Rapid Concerted Evolution of Nuclear Ribosomal DNA in Two Tragopogon Allopolyploids of Recent and Recurrent Origin
A. Kovarik, J. C. Pires, A. R. Leitch, K. Y. Lim, A. M. Sherwood, R. Matyasek, J. Rocca, D. E. Soltis, P. S. Soltis


We investigated concerted evolution of rRNA genes in multiple populations of Tragopogon mirus and T. miscellus, two allotetraploids that formed recurrently within the last 80 years following the introduction of three diploids (T. dubius, T. pratensis, and T. porrifolius) from Europe to North America. Using the earliest herbarium specimens of the allotetraploids (1949 and 1953) to represent the genomic condition near the time of polyploidization, we found that the parental rDNA repeats were inherited in roughly equal numbers. In contrast, in most present-day populations of both tetraploids, the rDNA of T. dubius origin is reduced and may occupy as little as 5% of total rDNA in some individuals. However, in two populations of T. mirus the repeats of T. dubius origin outnumber the repeats of the second diploid parent (T. porrifolius), indicating bidirectional concerted evolution within a single species. In plants of T. miscellus having a low rDNA contribution from T. dubius, the rDNA of T. dubius was nonetheless expressed. We have apparently caught homogenization of rDNA repeats (concerted evolution) in the act, although it has not proceeded to completion in any allopolyploid population yet examined.

IN allopolyploids, parental genomes may undergo numerous genetic changes (for reviews see Leitch and Bennett 1997; Matzke et al. 1999; Soltis and Soltis 2000; Rieseberg 2001; Osborn et al. 2003). Repetitive sequences frequently show evidence of homogenization, probably caused by processes such as unequal crossing over and gene conversion, mechanisms collectively referred to as concerted evolution (Zimmer et al. 1980; Dover 1982). The process of concerted evolution can best be studied in synthetic polyploids or in natural polyploids of clear and very recent ancestry. However, only a few naturally occurring polyploid species are known to have arisen spontaneously within the past 150 years: Cardamine schulzii (Urbanska et al. 1997), Spartina anglica (Ainouche et al. 2004), Senecio cambrensis, and Senecio eboracensis (Abbott and Lowe 2004), and Tragopogon mirus and T. miscellus (Ownbey 1950; Soltis et al. 2004).

Tragopogon (Asteraceae) comprises ∼100–150 species native to Eurasia. Most species are diploid (2n = 12). Three diploid species, T. dubius, T. porrifolius, and T. pratensis, were introduced into the Palouse region of eastern Washington and adjacent Idaho in the early 1900s (Ownbey 1950). The introduction of these diploid species into the Palouse brought them into close contact, something that rarely occurs in the Old World where the diploids are largely allopatric. Using morphology and cytology, Ownbey (1950) demonstrated that T. mirus and T. miscellus are allotetraploids (2n = 24) whose diploid (2n = 12) parents are T. dubius and T. porrifolius and T. dubius and T. pratensis, respectively. The ancestries of both tetraploids were subsequently confirmed through flavonoid, isozymic, and DNA studies (Ownbey and McCollum 1953; Brehm and Ownbey 1965; Roose and Gottlieb 1976; Soltis and Soltis 1989, 1991; Soltis et al. 1995; Cook et al. 1998). The allotetraploids have not formed in Europe, but are native to the Palouse, although their diploid parents are aliens in North America. Because the three diploids probably did not co-occur in the Palouse prior to the 1920s (the earliest collection, of T. dubius, the diploid parent common to both tetraploid species, is from 1928; Ownbey 1950), T. mirus and T. miscellus cannot be > ∼80 years old.

Ownbey (1950) discovered two populations of each new allotetraploid species, T. mirus and T. miscellus, and suggested that each was of independent origin. Subsequent morphological and cytological (Ownbey and McCollum 1954), isozymic (Roose and Gottlieb 1976; Soltis et al. 1995), and DNA (Soltis and Soltis 1989, 1991; Cook et al. 1998) evidence, when considered along with geographic distribution, demonstrated that each allopolyploid had formed repeatedly. In fact, there may be as many as 21 lineages of separate origin of T. miscellus and 12 of T. mirus in the Palouse alone. On a larger geographic scale, both tetraploids have also formed in Flagstaff, Arizona (Brown and Schaak 1972); T. miscellus has also formed in Gardiner, Montana, and Sheridan, Wyoming (M. Ownbey, unpublished results; D. E. Soltis and P. S. Soltis, personal observation).

Variable numbers of rRNA genes coding for 18S-5.8S-26S RNA occur in different plant species (between 1000 and >50,000 genes), forming multigene families in long tandem arrays (Hemleben and Zentgraf 1994). The intragenic (ITS) and intergenic spacers (IGS) associated with genic regions often vary among species and may be used to detect parentage of allotetraploids. However, similarly to other repeats the rRNA genes (rDNA) are subjected to concerted evolution leading to reduction of unit types. It has been proposed that this homogenization may significantly obscure the estimation of parental genome donors in allopolyploids (Álvarez and Wendel 2003). Perhaps one of the best examples of concerted evolution of rDNA was described in hybrid parthenogenetic lizards (Hillis et al. 1991). Some allopolyploid plants show some or considerable evidence of concerted evolution, including Gossypium (Wendel et al. 1995a), Nicotiana (Volkov et al. 1999; Lim et al. 2000; Kovarik et al. 2004), Cardamine (Franzke and Mummenhoff 1999), Triticum (Flavell and O'Dell 1976), some species of Brassica (Koch et al. 2003), and Glycine (Rauscher et al. 2004). Concerted evolution may be mediated by locus loss (Kotseruba et al. 2003) or interlocus gene conversion (Wendel et al. 1995a; Lim et al. 2000) or a combination of both. In contrast, some allopolyploids, including Brassica napus (Bennett and Smith 1991), Arabidopsis suecica (O'Kane et al. 1996), some species of Paeonia (Zhang and Sang 1999), and Krigia (Kim and Jansen 1994), do not homogenize rDNA, and the parental arrays show independent evolution and may undergo epigenetic rather than genetic changes (Chen and Pikaard 1997).

