers. No insertions, deletions or stop codons were detected in the reading frames.

Figure 1.

 Geographic distribution of Sasia ochracea (grey). Subspecies recognition and distribution follow Winkler & Christie (2002). The distribution ranges of the subspecies S. o. kinneari (B), and S. o. reichenowi (C) are delimited by the grey lines. The distribution range of the nominate subspecies S. o. ochracea (A) encompasses all the remaining distribution of the species. Sampling localities are indicated by dots. Numbers of sampled individuals are only indicated if different from one.

Data analyses

Analyses of the haplotypes were performed using Bayesian inference (BI), as implemented in MrBayes 3.1 (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003). The GTR + I model was selected as the best-fit model to our data set with the assistance of MrModeltest (Nylander, 2004) and the Akaike information criterion (Akaike, 1973). Four Metropolis-coupled Markov chains Monte Carlo (one cold and three heated) were run for five-million iterations, with trees sampled every 100 iterations. The first 300,000 iterations (3000 trees) were discarded (‘burn-in’ period), and the posterior probabilities were estimated for the remaining sampled iterations. Two independent Bayesian runs initiated from random starting trees were performed, and the log-likelihood values, posterior probabilities and average deviations of split frequencies were checked to ascertain that the chains had reached convergence.

Because many of the underlying assumptions of traditional tree-building methods (fully bifurcating trees, complete lineage sorting) are often violated when addressing intraspecific studies (Posada & Crandall, 2001), we also used networks to explore the phylogeographic structure of S. ochracea. We used tcs 1.18 (Clement et al., 2000) to construct a minimum-spanning network. The two individuals of S. abnormis were then excluded from the data set. We used Arlequin 2.0 (Schneider et al., 2000) to calculate the number of haplotypes, haplotype diversity (H), nucleotide diversity (π), Tajima’s D and Fu’s fs tests of selective neutrality (1000 replicates), and the mismatch distributions for each of the putative subspecies. The overall genetic structure of populations was investigated with an analysis of molecular variance (amova, 1000 permutations) using Tamura and Nei’s pairwise distances (φst, Excoffier et al., 1992). We performed two analyses of molecular variance, the other by partitioning the individuals by subspecies designation, and the other by partitioning the individuals according to the haplotype clusters found in the BI tree and haplotype network.

We used the coalescent-based program Mdiv (Nielsen & Wakeley, 2001; Nielsen, 2002) to distinguish between incomplete lineage sorting and ongoing gene flow between members of the Indochinese populations assigned to S. o. ochracea and S. o. kinneari, as well as between the main haplotype clusters that resulted from the BI and haplotype network; Mdiv simultaneously estimated the posterior distributions of the parameters theta (θ, effective population size scaled to the mutation rate), M (migration rate between populations since divergence) and T (time since population divergence). A series of simulations were conducted using various analytical parameters and prior combinations: the number of iterations varied between 10 and 50 million with a burn-in period corresponding to one-tenth of the sampled iterations, Mmax was set between 5 and 10, and Tmax was specified between 1 and 10. A finite-sites model (Hasegawa–Kishino–Yane model, or HKY model) was used in all Mdiv analyses. Values for θ and M were plotted, and the mode of the posterior distribution was considered the best estimate. A 95% credibility interval (CI) was estimated for the parameters θ and M.

