Am. J. Bot. Join the BSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bouillé, M.
Right arrow Articles by Bousquet, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Bouillé, M.
Right arrow Articles by Bousquet, J.
Agricola
Right arrow Articles by Bouillé, M.
Right arrow Articles by Bousquet, J.
(American Journal of Botany. 2005;92:63-73.)
© 2005 Botanical Society of America, Inc.

Genetics and Molecular Biology

Trans-species shared polymorphisms at orthologous nuclear gene loci among distant species in the conifer Picea (Pinaceae): implications for the long-term maintenance of genetic diversity in trees1

Marie Bouillé and Jean Bousquet2

Chaire de recherche du Canada en génomique forestière et environnementale and Centre de recherche en biologie forestière, Université Laval, Sainte-Foy, Québec, Canada G1K 7P4

Received for publication March 8, 2004. Accepted for publication September 9, 2004.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
For each of three nuclear gene loci, intraspecific- as well as trans-specific shared polymorphisms were detected in DNA among three distantly related species in the genus Picea. Few fixed interspecific polymorphisms were observed. Allele genealogies did not match species phylogenies, and species lineages were not reciprocally monophyletic. Based on molecular clocks and morphological evidence from the fossil record, the divergence time between species was estimated at 13–20 million years (my), and a mutation rate of 2.23 x 10–10 to 3.42 x 10–10 per site per year was estimated. Large historical population sizes in excess of 100 000 were inferred, which would have delayed the fixation of polymorphisms. These numbers translated into allele coalescence times in the order of 10 to 18 my, which implies the sharing of polymorphisms since common ancestry. These results suggest that trans-species shared polymorphisms might be frequent at plant nuclear gene loci, leading to high allelic diversity. Such a trend is more likely in trees and plants characterized by ecological and life-history determinants favoring large population sizes such as an outcrossing mating system, wind pollination, and a dominant position in ecosystem. These polymorphisms also call for caution in estimating congeneric species phylogenies from nuclear gene sequences in such plant groups.

Key Words: allele coalescence • expressed sequence tag polymorphisms • historical demography • intragenic recombination • nuclear gene phylogeny • species divergence • conifers • spruce


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
In plants, evidence of allele coalescence preceding ancient lineage divergence has been reported among widely divergent taxa in the Solanaceae for the self-incompatibility locus (Ioerger et al., 1990 ). The maintenance of trans-species shared polymorphisms among taxa such as Nicotiana, Petunia, and Solanum is thought to be locus-specific, involving frequency-dependent selection. Trans-species shared polymorphisms have also been observed between closely related taxa with incomplete reproductive isolation such as in Drosophila (Wang et al., 1997 ; Machado et al., 2002 ) and in Abies (Isoda and Shiraishi, 2001 ), which suggests the involvement of recent lateral gene flow. The maintenance of trans-species shared polymorphisms at orthologous nuclear gene loci could also depend on the fixation time of polymorphisms and thus, factors related to population size (Tajima, 1983 ; Rosenberg, 2003 ). For neutral or nearly neutral polymorphisms, large population sizes would retard their fixation and favor their persistence in populations (Kimura and Ohta, 1969 ; Hudson, 1990 ).

Most conifers including Picea sp. are characterized by an open-pollinated mating system and wind-dispersed pollen, which appear to promote extensive nuclear gene flow and large population sizes (Hamrick et al., 1992 ). This trend is best illustrated by the little differentiation in allozymes or nuclear DNA markers usually observed among spruce populations (Boyle and Morgenstern, 1987 ; Furnier et al., 1991 ; Isabel et al., 1995 ; Müller-Starck, 1995 ; Jaramillo-Correa et al., 2001 ; Perry and Bousquet, 2001 ; Collignon et al., 2002 ; Gamache et al., 2003 ). In the long-term, these factors should retard the fixation of neutral or nearly neutral polymorphisms, resulting in increased genetic diversity. Such factors have been suggested to account, at least partly, for the maintenance of high levels of allozyme diversity observed in conifers and trees in general, as compared to most other plants and organisms (e.g., Hamrick et al., 1992 ; Hamrick and Godt, 1996 ; Ledig, 1998 ).

Thus, an investigation of the genealogy of orthologous alleles at nuclear gene loci among distant conifer species might reveal deep coalescence, perhaps preceding species and lineage divergence. If so, shared ancestral polymorphisms would lead to allele genealogies that conflict with species phylogenies. While fixed polymorphisms accumulated because lineage divergence would favor the monophyly of species, ancient shared polymorphisms would tend to break up such monophyletic assemblages.

Codominant markers of expressed nuclear genes have been developed for genetic mapping and population genetics studies for a number of spruce species (Perry and Bousquet, 1998a , b ; Perry et al., 1999 ). These markers correspond to segregating alleles at arbitrarily chosen orthologous gene loci. Sampling allele diversity at the DNA sequence level for these gene loci offers the possibility to investigate allele coalescence in a set of reproductively isolated and divergent species characterized by high genetic diversity and reputed large effective population sizes. In this study, alleles for each of three nuclear gene loci were sampled, sequenced, and compared among three distantly related biological species in the genus Picea.


   MATERIAL AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Locus and taxa sampling
The core of this study is based on the sampling of three transcribed regions of the nuclear genome previously described: Sb16, part of a putative gene for the 60S ribosomal protein L13a spanning DNA sequences from two introns and three exons; Sb29, part of a putative gene for the "No Apical Meristem" (NAM) protein fragment comprising DNA sequences from one exon and one intron; and Sb62, part of a putative gene for the ribosomal protein L15 spanning DNA sequences from two introns and two exons (Perry and Bousquet, 1998a ). For each of these loci, the presence of length polymorphisms caused by insertions-deletions (indels) was previously observed among segregating alleles in at least one Picea species, and some of the length variant alleles were shared between species (Perry and Bousquet, 1998a , b ; Perry et al., 1999 ).