At least two major factors limit comparisons of the genetics and genomics of relatively old allotetraploids (e.g., Brassicas, tobacco) and modern populations of their diploid progenitor species. First, the time of origin of the polyploid is unknown, making inferences of molecular rates difficult. Second, the actual diploid populations (or genotype donors) contributing to the allopolyploid are also unknown. Extant diploid genomes may have diverged from the original genomes donated to the polyploid, or one or both diploid parents may now be extinct. In either case, comparisons of allopolyploids and extant diploids may be misleading with regard to patterns and rates of genetic change.

The recent allotetraploids T. mirus and T. miscellus represent a convenient natural system in which to study the early genetic changes associated with allopolyploidization. Given their recent origins, their parental genotypes are likely still extant in populations of the diploid progenitor species, and evidence from isozymes and DNA supports this inference. Early molecular studies of T. mirus and T. miscellus revealed the presence of both diploid parental rDNA types in these two allotetraploids (Soltis and Soltis 1991). However, ITS sequences of clones from the polyploids suggested that the parental rDNA types were not present in the allotetraploids in equal frequency. In fact, T. dubius appeared to be underrepresented in both allotetraploid species. To assess whether concerted evolution has been operating in the recently formed allotetraploids, T. mirus and T. miscellus, we investigated the rDNA cistron using cloning, Southern blots, slot blots, and single-stranded conformation polymorphism (SSCP) analysis. We investigated multiple individuals within populations and multiple populations of each tetraploid species that represent independent polyploidization events to assess whether the rDNA cistron was evolving comparably in populations of independent origin. In addition, we examined expression using reverse transcription PCR (RT-PCR) and investigated rDNA locus loss via the examination of chromosomes with fluorescent in situ hybridization (FISH).


Plant material:

Seeds of the three diploid and two allotetraploid species of Tragopogon were collected from natural populations in Idaho and Washington (Table 1). For each population sampled, we collected seeds from 50 individual plants; the seeds from each plant were kept separate. These seeds were subsequently planted in the greenhouses at Washington State University; plants were grown to maturity and allowed to self-pollinate. The resultant selfed seeds were collected from individual plants. Some of these seeds, representing one generation of selfing in the greenhouse, were later germinated in a greenhouse at the Royal Botanic Gardens, Kew, for use in this investigation. Additional seeds were germinated at the University of Florida to assess rDNA variation among individuals within each of the populations examined of the two tetraploids, T. mirus and T. miscellus. Root tips for cytogenetic studies and leaf tissue for isolation of DNA and RNA were harvested from 5- to 8-week-old plants.

View this table:

Summary of collections of Tragopogon used in this study

To represent the rDNA repeat condition close to the actual time of polyploidization, we used tissue from some of the earliest herbarium collections of T. mirus and T. miscellus (Table 1). Two of these collections were made by M. Ownbey in 1949, the year of his discovery of the allotetraploids (see Ownbey 1950). In fact, the specimens we used are isotypes; these represent duplicates of the first collections made of these two newly formed species. Thus, these two collections, one of T. miscellus from Moscow, Idaho, and one of T. mirus from Pullman, Washington, represent a point in time very close to the origin of these allotetraploids in these two towns (Ownbey 1950). Ownbey noted that these initial populations were “small and precarious” with each population consisting of <100 plants. A third early collection of T. miscellus, from Sheridan, Wyoming, collected in 1953 was also used. Little is known about the origin of T. miscellus in Sheridan, but we assume that this time of collection is also close to the time of polyploid formation in that area, given that the diploid progenitors were introduced into that area in the early 1900s (R. Hartman, personal communication).

Restriction site analysis and Southern blot hybridization:

We isolated total genomic DNAs from diploid seedlings of the same age (Table 1). To assess the extent, if any, of restriction site variation within populations of the two allotetraploids, we examined 21 plants of T. miscellus from three populations and 22 plants of T. mirus from three populations, along with samples of their diploid progenitors. We also examined rDNA variation among progeny resulting from three different heads (inflorescences) collected from the parent plant of T. mirus 2603-33. Three seedlings from head A, four seedlings from head B, and a single seedling from head C were compared.

Purified DNAs were digested with restriction enzymes, separated by gel electrophoresis, and blotted onto nylon membranes (Hybond XL, Amersham Biosciences, Buckinghamshire, UK). The Southern blot hybridization was carried out in modified Church-Gilbert buffer (Lim et al. 2000) using the 18S or 26S rDNA labeled with [α-32P]dCTP (ICN, Irvine, CA). The 18S probe was a cloned 1.7-kb fragment of the tomato 18S rRNA gene (Kiss et al. 1989); the 26S probe was an ∼280-bp PCR product derived from the 3′ end of the 26S rRNA tobacco gene (Lim et al. 2000). The bands were visualized by phosphorimaging [Storm, Molecular Dynamics (Sunnyvale, CA), and Amersham Biosciences].

For slot-blot hybridizations, 0.2 μg of genomic DNA were blotted onto a nylon membrane using a Schleicher & Schuell (Dassel, Germany) slot apparatus and hybridized with a 26S rDNA probe. The DNA concentrations were estimated using a spectrophotometer and determined to be nearly identical. Serial dilutions of plasmid inserts were loaded as amount standards. The relative proportion of parental ITS and IGS types in the tetraploids was determined by quantifying the radioactivity in diagnostic bands using a phosphorimager. Average values were calculated from triplicate experiments.

Chromosome preparation and fluorescent in situ hybridization:

Chromosome preparations and FISH were performed for multiple populations of the three parental diploid species (T. dubius, T. pratensis, and T. porrifolius) and two allopolyploid derivatives (T. mirus and T. miscellus) following Leitch et al. (2001) and Pires et al. (2004). The rDNA probe pTa71 (Gerlach and Bedbrook 1979) was labeled with biotin-16-dUTP (Sigma-Aldrich, St. Louis), and hybridization was detected with avidin-Cy3 (Amersham Pharmacia Biotech). Images were taken with a Leica CTRMIC epifluorescent microscope with an Orcha ER camera, and images were processed using Openlab software (Improvision, Amersham, Buckinghamshire, UK). Only those functions that apply equally to all pixels in the image were used. For every set of observations, between 5 and 10 metaphases were examined from each of one to three slides.