Recent reviews have highlighted the fact that the commonly used rates of 1.6% or 2.0% divergence per million years (often cited as the ‘mitochondrial molecular clock’) may not be generalizable owing to differences in rates of molecular evolution among lineages and between mitochondrial genes (Warren et al., 2003; Garcia-Moreno, 2004; Lovette, 2005; Pereira & Baker, 2006). In addition to this first problem is the fact that rates of molecular evolution vary through time. Indeed, rates within the 0–2 Myr bp window, which corresponds to the expected time frame of divergence within species, are faster than those in more ancient geological times, probably owing to the effect of purifying selection (Ho et al., 2005; Ho & Larson, 2006). The estimation of divergence times within species thus appears particularly challenging unless a rate of molecular evolution can be derived from an independent calibration point situated within the 0–2 Myr window. Such data are lacking within Sasia. In a previous study, we used the split between S. ochracea and S. abnormis, thought to have occurred between 5.5 and 4.5 Myr bp, as a calibration point to estimate the divergence times within the piculets (see Fuchs et al., 2006 and Woodruff, 2003 for more details). Here we use this calibration point again to estimate the timing of the first divergence time within S. ochracea. For this purpose, we used the topology of the Bayesian analyses as an input tree and estimated this divergence time using Paml 3.14β (Yang, 2003) under a molecular-clock hypothesis and a General Time Reversible (GTR) model. We manually rooted our tree with a sequence of Sasia africana (DQ188174; Fuchs et al., 2006), the sister-group of the S. ochracea/S. abnormis clade (Benz et al., 2006; Fuchs et al., 2006). We previously tested the applicability of the molecular-clock hypothesis to our data set using the likelihood ratio test (LRT) [inline image, where λ1 is the likelihood of the restricted model; Huelsenbeck & Rannala (1997)], which follows a chi-squared distribution with n − 2 degrees of freedom, where n is the number of taxa. For this purpose, we used Paml 3.14β (Yang, 2003) and the topology resulting from the BI. The results of the LRT analyses indicated that our data fitted a model with a constant rate of molecular evolution (−lnclock = 2203.46; −lnnon-clock = 2175.44; 2 ln = 56.04; d.f. = 44, = 0.11). Given that our calibration is situated outside the 0–2 Myr window, i.e. it should result in an underestimate of the rate of molecular evolution within S. ochracea, we will consider our estimate as the oldest possible age for the first split within S. ochracea and discuss it accordingly.

Results

We obtained a minimum of 801 bp of the ND2 gene for all individuals included in the study. These nucleotides correspond to the positions 5252–6052 of the chicken mitochondrial genome (Gallus gallus, X52392, Desjardins & Morais, 1990). All newly generated sequences have been deposited in the GenBank data base (accession numbers EU30447EU30483). Within Sasia ochracea, 48 variable sites were detected, resulting in 27 distinct haplotypes. Uncorrected p distances varied between 0% and 2.1%. The 2.1% distance was recovered between one individual from Sagaing (Myanmar, USNM B-06057) and two individuals sharing the same haplotype from Dac To (southern Vietnam, MNHN CG 1927-499/MNHN CG 1927-500). The number of haplotypes, gene and nucleotide diversities, and the results of the tests of selective neutrality for each putative subspecies are given in Table 1.

Table 1.   Number of haplotypes, haplotype diversity (H), nucleotide diversity (π), Tajima’s D and Fu’s fs statistics obtained for each subspecies.
 Sasia ochracea ochraceaSasia ochracea kinneari
Number of individuals3211
Number of haplotypes1910
Number of polymorphic sites3912
H0.9320.982
π0.008660.00367
Tajima’s D−1.03317 (> 0.16)−1.22056 (> 0.12)
Fu’s fs−4.81941 (= 0.05)−6.90375 (< 0.01)

The 50% majority-rule consensus tree (−ln = 1868.01) resulting from the Bayesian analysis was largely unresolved; only three nodes received posterior probabilities equal to or greater than 0.95 (Fig. 2). The haplotype network (maximum connection steps at 95% = 12) revealed five main haplotype lineages that are differentiated from each other by five to nine substitutions (Fig. 3): (1) southern Vietnam (South Annam and ‘Cochinchina’), (2) Thailand, (3) Nepal and India (West Bengal State), (4) India (Meghalaya State) and Myanmar, and (5) Laos, central and northern Vietnam, China and Thailand (hereafter referred as Indochina sensu stricto). The Indochinese sensu stricto cluster is sub-divided into two haplotype groups that are separated by two substitutions. Overall, the haplotype clusters have a biogeographical rather than a taxonomic component. In all but one case, individuals sampled in the same or close localities have identical or nearly identical haplotypes. The exception involves two individuals sampled in Umphang (central west Thailand), for which the two haplotypes differ by nine substitutions. One of these two is the only one sampled individual of a lineage that differs by at least eight substitutions from all other haplotype clusters (Fig. 3).

Figure 2.

 50% majority-rule consensus tree (−ln = 1868.01) obtained from the Bayesian analysis using a GTR + I model. Values represent posterior probabilities greater than 0.95. Acronyms represent tissue or voucher numbers. Shading types indicate individuals attributed to the subspecies ochracea (black) and kinneari (grey).

Figure 3.

 Unrooted haplotype network of the White-browed Piculet as computed using tcs 1.18 (Clement et al., 2000). Circled areas are proportional to the number of individuals possessing this haplotype. Shading corresponds to the subspecies Sasia ochracea ochracea (grey) and S. o. kinneari (white). Extinct or unsampled haplotypes are indicated by dots.

