Geography determines genetic relationships between species of mountain pine (Pinus mugo complex) in western Europe

Aim Our aims were to test whether morphological species of mountain pines were genetically supported in the western part of the distribution range of the Pinus mugo species complex (Pinus mugo Turra sensu lato), to resolve genetically homogeneous clusters of populations, to determine historical demographic processes, and to assess the potential hybridization of mountain pines with Scots pine, Pinus sylvestris L. Location Populations were sampled in the Iberian System, the Pyrenees, the French Mont Ventoux, Vosges and Jura mountains, the German Black Forest and throughout the Alps. This corresponded to a range‐wide sampling for mountain pine sensu stricto (Pinus uncinata Ram.) and to a sampling of the western parts of the ranges of dwarf mountain pine (Pinus mugo Turra sensu stricto) and bog pine/peatbog pine [Pinus rotundata Link/Pinus × pseudopumilio (Willk.) Beck]. Methods In total, 786 individuals of P. mugo sensu lato from 29 natural populations, and 85 individuals of P. sylvestris from four natural populations were genotyped at three chloroplast microsatellites (cpSSRs). Populations were characterized for standard genetic diversity statistics and signs of demographic expansion. Genetic structure was explored using analysis of molecular variance, differentiation statistics and Bayesian analysis of population structure (BAPS). Results One hundred haplotypes were identified in P. mugo sensu lato. There was a stronger differentiation between geographical regions than between morphologically identified taxa (P. mugo sensu stricto, P. uncinata and P. rotundata/P. ×pseudopumilio). Overall genetic differentiation was weak (G_ST = 0.070) and displayed a clear phylogeographic structure [N_ST = 0.263, N_ST > N_ST (permuted), P < 0.001]. BAPS identified a Pyrenean and an Alpine gene pool, along with several smaller genetic clusters corresponding to peripheral populations. Main conclusions The core regions of the Pyrenees and Alps were probably recolonized, respectively by P. uncinata and P. uncinata/P. mugo sensu stricto, from multiple glacial refugia that were well connected by pollen flow within the mountain chains. Pinus rotundata/P. × pseudopumilio populations from the Black Forest, Vosges and Jura mountains were probably recolonized from various glacial populations that kept their genetic distinctiveness despite late glacial and early Holocene expansion. Marginal P. uncinata populations from the Iberian System are compatible with elevational shifts and long‐term isolation. The causes of haplotype sharing between P. mugo sensu lato and P. sylvestris require further research.     

Allozyme investigations on the genetic differentiation between closely related pines —Pinus sylvestris, P. mugo, P. uncinata, andP. uliginosa (Pinaceae)

In eight natural European populations of four closely related taxa of pines (Pinus sylvestris, P. mugo, P. uncinata, andP. uliginosa) starch-gel electrophoreses revealed altogether 58 alleles at 15 loci from nine enzyme systems. With Nei's genetic distance (D) the largest mean genetic distance (D = 0.171) was found betweenP. sylvestris andP. mugo, a distance corresponding to that between other closely related pine species. Mean genetic distances between the other taxa were less than half that value and characteristic for subspecies or varieties rather than for species. On the basis of our results we suggest that both,P. uliginosa andP. uncinata, could be the result of the ancient hybridization betweenP. sylvestris andP. mugo.     

Volatiles as Chemosystematic Markers for Distinguishing Closely Related Species within the Pinus mugo Complex

Headspace solid‐phase microextraction (HS‐SPME) coupled to GC/MS analysis was used to identify the constituents of pine‐needle volatiles differentiating three closely‐related pine species within the Pinus mugo complex, i.e., P. uncinata Ramond ex DC., P. uliginosa G.E.Neumann ex Wimm ., and P. mugo Turra . Moreover, chemosystematic markers were proposed for the three analyzed pine species. The major constituents of the pine‐needle volatiles were α‐pinene (28.4%) and bornyl acetate (10.8%) for P. uncinata, δ‐car‐3‐ene (21.5%) and α‐pinene (16.1%) for P. uliginosa, and α‐pinene (20%) and δ‐car‐3‐ene (18.1%) for P. mugo. This study is the first report on the application of the composition of pine‐needle volatiles for the reliable identification of closely‐related pine species within the Pinus mugo complex.     