PCR and cloning experiments:

About 50 μg of total genomic DNA was isolated from leaf tissue of single plants or from herbarium specimens using the QIAGEN (Hilden, Germany) kit. For amplification and sequencing of the ITS region, we used ∼100 ng of template DNA, the forward N-nc18S10 and reverse C26A primers (Wen and Zimmer 1996). Purified DNAs were cloned using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA). For amplification of ITS-1, the following primers were used: 18Sfor, 5-GCGCTACACTGATGTATTCAA CGA G-3′ and 5.8Srev, 5′-CGCAACTTGCGTTCAAAGACTCGA-3′; for ITS-2, we used 5.8Sfor, 5′-GCATCGATGAAGAACGCA GC-3′ and 26Srev, 5′-CTTTTCCTCCGCTTATTGATATGC-3′. The primers were derived from conserved regions of the rDNA cistron. PCR was carried out on an MJ Research (Watertown, MA) thermocycler allowing an initial denaturation step (92°, 180 sec) and 35 cycles of 92° for 20 sec, 57° for 30 sec, and 72° for 30 sec, followed by a final 72° extension for 10 min. For ITS-2, the conditions were the same except for the annealing temperature (60°).

Individual colonies then served as template for a subsequent series of PCR reactions. Individual colonies were removed with a toothpick and placed in a standard PCR cocktail. PCR reactions were repeated as described above. Products were then sequenced in both directions with the PCR primers as sequencing primers using Beckman-Coulter (Fullerton, CA) dye terminator cycle sequencing quick-start kits and a Beckman-Coulter automated sequencer following the manufacturer's protocols.

The ITS sequences were initially aligned using the program Clustal X (Thompson et al. 1997). Following the initial alignment, the sequences were then adjusted manually using the program Se-Al v.2.0 a11 (Rambaut 2003). Sequences from allotetraploids consisting of ITS-1 (complete), 5.8S, and ITS-2 (partial) were deposited in the EMBL/GenBank database under accession nos. AY458586, AY458588, AY458589, and AY458587.

We also conducted experiments to test for PCR bias. We mixed genomic DNAs from pairs of the diploids (T. dubius 2613, T. pratensis 2598, and T. porrifolius 2612) in ratios of 0, 20, 40, 60, 80, and 100% of each. We then amplified the ITS-1 region. The PCR products were digested with BstNI and HpaII restriction enzymes that yielded species-specific patterns (Figure 5). The experiment showed no evidence of PCR bias, and the products reflect the status of the starting material.

RNA isolation and RT-PCR assay:

To study expression of parental rDNA, we analyzed the ITS-1 region in the rRNA precursors of randomly selected progeny of T. miscellus 2605-6 and T. mirus 2602-6. We applied two-step RT-PCR using total RNA isolated from leaves as template. RNA was isolated from leaves of ∼60-day-old T. mirus 2602-6 and T. miscellus 2605-6 plants by the TRISOL (Sigma-Aldrich, St. Louis) method. RNA samples were treated with RQ1 Dnase I (Promega, Madison, WI; 0.1 units/μg RNA; 60 min at 37°; 15 min at 65°) to eliminate any contaminating genomic DNA. For each RNA isolate we performed several control experiments involving amplification reactions on DNA templates and RNA templates prior to reverse transcription.

Reverse transcription reactions (20 μl) typically contained 1 μg of total RNA, 2 pmol of 5.8Srev primer, and 200 units of reverse transcriptase (Invitrogen Superscript II RNAse H) by using conditions recommended by the supplier (Invitrogen, Paisely, UK). For PCR (50 μl) amplification we used 100 ng of input RNA or 100 ng of genomic DNA as template, 40 pmol of each primer (18Sfor and 5.8Srev), 12 nmol of each dNTP, and 1.9 units of DyNAzyme II DNA polymerase (FINNZYMES, Espoo, Finland). Cycling conditions were as follows: initial denaturation step (92°, 180 sec) and 35 cycles of 92° for 20 sec, 57° for 30 sec, and 72° for 30 sec, followed by a final 72° extension for 10 min. To discriminate among ITS-1 variants, the PCR products were digested with BstNI (New England Biolabs, Beverly, MA), which produces species-specific DNA fragments for each diploid progenitor (Figure 3A); the resulting fragments were separated on a 7% polyacrylamide gel. The gel was stained after electrophoresis with ethidium bromide and photographed. The analyses were carried out on two RNAs isolated from a single plant.

SSCP analysis:

An ∼700-bp DNA fragment, composed of the 3′-part of the 18S rRNA gene, ITS-1, and the 5′-part of the 5.8S rRNA gene, was amplified using 18Sfor and 5.8Srev primers. The 50-μl PCR reaction mixture contained 1× buffer for DyNAzyme II, 250 μm each dNTP, 0.4 μm each primer, 10 ng of genomic DNA, and 2 units DyNAzyme II DNA polymerase under the conditions described above. The PCR mixture was ethanol precipitated, redissolved, treated with HaeIII restriction endonuclease, and size separated on 2% agarose, and the 300-bp fragment was eluted and resuspended in denaturing loading buffer (30% dimethyl sulfoxide, 0.1% bromphenol blue, and 0.1% xylene cyanol). The samples were heat denatured (10 min at 99°), immediately cooled on ice, and loaded onto preelectrophoresed and precooled (5°) gels. The samples were electrophoresed in 10% polyacrylamide gels with 5/9 × TBE supplemented with 7% urea. Gels were run at 7.4 V/cm at 5° for 90 hr and then silver stained or stained with ethidium bromide.


rDNA copy number:

Slot-blot experiments of multiple populations of each of the three diploid species revealed that T. dubius, T. porrifolius, and T. pratensis have, on average, comparable numbers of rDNA repeats (Figure 1). However, estimates vary among populations of both T. dubius and T. porrifolius. Populations of T. dubius from Rosalia (2614) and Spokane (2615) have ∼900 rDNA copies per 1C genome, whereas the population from Pullman (2613) has nearly twice that many—1700. The two accessions of T. porrifolius from Troy (2607 and 2612) had identical numbers of rDNA repeats (1500/1C genome), whereas the population from Pullman (2611) had only 1000 copies. All populations of T. pratensis had similar estimates of rDNA copy number, ranging from 1000 to 1300. In addition, we found that multiple populations of the two polyploids, T. mirus and T. miscellus, contain nearly double the amount of rDNA repeats present in any of the diploids (Figure 1). Variation among populations of the diploid Tragopogon species is an important consideration in the formation of allotetraploids. For example, a polyploid plant of T. mirus that was formed from a “low rDNA copy” T. dubius parent (such as population 2614; Figure 1) and a “high rDNA copy” T. porrifolius (such as population 2607; Figure 1) would yield a raw allopolyploid at the point of origin with more T. porrifolius rDNA repeats than T. dubius repeats. Conversely, a polyploid T. mirus that formed from a “high rDNA copy” T. dubius parent (such as 2613) and a “low copy” T. porrifolius parent (such as 2611) would result in the formation of a T. mirus with more rDNA repeat units of T. dubius than of T. porrifolius. Taking into account this variation in rDNA repeat number among diploid populations noted above (Figure 1), the ratios of parental rDNA types in a newly formed allopolyploid might be as much as 1.6/1.

Figure 1.—

Copy numbers of rDNA units in Tragopogon haploid genomes. Quantification was performed by slot-blot analysis using the 26S rDNA hybridization probe. The signal in allotetraploids was nearly the same as in diploids. The exact copy numbers were calculated from known nuclear content for each species (Pires et al. 2004), considering ∼10 kb as the length of rDNA operon. Note that copy numbers in tetraploids were slightly less than double those in respective diploid progenitors.

With this caveat in mind, both cloning and Southern blot experiments were used to examine the diploid parental rDNA contributions in a suite of populations (Table 1) of the two polyploids, T. mirus and T. miscellus. In the Southern blot experiments, we examined multiple plants from each of the allotetraploid populations. Below we targeted the ITS and IGS regions, which are known to differ in sequence among the three diploids (Soltis and Soltis 1991; Mavrodiev et al. 2005).

Cloning of ITS sequences in present-day populations:

ITS analysis involved the sequencing of 15 or more clones for each accession. Eleven polymorphic sites were found in ITS-1, 13 in ITS-2, and a single site toward the 3′ end of the 5.8S subunit (Table 2). The lengths of completely sequenced ITS-1 regions were 273 bp in T. porrifolius and T. pratensis and 272 bp in T. dubius. The single change detected in the 5.8S region is inferred to be a C → T substitution on the basis of sequences available from this region for >50 species of the genus Tragopogon and closely related genera (Mavrodiev et al. 2005). SSCP analysis of primary PCR products was used to test further the statistical significance of our cloning experiments and confirmed the high homogeneity of rDNA arrays in the parental species (Figure 2).

View this table:

Summary of polymorphic sites within the ITS1, 58S, and ITS2 regions in T. dubius, T. porrifolius, and T. pratensis

Figure 2.—

SSCP analysis of the ITS1 region in allotetraploid species of Tragopogon and their diploid parents. The amplified region is shown in Figure 3A. The high-resolution gel resolved the profiles in the parental species. Two to three bands were visible in T. dubius, two bands in T. pratensis, and a single band in T. porrifolius under the conditions used. The parental bands are found in all lanes loaded with DNA from the tetraploid species. Note the significant reduction of T. dubius-specific bands in the populations of the two allotetraploids, with the exception of T. mirus 2602 (comparable results for all populations were also obtained with cloning and Southern blots).

In 8 of 10 polyploid populations that we examined (all except T. mirus 9, 2602), there were far fewer rDNA units of T. dubius than of the other diploid parent (Table 3). The ratios that typically skewed away from T. dubius were further confirmed by SSCP analysis of the PCR products (Figure 2). We detected a variable number of ITS-1-5.8S-ITS-2 clones that contained features of both parental types (Table 3). Typically, the chimeric sequences exhibited the ITS-1 region of one of the parental diploids and the ITS-2 sequence of the second diploid parent. Only a few clones contained substitutions that did not correspond to interspecific polymorphisms. To reduce the probability of PCR-mediated recombination (Cronn et al. 2002), we reduced the size of the amplicon to ∼700 bp by using ITS-1 primers (18Sfor; 5.8Srev). Essentially the same results were obtained with this combination of primers. A relatively homogeneous “crossover” family was recovered from T. mirus 2602, accounting for ∼19% of the ITS-1 clones. The 5′ half of the gene was mostly similar to the T. dubius ITS-1 sequence, whereas the 3′ half of the sequence was similar to that of T. porrifolius (Table 2).

View this table:

Cloning results for populations of T. mirus and T. miscellus

Cloning of ITS sequences in herbarium specimens:

Resynthesized allotetraploids of T. mirus and T. miscellus would be ideal for assessing the “starting point” ratio of ITS diploid types in a raw allopolyploid. As a substitute for resynthesized polyploids, we used herbarium specimens collected by M. Ownbey in 1949 and 1953 (see materials and methods) to represent the early state of the allotetraploids. Samples from all three herbarium specimens exhibited a nearly 1:1 ratio of diploid parental ITS clones (Table 3), in contrast to ratios obtained from most extant samples.