The results of the amova indicated significant structuring of the genetic variability when partitioning by subspecies (d.f. = 42, φst = 0.17, < 0.001), even though most of the molecular variability was found within subspecies. Partitioning the individuals into the five main haplotype clusters recovered in the BI tree and statistical parsimony network yielded a φst value of 0.76 (d.f. = 42, < 0.001), suggesting that grouping the individuals by geography has a better fit to the data than grouping them by subspecies (note that some groups contain fewer than five individuals).

The bell shape of the S. o. kinneari mismatch distribution (P-values of fit to the sudden expansion model could not be calculated for kinneari owing to a lack of convergence in the least-squares procedures) indicates a recent population expansion, a result in agreement with the significant value of Fu’s fs test (Fig. 4, Table 1). In contrast, the multimodal shape of the mismatch distribution of S. o. ochracea indicates that this taxon is composed of several genetically differentiated populations (Fig. 4), albeit the mismatch distribution does not differ significantly from a sudden expansion model (test of goodness of fit: sum of squared deviation = 0.02, = 0.26; Harpending’s raggedness index = 0.02, = 0.58). The mismatch distribution of the Indochinese sensu stricto cluster (28 individuals, Fig. 4) as well as the significantly large negative value of Fu’s fs (Fs = −11.42, < 0.001) are also in accordance with a hypothesis of population expansion (test of goodness of fit: sum of squared deviation = 0.004, = 0.99; Harpending’s raggedness index = 0.016, = 0.96). We did not investigate the mismatch distributions for the four remaining haplotype clusters owing to the limited number of sampled individuals.

Figure 4.

 Mismatch distributions for Sasia ochracea ochracea, S. o. kinneari and Indochinese sensu stricto populations obtained with the program Arlequin version. 2.0 (Schneider et al., 2000). Grey bars represent observed data, and lines represent the simulated data used to test the goodness of fit to the sudden expansion model.

Estimates for θ and M were 9.98 (95% CI = 6.50–16.75) and 0.98 (95% CI = 0.34–5.58), respectively, when the individuals were grouped by subspecies. Thus, the coalescent-based analyses suggest a non-zero migration rate between the two subspecies (Fig. 5). The pairwise comparison analyses between the main haplotypes clusters recovered in the BI tree and haplotype network always yielded estimates of gene flow between clusters close to zero (Table 2). The estimates for T (time since population divergence) were difficult to ascertain for all the pairwise comparisons that did not involve the Indochinese sensu stricto cluster, as the posterior distribution for this parameter never converged, even if the chains were run for 50-million iterations (data not shown). Estimates for T for all the pairwise comparisons involving the Indochina sensu stricto cluster were very similar to each other (Indochina sensu stricto-Myanmar/India = 1.14, Indochina sensu stricto-India/Nepal = 1.2, Indochina sensu stricto-southern Vietnam 1.2), suggesting that all the splits involving the Indochina sensu stricto cluster occurred synchronously.

Figure 5.

 Posterior distributions of M (migration rate) for the comparison Sasia ochracea ochracea/S. o. kinneari based on analysis of 801 bp of mtDNA using the program mdiv (Nielsen, 2002). Two simulations were conducted (10- and 50-million iterations), and the result from the longer run is presented here. Priors were set as Tmax = 10 and Mmax = 10, and a Hasegawa-Kishino-Yane (HKY) model was used in all analyses. The posterior distribution of the migration rate indicates the occurrence of limited gene flow between populations assigned to S. o. ochracea and S. o. kinneari.

Table 2.   Mdiv estimates of θ and migration rate (M) between Sasia ochracea ochracea and S. o. kinneari and between the main haplotype clusters recovered in the BI tree and haplotype network.
Population 1Population 2θ (95% CI)M (95% CI)
  1. 95% credibility intervals are given in brackets for each parameter. We did not perform Mdiv analyses with the haplotype lineage represented by only one individual.