Molecular signatures of divergence and selection in closely related pine taxa

Efforts to detect loci under selection in plants have mostly focussed on single species. However, assuming that intraspecific divergence may lead to speciation, comparisons of genetic variation within and among recently diverged taxa can help to locate such genes. In this study, coalescent and outlier detection methods were used to assess nucleotide polymorphism and divergence at 79 nuclear gene fragments (1212 SNPs) in 16 populations (153 individuals) of the closely related, but phenotypically and ecologically distinct, pine taxa Pinus mugo, P. uliginosa and P. uncinata across their European distributions. Simultaneously, mitochondrial DNA markers, which are maternally inherited in pines and distributed by seeds at short geographic distance, were used to assess genetic relationships of the focal populations and taxa. The majority of nuclear loci showed homogenous patterns of variation between the taxa due to a high number of shared SNPs and haplotypes, similar levels of polymorphism, and low net divergence. However, against this common genetic background and an overall low population structure within taxa at mitochondrial markers, we identified several genes showing signatures of selection, accompanied by significant intra- and interspecific divergence. Our results indicate that loci involved in species divergence may be involved in intraspecific local adaptation.     

Contribution to the taxonomy of Pinus uncinata (Pinaceae) based on cone characters

Eight natural populations of Pinus uncinata Ramond from the Spanish Pyrenees, five of P. mugo Turra from the Carpathians, the Sudethians and Abruzzian Apennines, three of P. sylvestris L. from the Central-European Lowland and the Iberian Peninsula and one of P. uliginosa Neumann from SW Poland have been compared biometrically in 16 cone characters. P. uncinata appears the most distant from all other taxa, differing by having the largest and most asymmetric cones and the thickest apophyses with the most prominent umbo. In cone characters, P. mugo is more similar to P. sylvestris than to P. uncinata. Nevertheless, P. mugo and P. sylvestris remain distant from each other in a number of morphological characters examined. P. uliginosa has cones intermediate between the other three taxa and is moreover very variable. The locality of P. uliginosa in the Silesian Lowland may be of a relict character and conserve genes from P. uncinata.     

Pinus uncinata Ramond taxonomy based on needle characters

 11 needle characters of Pinus uncinata Ramond from the Spanish Pyrenees, P. mugo Turra from the Tatra Mountains, and P. sylvestris L. from N.E. Poland were analysed. It was shown that P. uncinata is much more closely related to P. mugo than to P. sylvestris. Nevertheless P. uncinata remains distinct from both in a number of the characters examined. The species differs from P. mugo especially in the length of the needles and in the distance between the vascular bundles. It differs from P. sylvestris in such characters as the number of resin canals, thickness of epidermis cells and the distance between the vascular bundles. Width of epidermis cells was similar in all taxa studied. Received May 31, 2000 Accepted February 13, 2001     

Climatic niche and neutral genetic diversity of the six Iberian pine species: a retrospective and prospective view