The three spruce species sampled, the North American, sympatric Picea glauca (Moench) Voss and P. mariana (Mill.) B.S.P., and the North Eurasian P. abies (L.) Karst. are dominant in their ecosystems with large natural ranges (Wright, 1955 ). They harbor an open-pollinated mating system and extensive intraspecific gene flow for nuclear genes (e.g., Perry and Bousquet, 2001 ; Gamache et al., 2003 ). These various parameters should concur in large effective population sizes. On the other hand, they are not closely related: contrary to closely related taxa in the genus, they do not cross naturally and show little compatibility with each other (Wright, 1955 ; Mikkola, 1969 ), so they warrant recognition as distinct biological species. They also represent divergent taxa in terms of morphology (Wright, 1955 ; Weng and Jackson, 2000 ), with deep phylogenetic branching in the genus, as shown by phylogenies based on paternally inherited cpDNA (Sigurgeirsson and Szmidt, 1993 ; Bouillé and Bousquet, unpublished data). Hence, it is highly unlikely that if trans-species shared polymorphisms were observed, they could be attributed to recent lateral gene flow between species.

For each species and each nuclear locus, around 20 alleles (haplotypes) were sampled from a panel of individuals representative of multiple populations from diverse areas of the species ranges so that most common alleles would be sampled for each locus in each species. To ensure that only one allele was sequenced at a time, haploid megagametophytes were used. Sequences from outgroup taxa belonging to the Pinaceae family were also determined for rooting the three gene trees: Abies lasiocarpa, Pinus cembra, Pseudotsuga mensiezii, and Tsuga canadensis. In order to estimate the divergence time between the three spruce species (see next), the chloroplast gene rbcL (large subunit of ribulose-1,5-bisphosphate carboxylase) was also sequenced for each of Picea abies, P. glauca, and P. mariana. Other conifer rbcL sequences were retrieved from Genbank (www.ncbi.nlm.nih.gov/Genbank).

Biochemical methods
Genomic DNA was extracted with the DNeasy Plant mini kit (Qiagen, Mississauga, Ontario). PCR and sequencing were carried out using the following primers: for Sb16, forward primer 5'-GTTCCGCCACCATATGAC-3' and reverse 5'-GCTCATTCAGCTACAAAAGC-3'; for Sb29 forward 5'-AGCGGCATTGAACAGAGTAAC-3' and reverse 5'-AATGGAAATGAAGGCAGACTC-3'; and for Sb62, forward 5'-TGAGATCCGTGGCTGAAGAG-3' and reverse 5'-GATAACGCCGGAGAGATAGAG-3'. PCR was performed with Platinum Taq DNA polymerase (Gibco BRL, Carlsbad, California, USA) according to the manufacturer's recommendations: 4 min at 95°C for denaturation; 40 cycles consisting of 30 s at 95°C, 30 s at annealing temperatures adjusted to 60°C for Sb16, and 55°C for Sb29 and Sb62, and 1 min at 72°C for extension; and final elongation for 10 min at 72°C. For rbcL, the primers and PCR protocol followed previously published procedures (Wang et al., 1999 ). Amplified fragments were purified and concentrated with Amicon Microcon-PCR filter units (Millipore, Bedford, Massachusetts, USA). Both DNA strands were directly sequenced with Perkin-Elmer ABI DNA sequencers (3100 and 3700, Applied Biosystems, Foster City, California, USA), using BigDye Terminator cycle sequencing kits. Genbank accession numbers are as follows: for Sb16, AY606806 to AY606815; for Sb29, AY6117037 to AY611050; for Sb62, AY611051 to AY611064; for rbcL, AY611034 to AY611036.

Data analysis
For each nuclear locus, indels as well as substitutions were identified among alleles within and across species. Because of the small number of substitutions observed between alleles within and among species, correcting for multiple substitutions had no effect on the observed numbers of substitutions. In addition, all indels were treated as single events because there was no indication that the observed indels were associated with simple sequence repeats or other potential source of homoplasious variation. Nucleotide diversity ({pi}) was estimated for each species using the program DnaSP (Rozas and Rozas, 1999 ). The average numbers of pairwise differences per site, including substitutions and indels, were estimated between alleles within species (dw) and between species (db). Net between-species divergence values per site were also obtained by correcting for within-species diversity according to Nei (1987 , equation 10.21). For each nuclear locus, a site-by-site analysis was also conducted, and sites were classified as those carrying (1) shared polymorphisms between Picea species (trans-species shared polymorphisms, those derived polymorphisms shared by at least two species which do not support species monophyly), (2) reciprocally fixed differences between Picea species (interspecific fixed polymorphisms, which support species monophyly), and (3) apomorphies (intraspecifically derived polymorphisms limited to one species). The derived or ancestral nature of each state at a polymorphic site was determined by comparison with outgroup sequences from other genera of the Pinaceae. A state was assumed as derived when it was different than that observed in the outgroup sequences.

Gene/allele genealogies were estimated for each locus using parsimony analysis and the neighbor-joining method as implemented in PAUP* 4.0b10 (Swofford, 2002 ). For both methods, pairwise deletion of indel sites was enforced so the number of characters corresponded to the length of the consensus sequence between any two taxa, plus a number of additional binary characters for scoring the presence or absence of each indel. If a substitution and an indel were superimposed on a same site, they were considered independently, and a missing value was assigned to the substitution site(s) for those sequences with the deletion. Parsimony analyses consisted of heuristic searches with 100 replicates of random additions of sequences, Tree-bisection-reconnection branch swapping and saving multiple trees. Scenarios constraining alleles of the same species to monophyly (reciprocal monophyly) were tested: differences in tree length between constrained trees and minimum trees were tested for statistical significance using the T-PTP test of Faith (1991) as implemented in PAUP* with 5000 permutations for each gene and by considering both substitutions and indels. For neighbor-joining, pairwise numbers of differences per site were used, including substitutions and indels. The robustness of internal nodes was assessed by means of 500 bootstrap replicates for each method. Nodes supported by a frequency equal or larger than 50% were retained.

For each locus, intragenic recombination was tested among Picea sequences following the "four-gamete test" (Hudson and Kaplan, 1985 ) as implemented in DnaSP (Rozas and Rozas, 1999 ). The homoplasy test (Maynard Smith and Smith, 1998 ) was also conducted with the START 1.0.6 software (Jolley et al., 2001 ) and corroborated the results obtained earlier in all cases. If recombinant sites were identified for a given gene, the putative recombinant types were identified as follows. The ancestral state of character at each recombining position was determined from the outgroup sequences, thus allowing the identification of the allele of the ancestral type. Then, both allele types, each different from the ancestral allele at only one of the two recombinant sites, were assumed derived, thus giving rise to two derived types. The recombinant type was identified as the one resulting from recombination between these two derived types. Recombination between the ancestral type and a derived type at the two recombinant sites is also possible but less likely, especially if the double-derived type is rare or unique, such as an apomorphy. Most likely, this double-derived type is rather the recombinant type. Once a putative recombinant type allele was identified for a given gene, it was substracted from the data set, and the gene tree was estimated again to verify the robustness of tree topology.

