Portal de Periódicos da CAPES

Nucleotide diversity among natural populations of a North American poplar ( Populus balsamifera , Salicaceae)

2009; Wiley; Volume: 182; Issue: 3 Linguagem: Inglês

10.1111/j.1469-8137.2009.02779.x

1469-8137

Amy Breen, Elise Glenn, Adam Yeager, Matthew S. Olson,

Chromosomal and Genetic Variations

New PhytologistVolume 182, Issue 3 p. 763-773 Free Access Nucleotide diversity among natural populations of a North American poplar (Populus balsamifera, Salicaceae) Amy L. Breen, Amy L. Breen Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorElise Glenn, Elise Glenn Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorAdam Yeager, Adam Yeager Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorMatthew S. Olson, Matthew S. Olson Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this author Amy L. Breen, Amy L. Breen Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorElise Glenn, Elise Glenn Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorAdam Yeager, Adam Yeager Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this authorMatthew S. Olson, Matthew S. Olson Institute of Arctic Biology, and Department of Biology and Wildlife, University of Alaska, 311 Irving 1, 902 N. Koyukuk Dr., Fairbanks, AK 99775, USASearch for more papers by this author First published: 16 April 2009 https://doi.org/10.1111/j.1469-8137.2009.02779.xCitations: 18 Author for correspondence:Amy L. BreenTel: +1 (907) 474 1175Email: [email protected] AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Summary • Poplars (Populus spp.) comprise an important component of circumpolar boreal forest ecosystems and are the model species for tree genomics. In this study, we surveyed genetic variation and population differentiation in three nuclear genes among populations of balsam poplar (Populus balsamifera) in North America. • We examined nucleotide sequence variation in alcohol dehydrogenase 1 (Adh1) and glyceraldehyde 3-phosphate dehydrogenase (G3pdh), two well-studied nuclear loci in plants, and abscisic acid insensitivity 1B (ABI1B), a locus coincident with timing of seasonal dormancy in quantitative trait locus (QTL) studies of hybrid poplars. We compared estimates of baseline population genetic parameters for these loci with those obtained in studies of other poplar species, particularly European aspen (Populus tremula). • Average pairwise nucleotide diversity (πtot = 0.00216–0.00353) was equivalent to that in Populus trichocarpa, but markedly less than that in P. tremula. Elevated levels of population structure were observed in ABI1B between the northern and southern regions (FCT = 0.184, P < 0.001) and among populations (FST = 0.256, P < 0.001). • These results suggest that geographic or taxonomic factors are important for understanding patterns of variation throughout the genus Populus. Our findings have the potential to aid in the design of sampling regimes for conservation and breeding stock and contribute to historical inferences regarding the factors that shaped the genetic diversity of boreal plant species. Introduction The North America boreal forest spans the continent, extending > 10° of latitude from central Labrador westward through Canada to interior Alaska. Climatological gradients across the region largely reflect latitude, with temperature and net radiation decreasing to the north (Elliott-Fisk, 2000). Periods of major climate fluctuations, including alternating glacial and interglacial cycles, predominate in the history of the boreal forest. During the late-Quaternary and previous glaciations, the boreal region was covered with glacial ice. Boreal forest organisms were largely displaced south of their current limits in North America during glaciation events and migrated northward when the climate warmed and glaciers receded (Juday et al., 2005). Genetic diversity in key boreal forest organisms may reveal signals of historical demography and adaptation to climate, thus providing a context for understanding species responses to future