Quaternary climatic fluctuations have left contrasting historical footprints on the neutral genetic diversity patterns of existing populations of different tree species. We should expect the demography, and consequently the neutral genetic structure, of taxa less tolerant to particular climatic extremes to be more sensitive to long‐term climate fluctuations. We explore this hypothesis here by sampling all six pine species found in the Iberian Peninsula (2464 individuals, 105 populations), using a common set of chloroplast microsatellite markers, and by looking at the association between neutral genetic diversity and species‐specific climatic requirements. We found large variation in neutral genetic diversity and structure among Iberian pines, with cold‐enduring mountain species (Pinus uncinata, P. sylvestris and P. nigra) showing substantially greater diversity than thermophilous taxa (P. pinea and P. halepensis). Within species, we observed a significant positive correlation between population genetic diversity and summer precipitation for some of the mountain pines. The observed pattern is consistent with the hypotheses that: (i) more thermophilous species have been subjected to stronger demographic fluctuations in the past, as a consequence of their maladaptation to recurrent glacial cold stages; and (ii) altitudinal migrations have allowed the maintenance of large effective population sizes and genetic variation in cold‐tolerant species, especially in more humid regions. In the light of these results and hypotheses, we discuss some potential genetic consequences of impending climate change.     

Geographical patterns of nucleotide diversity and population differentiation in three closely related European pine species in the Pinus mugo complex

Nucleotide polymorphism at 12 nuclear loci and two mitochondrial gene fragments was studied in three closely related pine species from the Pinus mugo complex in populations across the species distributional range in Europe. Despite large differences in the census sizes of the populations, high and similar levels of nucleotide diversity (θ_sil = ～0.013–0.017) were found at nuclear loci in the three pine species. More rapid decay of overall linkage disequilibrium (LD) and recombination to diversity ratio (ρ/θ) was observed across the species distributional range in P. mugo (ρ = 0.0369 ± 0.0028; ρ/θ = ～2.2) than in P. uncinata (ρ = 0.0054 ± 0.0011; ρ/θ = ～0.4) and P. uliginosa (ρ = 0.0051 ± 0.0010, ρ/θ = ～0.4). However, regional groups of P. mugo showed similar levels of LD and ρ/θ ratio to the other species. An excess of rare nucleotide variants was found in P. mugo at four loci, but, overall, the allelic frequency spectrum in the three species did not deviate significantly from neutrality (multilocus Tajima's D = ?0.681, D = ?0.118 and D = ?0.266, P > 0.05, respectively). Bayesian clustering methods showed no clear correspondence of clusters to species or geographical regions. Some differences between populations and species were found in a hierarchical analysis of molecular variance (AMOVA) and in the distribution of the mitochondrial DNA haplotypes, suggesting rather limited gene flow between the taxa and ongoing divergence. As all three pine taxa have similar genetic backgrounds, they form an excellent system for searching for loci involved in adaptive variation and speciation. © 2013 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 172 , 225–238.     

Cytoplasmic composition in Pinus densata and population establishment of the diploid hybrid pine

Sequence and restriction site analyses of the paternally inherited chloroplast rbcL gene and maternally inherited mitochondrial nad1 fragments from the same set of populations and individuals were used to investigate cytoplasmic composition and population establishment of Pinus densata, a diploid pine that originated through hybridization between P. tabuliformis and P. yunnanensis. Two variable sites and three chlorotypes (TT, TC and GC) were detected on the rbcL gene of the three pines. P. densata harboured the three chlorotypes, two of which (TT, GC) were characteristic of the parental species, respectively. The third chlorotype (TC) was distributed extensively in seven of the 10 P. densata populations analysed, and might represent a mutation type or have been derived from an extinct parent. The distribution of chlorotypes, together with that of mitotypes, indicated that significant founder effect and backcross happened during the population establishment of the hybrid pine. P. tabuliformis and P. yunnanensis had acted as both mother and father donors, i.e. bi-directional gene flow existed between the two parental species in the past. Population differentiation of P. densata is high, as detected from the cytoplasmic genomes: GST = 0.533 for cpDNA and GST = 0.905 for mtDNA. The differences in cytoplasmic composition among the hybrid populations suggest that the local populations have undergone different evolutionary histories.     