A mutation rate (per site per year) was estimated for the three nuclear gene loci studied by averaging the divergence values per site (substitutions only) estimated between the three spruce species and then dividing by 2T where T is the divergence time between the spruce species under study. The first approach to estimate T relied on an analysis of the fossil record, where the earliest fossils morphologically representative of extant taxa could be interpreted as evidence for the diversification of the genus Picea in its major extant lineages. The second approach relied on estimating a molecular clock from rates of protein divergence, as estimated by antigenic distances between the genus Picea and both subgenera of Pinus in the Pinaceae (Prager et al., 1976 ) (Fig. 1). The lineage leading to Picea split early from the lineage leading to Pinus in the history of the Pinaceae (e.g., Magallón and Sanderson, 2002 ). Thus, the average distance between Picea and Pinus was calibrated by using as landmark the date of early diversification of the Pinaceae in the Early Cretaceous, 120–140 my ago (mya) (Florin, 1963 ; Miller, 1988 ). No test of rate constancy could be performed and the relationship between time and antigenic distance was assumed to be linear (Prager et al., 1976 ). The third approach was based on a molecular clock constructed using 32 rbcL sequences from both subgenera of Pinus (Wang et al., 1999 ) and rbcL sequences determined herein for Picea abies, P. glauca, and P. mariana. To construct the molecular clock, the average pairwise number of substitutions was estimated from all pairwise sequence comparisons between Pinus and Picea using both synonymous and nonsynonymous substitutions. The same landmark as before (120–140 mya) was used to calibrate the clock (Fig. 1). Before estimating the rbcL clock, lineage relative rate tests (Li and Bousquet, 1992 ) were used to assess rate homogeneity between sequences of Pinus and Picea, using as outgroups rbcL sequences from three sister taxa of the Pinaceae: Cupressus corneyana (Cupressaceae), Podocarpus gracilior (Podocarpaceae), and Taxus baccata (Taxaceae) (Wang et al., 1999 ). The lineage relative rate test follows approximately the standardized normal distribution. The tests were conducted with an application developed by J. Laroche (Centre for Bioinformatics, Univ. Laval). The rbcL rates were corrected for multiple substitutions following the two-parameter method of Kimura (1980) because divergence rates ranged from 5 to 10% between conifer families.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 1. Estimating the divergence time between Picea abies, P. glauca, and P. mariana (node T). A first estimate was directly obtained from analysis of the fossil record, and two other estimates were obtained from molecular clocks (antigenic distances and rbcL sequences; see Materials and Methods), which were both calibrated using the landmark L, the divergence time between the genera Pinus and Picea, as derived from analysis of the fossil record. Date estimates at node T were obtained by linear intrapolation for each of the antigenic distances and rbcL clock. For rbcL, relative rate tests were conducted, and no significant rate heterogeneity was detected (see Results). Figure not drawn to scale

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Locus Sb16
This locus harbored the lowest haplotype/allelic diversity with eight distinct haplotypes/alleles observed among 23 haploid megagametophytes sampled (Table 1). Of 845 positions sampled, 28 sites were variable (Table 1). Neither substitutions nor indel polymorphisms were observed in P. abies in spite of repetitive sampling (Table 2, Table 3). Picea glauca was the most diverse species for this gene, in terms of the number of alleles detected and number of polymorphic sites observed along the gene (Table 2). The largest number of pairwise differences per site (0.0198) was noted between alleles 847-P. glauca and 802-P. mariana2, and the smallest number (0.0025) was observed between the same size alleles 804-P. abies, 804-P. glauca1, and 804-P. glauca2 from different species (see Table 3). Except for P. abies, the numbers of pairwise differences per site within species (values dw, Table 2) were in the same range as those observed between species (values db, Table 4). The net divergence values between species (Table 4) were smaller, on average, than within-species divergence values (Table 2), indicating that allele divergence would precede species divergence, on average. Trans-species shared polymorphisms were observed at several sites (sites S, Table 3), which does not support the hypothesis of monophyly of P. glauca alleles.


View this table:
[in this window]
[in a new window]
  		Table 1. Parameters derived from the sequence alignment of alleles from three distantly related Picea species at three nuclear gene loci (excluding outgroup sequences from other Pinaceae)

 

View this table:
[in this window]
[in a new window]
  		Table 2. Intraspecific diversity parameters for three nuclear gene loci in three distantly related Picea species

 

View this table:
[in this window]
[in a new window]
  		Table 3. Variable sites for the locus Sb16 (60S ribosomal protein L13A) in Picea spp

 

View this table:
[in this window]
[in a new window]
  		Table 4. Pairwise divergence per site and net pairwise divergence per site between species for each of three nuclear gene loci

 
A total of two equally parsimonious gene topologies of 30 steps were found, with alleles from P. glauca not regrouped in a monophyletic assemblage (Fig. 2A). The general pattern was one of P. glauca alleles coalescing before the divergence of lineages leading to extant species. The neighbor-joining tree was similar (Fig. 2B). Constraining the tree topology to render P. glauca monophyletic resulted in a statistically significant increase (P < 0.01) of nine steps in tree length, as compared to the minimum trees. Tests of intragenic recombination among Picea alleles were positive for two pairs of sites (Table 3). Both cases involved a trans-species shared polymorphism at site 734. Thus, intragenic recombination might have contributed to trans-species shared polymorphism in at least one case. As in the Methods, allele 802-P. mariana1 was identified as a potential recombinant, likely from ancient recombination preceding species divergence. However, excluding this sequence from the phylogenetic analyses did not change the basic pattern of P. glauca alleles coalescing before the divergence of lineages leading to extant species.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 2. Phylogeny of spruce alleles at locus Sb16 (60S ribosomal protein L13A). (A) majority-rule consensus tree obtained by parsimony. (B) Tree obtained by the neighbor-joining method. Only nodes minimally supported by 50% bootstrap estimates are shown. Numbers above the branches indicate majority-rule consensus, and numbers below the branches, bootstrap estimates from 500 replicates. Dashed lines indicate putative recombinant alleles