climate change and development of conservation strategies. The genus Populus (aspen, cottonwood and poplar; collectively referred to as poplars) comprises an important, and sometimes dominant, component of circumpolar boreal forest ecosystems. Poplars also are firmly established as the model species for tree genomics, with the recent publication of the full genome of western black cottonwood, Populus trichocarpa (Tuskan et al., 2006). The general picture of nucleotide variation in poplar is underdeveloped and at present is based almost exclusively on European aspen, Populus tremula (Ingvarsson, 2005a,b, 2008; Ingvarsson et al., 2006; Garcia & Ingvarsson, 2007); the single exception is a study of nucleotide variation in P. trichocarpa across its coastal range in northwestern North America (Gilchrist et al., 2006). Poplars are long-lived trees characterized by a dioecious breeding system, wind dispersal of pollen and seeds, clonality, and often continental-scale distribution. As a result, poplars are potentially comprised of interbreeding populations of immense size. These life history traits typify a plant expected to exhibit abundant genetic variation and little population differentiation (Hamrick & Godt, 1996; Brunner et al., 2004). Undoubtedly, other aspects of individual poplar species' biology such as hybridization, introgression, migration and demographic history also influence genetic diversity and effective population size. Recent estimates of population genetic parameters in P. tremula indicate relatively high levels of genetic variation (77 loci; average pairwise nucleotide diversity for synonymous and noncoding sites (πsil) = 0.0120; Ingvarsson, 2008) compared with coniferous trees (mean πsil ranges from 0.0038 in Cryptomeria japonica to 0.0064 in Pinus taeda; Kado et al., 2003, Brown et al., 2004). This species-wide level of silent polymorphism is equivalent to that found in other outcrossing plant species including Arabidopsis lyrata ssp. petrea (mean πsil = 0.029; Wright et al., 2003) and Arabidopsis halleri (πsil = 0.015; Ramos-Onsinsa et al., 2004). The extent of the influence of geographic or taxonomic factors on patterns of variation throughout the genus Populus is an open question. Populus tremula exhibits levels of nucleotide variation over 5-fold greater than those found in P. trichocarpa (mean πsil = 0.0029; Gilchrist et al., 2006), a North American poplar. Moreover, moderate population differentiation exists in P. tremula (for 11 loci, FST was found to range from 0.040 to 0.214; Ingvarsson, 2005a,b). By contrast, isozyme and microsatellite surveys in North American poplars have found little to no population differentiation (Populus balsamifera: for eight isozyme loci, FST ranged from 0.008 to 0.023 (Farmer et al., 1988); Populus deltoides: the mean FST for 22 isozyme loci was 0.064 (range not reported) (Marty, 1984); P. trichocarpa: the mean FST for 12 isozyme loci was 0.063 (range not reported) (Weber & Stettler, 1981); Populus tremuloides: for 10 isozyme loci, FST ranged from −0.006 to 0.061 (Lund et al., 1992); for 16 microsatellite loci, FST ranged from 0.006 to 0.045 (Cole, 2005); the mean FST for 15 isozyme loci was 0.068 (range not reported) (Hyun et al., 1987); and the mean FST for four microsatellite loci was 0.032 (range not reported) (Wyman et al., 2003)). Although these estimates for North American poplars are approximately 50% lower than estimates for P. tremula, several of the studies were conducted across small geographic scales relative to the entire range of the study species and may not capture a species-wide estimate of diversity or structure. Comprehensive knowledge of the levels of genetic variation and population structure is crucial for informed decisions concerning breeding stocks, conservation, and responses to future environmental change. The disparity in nucleotide diversity and population structure estimates among different poplar species suggests that genetic parameters of a single poplar species, such as P. tremula, may