Maternal lineages of Pinus densata,a diploid hybrid

Previous morphological, allozyme and chloroplast DNA data have suggested that Pinus densata originated through hybridization between P. tabuliformis and P. yunnanensis. In the present study, sequence and restriction site analyses of maternally inherited mitochondrial nad1 intron were used to detect variation patterns in 19 populations of P. tabuliformis, P. yunnanensis and P. densata. A total of three mitotypes (A, B, C) were detected. All but one of the populations of P. yunnanensis possessed mitotype B while all populations of P. tabuliformis had mitotype A. Pinus densata populations, on the other hand, harboured both mitotypes A and B, which are characteristic of P. tabuliformis and P. yunnanensis, respectively. This result gives strong additional evidence supporting the hybrid origin of this diploid pine. The distribution of mitotypes indicated very different mating compositions and evolutionary history among P. densata populations. It seems that local founder populations and backcrosses may have played important roles in the early establishment of P. densata populations. The uplift of the Tibetan Plateau had a significant impact on the distribution of maternal lineages of P. densata populations.     

Speciation history of three closely related pines Pinus mugo (T.), P. uliginosa (N.) and P. sylvestris (L.)

Nucleotide polymorphisms at genomic regions including 17 nuclear loci, two chloroplast and one mitochondrial DNA fragments were used to study the speciation history of three pine species: dwarf mountain pine (Pinus mugo), peat‐bog pine (P. uliginosa) and Scots pine (P. sylvestris). We set out to investigate three specific speciation scenarios: (I) P. uliginosa is a homoploid hybrid between the other two, (II) the species have evolved without gene flow after divergence and (III) there has been substantial gene flow between the species since their divergence. Overall, the genetic data suggest that P. mugo and P. uliginosa share the same gene pool (average net divergence of 0.0001) and that the phenotypic differences (e.g. growth form) are most likely due to very limited areas of the genome. P. mugo and P. uliginosa are more diverged from P. sylvestris than from each other (average net divergence of 0.0027 and 0.0026, respectively). The nucleotide patterns can best be explained by the divergence with migration speciation scenario, although the hybrid speciation scenario with small genomic contribution from P. sylvestris cannot be completely ruled out. We suggest that the large amount of shared polymorphisms between the pine taxa and the lack of monophyly at all loci studied between P. sylvestris and P. mugo–P. uliginosa can largely be explained by relatively recent speciation history and large effective population sizes but also by interspecific gene flow. These closely related pine taxa form an excellent system for searching for loci involved in adaptive variation as they are differentiated in phenotype and ecology but have very similar genetic background.     

Different patterns of genetic structure of relict and isolated populations of endangered peat-bog pine (Pinus uliginosa Neumann)

Recent changes in environmental conditions in populations of peat-bog pine (Pinus uliginosa Neumann) caused rapid decline or even extinction of the species in several stands in Central Europe. Conservation strategies forP. uliginosa require information about the evolutionary history and genetic structure of its populations. Using isozymes we assessed the genetic structure ofP. uliginosa from four isolated stands in Poland and compared the results to genetic structures of other closely related pine species including eight populations ofPinus mugo, ten ofPinus sylvestris and one ofPinus uncinata. The level of genetic variability ofP. uliginosa measured by the mean number of alleles per locus and average heterozygosity was similar to others related toP. uliginosa taxa from the reference group but it differs among populations. High genetic similarity was found between two populations ofP. uliginosa from Low Silesian Pinewood. The populations were genetically distinct as compared to other populations includinglocus classicus of the species from the peat bog at Batorów Reserve. Very low genetic distance (D_n = 0.002) and small genetic differentiation (GST = 0.003) were found betweenP. uliginosa andP. mugo in the sympatric populations of the species from Zieleniec peat bog suggesting the ongoing natural hybridisation and genetic contamination of peat-bog pine from this area. Some evidence for skew in allele frequency distribution potentially due to recent bottleneck was found in population from Low Silesian Pinewood. The analysed open pollinated progeny derived from twoP. uliginosa stands from Low Silesian Pine-wood showed the excess of homozygotes as compared to the maternal trees indicating high level of inbreeding (F = 0.105,F = 0.081). The results are discussed in the context of evolution ofP. uliginosa populations, taxonomic relationships between the analysed species and conservation strategies for active protection of peat-bog pine.     