 
Locus Sb29
The haplotype/allelic diversity observed for Sb29 was higher than that detected for Sb16 (Table 1). The gene was also more variable, with a total of 28 variables sites of the 518 positions sampled (Table 1). Contrary to the trend observed for Sb16, P. abies appeared to be the most diverse species in terms of haplotype/allelic diversity, together with P. glauca (Table 2). The largest number of pairwise differences per site (0.0238) was between same-species alleles 498-P. abies and 511-P. abies, and the smallest number (0.0059) between alleles 503-P. glauca and 512-P. glauca1. The numbers of pairwise differences per site within species (values dw, Table 2) were in the same range as those observed between species (values db, Table 4). As for Sb16, the net divergence values between species (Table 4) were smaller than within-species divergence values (Table 2), indicating that allele divergence would precede species divergence, on average. Trans-species shared polymorphisms were observed at several sites between haplotypes from different Picea taxa (sites S, Table 5). No sites with fixed interspecific differences favoring reciprocal species monophyly (sites F) were noted (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 5. Variable sites for the locus Sb29 (NAM protein fragment) in Picea spp

 
A total of 20 most parsimonious trees of 38 steps was obtained. Alleles from P. abies and P. glauca did not form reciprocally monophyletic groups (Fig. 3A). The pattern was again one of allele coalescence preceding the divergence of lineages leading to extant species. The neighbor-joining tree was similar (Fig. 3B). Gene genealogies constraining alleles from the same species to monophyly (reciprocal monophyly) were four steps longer than the minimum trees, which was a statistically significant increase (P < 0.01). Tests of intragenic recombination within the genus Picea recovered five putative recombinant sites, of which three involved trans-species shared polymorphisms. Based on the Methods, two recombinant alleles were tentatively identified: 503-P. glauca and 511-P. abies. Both were likely the result of ancient recombination at or before species divergence. Removing these putative recombinants from the phylogenetic analyses did not alter the basic pattern of allele coalescence preceding the divergence of lineages leading to extant species.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3. Phylogeny of spruce alleles at locus Sb29 (NAM protein fragment) in Picea spp. (A) Majority-rule consensus tree obtained by parsimony. (B) Tree obtained by the neighbor-joining method. Only nodes minimally supported by 50% bootstrap estimates are shown. Numbers above the branches indicate majority-rule consensus, and numbers below the branches, bootstrap estimates from 500 replicates. Dashed lines indicate putative recombinant alleles

 
Locus Sb62
This locus harbored the highest haplotype/allelic diversity (Table 1). As for variation along the gene, of 650 positions sampled, a total of 29 variables sites were detected. Only 10 of these were informative, those with polymorphisms shared by at least two alleles (Table 1). Picea mariana was by far the most diverse species for this gene, in terms of number of alleles detected and number of polymorphic sites along the gene (Table 2). The largest number of pairwise differences per site (0.0231) was between alleles 556-P. abies and 578-P. mariana1, and the smallest number (0.0017) was between alleles 577-P. glauca and 578-P. abies, therefore from two different species. The numbers of pairwise differences per site within species (values dw, Table 2) were in the same range as those observed between species (values db, Table 4). As for the two previous loci, the net divergence values between species (Table 4) were smaller than within-species divergence values (Table 2), again indicating that allele divergence would precede species divergence, on average. Trans-species shared polymorphisms involving alleles from different Picea species were at several sites (sites S, Table 6). As for Sb29, no sites with fixed interspecific polymorphism favoring reciprocal species monophyly (sites F) were observed (Table 6).


View this table:
[in this window]
[in a new window]
  		Table 6. Variable sites for the locus Sb62 (60S ribosomal protein L15) in Picea spp

 
A single most parsimonious gene tree of 35 steps was obtained, with alleles from the three species coalescing before the divergence of lineages leading to extant species (Fig. 4A). The neighbor-joining tree was essentially the same (Fig. 4B). Gene genealogies constraining species to monophyly (reciprocal monophyly) were five steps longer than the minimum trees, which was a statistically significant increase (P < 0.01). No intragenic recombination was detected for this locus among Picea sequences.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 4. Phylogeny of spruce alleles at locus Sb62 (60S ribosomal protein L15). (A) Majority-rule consensus tree obtained by parsimony. (B) Tree obtained by the neighbor-joining method. Only nodes minimally supported by 50% bootstrap estimates are shown. Numbers above the branches indicate majority-rule consensus, and numbers below the branches are bootstrap estimates from 500 replicates

 
Trans-species shared polymorphisms
The number of polymorphisms per type (S, F, I) may vary between Tables 3, 5, and 6 when all taxa were considered simultaneously, and Table 7 where mutually exclusive pairwise taxa comparisons are considered. Trans-species shared polymorphisms (sites S in Tables 3, 5, 6) did not show any distribution pattern suggestive of maintenance by selection. These polymorphisms were mostly found in introns and involved substitutions as well as indels without a clear clustering pattern, and they were found in exons mainly for Sb29, at synonymous, nonsynonymous, and indel sites. Tajima's D test values (Tajima, 1989 ) as implemented in DnaSP (Rozas and Rozas, 1999 ) did not deviate significantly from neutral expectations for any of the three gene regions surveyed (D = –0.90 for Sb16, –1.01 for Sb29 and –1.67 for Sb62, all P > 0.05).


View this table:
[in this window]
[in a new window]
  		Table 7. Types of polymorphisms detected between Picea species at each of three nuclear gene loci

 
In pairwise comparisons of species (Table 7), trans-species shared polymorphisms were the most prevalent between P. abies and P. glauca with 19 instances. Also, these taxa had the fewest apomorphies and no fixed interspecific polymorphisms, suggesting a higher phylogenetic proximity between them. Locus-to-locus variation in the diverse types of polymorphism was high, reinforcing the need to span a larger number of nuclear loci or scanning cpDNA and mtDNA regions before reaching any definitive conclusion about their higher phylogenetic proximity.

Overall, Sb16 contributed 11 sites to trans-species shared polymorphisms (Table 3), while Sb29 contributed nine such sites (Table 5) and Sb62 seven such sites (Table 6). If we were to remove putative recombinant sites involving trans-species shared polymorphisms for loci Sb16 and Sb29, there would remain eight, six, and seven sites involving trans-species shared polymorphisms, respectively, for Sb16, Sb29, and Sb62.