not be generalized across all poplars. Here we report the results of a genetic variation and population differentiation survey of three nuclear genes sampled in balsam poplar (Populus balsamifera) at the northernmost and southernmost limits of its distribution in North America (from ∼40° to 70° N; Table 1 and Fig. 1). We surveyed two well-studied nuclear loci in plants, alcohol dehydrogenase 1 (Adh1) and glyceraldehyde 3-phosphate dehydrogenase (G3pdh), and a gene coincident with the timing of seasonal dormancy in quantitative trait locus (QTL) studies of hybrid poplars, abscisic acid insensitivity 1B (ABI1B) (Frewen et al., 2000). We expected that, of these loci, ABI1B would be most likely to exhibit differences between northern and southern populations because it putatively controls the timing of bud set, a trait linked to dormancy adaptation across latitude (Chen et al., 2002). Moreover, Garcia & Ingvarsson (2007) recently reported an excess of nonsynonymous site diversity and extensive haplotype structure at the ABI1B locus in P. tremula suggestive of balancing selection. We compare our results with expectations from previous poplar studies. Table 1. Sample sizes and locations of the study populations of North American Populus Species Population State or Province Latitude (° N) Longitude (° W) No. of individuals Adh1 G3pdh ABI1B Populus balsamifera Cache Creek Alaska 69.41 145.88 14 16 15 Cottonwood Creek Alaska 69.10 147.89 15 15 15 Yukon River bridge Alaska 65.88 149.72 15 16 15 Chena River Alaska 65.07 146.08 17 17 15 Grand Portage Minnesota 47.98 89.66 5 5 5 Grand Forks Minnesota 47.93 97.02 15 16 18 Ridges State Park Wisconsin 45.07 87.11 5 5 5 Guelph Lake Ontario 43.60 80.26 16 15 15 Populus deltoides Hubbard State Park Connecticut 41.55 72.83 5 5 5 Populus trichocarpa Valdez Alaska 61.13 146.35 15 15 15 Adh1, alcohol dehydrogenase 1; G3pdh, glyceraldehyde 3-phosphate dehydrogenase; ABI1B, abscisic acid insensitivity 1B. Figure 1Open in figure viewerPowerPoint Distribution of alcohol dehydrogenase 1 (Adh1), glyceraldehyde 3-phosphate dehydrogenase (G3pdh) and abscisic acid insensitivity 1B (ABI1B) haplotypes observed in Populus balsamifera (circles), Populus trichocarpa (stars) and Populus deltoids (triangles). Pie charts indicate the frequency of haplotypes within each population and unique alleles are indicated by different colors. The range of P. balsamifera in North America is shown in green (US Geological Survey, 2006). Significant variance components between northern and southern regions (FCT) and among populations within regions (FST) are indicated in bold and denoted with an asterisk (P≤ 0.001). Materials and Methods Study system and tissue collection Balsam poplar is among the most widely distributed species of Populus in North America (Little, 1971), ranging from New Foundland northwest to Alaska (Fig. 1). A recent phylogeny of Populus placed P. balsamifera and P. trichocarpa as sister species (Hamzeh & Dayanandan, 2004) and the forthcoming volume of Flora of North America denotes these species as infraspecific taxa of P. balsamifera (P. balsamifera=P. balsamifera subsp. balsamifera L. and P. trichocarpa=P. balsamifera subsp. trichocarpa (Torr. & A. Gray) Hultén; http://www.tropicos.org). Despite the accepted nomenclature, the two taxa are regarded as separate species within the literature and we will follow that precedent here. Leaf or bud tissues for genetic analyses were collected from 5 to 18 individuals from each of eight populations of P. balsamifera (Table 1). Of the eight study populations, four were from the northernmost and four from the southernmost limits of balsam poplar's distribution in North America (Fig. 1). This design was chosen in an attempt to estimate the upper bound for diversity and population structure for this species. We also sampled a single population of western black cottonwood (Populus trichocarpa Torr. & A. Gray ex Hook.) and eastern cottonwood (Populus deltoides Bartram ex Marsh.) to determine whether