Early phenology and growth trait variation in closely related European pine species

Closely related taxa occupying different environments are valuable systems for studying evolution. In this study, we examined differences in early phenology (bud set, bud burst) and early growth in a common garden trial of closely related pine species: Pinus sylvestris, P. mugo, and P. uncinata. Seeds for the trial were sourced from populations across the ranges of each species in Europe. Over first 4 years of development, clear differences were observed between species, while the most significant intraspecific differentiation was observed among plants from P. sylvestris populations from continental European locations. Trait differences within P. sylvestris were highly correlated with altitude and latitude of the site of origin. Meanwhile, P. mugo populations from the Carpathians had the earliest bud set and bud flush compared to other populations of the species. Overall, populations from the P. mugo complex from heterogeneous mountain environments and P. sylvestris from the Scottish Highlands showed the highest within‐population variation for the focal traits. Although the three species have been shown to be genetically highly similar, this study reveals large differences in key adaptive traits both among and within species.     

The postglacial history of Scots pine (Pinus sylvestris L.) in western Europe: evidence from mitochondrial DNA variation

The geographical structure of mitochondrial (mt)DNA variants (mitotypes) was investigated in 38 western European populations of Scots pine Pinus sylvestris using restriction fragment length polymorphism (RFLP) analysis of total DNA and a homologous cox1 probe. Three major mitotypes (designated a, b and d ) were detected. Within Spain all three major mitotypes were found, gene diversity was high, H_T = 0.586, and this diversity was distributed predominantly among rather than within populations (F_ST(M) = 0.813 for the seven Spanish populations). Mitotype d was present only in the most southerly population from the Sierra Nevada . Elsewhere in Europe, populations showed little or no mtDNA diversity within regions, but there were marked differences between regions. Italian populations were fixed for mitotype b ; populations from northern France, Germany, Poland, Russia and southern Sweden were fixed for mitotype a ; while populations in northern Fennoscandia were fixed for mitotype b . The isolated Scottish populations were predominantly of mitotype a , but mitotype b was present in three of the 20 populations scored. In Scotland, UK gene diversity (H_T = 0.120) and genetic differentiation among populations (F_ST(M) = 0.37) was much lower than in Spain. When interpreted in the light of complementary data from pollen analysis and nuclear genetic markers, the results suggest that present-day populations of P. sylvestris in western Europe have been derived from at least three different sources after glaciation.     

Reconstructing the plant mitochondrial genome for marker discovery: a case study using Pinus

Whole‐genome‐shotgun (WGS) sequencing of total genomic DNA was used to recover ~1 Mbp of novel mitochondrial (mtDNA) sequence from Pinus sylvestris (L.) and three members of the closely related Pinus mugo species complex. DNA was extracted from megagametophyte tissue from six mother trees from locations across Europe, and 100‐bp paired‐end sequencing was performed on the Illumina HiSeq platform. Candidate mtDNA sequences were identified by their size and coverage characteristics, and by comparison with published plant mitochondrial genomes. Novel variants were identified, and primers targeting these loci were trialled on a set of 28 individuals from across Europe. In total, 31 SNP loci were successfully resequenced, characterizing 15 unique haplotypes. This approach offers a cost‐effective means of developing marker resources for mitochondrial genomes in other plant species where reference sequences are unavailable.     