S. o. ochraceaS. o. kinneari9.98 (6.50–16.75)0.98 (0.34–5.58)
Indochina sensu strictoNepal/Myanmar7.32 (4.30–14.21)0.10 (0.01–1.19)
Indochina sensu strictoIndia/Nepal6.29 (3.36–11.29)0.08 (0.01–1.08)
Indochina sensu strictoCochinchina7.46 (4.21–14.41)0.07 (0.01–1.61)
Nepal/MyanmarIndia/Nepal1.84 (1.05–7.10)0.04 (0.01–4.66)
Nepal/MyanmarCochinchina1.91 (1.04–7.27)0.02 (0.01–7.76)
India/NepalCochinchina3.81 (1.73–12.16)0.1 (0.015–4.60)

We estimated that the first split within S. ochracea occurred during the Pleistocene, between 400,000 ± 76,900 yr bp (using 5.5 Myr bp as a calibration point) and 333,000 ± 64,600 yr bp (using 4.5 Myr bp as a calibration point). These values yielded rates of molecular evolution of between 0.021 ± 0.002 substitution/site/Myr (using 4.5 Myr bp as a calibration) and 0.017 ± 0.001 substitution/site/Myr (using 5.5 Myr bp as a calibration point).

Discussion

Our study has provided a detailed analysis of genetic variation within a widespread Indochinese species of bird, the White-browed Piculet.

Genetic variation and endemicity

Our analyses revealed that S. ochracea is composed of at least five lineages that strongly correlate with their geographic locations: (1) south Vietnam (South Annam and ‘Cochinchina’), (2) India (West Bengal State) and Nepal, (3) Myanmar and India (Meghalaya State), (4) Indochina sensu stricto, and probably (5) Thailand (Fig. 3). The analysis of molecular variance indicated that the current taxonomy does not adequately reflect the amount of genetic variation within S. ochracea. Indeed, the vast majority of the genetic variation within S. ochracea is nested within the nominal subspecies, which occupies a large geographic area, ranging from Nepal to southern Vietnam. The fact that grouping the individuals by geography provides a better fit to the data than grouping them by subspecies reinforces our primary assumption, based on ethological and morphological data (tiny size, short tail, rounded wings), that this species has limited dispersal capabilities.

The shape of the mismatch distribution, the significant P-value of Fu’s fs test, and the fact that some S. o. kinneari individuals share haplotypes with members of the Indochinese sensu stricto cluster indicate that the subspecies kinneari, endemic to North Vietnam (‘Tonkin’) and South China, could represent a recent population expansion from the Indochinese cluster (Laos, North Annam). Our coalescent-based analyses further suggest that gene flow (one female per generation) has not been interrupted between kinneari and the remaining individuals assigned to the Indochinese sensu stricto cluster.

Biogeography

We estimated that the first split within S. ochracea occurred during the Pleistocene, between 400,000 ± 76,900 and 333,000 ± 64,600 yr bp, depending on the value of the calibration point used, i.e. 4.5 or 5.5 Myr bp. Our estimate should be regarded as the oldest age possible for this divergence since we used a calibration point that is outside the 0–2 Myr window, for which the rates of molecular evolution tend to be faster than in the time window within which our calibration point is situated (Ho et al., 2005; Ho & Larson, 2006). Therefore, it is likely that the first split within S. ochracea occurred more recently than our estimate. Interestingly, our dates are overall very similar to those of Zou et al. (2007), who cautiously estimated that the main lineages among Alcippe morrisonia, a species with similar habitat requirements to those of S. ochracea, appeared c. 127,329 years ago (95% CI = 55,900–425,000 yr bp). At this point, it is worth noting that Southeast Asia has experienced several important climatic oscillations during the last 400,000 years (Hope et al., 2004). About 355,000 yr bp, trees from Fagaceae (mainly Quercus and Castanopsis) began to expand in South China and have become dominant in forests since then. The sudden expansion of Fagaceae in the pollen assemblage about 355,000 yr bp might imply strengthened seasonality and a cooler climate than previously (Sun et al., 2003). The relative climatic cooling during this late period is confirmed by the rise in pollen percentages from boreal conifers (Sun et al., 2003). Most of the differentiated lineages highlighted throughout this paper involve mountainous areas (Annamites, Nepal, Myanmar, Tenasserim). Thus, the expansion of the Fagaceae forest type may have restricted the various lineages of White-browed Piculet to mountainous refugia where suitable habitat persists. It is also worth noting that all the splits between the main haplotype clusters occurred synchronously, suggesting that the same external factor, which is likely to be climate-driven, promoted the isolation of these lineages. Nevertheless, the exact timing of these events remains to be determined with the addition of data both on rates of molecular evolution during the 0–2 Myr window and on palaeoenvironmental reconstructions during the last 400,000 years in Nepal, Myanmar, Laos, and Vietnam (Hope et al., 2004).