Fixed interspecific polymorphisms are indicative of a separate lineage evolution. There were no fixed interspecific polymorphisms favoring species monophyly for loci Sb29 and Sb62, while there were only three such sites for Sb16, all favoring the monophyly of P. mariana alleles.

Divergence time between the three Picea species and mutation rate at nuclear gene loci
Because of uncertainty regarding the exact divergence history among the lineages leading to Picea abies, P. glauca, and P. mariana, a trichotomy was assumed for estimating their divergence time T. A first approximation of T is given by the fossil record for which the earliest fossils morphologically representative of extant spruce taxa could be interpreted as evidence for the diversification of the genus in its major extant lineages. These fossils date back to the middle Miocene, around 15 mya, and more questionable fossils were reported from the early Miocene, back to 23 mya (Wolfe, 1964 ). While P. abies, P. glauca, and P. mariana represent morphologically divergent taxa (Wright, 1955 ; Weng and Jackson, 2000 ) with deep phylogenetic branching (Sigurgeirsson and Szmidt, 1993 ; Bouillé and Bousquet, unpublished data), other extant taxa might have branched out earlier and thus, this period should be considered as a lower bound estimate of the divergence between the lineages leading to extant taxa. This period is much latter than the first reliable occurrence of Picea fossils, that is, from the middle Eocene, around 45 mya (Axelrod, 1998 ; LePage, 2001 ).

A second estimate of divergence time T was obtained from antigenic distance data (Prager et al., 1976 ). With an average intergeneric distance of 3.5 units for the divergence between Picea and Pinus and a landmark divergence time of 120–140 mya between these two genera, the maximum distance of 0.5 unit reported between Picea mariana, P. glauca, and P. abies (Prager et al., 1976 ) converts to a divergence time T of 17– 20 mya, assuming a linear relationship (Fig. 1).

A third estimate of divergence time T could be obtained from rbcL sequences, which behaved essentially as a molecular clock for the set of sequences sampled: lineage relative rate tests indicated that there was no rate heterogeneity between the Picea and Pinus lineages using three distinct sister groups to the Pinaceae (test value = 1.16 with outgroup Cupressus, test value = 1.92 with outgroup Podocarpus, test value = 0.97 with outgroup Taxus, all not significant at P = 0.05). With an average rate (K) of 0.0276 substitution per site (synonymous and nonsynonymous) for the divergence between Picea and Pinus and a landmark divergence time of 120–140 mya (L), and by dividing K by 2L, an overall rate of substitution per site per year of 0.99 x 10–10 to 1.15 x 10–10 was obtained. Applying these rates to half of the average pairwise rate of 0.0031 substitution per site estimated between Picea abies, P. glauca, and P. mariana translates to a divergence time T of 13 to 16 mya (Fig. 1).

An estimate of mutation rate µ for the three nuclear gene loci was obtained by calculating the average divergence per site between species (substitutions only) (Table 4) and then dividing by 2T where T was estimated between 13 to 20 mya, the range of possible values determined earlier for the split between Picea abies, P. glauca, and P. mariana. With an average substitution rate per site of 0.0089 between spruce species, µ was estimated as 2.23 x 10–10 to 3.42 x 10–10 per site per year. This mutation rate is used later for estimating parameters of historical demography.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
Diversity at nuclear gene loci within spruce taxa
Low levels of nucleotide diversity (Table 2, {pi} values) were observed within spruce species. However, the values of nucleotide diversity were in the same range as those reported for synonymous sites of protein coding sequences in the conifer Pinus sylvestris L. and in Arabidopsis (Dvornyk et al., 2002 ). In contrast, haplotype/allelic diversity was quite high for the three spruce species studied at each of the three gene loci investigated, except for the locus Sb16 in Picea abies, which was monomorphic. A similar trend of high haplotype diversity was observed for the gene pal1 in the conifer Pinus sylvestris, which was mostly caused by synonymous substitutions (Dvornyk et al., 2002 ). Altogether, the evidence suggests that large effective population sizes are necessary to maintain such important cohorts of neutral or nearly neutral alleles.

There was no clear trend in the current study as to which of the three Picea species analyzed was the most diverse in terms of haplotype/allelic diversity: P. glauca was the most diverse for Sb16 while P. abies was fixed, but P. abies and P. glauca were the most diverse for Sb29, and P. mariana was the most diverse for Sb62. When analyzing diploid genotypes for a dozen of expressed nuclear gene loci, average estimates of observed heterozygosity were similar among the three spruce species analyzed herein, but locus-to-locus variance was high (Perry and Bousquet, 1998a , b ; Perry et al., 1999 ). Overall, our results suggest that the gene-to-gene variance is high and that more loci will need to be investigated in order to compare adequately the diversity residing in the various species. Such gene-to-gene variance is expected because of stochastic factors (Tajima, 1983 ; Arbogast et al., 2002 ; Rosenberg and Nordborg, 2002 ).

Allele coalescence and historical population size
For each of the three nuclear gene loci analyzed, the average numbers of pairwise differences between alleles from different species (db values) were small and in the same range as those observed between alleles within species (dw values). When correcting for within-species diversity, the resulting net divergence values between species were smaller than those observed within-species, indicating that, on average, allele divergence would precede species divergence. This trend is surprising, given a divergence time of at least 13 mya between these biological species. This unexpected trend is also supported by the analysis of gene sequences site by site in which no fixed interspecific polymorphisms were found for two of the three nuclear gene loci analyzed, Sb29 and Sb62. These polymorphisms would support reciprocal species monophyly. On the other hand, several trans-species shared polymorphisms were observed for each of the three nuclear gene loci. The trend towards trans-species shared polymorphisms was confirmed by a phylogenetic analysis in which both distance and character-state approaches indicated that there were no instances where alleles from the various species were reciprocally monophyletic. Rather, alleles from the various species were intermingled for Sb29 or Sb62. For Sb16, for which no allelic variation was observed for P. abies and the two alleles detected in P. mariana were monophyletic, alleles of P. glauca were not monophyletic and many instances of trans-species shared polymorphisms were identified.