Extremely low nucleotide polymorphism in Pinus krempfii Lecomte,a unique flat needle pine endemic to Vietnam

Pinus krempfii Lecomte is a morphologically and ecologically unique pine, endemic to Vietnam. It is regarded as vulnerable species with distribution limited to just two provinces: Khanh Hoa and Lam Dong. Although a few phylogenetic studies have included this species, almost nothing is known about its genetic features. In particular, there are no studies addressing the levels and patterns of genetic variation in natural populations of P. krempfii. In this study, we sampled 57 individuals from six natural populations of P. krempfii and analyzed their sequence variation in ten nuclear gene regions (approximately 9 kb) and 14 mitochondrial (mt) DNA regions (approximately 10 kb). We also analyzed variation at seven chloroplast (cp) microsatellite (SSR) loci. We found very low haplotype and nucleotide diversity at nuclear loci compared with other pine species. Furthermore, all investigated populations were monomorphic across all mitochondrial DNA (mtDNA) regions included in our study, which are polymorphic in other pine species. Population differentiation at nuclear loci was low (5.2%) but significant. However, structure analysis of nuclear loci did not detect genetically differentiated groups of populations. Approximate Bayesian computation (ABC) using nuclear sequence data and mismatch distribution analysis for cpSSR loci suggested recent expansion of the species. The implications of these findings for the management and conservation of P. krempfii genetic resources were discussed.     

Lack of evidence on hybrid swarm in the sympatric population of Pinus mugo and P. sylvestris

Natural hybridisation between the taxa from Pinus mugo complex and P. sylvestris was postulated in several sympatric populations of the species in Europe. However, due to the absence of precise methods for identification of hybrid seeds and hybrid trees, the frequency of hybridisation and its influence on the genetic structure of relict P. mugo populations has not been clarified so far. In the present study, the species-specific chloroplast DNA (cpDNA) marker and isozymes were applied to test the hybridisation hypothesis in a postulated hybrid swarm population of the species from “Bór na Czerwonem” reserve at the northern foothills of Tatra Mts., Poland. The cpDNA marker was used to identify hybrids within two groups of polycormic (multi-stemmed) and monocormic (single-stemmed) pines from this population. Allelic frequencies at isozyme loci from both groups were compared to frequencies found in allopatric populations of the species. Additionally, cpDNA haplotypes of seedlings derived from open pollinated seeds were studied to detect the possibility of successful cross-pollination. The mixture of seedlings with P. sylvestris and P. mugo cpDNA haplotypes was derived from one parental tree that indicates hybridisation. However, all the mature polycormic pines had cpDNA haplotypes species specific to P. mugo and the isozyme frequencies were similar to frequencies found in three allopatric populations of P. mugo from Tatra Mts. (mean genetic distance, Dn=.027). The differences were much larger in comparisons with monocormic pines from this area (Dn=.085) and two P. sylvestris samples from distant allopatric populations (Dn=.077). Nearly all monocormic pines had cpDNA species specific to P. sylvestris and isozyme frequencies similar to other populations of this species (mean Dn=.004). Only one P. sylvestris-like monocormic tree had cpDNA of P. mugo and can be considered as a hybrid. The results do not provide evidence supporting the hybrid swarm hypothesis. Rather, the results suggest that mature hybrids have a low frequency within this population and (rare) hybridisation is not reciprocal but unidirectional with P. mugo as pollen donor.     

DNA marker based identification of spontaneous hybrids between Pinus mugo and P. sylvestris at the Lithuanian sea‐side

The objective was to identify spontaneous hybrids between P. mugo and P. sylvestris using organelle DNA markers in sympatric zones at the sea‐side spit of Kursiu Nerija in western Lithuania. A field inventory was carried out over the entire Lithuanian part of the spit and 203 individuals morphologically intermediate between P. sylvestris and P. mugo were tested for their male parent with chloroplast DNA PCR‐RFLP markers and for their female parent with mitochondrial DNA PCR markers. Unfortunately, the mitochondrial DNA Nad7 marker failed to identify the female parent species in our study. However, the chloroplast DNA PCR‐RFLP marker revealed that out of 203 tested individuals only 23 had a paternity different than that indicated by morphology. Of these, 13 individuals were morphologically identified as P. sylvestris but possessed cpDNA of P. mugo (putative hybrids with P. sylvestris (female) ×P. mugo (male parent), and 10 individuals morphologically identified as P. mugo possessed cpDNA of P. sylvestris and may be hybrids with P. mugo (female) ×P. sylvestris (male parent). The remainder of the 177 individuals identified in the field inventory were were considered as pure species. In conclusion, our study indicates ongoing spontaneous hybridization between P. mugo and P. syvestris in Kursiu Nerija. Human impact via seed transfer on altered hybridization rates as well as evolutionary consequences are discussed.     