Restriction site/Southern blot analysis:

We used Southern blot analysis of restriction site variation in all rDNA repeats to provide another perspective on the distribution of rDNA repeat types in allotetraploid Tragopogon. We isolated genomic DNAs from 5 to 10 plants from each of six populations (three T. mirus and three T. miscellus) and subjected DNAs to restriction enzyme analysis. BstNI discriminates T. dubius from the other two diploid species (Figure 3, A and B). In diploid species the ITS-1 probe hybridized to the ∼700-bp fragment in T. pratensis and T. porrifolius and to the ∼500-bp fragment in T. dubius. Both fragments appeared to be present in all populations of both allotetraploids. However, the intensity of the ∼500-bp T. dubius-specific fragment was reduced in all of the populations that we examined of both T. mirus and T. miscellus, with the exception of population T. mirus 2602. Typically, the T. dubius rDNA units represented ∼22–32% and 7–33% of the total rDNA present in T. miscellus (all populations) and T. mirus (except 2602), respectively (Table 4). Some individuals from population T. mirus 2603 (33) had only a faint 500-bp band, representing only ∼7% of the total signal in the lane, suggesting exceptionally low content of T. dubius rDNA. In T. mirus 2602, the ITS repeat of T. dubius was better represented, representing 42% to as much as 80% of the total rDNA.

Figure 3.—

Southern blot hybridization and clone analysis of the ITS1 region. (A) Restriction enzyme maps of the major T. dubius, T. pratensis, and T. porrifolius rDNA units. The position of probe hybridization is indicated by a thick line. The sizes of hybridization fragments are indicated. 18S and 5.8S are genes coding for rRNA molecules. Distances are approximately to scale. Thick arrows indicate positions of 18Sfor and 5.8Srev primers. (B) Southern blot hybridization of genomic DNAs showing variability in the parental ITS1 families content among the populations of T. mirus. The DNAs were digested with the BstNI restriction enzyme and hybridized with the ITS1 probe. The diagnostic fragments representing T. dubius (du) and T. porrifolius (po) families are indicated. Sample 2603-33 is shown after longer exposure to reveal a particularly faint ∼500-bp T. dubius-specific band in this plant. Numbers in italics below the tracts indicate relative proportions of the T. dubius-specific band expressed as a percentage of total signal in tract (also for C and D). (C) Southern blot hybridization of genomic DNAs showing variation in the ITS1 families among populations of T. miscellus. The digestion and hybridization conditions were the same as in B. The diagnostic fragments representing T. dubius (du) and T. pratensis (pr) families are indicated. (D) Analysis of unit inheritance in unusual 2603-33 plants. The sibling plants were grown from seeds collected from different heads (A–C). The Southern blot hybridization and probe was the same as in B. Note variability in the ∼500-bp band of T. dubius origin. (E) Southern blot hybridization showing both inheritance of parental and appearance of novel population-specific bands in population T. mirus 2602 (novel bands not present in any of the parental diploid accessions are indicated with asterisks).

View this table:

Presence of parental rDNA types in T. mirus and T. miscellus

To assess the extent of variation among plants further, we examined rDNA variation among progeny derived from three different inflorescences collected from a single plant of T. mirus 2603-33 (Figure 3D). Plants A1–A3, B3, and C all have negligible amounts of T. dubius rDNA (5–7%). However, in plants B1, B2, and B4, T. dubius represents 18–20% of the total, which is more typical of that observed in other plants of both T. mirus and T. miscellus (Table 4).

The IGS region was analyzed with several methylation-insensitive enzymes. The profiles obtained for diagnostic KpnI sites (Figure 4A) using the 18S probe are shown in Figure 4, B and C. The enzyme digestion produced a single KpnI band of ∼11 kb in all T. dubius accessions, an ∼8-kb band in T. pratensis, and multiple bands in the ∼8-kb region in T. porrifolius. In T. mirus 2601 and 2603 the 11-kb band originating from the T. dubius parent was weaker than the bands in the 8-kb region. As for the ITS-1 hybridization, the 18S probe showed the weakest hybridization to the T. dubius-specific band in the lane loaded with the DNA isolated from plant 2603-33 (Figure 4B). However, T. mirus 2602 was unique among polyploid populations surveyed in having a predominant 11-kb band (corresponding to T. dubius) and a relatively weaker pair of 8-kb bands (Figure 4B). In the allotetraploid T. miscellus, both parental bands were identified; however, the intensity of the 11-kb band was reduced from the expectations of simply combining the two diploid parents (Figure 4C). Other enzymes produced similar results (not shown). We also attempted to conduct Southern blot experiments on DNAs from the three herbarium collections used above (Table 2), but because the DNAs were somewhat degraded, we were unable to quantify precisely the diploid contributions using this approach.

Figure 4.—

Southern blot hybridization analysis of the IGS region. (A) Restriction enzyme maps of the major T. dubius, T. pratensis, and T. porrifolius rDNA units. The position of probe hybridization is indicated by a thin line. The sizes of hybridization fragments are indicated. 18S, 5.8S, and 26S are genes coding for rRNA molecules. (B) Southern blot hybridization of genomic DNAs showing variation in the parental IGS regions among populations of T. mirus. The DNAs were digested with the KpnI restriction enzyme and hybridized with the 18S rDNA probe. The diagnostic fragments representing T. dubius (du) and T. porrifolius (po) families are indicated. T. porrifolius had a variable number of bands in the 8-kb region, indicating the presence of multiple gene families displaying sequence polymorphisms in the IGS region. Numbers in italics below the tracts indicate relative proportions of the T. dubius-specific band expressed as a percentage of total signal in the tract (also for C). (C) Southern blot hybridization of genomic DNAs showing variation in the IGS families among populations of T. miscellus. The digestion and hybridization conditions were the same as in B. The diagnostic fragments representing T. dubius (du) and T. pratensis (pr) families are indicated. (D) The presence of novel polymorphic sites in the IGS spacer of T. mirus, population 2602. Genomic DNAs from T. dubius, T. porrifolius, and T. mirus 2602-6 were digested with BstYI/SspI and VspI methylation-insensitive restriction enzymes and the blots were hybridized to the 26S probe. Hybrid-specific bands not seen in any progenitor populations are indicated by asterisks.