While introgressive hybridization represents a possible cause for the presence of shared polymorphisms among closely related taxa (e.g., Wang et al., 1997 ; Isoda and Shiraishi, 2001 ; Machado et al., 2002 ), and while natural hybridization has been reported between closely related spruce taxa (e.g., between P. mariana and P. rubens Sarg., Perron and Bousquet, 1997 ), more or less recent interspecific gene exchanges appear highly unlikely between the three spruce species analyzed herein. Contrary to closely related taxa in the genus, these species do not cross naturally, and they are hardly compatible with each other (Wright, 1955 ; Mikkola, 1969 ), which warrant their actual recognition as distinct biological species. In such a slowly evolving genus as Picea, achieving reproductive isolation is generally indicative of large divergence (Wright, 1955 ). These species are also divergent with respect to several morphological characters, which is notable for this rather morphologically uniform genus (Wright, 1955 ; Weng and Jackson, 2000 ). In phylogenies based on cpDNA, these species are also present in distinct lineages with deep phylogenetic branching in the genus (Sigurgeirsson and Szmidt, 1993 ; Bouillé and Bousquet, unpublished data). Thus, it is highly unlikely that these biological species have exchanged genes in the recent past. If they had done so, cpDNA phylogenies would reveal high phylogenetic affinity between these taxa, because cpDNA is paternally transmitted by pollen in conifers (e.g., Stine et al., 1989 ).

Hence, under such conditions, trans-species shared polymorphisms are likely to be of shared ancestry. For all three loci, Tajima's test for selection was inconclusive and trans-specific-shared polymorphisms did not show any distribution pattern suggestive of maintenance by selection. Many shared polymorphisms were observed in introns, and there was an excess of shared polymorphisms over fixed polymorphisms for the three loci investigated (Table 7), which is not suggestive of locus-specific selective effects (Wang et al., 1997 ). Even if neutral tests are known for their lack of power, it seems reasonable to assume, given the overall evidence at hand, that the observed polymorphisms are neutral or nearly neutral and that allele coalescence time would be governed largely by demographic factors.

Historical population size (Ne) was estimated by solving the equation {pi} = 4Neµ (Tajima, 1983 ), where µ is the mutation rate per generation for the nuclear gene loci examined. The mutation rate µ was estimated earlier as 2.23 x 10–10 to 3.42 x 10–10 per site per year. This estimate is in the same range as the estimated rate of synonymous substitutions per site per year obtained from comparing spruce and pine nuclear phytochrome PHYO gene sequences (µ = 4.8 x 10–10, Garcia-Gil et al., 2003 ). By assuming 50 years as the average generation time in these species, which is about three to five times the age at first reproduction but which is less than the maximum life expectancy in these species (Burns and Honkala, 1990 ), µ becomes 1.11 to 1.71 x 10–8 per site per generation. Then, by using average {pi} values deduced from Table 2 for each species, Ne estimates from 96 000 to 182 000 were obtained by applying the formula above, depending of the species (Table 8). The Ne estimates are large and well above estimates for animal species for which numbers in the range of 10 000 to 50 000 have been estimated (e.g., Sherry et al., 1997 ; Hare et al., 2002 ). Even if Ne estimates were underestimated or overestimated due to a number of factors, including the value of generation time used and the fact that trees have overlapping generations (Caballero, 1994 ), the order of magnitude of the estimates obtained appears robust. More precise estimates for each species would require better estimates of generation time that would take into account temporal and spatial heterogeneity as well as generation overlap. As well, they would require allele frequencies estimated from several more loci in order to stabilize the gene-to-gene variance in nucleotide diversity (Tajima, 1983 ; Arbogast et al., 2002 ).


View this table:
[in this window]
[in a new window]
  		Table 8. Estimates of historical population size and allele coalescence time from data at three nuclear gene loci in three distantly related Picea species

 
The average coalescence time of two randomly picked alleles at the same gene locus is estimated by 2Ne generations (Tajima, 1983 ). Hence, the previously estimated values of Ne, apply to a 2Ne period of around 200 000 to 360 000 generations or 10 to 18 mya, assuming a generation time of 50 years (Table 8). These estimates are in the same range as those of species divergence time (13 to 20 mya). They indicate the ancient nature of the allelic polymorphisms observed in these species and the sharing of neutral or nearly neutral variation since common ancestry. These estimates are in agreement with the coalescence of many alleles before species divergence on the gene trees and the lack of reciprocal monophyly between alleles from different species.

Implications for the long-term maintenance of genetic diversity in trees
The order of magnitude of these values of historical population size (Ne) reconciles well with much lower estimates obtained at the time of speciation for Picea rubens, a recent derivative of P. mariana during the Pleistocene, which harbors reduced gene diversity at expressed gene loci (in terms of observed heterozygosity, proportion of polymorphic loci and number of alleles, Perron et al., 2000 ) compared to that of P. abies (Perry et al., 1999 ), P. glauca (Perry and Bousquet, 1998b ), or P. mariana (Perry and Bousquet, 1998a ; Perron et al., 2000 ). To account for the lower genetic diversity observed in P. rubens, a mild bottleneck of Ne in the range of 10 000 was estimated from genetic drift simulations (Perron et al., 2000 ). Hence, the estimates of historical population size in this study, which are more than an order of magnitude higher than those inferred for the genetically depauperate P. rubens, support the idea that large historical population sizes in P. abies, P. glauca, and P. mariana led at least partly to the long-term maintenance of higher levels of gene diversity. Indeed, these three species do not seem to have suffered much from the Pleistocene glaciations, even if they were displaced repeatedly (e.g., Davis, 1983 ).

Such large historical population sizes also reconcile well with our knowledge of the ecological determinants and reproductive biology of these species. Spruces harbor archaic wind-pollinated mating systems with high outcrossing rates (e.g., Perry and Bousquet, 2001 ). It is unlikely that these traits have changed during their history. These taxa are also abundant and occupy a dominant position in their ecosystems (Wright, 1955 ). All these factors contribute to maintaining large effective population sizes. Such population sizes and extensive gene flow are also suggested by the low levels of population differentiation in spruces, as measured by nuclear markers of a molecular or biochemical nature (Boyle and Morgenstern, 1987 ; Furnier and Stine, 1991 ; Isabel et al., 1995 ; Müller-Starck, 1995 ; Jaramillo-Correa et al., 2001 ; Perry and Bousquet, 2001 ; Collignon et al., 2002 ; Gamache et al., 2003 ). Exceptions to this rule exist in which spruce taxa with more scattered populations and restricted ranges show more population differentiation and overall smaller genetic diversity as a result of more or less recent bottlenecks, such as for P. rubens (Hawley and DeHayes, 1994 ; Perron et al., 2000 ; Rajora et al., 2000 ) or for P. martinezii T. F. Patterson (Ledig et al., 2000 ).