Cryptic hybrids between Pinus uncinata and P. sylvestris

We tested the performance of molecular markers and biometric traits in the identification of hybrids between closely related mountain pine (Pinus uncinata) and Scots pine (Pinus sylvestris). A plastid DNA marker and a set of morphological and anatomical needle traits were applied in analyses of individuals from several sympatric stands of the species and a single‐species' population from southern Europe, used as a reference. A polymerase chain reaction‐restriction fragment length polymorphism (PCR‐RFLP) marker from the plastid trnL–trnF region and morphological and anatomical traits clearly discriminated between the pure species. Significant differences were found between P. uncinata and P. sylvestris, mostly in the shape of epidermal cells and the number of stomata. Four putative hybrids with P. sylvestris morphology, but with P. uncinata plastid DNA haplotypes, were found in a population from Sierra de Gúdar near Valdelinares, the southernmost locality of the latter species in eastern Spain. Discrimination analyses between and within populations placed these individuals on the edge of an agglomeration of P. sylvestris individuals. The results suggest that hybridization between the species is rare, but can result in cryptic hybrids morphologically similar to the maternal species. © 2010 The Linnean Society of London, Botanical Journal of the Linnean Society, 2010, 163 , 473–485.     

Phylogeography of Pinus tabulaeformis Carr. (Pinaceae), a dominant species of coniferous forest in northern China

How coniferous trees in northern China changed their distribution ranges in response to Quaternary climatic oscillations remains largely unknown. Here we report a study of the phylogeography of Pinus tabulaeformis, an endemic and dominant species of coniferous forest in northern China. We examined sequence variation of maternally inherited, seed‐dispersed mitochondrial DNA (mtDNA) (nad5 intron 1 and nad4/3–4) and paternally inherited, pollen‐ and seed‐dispersed chloroplast DNA (cpDNA) (rpl16 and trnS‐trnG) within and among 30 natural populations across the entire range of the species. Six mitotypes and five chlorotypes were recovered among 291 trees surveyed. Population divergence was high for mtDNA variation (G_ST = 0.738, N_ST = 0.771) indicating low levels of seed‐based gene flow and significant phylogeographical structure (N_ST > G_ST, P < 0.05). The spatial distribution of mitotypes suggests that five distinct population groups exist in the species: one in the west comprising seven populations, a second with a north–central distribution comprising 15 populations, a third with a southern and easterly distribution comprising five populations, a fourth comprising one central and one western population, and a fifth comprising a single population located in the north‐central part of the species’ range. Each group apart from the fourth group is characterized by a distinct mitotype, with other mitotypes, if present, occurring at low frequency. It is suggested, therefore, that most members of each group apart from Group 4 are derived from ancestors that occupied different isolated refugia in a previous period of range fragmentation of the species, possibly at the time of the Last Glacial Maximum. Possible locations for these refugia are suggested. A comparison of mitotype diversity between northern and southern subgroups within the north‐central group of populations (Group 2) showed much greater uniformity in the northern part of the range both within and between populations. This could indicate a northward migration of the species from a southern refugium in this region during the postglacial period, although alternative explanations cannot be ruled out. Two chlorotypes were distributed across the geographical range of the species, resulting in lower levels of among‐population chlorotype variation. The geographical pattern of variation for all five chlorotypes provided some indication of the species surviving past glaciations in more than one refugium, although differentiation was much less marked, presumably due to the greater dispersal of cpDNA via pollen.