diversity was influenced by introgression from these closely related species at the margins of the range of P. balsamifera (Table 1, Fig. 1; Hamzeh & Dayanandan, 2004). Because poplars are clonal, within each population we sampled trees separated by at least 15 m to limit sampling of multiple ramets from a single genet. Nucleotide genotype determinations indicated that this distance was sufficient for collecting from genetically different individuals. Specimens were stored at −80°C until DNA extraction at the University of Alaska Fairbanks, USA. Three loci were studied. These included portions of the Adh1, G3pdh and ABI1B genes. The Adh1 enzyme allows plants to compensate for low-oxygen stress, whereas the G3pdh enzyme plays an important role in glycolysis and glyconeogenesis. ABI1B is involved in transduction of abscisic acid response signals and is putatively associated with seasonal leaf dormancy in poplar (Frewen et al., 2000). These loci are well studied in other plant species, including European aspen, allowing us to compare nucleotide diversity among poplar species (Ingvarsson, 2005a; Garcia & Ingvarsson, 2007). DNA extraction, polymerase chain reaction (PCR) and sequencing We extracted DNA from frozen leaf or bud tissue with DNeasy Plant Extraction kits (Qiagen, Valencia, CA, USA). PCR primers were designed to amplify ~600-bp segments of the Adh1 and G3pdh loci using published sequences of P. tremula in the EMBL/Genbank database (accession numbers AJ580717 and AJ843581). To amplify ABI1B, we designed primers from the genomic sequence of P. trichocarpa obtained from the Bradshaw laboratory at the University of Washington, USA (Frewen et al., 2000). Primers for each locus were: (1) Adh1, 5′-ATA AGT TAC AAC CAT CAG CGA TTA GTG-3′ and 5′-GTG AAT ACA CCG TCT GCC ATA TTG-3′; (2) G3pdh, 5′-TGC AGC GTG AAA CAC AAC MAT T-3′ and 5′ATG GGC TAC TTA TTT AAC AAT CAT-3′; and (3) ABI1B, 5′-GGC CTG AGT GAT GGA AGT AT-3′ and 5′-CGT CTT ATG ATT ATG AAC AT-3′. We generated bidirectional sequence data from PCR fragments of Adh1, G3pdh and ABI1B. Loci were amplified using TaKaRa Ex Taq polymerase (Takara Bio Inc., Madison, WI, USA) and column-purified (Qiagen). PCR products were sequenced using the above primers and BigDye Terminator Cycle Sequencing chemistry (v. 3.1; Applied Biosystems, Foster City, CA, USA) on an ABI3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Analyses aligner (v. 2.0.5; CodonCode Corporation, Dedham, MA, USA) was used for curating and trimming sequences based on Phred quality scores (Ewing & Green, 1998; Ewing et al., 1998). Sequence ends were trimmed until the average quality value was Phred > 25 in a window of 10 bases. We aligned bidirectional sequences for each individual separately and visually inspected the chromatograms using sequencher (v. 4.7; Gene Codes, Ann Arbor, MI, USA). Heterozygous sites were scored using the 'call secondary peaks' function in sequencher with the minimum lower peak height set at > 60% and manually confirmed (Weckx et al., 2005). If the bidirectional reads for a single individual differed, the final call was either made visually or was based on the higher quality chromatogram (Phred > 30) which was almost always homozygous. Although this method was among the least likely to introduce unknown bias, we are cognizant that, if anything, it may have slightly underestimated the frequency of singletons and uncommon alleles in our samples. Nonetheless, our curatorial methods were consistent across the loci presented herein. A polymorphic insertion/deletion (indel) in the middle of the G3pdh region resulted in high-quality (Phred > 30) chromatogram in one direction up to the indel region and lower quality thereafter for some heterozygous individuals. For these few individuals, we changed the base calls of the indel region to missing data and base calls in the regions flanking the indel were based on Phred > 30 scores from one directional sequence. Levels of genetic variation were estimated as average per site pairwise nucleotide