These results bear serious implications for our understanding of the maintenance of neutral or nearly neutral genetic diversity at nuclear genes in spruces and, more generally, in conifers and other tree species harboring ecological and reproductive strategies promoting large population sizes. On average, one would expect large population sizes to favor the maintenance of high levels of neutral or nearly neutral genetic diversity. Under such conditions, novel alleles would be less likely to be lost through drift and their time to fixation would be longer (Kimura and Ohta, 1969 ).

If effective, this process would lead to the disproportionate coexistence of neutral or nearly neutral polymorphisms at nuclear gene loci and high levels of haplotype/allelic diversity, such as observed in this study and in previous studies (Perry and Bousquet, 1998a , b ; Perry et al., 1999 ). The comparative analysis of allozyme diversity among plant taxa with contrasting life-history and ecological determinants appears to support such a neutral/demographic process (Hamrick et al., 1992 ; Hamrick and Godt, 1996 ). On average, taxa with higher levels of heterozygosity and higher number of alleles per locus are characterized by ecological, population, and demographic features conferring large population sizes, such as outcrossing, wind pollination or dominant position in the ecosystem. The three spruce species analyzed in this study are good examples, with high levels of diversity at the population level (He typically in the range of 0.20–0.30 and above) for allozymes (Boyle and Morgenstern, 1987 ; Bergmann and Ruetz, 1991 ; Furnier et al., 1991 ; Hamrick et al., 1992 ; Isabel et al., 1995 ; Müller-Starck, 1995 ; Jaramillo-Correa et al., 2001 ). Such a link between population determinants and diversity at allozyme loci is not unexpected, because of the essentially neutral or nearly neutral nature of allozyme polymorphisms (e.g., in black spruce, Isabel et al., 1995 ; in white spruce, Jaramillo-Correa et al., 2001 ). Recently, this diversity trend has further been confirmed at the DNA level, for which the highest levels of within-population gene diversity for RAPD and microsatellite markers were detected in outcrossing plant taxa and long-lived perennials, which included tree species (Nybom, 2004 ).

Implications for estimating species phylogenies from nuclear genes
The detection of trans-species shared polymorphisms at orthologous gene loci calls for caution in estimating congeneric species phylogenies from nuclear genes in plants with life-history and reproductive determinants favoring large effective population sizes, even when gene orthologues can be unambiguously distinguished from putative paralogs. In this study, well-characterized regions of the nuclear genome were used and paralogous sequences were avoided: primers were designed to be specific to a single gene region (Perry and Bousquet, 1998a ), the mendelian segregation of allelic variants had been previously documented (Perry and Bousquet, 1998a , b ; Perry et al., 1999 ), and orthology was further validated by sequencing alleles from haploid tissues. Even so, allele orthology did not appear to be a sufficient criterion to guarantee a gene tree matching a species tree. The detection of several trans-species shared polymorphisms at the three gene loci investigated in this study warns about potential pitfalls in estimating species phylogeny from nuclear genes in such taxa. It emphasizes the need for data from many nuclear loci (Arbogast et al., 2002 ; Rosenberg and Nordborg, 2002 ). Ideally, validation should be sought with phylogenies derived from chloroplast and mitochondrial genes for which allele coalescence time and the frequency of trans-species shared polymorphisms would be reduced as compared to nuclear genes (Bouillé and Bousquet, unpublished data). This is because of the uniparental transmission of organellar genomes for which the average coalescence time of two randomly picked cpDNA or mtDNA alleles is not 2Ne but Ne generations. Such recommendations of caution appear even more appropriate when little is known about the demography and the history of a species or group of species.

Trans-species shared polymorphisms of two different types were detected. The ones shared by ancestry were the most abundant. But for two of the three nuclear genes analyzed, a few trans-species shared polymorphisms were detected, which were likely the result of ancient intragenic recombination. The genome-wide frequency of intragenic recombination is unknown for plant nuclear genes. But intragenic recombination in itself constitutes another factor that can lead to biased phylogenies (Shierup and Hein, 2000 ; Posada and Crandall, 2002 ; Rosenberg and Nordborg, 2002 ). In our study, this factor was taken into account by estimating gene trees with and without the recombinant alleles.

The presence of trans-species shared polymorphisms also calls for caution in interpreting gene genealogies. In plant and tree species suspected of large historical population sizes, nonmonophyletic patterns of allelic variation should persist for tens or hundreds of thousands generations. In such cases, it has been pointed out that extreme caution should be exercised when inferring reticulate evolution from polyphyletic patterns at nuclear loci (Hare et al., 2002 ). Such patterns could also be caused by the long-term maintenance of trans-species-polymorphisms of shared ancestry. More generally, our study emphasizes the limitations of single or oligo-gene genealogies at the nuclear level to circumscribe useful taxonomical or ecological units when large historical population sizes are suspected.


   FOOTNOTES
 
1 We thank S. Senneville for support in the laboratory, J. Laroche (Bioinformatics Centre, Univ. Laval) for relative rate tests, B. Lepage (Univ. Pennsylvania) for discussions about the spruce fossil record, D. J. Perry (Canadian Grain Commission) for fruitful discussions at the onset of the study, and O. Savolainen (Univ. of Oulu), J. Hey (Rutgers Univ.), and two anonymous reviewers for constructive comments on previous drafts of this manuscript. This study was supported by a discovery grant from the National Science and Engineering Research Council of Canada to J.B. Back

2 bousquet{at}rsvs.ulaval.ca Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Arbogast B. S. S. V. Edwards J. Wakeley P. Beerli J. B. Slowinski 2002 Estimating divergence times from molecular data on phylogenetic and population genetic timescales. Annual Review of Ecology and Systematics 33: 707-740

Axelrod D. I. 1998 The Eocene Thunder Mountain flora of central Idaho. University of California Publications in Geological Sciences 142: 1-61

Bergmann F. W. Ruetz 1991 Isozyme variation and heterozygosity in random tree samples and selected orchard clones from the same Norway spruce populations. Forest Ecology and Management 46: 39-47[CrossRef][ISI]