Quantification of the relative proportion of parental IGS types in the tetraploids with a phosphorimager (indicated by du, po, and pr in Figure 4, C and D) (Table 4) yielded results in close agreement with the results for ITS (above; Table 4) and in good agreement with the findings of our cloning experiments. Our analyses again revealed that rDNAs of T. dubius origin were in low abundance. Again, population T. mirus 2602 differed significantly (P < 0.01) from all other present-day polyploid populations examined in having a relatively high T. dubius rDNA content. All nine plants examined from this population exhibited this same general pattern. Another distinctive feature of this population was a unique hybridization pattern; genomic DNA of this collection displayed several novel bands not present in any of the diploid species after digestion with several enzymes (Figures 3E and 4D). The appearance of the 170-bp BstNI/NlaIII fragment following hybridization with the ITS-1 probe (Figure 3E) was in good agreement with the predicted restriction pattern from the sequence analysis.

Expression patterns:

No amplification products resulted when RNA was used alone, indicating absence of contamination of RNA samples with DNA (not shown). To exclude possible primer bias toward either sequence type, we mixed RNA or DNA from progenitor species at stoichiometric ratios and subjected the mixture to RT-PCR and PCR. The mixed samples produced full additivity of parental bands (Figure 5, left and right margins).

Figure 5.—

35S rRNA gene expression in diploid and allotetraploid species of Tragopogon. Primary RNA transcripts were analyzed in the ITS-1 region by RT-PCR using BstNI restriction site polymorphisms. The digestion products of PCR reactions were separated on a 7% polyacrylamide gel and stained with ethidium bromide. The profiles after reverse transcription and subsequent cDNA amplification are indicated as “RNA”; the profiles obtained after DNA amplification are indicated as “DNA.” Control mixing experiments containing RNA and DNA samples are shown on the right and left margins. The sizes of the diagnostic bands were as follows: 90 bp (T. dubius), 500 bp (T. dubius), 680 bp (T. porrifolius), and 680 bp (T.pratensis).

The cDNA analysis in the two allotetraploid species revealed two different expression phenotypes. In T. miscellus 2605-6 both parental T. dubius and T. pratensis bands were detected after reverse transcription, suggesting expression of both parental rDNA types (Figure 5, lane “RNA”). However, in T. mirus 2602-6, amplification of DNA produced nearly equal fluorescence signals in bands corresponding to T. dubius and T. porrifolius units, but amplification of cDNA yielded a strong T. porrifolius-specific band (∼680 bp) and negligible amounts of T. dubius-specific bands (∼500 and ∼90 bp). This finding indicates dominant transcription of T. porrifolius units and silencing of those units of T. dubius origin. The same result was observed in analysis of the ITS-2 region of primary transcripts (not shown).

Localization of rDNA on chromosomes:

To determine whether the rDNA units of T. dubius have been lost or replaced in Tragopogon allotetraploids, we examined the chromosomes of the tetraploids and their diploid progenitors by FISH (Figure 6). The pTa71 probe carrying the entire 18S-5.8S-26S genic region hybridized strongly to one chromosome pair in T. dubius and T. pratensis and to two chromosome pairs in T. porrifolius. The probe hybridized to two chromosome pairs in T. miscellus (which combines the T. dubius and T. pratensis parental loci) and to three chromosome pairs in T. mirus (which combines the T. dubius and T. porrifolius parental loci). Thus, the application of FISH to the tetraploids shows complete additivity in locus number found in the progenitor diploids, consistent with our previous studies (Soltis and Soltis 1991; Pires et al. 2004). We estimate that a loss of 75–90% of the rDNA units would have been detected by FISH, but no such substantial reduction was observed. It is therefore likely that some rDNA of T. dubius origin was converted to the T. pratensis and T. porrifolius type in the allotetraploids.

Figure 6.—

FISH to root-tip metaphase (A, C, D, G, and I), interphase (B, E, and F), and prophase (H) cells of Tragopogon using biotin-labeled pTa71 detected with avidin-Cy3 (red fluorescence) for 18-5.8-26S rDNA and counterstained with DAPI for DNA (blue fluorescence). (A and B) T. dubius at metaphase. (A) One pair of rDNA loci, which show variable levels of decondensation at interphase (B), giving strings and foci of signal. (C) T. pratensis has one pair of rDNA loci; the apparently remote signal on the metaphase chromosomes (arrows) indicates condensed rDNAs distal to secondary constrictions. (D and E) T. porrifolius. The metaphase preparation (D) shows two pairs of rDNA loci, one on chromosome A and one on chromosome D; each pair is potentially active, but often only the sites on chromosome D are decondensed, resulting in secondary constrictions (D, arrows). At interphase (E), the less active pair on chromosome A is regularly condensed and outside the nucleolus (arrowheads). (F and G) T. miscellus 2605, with two pairs of rDNA sites; both are decondensed at interphase (F), giving strings and foci of signal, and both have secondary constrictions at metaphase (G). (H and I) T. mirus 2602 has three pairs of rDNA loci. All are potentially active, but one pair is typically condensed and lacks secondary constrictions, identified at prophase (H, arrowheads) and metaphase (I, small arrowhead). In I, the arrows identify the rDNA loci of chromosome D of T. porrifolius. Assuming that rDNA condensation patterns are inherited in a manner that resembles those of the diploids, then the large arrowhead designates the T. dubius-derived locus and the small arrowhead the T. porrifolius-derived locus from chromosome A.

Locus decondensation at interphase and secondary constrictions at metaphase signify rDNA expression at the locus. Secondary constrictions were observed at the single rDNA locus in both T. dubius (Figure 6B) and T. pratensis (Figure 6C). Both of these loci are decondensed and probably expressed in T. miscellus (Figure 6, F and G). T. porrifolius has two rDNA loci (Ownbey and McCollum 1954) on chromosomes A and D (Pires et al. 2004), and the locus on A is regularly condensed and presumably transcriptionally silent in root-tip meristematic cells (Figure 6, D and E). The pattern of activity observed in T. mirus reflects the pattern found in its progenitor diploids (Figure 6, H and I): that is, the rDNA loci from T. dubius and from chromosome D in T. porrifolius are active, whereas the locus from chromosome A in T. porrifolius is inactive.