diversity (π; Nei, 1987) and as the relationship between segregating sites and alleles sampled (θW; Watterson, 1975). To determine the extent of linkage disequilibrium across each locus, we plotted r2 values against pairwise distances between polymorphic sites and calculated the nonlinear quadratic regression slope using jmp (v. 7; SAS Institute, Cary, NC, USA; Hill & Robertson, 1968). To test whether the folded site frequency spectrum was consistent with expectations derived from neutral evolution, we estimated Tajima's D (Tajima, 1989) and Fu and Li's F* (Fu & Li, 1993) separately for each population. Estimates of genetic diversity, neutrality and linkage disequilibrium were calculated using DnaSP software (v. 4.50.3; Rozas et al., 2003). We obtained the associated one-tailed P-values for Tajima's D and Fu and Li's F* by computing 10 000 coalescent simulations based on θ from the observed data and assuming free recombination in DnaSP (Hudson, 1990). Because we performed multiple tests for neutrality across eight populations, we applied a standard Bonferroni correction to levels of significance for each locus (8 tests/neutrality statistic, Bonferroni critical value ∝ = 0.006; Rice, 1989). We examined the population structure of P. balsamifera using an analysis of molecular variance (AMOVA) implemented in the arlequin software package (v. 3.11; Excoffier et al., 1992; Schneider et al., 1997). This method partitioned the genetic variance among northern and southern regions (FCT), among populations within those regions (FST) and within populations (FSC). Significance levels were determined using uncorrected pairwise differences between haplotypes through 1000 random permutation replicates. Using phase (v. 2.1.1; Stephens et al., 2001; Stephens & Donnelly, 2003), we calculated ρ, the recombination parameter, and inferred the haplotype phase for heterozygous alleles to present a geographical display of patterns of variation for each locus (see Fig. 1). We used the default model (-MR0), which is the general model for recombination rate variation. Separate phase runs for each locus were performed with a burn-in period of 100 followed by 10 000 iterations to ensure convergence of haplotype estimation. Over 97% of haplotypes were determined at a confidence probability of ≥ 95%. The remaining haplotypes contained only a single polymorphic site of uncertain phase. ρ was estimated for each locus as the median of the results for the posterior distribution of the recombination parameter across 10 000 data sets generated by coalescence (Crawford et al., 2004). All sequences were deposited in the European Molecular Biology Laboratory (EMBL)/GenBank nucleotide sequence database (accession numbers FJ581048–FJ581417). We compared average per site pairwise nucleotide diversity (π) between our North American collections of poplar (P. balsamifera, P. deltoides and P. trichocarpa) and three Eurasian Populus species (Populus alba, Populus nigra and P. tremula) using samples available from EMBL/GenBank. For Adh1 we were able to compare diversity among P. balsamifera, P. deltoides, P. trichocarpa (data presented herein), P. tremula (accession numbers AJ842873–AJ842906; Ingvarsson, 2005a), P. nigra (accession numbers AJ580714–AJ580723; unpublished) and P. alba (accession numbers AJ580702–AJ580713; unpublished). For G3pdh and ABI1B, however, we were able to compare only P. balsamifera, P. deltoides, P. trichocarpa and P. tremula (G3pdh: accession numbers AJ843576-AJ843623, Ingvarsson, 2005a; ABI1B: accession numbers AM690392–AM690435, Garcia & Ingvarsson, 2007) because data for the other two Eurasian poplar species were not available. Sequence data were obtained from EMBL/Genbank, aligned in sequencher and trimmed to the length of our partial sequence before analysis. Results Nucleotide diversity in P. balsamifera We sequenced regions of Adh1, G3pdh and ABI1B from 5 to 18 individuals within eight populations of P. balsamifera for a total sample of 102–105 trees and 204–210 alleles per