Boyle T. J. B. E. K. Morgenstern 1987 Some aspects of the population structure of black spruce in central New Brunswick. Silvae Genetica 36: 53-60[ISI]

Burns R. M. B. H. Honkala 1990 Silvics of North America, vol. 1, Conifers. Agriculture Handbook 654. US Department of Agriculture, Forest Service, Washington, D.C., USA

Caballero A. 1994 Developments in the prediction of effective population size. Heredity 73: 657-679

Collignon A.-M. H. Van de Sype J.-M. Favre 2002 Geographical variation in random amplified DNA and quantitative traits in Norway spruce. Canadian Journal of Forest Research 32: 266-282[CrossRef]

Davis M. B. 1983 Quaternary history of deciduous forests of eastern North America and Europe. Annals of the Missouri Botanical Garden 70: 550-563

Dvornyk V. A. Sirviö M. Mikkonen O. Savolainen 2002 Low nucleotide diversity at the pal1 locus in the widely distributed Pinus sylvestris. Molecular Biology and Evolution 19: 179-188[Abstract/Free Full Text]

Faith D. P. 1991 Cladistic permutation tests for monophyly and nonmonophyly. Systematic Zoology 40: 366-375[CrossRef]

Florin R. 1963 The distribution of conifer and taxad genera in time and space. Acta Horti Bergiani 20: 121-312

Furnier G. R. M. Stine C. A. Mohn M. A. Clyde 1991 Geographic patterns of variation in allozymes and height growth in white spruce. Canadian Journal of Forest Research 21: 707-712[CrossRef]

Gamache I. J. P. Jaramillo-Corea S. Payette J. Bousquet 2003 Diverging patterns of mitochondrial and nuclear DNA diversity in subarctic black spruce: imprint of a founder effect associated with postglacial colonization. Molecular Ecology 12: 891-901[CrossRef][Medline]

Garcia-Gil M. R. M. Mikkonen O. Savolainen 2003 Nucleotide diversity at two phytochrome loci along a latitudinal cline in Pinus sylvestris. Molecular Ecology 12: 1195-1206[CrossRef][Medline]

Hamrick J. L. M. J. W. Godt S. L. Sherman-Broyles 1992 Factors influencing levels of genetic diversity in woody plant species. New Forests 6: 95-124[CrossRef]

Hamrick J. L. M. J. W. Godt 1996 Effects of life history traits on genetic diversity in plant species. Philosophical Transactions of the Royal Society of London, B, Biological Sciences 351: 1291-1298[CrossRef]

Hare M. P. F. Cipriano S. R. Palumbi 2002 Genetic evidence on the demography of speciation in allopatric dolphin species. Evolution 56: 804-816[CrossRef][ISI][Medline]

Hawley G. J. D. H. DeHayes 1994 Genetic diversity and population structure of red spruce (Picea rubens). Canadian Journal of Botany 72: 1778-1786[CrossRef]

Hudson R. R. 1990 Gene genealogies and the coalescent process. In D. Gutuyama and J. Antonivics [eds.], Oxford surveys in evolutionary biology, vol. 7, 1–44. Oxford University Press, Oxford, UK

Hudson R. R. N. L. Kaplan 1985 Statistical properties of the number of recombination events in the history of a sample of DNA samples. Genetics 111: 147-164[Abstract/Free Full Text]

Ioerger T. R. A. G. Clark T.-H. Kao 1990 Polymorphism at the self-incompatibility locus in Solanaceae predates speciation. Proceedings of the National Academy of Sciences, USA 87: 9732-9735[Abstract/Free Full Text]

Isabel N. J. Beaulieu J. Bousquet 1995 Complete congruence between gene diversity estimates derived from genotypic data at enzyme and random amplified polymorphic DNA loci in black spruce. Proceedings of the National Academy of Sciences, USA 92: 6369-6373[Abstract/Free Full Text]

Isoda K. S. Shiraishi 2001 Allelic sequence polymorphisms in the intron region of the nuclear-encoded GapC gene preceded the speciation of three closely related Abies species (Pinaceae). Theoretical and Applied Genetics 102: 244-250[CrossRef][ISI]

Jaramillo-Correa J. P. J. Beaulieu J. Bousquet 2001 Contrasting evolutionary forces driving population structure at expressed sequence tag polymorphisms, allozymes and quantitative traits in white spruce. Molecular Ecology 10: 2729-2740[CrossRef][Medline]

Jolley K. A. E. J. Feil M.-S. Chan M. C. J. Maiden 2001 Sequence type analysis and recombinational tests (START). Bioinformatics 17: 1230-1231[Abstract/Free Full Text]

Kimura M. T. Ohta 1969 The average number of generations until fixation of a mutant gene in a finite population. Genetics 61: 763-771[Free Full Text]

Kimura M. 1980 A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16: 111-120[CrossRef][ISI][Medline]

Ledig F. T. 1998 Genetic variation in Pinus. In D. M. Richardson [ed.], Ecology and biogeography of Pinus, 251–280. Cambridge University Press, Cambridge, UK

Ledig F. T. B. Bermejo-Velazquez P. D. Hodgskiss D. R. Johnson C. Flores-Lopez V. Jacob-Cervantes 2000 The mating system and genus diversity in Martinez spruce, an extremely rare endemic of Mexico's Sierra Madre Oriental: an example of facultative selfing and survival in interglacial refugia. Canadian Journal of Forest Research 30: 1156-1164[CrossRef]

LePage B. A. 2001 New species of Picea A. Dietrich (Pinaceae) from the middle Eocene of Axel Heiberg Island, Arctic Canada. Botanical Journal of the Linnean Society 135: 137-167[CrossRef]

Li P. J. Bousquet 1992 Relative-rate test for nucleotide substitutions between two lineages. Molecular Biology and Evolution 9: 1185-1189[ISI]

Machado C. A. R. M. Kliman J. A. Markert J. Hey 2002 Inferring the history of speciation from multilocus DNA sequence data: the case of Drosophila pseudoobscura and close relatives. Molecular Biology and Evolution 19: 472-488[Abstract/Free Full Text]

Magallón S. M. J. Sanderson 2002 Relationships among seed plants inferred from highly conserved genes: sorting conflicting phylogenetic signals among ancient lineages. American Journal of Botany 89: 1991-2006[Abstract/Free Full Text]