Comparisons of our results with other studies of Southeast Asian phylogeography are difficult. In some cases, these other studies were performed at different geographical scales or were focused on the Sunda shelf (e.g. Enicurus leschenaultii, Moyle et al., 2005; Cynopterus fruit bats, Campbell et al., 2006; Alcippe morrisonia, Zou et al., 2007). In other cases, the distribution of the study species is too restricted (e.g. Garrulax canorus, Li et al., 2006), or the habitats of the study organisms are different from that of S. ochracea (e.g. rice, Londo et al., 2006; dhole, Iyengar et al., 2005). However, some congruent patterns appear concerning regions with high genetic distinctiveness.

Our results indicate that two highly divergent haplotypes (differing by nine substitutions) occur at Umphang (central western Thailand). The first haplotype belongs to the Indochinese sensu stricto cluster (Laos, China, Thailand, North Annam and Tonkin), whereas the second is the only representative of another well-differentiated lineage. Interestingly, the geographic localization of the sampling point (close to the Thailand/Myanmar border) is adjacent to the range of S. o. reichenowi (Tenasserim and extreme south Thailand), a taxon that we have not been able to sample. An explanation of the coexistence of two differentiated haplotypes in central western Thailand could be the presence of a secondary contact zone as a result of population expansions of the Indochinese and Tenasserim clusters. Interestingly, Stuart et al. (2006), using mitochondrial DNA sequence data, suggested the syntopic occurrence of two differentiated and phylogenetically unrelated populations of frogs, previously attributed to Odorrana livida, close to the Myanmar/Thailand border. The localization of their sampling site is slightly south of ours. Likewise, Veron et al. (2007) highlighted a genetic break in Myanmar between populations of mongooses (Herpestes auropunctatus and Herpestes javanicus) previously considered conspecific or closely related. Thus, these results suggest that the occurrence of a contact zone between well-differentiated populations in the Tenasserim mountains may be applicable for several groups of vertebrates. Furthermore, the distributional limits of several other closely related Indo-Malayan bird species (e.g. Picus vittatusPicus viridanus, Iole olivaceaIole virescens, and Alophoixus pallidusAlophoixus flaveolus) coincide with the Thailand/Myanmar border. Further sampling along the Thailand/Myanmar border as well as sequencing of nuclear genes (both autosomal and sex-linked) will be necessary to quantify the extent of this secondary contact zone and the amount of gene flow between these populations/taxa.

The other area for which high genetic distinctiveness has been suggested is Vietnam, in the southern parts of the Annamites mountains (formerly ‘Cochinchina’). This has been suggested for birds (this study, Zou et al., 2007) and also for frogs, for which syntopic populations previously thought to belong to the same species (Odorrana livida) and that are well differentiated genetically occur a few metres apart (Stuart et al., 2006). These results suggest that this area might also act as a contact zone between various expanding populations. The genetic peculiarity of the southern Vietnam area has also been suggested by Gorog et al. (2004), who showed, using mitochondrial DNA sequences, that populations from southern Vietnam assigned to Maxomys surifer (Rodentia) are closer to conspecific populations from the Malay Peninsula and the Sunda shelf than to populations from northern Vietnam (Ha Tinh Province). Interestingly, some taxonomic and distributional data for birds are also in accordance with such a hypothesis. For example, the distributional range of Alophoixus ochraceus (Passeriformes) is thought to be Borneo, Java, Sumatra, south-west Cambodia and southern Vietnam, whereas that one of its closest relative, Alophoixus pallidus, encompasses the rest of Indochina. Thus, the southern parts of Vietnam are notable not only because they hold populations that are genetically differentiated from those of the rest of Southeast Asia but also because the relationships of these populations are either with populations from continental Indochina or with populations from the Sunda shelf. Further work, including additional sampling of both taxa and genes (autosomal and sex-linked) may help to address in more detail the exact patterns of genetic structure in these regions.

While our comparisons yielded some congruent patterns in space across taxa with very different life-history traits and biology (frogs, mammals and birds), it is still necessary to evaluate whether all these splits occurred synchronously in response to the same external factor (e.g. climatic variations) or whether they occurred randomly. The state of knowledge on fossils from the region as well as our still only partial understanding of rates of molecular evolution at that time-scale currently prevent such comparisons.