locus. For each individual, we aligned a total of 1827 bp. Overall, this included 858 bases from coding regions and 969 bases from introns and untranslated regions. The average sequence length was 609 bp and included both coding and noncoding sites for each locus (Table 2). The complete coding regions were obtained for exons II–IV, exons V–V11 and exon III for Adh1, G3pdh and ABI1B, respectively. Table 2. Loci studied and length of regions analyzed in Populus balsamifera Locus Alleles Totala Coding Noncodinga Codingb Nonsynonymous Synonymous Adh1 204 567 225 342 169.17 55.83 G3pdh 210 599 324 275 245.95 78.05 ABI1B 206 661 309 352 231.11 77.89 Total 1827 858 969 646.23 211.77 a Including indels. b The total numbers of synonymous and nonsynonymous sites were computed as Nei & Gojobori (1986). Adh1, alcohol dehydrogenase 1; G3pdh, glyceraldehyde 3-phosphate dehydrogenase; ABI1B, abscisic acid insensitivity 1B. We observed from five to eight segregating sites (single nucleotide polymorphisms (SNPs)) per locus; values of θW ranged from 0.00150 in Adh1 to 0.00205 in both G3pdh and ABI1B (Table 3). We observed only synonomous SNPs in coding regions for Adh1. By contrast, only nonsynonymous SNPs were observed in coding regions of G3pdh and ABI1B. A total of three singletons were detected across all loci. Our estimates of nucleotide diversity for P. trichocarpa (θW ranged from 0.00131 to 0.00401) and P. deltoides (θW ranged from 0.00160 to 0.00245) were similar to those observed in P. balsamifera. Table 3. Estimates of nucleotide diversity in North American Populus Species Locus Region Alleles S H d Polymorphism (×10−3) θW πtot πsil πsyn πnonsyn Populus balsamifera Adh1 North 122 4 0.574 1.31 3.48 4.96 8.94 0.00 South 82 5 0.669 1.77 3.27 4.67 9.02 0.00 Total 204 5 0.630 1.50 3.41 4.87 8.93 0.00 SD 0.022 0.07 0.07 G3pdh North 128 6 0.735 1.91 2.26 1.58 0.00 3.20 South 82 4 0.717 1.39 2.01 1.89 0.00 2.20 Total 210 7 0.730 2.05 2.16 1.70 0.00 2.81 SD 0.017 0.77 0.08 ABI1B North 120 8 0.696 2.26 3.67 3.35 0.00 4.28 South 86 5 0.614 1.51 2.38 2.41 0.00 2.33 Total 206 8 0.734 2.05 3.53 3.26 0.00 4.03 Populus deltoides Adh1 10 0.012 0.73 0.08 G3pdh 10 4 0.644 2.45 2.58 2.84 2.56 2.26 ABI1B 10 3 0.356 1.60 1.61 2.49 9.13 0.00 Populus trichocarpa Adh1 30 9 0.655 4.01 4.83 6.88 4.28 0.00 G3pdh 30 3 0.432 1.31 1.50 2.24 0.00 0.52 ABI1B 30 10 0.779 3.82 3.46 4.35 2.39 1.84 Adh1, alcohol dehydrogenase 1; G3pdh, glyceraldehyde 3-phosphate dehydrogenase; ABI1B, abscisic acid insensitivity 1B; S, segregating sites; Hd, haplotype diversity; SD, standard deviation; θW, Watterson estimator. Average pairwise nucleotide diversity: πtot, total (nonsynonymous, synonymous and noncoding); πsil, silent (synonymous andnoncoding); πsyn, synonymous; πnonsyn, nonsynonymous. A decline in linkage disequilibrium as pairwise nucleotide distance increased, represented by the relationship between r2 and the distance in base pairs between polymorphic sites, was not apparent for any of the loci, probably because the average distance between sites was < 250 bp (quadratic regression r2 ranged from 0.02 to 0.24, P > 0.35 for all regions). Estimates of the recombination parameter generated from phase varied by 3 orders of magnitude among loci (ρ = 1.89 × 10−3 for Adh1; 6.03 × 10−4 for G3pdh; 4.48 × 10−6 for ABI1B). The recombination rate (r) relative to mutation (µ) estimated as ρ/θW (= 4Ner/4Neµ=r/µ (Ne, effective population size)) varied from 1.26 in Adh1 to 0.294 in G3pdh and 0.0022 in ABI1B. Based on a heuristic comparison of the average ρ:θ ratio in P. tremula (ρ:θ ≈ 1; Ingvarsson, 2008), the only locus that appears to be an outlier is ABI1B. Haplotype diversity (Hd) was similar across all loci in P. balsamifera, ranging from 0.630 in Adh1 to 0.730 in G3pdh and 0.734 in ABI1B (Fig. 1, Table 3). Populus balsamifera and P. trichocarpa shared no haplotypes with P. deltoides. By contrast, P. balsamifera and P. trichocarpa shared haplotypes for all three loci. Of eight Adh1 haplotypes observed in P. balsamifera