We have studied the evolution of the rDNA sequences in T. mirus and T. miscellus, two recently and recurrently formed allotetraploids that originated in the past 80 years or less. We analyzed present-day populations, as well as herbarium specimens collected in 1949 and 1953, very close to the time of the origin of the allopolyploids. Our data indicate that concerted evolution may occur rapidly (within a few decades) in newly formed polyploids.

Biased elimination of T. dubius rDNA in allotetraploids:

Nearly equal numbers of parental rDNA types were found in the samples of “paleo” DNAs isolated from herbarium specimens of the two allotetraploids collected in 1949 and 1953. It is clear that Ownbey's early collections of T. mirus and T. miscellus from 1949 and 1953 are very close to the time of origin of these newly formed species (see Ownbey 1950). Thus, if the populations Ownbey described actually formed in the early 1940s, concerted evolution must have occurred within the past 60 years. Given that these plants appear to be biennials, the time frame is actually closer to only 30 generations.

Relatively balanced ratios of parental ITS types were also found in the Albion population of T. mirus (population 9). Significantly, the Albion population may represent a very recently formed population of T. mirus for the following reasons: (i) D. and P. Soltis first visited the Albion site in 1990 and made the first known collection of T. mirus from this town, finding one relatively small population (<200 individuals) as well as both parental diploids; (ii) Ownbey did not detect either tetraploid species in Albion during his collecting of Tragopogon through the 1960s and early 1970s (Novak et al. 1991); and (iii) the Albion population was an unusual population of T. mirus in that it exhibited considerable additivity in RAPD markers, suggesting low genome divergence from its ancestral diploids (Cook et al. 1998). Thus, the Albion collection of T. mirus appears to be more recently formed than the other populations of living plants included in this study. Its rDNA ratios might therefore be more balanced because insufficient time has elapsed to allow homogenization of units across the population.

In 8 of 10 polyploid populations that we examined (all except T. mirus 9, 2602), there are far fewer rDNA units of T. dubius than of the other diploid parent. A population-level analysis of unit-specific restriction sites revealed that the direction of homogenization was uniform in all individuals analyzed (Table 4). Our application of FISH to allotetraploids indicates that the parental rDNA loci remain intact, suggesting that the rDNA sequences of T. dubius origin might have been converted to the T. pratensis or T. porrifolius types. It is unclear why the T. dubius variants typically are less favored than the rDNAs of the other two diploid progenitor species and why other loci, including 5S rDNA (data not shown) and satellite repeats (Pires et al. 2004), do not seem to undergo concerted evolution. Structural, chromosomal, and epigenetic factors need to be considered when evaluating why concerted evolution may occur at some loci but not at others.

An important comparison is provided by populations 2604 (Moscow, ID) and 2605 (Pullman, WA), which represent reciprocally formed populations of T. miscellus. We compared the rDNA copy number in these two polyploid accessions to their putative diploid parental genotypes [T. pratensis (2608 Moscow, ID) and T. dubius (2613 Pullman, WA)] (Soltis et al. 1995). Importantly, T. dubius 2613 has more rDNA units (by 30%) than found in T. pratensis, 2608 (Figure 1). Yet, in populations 2604 and 2605 of the allotetraploid T. miscellus, the T. dubius rDNA units are severely underrepresented compared to the contribution of T. pratensis. Molecular evolution of the rDNA cistron in these plants typically follows the same trajectory (i.e., is repeatable). Similarly, as in allopolyploids in Nicotiana (Kovarik et al. 2004) and Glycine (Joly et al. 2004), the direction of rDNA fixation in Tragopogon does not seem to be determined by direction of the cross, suggesting that nuclear-cytoplasmatic interactions (Gill 1991) might not play a significant role in elimination of parental rDNAs in these allotetraploids.

Bidirectional concerted evolution:

T. dubius was consistently the underrepresented rDNA type in all plants from all populations of T. miscellus examined, as well as in all plants from three of the five populations of T. mirus examined. However, in two populations of T. mirus (9, 2602), the rDNA units of T.dubius and T. porrifolius were more balanced in their representation. In fact, in one of the nine plants examined (2602–10) from this population, 80% of the rDNA units were from T. dubius (Figures 3B and 4B), suggesting that the typical trend of bias away from the rDNA of T. dubius is occasionally reversed. Thus, bidirectional concerted evolution of rDNA, as reported for different allopolyploid species of cotton (Wendel et al. 1995a), has also occurred in T. mirus. However, this bidirectional concerted evolution is noteworthy in that it is occurring in different populations of the same species (T. mirus) rather than in different species of the same genus (as in Gossypium).

Within-population heterogeneity—implications for rDNA evolution:

Although the direction of homogenization seems consistent for all individuals of a given population, there was some plant-to-plant variability in the magnitude of repeat loss. For example, progeny derived from three different inflorescences collected from a single plant of T. mirus 2603-33 seem to display considerable heterogeneity in proportions of parental units: five plants had a low level of T. dubius repeats (5–7%), while the other three plants exhibited a percentage of T. dubius units (18–20%) more typical of other T. mirus and T. miscellus populations, suggesting that substantial variation may exist even between genetically “similar” plants. The exceptionally low content of T. dubius units in some 2603-33 plants closely resembles the monomorphic phenotype often seen in relatively ancient allopolyploids (Wendel et al. 1995a; Chase et al. 2003; Rauscher et al. 2004). Thus, it seems that individuals displaying nearly homogenized rDNA arrays may occur in low frequency within allopolyploid populations of Tragopogon, which typically comprise hundreds to >1000 individuals.

In general, populations of T. mirus displayed higher within-population heterogeneity than populations of T. miscellus (Table 4). There is also evidence that in population 2602 of T. mirus some units may have experienced structural change during the subsequent course of allopolyploid evolution: a mutation created a novel NlaIII site within the “novel” ITS-1 type (Figure 3A and Table 3; nucleotide position 223) and novel gene families were identified by Southern blot analysis of the IGS region (Figure 4D). In T. miscellus we did not observe changes in sequences of ITS and IGS in any populations examined. T. mirus contains three rDNA loci (two inherited from T. porrifolius and one from T. dubius) while T. miscellus inherited a single rDNA locus from each parent (Figure 6; Ownbey and McCollum 1954; Soltis and Soltis 1991; Pires et al. 2004). Perhaps the higher number of rDNA loci in T. mirus increases the chances of genetic interactions contributing to within- and among-population variability.