Taxonomic recommendations at the subspecies level

Overall, the current number of recognized subspecies does not fit the number of genetically differentiated lineages as revealed by our molecular data. This agrees with many previous observations of discrepancies between morphological and molecular data at the subspecies level in birds (Zink, 2004; but see also Phillimore & Owens, 2006).

Our analyses identify at least five genetically differentiated lineages within the nominate subspecies S. o. ochracea. One of these lineages (the Indochinese cluster sensu stricto) also contains members of the kinneari lineage, and we found evidence for ongoing gene flow among the ochracea and kinneari populations of the Indochinese sensu stricto cluster. In its extreme, this result could be used to suggest that the nominate subspecies may be split into five taxonomic units, especially if we consider that the amount of ongoing gene flow between the five main haplotype clusters is zero or thereabouts. Furthermore, there is considerable variation in the plumage colour of museum skins from the extreme south (southern Vietnam) and north (China) of the breeding range (J.F., personal observation). Birds from the south are clearly lighter (light orange) than birds from the north, which are more orange-brown. However, birds collected in the intermediate areas (central Vietnam and Laos) are intermediate in plumage colouration intensity, suggesting a clinal variation of this character. In addition, we find no other character that correlates with the genetic data. Consequently, we refrain from proposing taxonomic changes at the subspecies level until (1) more complete geographical sampling is achieved in some key areas (e.g. southern Vietnam, Tenasserim) and in the area where a potential secondary contact zone occurs (central west Thailand), and (2) nuclear genes are added to the data set.

Acknowledgements

We are very grateful to P. Sweet and J. Cracraft (AMNH), J. Bates, S. Hackett and D. Willard (FMNH), M. Braun and J. Dean (USNM), J. Fjeldså and J.B. Kristensen (ZMUC), and M.V. Kalyakin (ZMMU) for kindly sending us tissues or toe-pad samples, and to M.B. Robbins and A.T. Peterson for information on the KU specimen. Help during laboratory work was kindly provided by A. Tillier, C. Bonillo and J. Lambourdière at MNHN and by D. Zuccon, M. Irestedt and P. Eldenäs at NRM. Laboratory work was supported at MNHN by the ‘Service Commun de Systématique Moléculaire’, IFR CNRS 101, MNHN and by the Plan Pluriformation ‘Etat et structure phylogénétique de la biodiversité actuelle et fossile’, and at NRM by the Swedish Research Council (grant no. 621-2004-2913 to P.E.). We acknowledge support from a SYNTHESYS grant made available to J.F. by the European Community - Research Infrastructure Action under the FP6 ‘Structuring the European Research Area’ programme (SE-TAF-746), during which I. Bisang provided invaluable help. The Phongsaly Forest Conservation and Rural Development Project, a Lao-European cooperation, is acknowledged, and its staff, especially P. Rousseau, C. Hatten, Y. Varelides, R. Humphrey and Y. Tipavanh, are thanked for their assistance and company during the fieldwork of J.F. with A. Cibois, M. Ruedi and R. Kirsch. Y. Laissus, R. Pujol and the ‘Société des Amis du Muséum’ provided financial support for the fieldwork of J.F. and are sincerely acknowledged. Fieldwork in Vietnam was made in collaboration with the Institute of Ecology and Biological Resources. Nguyen Cu, Le Manh Hung, P. Alström, I. Cederholm, M. Irestedt, P. Mortensen, P. Nilsson and J. Ohlson participated in this work, which was supported by a Swedish Research Links grant from the Swedish Research Council. We thank M. Patten and two anonymous referees for providing helpful comments on a previous draft of this manuscript, and Hannah Thomas for improving the English.

Biosketches

Jérôme Fuchs completed his PhD degree at the University Pierre and Marie Curie, Paris, under the supervision of Professor Eric Pasquet. His thesis work focused on the relative contributions of dispersal and vicariant events to bird evolution as found from molecular phylogeny and dating. His research interests include historical biogeography, avian evolution and molecular phylogeny.

Per Ericson is Head of the Department of Vertebrate Zoology at the Swedish Museum of Natural History and an associate professor of zoology at the University of Stockholm. His research interests include avian evolution, systematics and biogeography.

Eric Pasquet is a research officer at the Department of Systematics and Evolution (MNHN, Paris). He is curator of the bird collection at the Museum National d’Histoire Naturelle, Paris, and manager of the Molecular Systematic Facility of the Museum. His research focuses on the molecular phylogeny of birds.

Editor: Michael Patten

Ancillary