and P. trichocarpa, one was found in both species. Two of eleven G3pdh haplotypes and five of 10 ABI1B haplotypes also were shared. The most common ABI1B variant in P. trichocarpa (12/30 alleles), however, was unique to that taxa. Levels of genetic variation as estimated by segregating sites (S), θW, π and Hd in G3pdh and ABI1B were greater for populations in the north than in the south (Table 3). This trend also was observed in Adh1 for π, but not for S, θW, and Hd. Based on Tajima's D and Fu and Li's F*, we found evidence for departure from neutral evolution in a few populations for G3pdh and ABI1B, although Tajima's D was significant only in the Alaskan Chena River population for ABI1B after Bonferroni correction (Table 4; D= 1.881, P = 0.002; Bonferroni critical value ∝ = 0.006). For Adh1, both Tajima's D (range 1.242–2.495, P = 0.001–0.052) and Fu and Li's F* (range 0.646–1.715, P = 0.003–0.277) were elevated in most populations. For P. trichocarpa and P. deltoides, we found evidence for nonneutral evolution only in Adh1 for P. trichocarpa (Table 4; F* = 1.337, P = 0.033). Table 4. Estimates of neutrality observed in North American Populus species Species Region Population D Tajima F* Fu&Li Adh1 G3pdh ABI1B Adh1 G3pdh ABI1B Populus balsamifera North Cache Creek 2.266* (0.001) 0.214 (0.360) 1.707 (0.010) 1.639* (0.009) 0.279 (0.424) 0.952 (0.113) Cottonwood Creek 2.495* (0.001) 0.061 (0.458) 0.334 (0.334) 1.715* (0.003) 0.804 (0.262) 1.108 (0.099) Yukon River bridge 2.109* (0.001) 1.530 (0.039) 1.213 (0.053) 1.584* (0.003) 1.289 (0.042) 0.781 (0.178) Chena River 2.491* (0.001) 0.383 (0.286) 1.881* (0.002) 1.712* (0.005) −0.119 (0.582) 1.013 (0.096) South Grand Portage 1.953* (0.001) 0.096 (0.424) 1.471 (0.019) 1.582* (0.004) 0.174 (0.426) 1.450 (0.021) Grand Forks 1.242 (0.052) 1.573 (0.026) 1.114 (0.080) 0.646 (0.277) 1.303 (0.055) 1.234 (0.085) Ridges State Park 1.471 (0.014) 1.438 (0.042) −0.329 (0.646) 1.450 (0.022) 1.255 (0.067) 0.450 (0.322) Guelph Lake 2.154* (0.003) 1.190 (0.082) 0.755 (0.159) 1.597* (0.009) 1.180 (0.089) 1.197 (0.080) Populus deltoides Hubbard State Park −1.562 (0.986) 0.204 (0.379) 0.021 (0.475) −1.934 (0.986) −0.231 (0.599) 0.982 (0.216) Populus trichocarpa Valdez 0.638 (0.163) 0.333 (0.338) −0.300 (0.673) 1.337 (0.033) 0.894 (0.208) 0.595 (0.226) One-tailed P-values for Tajima's D and Fu and Li's F* calculated by coalescent simulation are in parentheses. Significant neutrality estimates are in bold type (P≤ 0.050). Estimates that were significant after correction for multiple inference are indicated with an asterisk (∝= 0.006). Adh1, alcohol dehydrogenase 1; G3pdh, glyceraldehyde 3-phosphate dehydrogenase; ABI1B, abscisic acid insensitivity 1B. Because positive values of Tajima's D and Fu and Li's F* can be generated by sampling across paralogs of a duplicated gene, we examined this possibility for Adh1 in P. balsamifera. Populus trichocarpa is an ancient polyploid (Sterk et al., 2005; Tuskan et al., 2006); so we performed BLAST searches for Adh duplicates within the assembled genome of P. trichocarpa, available at http://genome.jgi-psf.org/ (Tuskan et al., 2006). This search yielded only one copy of Adh1. Sampling across paralogs also is expected to yield a phylogenetic tree with two distinct clades and long internal branches. To address this possibility, we constructed a neighbor-joining tree using paup* (v. 4.0; Swofford, 2002) assuming a Jukes–Cantor model of evolution. The topology we observed (not shown) lacked deep coalescence and long internal branches, which is inconsistent with patterns expected from sampling across paralogs. Finally, if we had sampled across paralogs we would have expected most polymorphic sites to be heterozygous for nearly all individuals; however, this pattern was not detected. Our analyses, combined with the observations of low genetic diversity and high recombination, collectively suggest that we did not sample paralogs from an ancient gene duplication event. Population differentiation in P. balsamif