Because sequence elimination and/or replacement is likely to be an irreversible process, we can envision a scenario in which a novel allopolyploid population is gradually enriched for individuals with increasingly homogenized rDNA repeats. As a result, older populations are more likely to have a higher proportion of individuals with one predominant rDNA type. Our ongoing investigations of rDNA evolution in several allotetraploid species of Tragopogon native to Europe or Eurasia reveal that rDNA units appear to be largely homogenized. In most of the allotetraploids from the Old World, the rDNA of one of the two diploid parents could not be detected on sequencing chromatograms; only via the sequencing of numerous clones could this “rare” parental rDNA be detected (E. V. Mavrodiev, P. S. Soltis and D. E. Soltis, unpublished results). In contrast to the newly formed allopolyploids examined here, in older polyploids in Tragopogon, concerted evolution has gone to completion, or nearly so. Certainly, in the future it will be interesting to follow plants from the populations used in this study to address the question of time frame and dynamics of sequence homogenization.

Recombinant ITS clones:

Recombinant clones of the ITS1-5.8S-ITS2 sequences were detected in the allotetraploids at a frequency between 0 and 25%, depending on the population analyzed (Table 4). These “recombinant” clones could reflect either true biological recombination events or artifacts of PCR (Cronn et al. 2002). However, several lines of evidence indicate the presence of at least some recombinant genomic rDNA molecules. First, the frequency of recombinant clones was low (3 of 171 sequenced clones, i.e., 1.8%) in experiments in which DNAs from diploid plants were mixed at different ratios for subsequent PCR and cloning. This proportion of PCR-mediated recombination is similar to that reported for the 5S locus of Gossypium (2.4%; Cronn et al. 2002) and is in a range expected for small amplicons of <1.0 kb in length. Second, a lower proportion of recombinant molecules (6%) was recovered from polyploid plants collected close to the time of their origin (i.e., the earliest herbarium specimens of T. mirus and T. miscellus and T. mirus from Albion) than from present-day populations (12%). Third, a PCR-independent Southern hybridization method confirmed the presence of a relatively homogeneous crossover (i.e., recombinant) ITS family in T. mirus 2602. The restriction polymorphisms in the IGS region suggest that gene conversion might have altered other parts of the rDNA operon in this accession. There is also evidence in the literature indicating that novel and abundant rDNA variants were “recombinants,” amplified following allopolyploidization in synthetic polyploid plants (Skalická et al. 2003) and hybrid algae (Coleman 2002). Coleman (2002) proposed that ITS crossover and other variants in the rDNA cistron could be the hallmark of recent hybridization events. Recombinant ITS clones have also been isolated from both diploid and polyploid plants (e.g., Vanhouten et al. 1993; Wendel et al. 1995b; Buckler et al. 1997; Franzke and Mummenhoff 1999) and fungi (Hughes and Petersen 2001), pointing to physical interactions between rDNA loci that occupy different genomic locations (Leitch and Bennett 1997). Cloning of rDNA units from a nonamplified genomic DNA library will be needed to exclude putative PCR-mediated artifacts and to confirm the identity and origin of recombinant clones.

Evidence for nucleolar dominance:

The nearly equal FISH signals detected on parental homeologous chromosomes of the allotetraploid species (Figure 6) indicate that interlocus homogenization of rDNA sequences has occurred without material changes in rDNA unit copy number in at least some allotetraploid Tragopogon populations. Surprisingly, given the sequence changes that must be occurring at some loci, unit homogenization does not appear to have influenced condensation/decondensation patterns at rDNA loci in the allotetraploids because the condensation patterns in the polyploids appear to be the sum of those observed in the parental diploids. Moreover, T. miscellus 2605-6, which contains rDNA units highly skewed toward the T. pratensis type, showed balanced transcription of both parental (i.e., T. dubius and T. pratensis) rDNAs, and inspection of mitotic nuclei revealed secondary constrictions (a chromosomal hallmark of rRNA gene activity) at both loci. These data suggest gene conversion or elimination of the condensed, but not the decondensed, active units of T. dubius origin. However, in another allotetraploid, T. mirus 2602-6, which contains nearly equal ratios of parental diploid homeologues (i.e., T. dubius and T. porrifolius), the rDNA of T. porrifolius appeared to be preferentially transcribed.


Our data strongly suggest that rapid concerted evolution is occurring in the recently formed allotetraploids T. mirus and T. miscellus; we have apparently caught concerted evolution homogenizing rDNA units between parental genomes in the act. The data tend to favor replacement rather than elimination of units, but some uncertainty remains. Ideally, a direct method allowing us to distinguish parental gene families on the chromosomes should be developed. The loss of T. dubius rDNA units was not correlated with their transcriptional inactivation in T. miscellus, suggesting that silencing and elimination could be independent events. More detailed population-level studies are needed to determine whether the silencing at nucleolus organizer region loci is stochastic or directed toward a specific rDNA type in Tragopogon allopolyploids and whether epigenetic modification influences homogenization of units.


We thank B. Koukalova (Institute of Biophysics Academy of Science, Czech Republic) for helpful comments. This work was supported by the Grant Agency of the Czech Republic (521/04/0775, 204/05/0687), Czech Academy of Sciences (Z5004010), and National Environment Council, UK. P.S. and D.S. acknowledge funding from the U.S.-U.K. Fulbright Distinguished Professorship Program, the University of Florida Research Foundation, and National Science Foundation (NSF) grant MCB-034659. J.C.P. acknowledges support from the U.S. NSF-North Atlantic Treaty Organization Postdoctoral Fellowship (DGE-0000658) Program.


  • Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY458586, AY458588, AY458589, and AY458587

  • Communicating editor: O. Savolainen

  • Received June 25, 2004.
  • Accepted October 25, 2004.


View Abstract