Polyploidy, frequently termed “whole genome duplication”, is a major force in the evolution of many eukaryotes. Indeed, most angiosperm species have undergone at least one round of polyploidy in their evolutionary history. Despite enormous progress in our understanding of many aspects of polyploidy, we essentially have no information about the role of chromosome divergence in the establishment of young polyploid populations. Here we investigate synthetic lines and natural populations of two recently and recurrently formed allotetraploids Tragopogon mirus and T. miscellus (formed within the past 80 years) to assess the role of aberrant meiosis in generating chromosomal/genomic diversity. That diversity is likely important in the formation, establishment and survival of polyploid populations and species.
Applications of fluorescence in situ hybridisation (FISH) to natural populations of T. mirus and T. miscellus suggest that chromosomal rearrangements and other chromosomal changes are common in both allotetraploids. We detected extensive chromosomal polymorphism between individuals and populations, including (i) plants monosomic and trisomic for particular chromosomes (perhaps indicating compensatory trisomy), (ii) intergenomic translocations and (iii) variable sizes and expression patterns of individual ribosomal DNA (rDNA) loci. We even observed karyotypic variation among sibling plants. Significantly, translocations, chromosome loss, and meiotic irregularities, including quadrivalent formation, were observed in synthetic (S0 and S1 generations) polyploid lines. Our results not only provide a mechanism for chromosomal variation in natural populations, but also indicate that chromosomal changes occur rapidly following polyploidisation.
These data shed new light on previous analyses of genome and transcriptome structures in de novo and establishing polyploid species. Crucially our results highlight the necessity of studying karyotypes in young (<150 years old) polyploid species and synthetic polyploids that resemble natural species. The data also provide insight into the mechanisms that perturb inheritance patterns of genetic markers in synthetic polyploids and populations of young natural polyploid species.
Citation: Lim KY, Soltis DE, Soltis PS, Tate J, Matyasek R, Srubarova H, et al. (2008) Rapid Chromosome Evolution in Recently Formed Polyploids in Tragopogon (Asteraceae). PLoS ONE 3(10): e3353. doi:10.1371/journal.pone.0003353
Editor: Samuel P. Hazen, University of Massachusetts Amherst, United States of America
Received: July 3, 2008; Accepted: September 8, 2008; Published: October 9, 2008
Copyright: © 2008 Lim et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors thank NERC, Grant Agency of the Czech Republic (521/07/0116), and the USA National Science Foundation (grants MCB-0346437, DEB-0608268, DBI 0638536 and DBI 0501712) for support. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Polyploidy has played a major role in generating angiosperm biodiversity. Chromosome counts suggest that between 30 and 80% of angiosperm species are polyploid, while genomic studies of selected model and crop species reveal evidence of extensive ancient genome-wide multiplications. Indeed recent genomic investigations indicate that most, if not all, angiosperm species have undergone at least one genome duplication event in their evolutionary history, and several have evidence of multiple polyploidy-diploidisation-polyploidy cycles , , .
Angiosperm genomes are astonishingly plastic in their ability to tolerate considerable karyotypic (e.g. chromosome number variation, translocations), genetic (mutations, retroelement transpositon, deletions) and epigenetic (DNA methylation, histone methylation/acetylation) variability. This tolerance enables polyploids to form and establish and has contributed significantly to their widespread occurrence . Large-scale genetic changes induced by polyploidy are thought to influence the transcriptome, metabolome and proteome, which can concomitantly alter the phenotype and ecology of the individuals. Most of the new genetic changes are probably maladaptive, but in a few rare instances individuals arise that are able to outcompete the parental diploids or colonize new niches.
There are several examples of recent speciation via polyploidy that occurred within the last 150 years: Spartina anglica , Senecio cambrensis and S. eboracensis , Cardamine schulzii , Tragopogon mirus and T. miscellus . Studies on the genetic consequences of allopolyploidy in these de novo polyploid species reveal some significant differences. In S. anglica allopolyploidy induced few changes in genome structure, but there is epigenetic reprogramming , , while in Senecio ,  and Tragopogon  allopolyploids there are substantial genetic changes including loss of sequence (genomic DNA profiles) and perturbations to the transcriptome (cDNA profiles). However, little attention has been paid to the chromosomal content of any of these allopolyploids, and it is unknown what mechanisms drive the observed genetic changes. For example, genetic change could be driven by local mutation, small-scale deletion or insertion of a particular sequence, or via major chromosomal changes, including whole-arm transposition and chromosome losses or duplications. Because there is little or no understanding of chromosomal variation in these recently formed, natural polyploids, we have embarked on a characterization of karyotypic variation between individuals and populations of the two Tragopogon allopolyploids from North America.
Tragopogon mirus and T. miscellus have proved to be excellent evolutionary model systems for understanding early allopolyploid formation. These allopolyploids are derived from three diploid progenitors (each 2n = 2x = 12), T. pratensis, T. porrifolius and T. dubius, the latter being shared by both allopolyploids (T. mirus derived from T. dubius×T. porrifolius, 2n = 4x = 24; T. miscellus derived from T. dubius×T. pratensis, 2n = 4x = 24, Figure 1) . The diploid parents of both polyploids are in well separated clades  and are not closely related based on ITS/ETS sequences, allozymes and other genetic markers , . There are many reports of natural F1 hybrids involving these diploids and we have also produced them in the glasshouse, but these hybrids are highly sterile suggesting minimal homeologue pairing at meiosis . In contrast the allopolyploids T. mirus and T. miscellus are fertile and expanded their ranges rapidly after their initial formations, in large part via multiple origins . The two allotetraploids now occupy a large geographic area of eastern Washington and adjacent Idaho, USA, and comprise many thousands of individuals in several populations. Molecular analyses have revealed that T. mirus has recurrently formed at least 13 times and T. miscellus possibly as many as 21 times [13,16,18,19,20, Symonds et al. unpubl.,21], reviewed in Soltis et al. . Furthermore, there are genetic differences between populations of each tetraploid, most of which likely reflect variation found in the diploids and inherited in the polyploid populations through recurrent formation, and others of which may have arisen through subsequent divergence of the polyploids. Examples of the latter include the divergence of rRNA gene copy number, sequence homogeneity and expression patterns , . Here we show that population differences are also reflected in substantial variability in karyotypes among individuals, differences that appear correlated with irregular meiosis.
Genome structure of Tragopogon allotetraploids
The chromosome sizes, numbers and centromere indices are similar between the parental diploid species, and three tandem repeats characterized by us do not generate differences in chromosomal distribution. For these reasons we are unable to identify the parental origin of the chromosomes in the derived allotetraploid species via morphology alone . We therefore used FISH with total genomic DNA probes (called Genomic In Situ Hybridization or GISH) to determine the genomic composition of karyotypes of T. mirus and T. miscellus individuals (Figures 2 and 3). We analysed 12 plants, nine of T. mirus from three populations (Table 1) and three of T. miscellus, two from one population and a third individual from a second population (Table 2). Five of these plants, including representatives of each tetraploid, were chosen because we had previously observed ,  that they had particular 45S nuclear ribosomal DNA (45S rDNA) compositions and expression characteristics (Table 3). GISH labelling enabled the genomic origin of the chromosomes to be determined by fluorescence colour: digoxingenin-labelled genomic DNA of T. dubius labelled chromosomes of T. dubius origin green or yellow (D-genome) and biotin-labelled genomic DNA of either T. pratensis (to T. miscellus) or T. porrifolius (to T. mirus) labelled the P-genome (either T. pratensis or T. porrifolius) orange or red. However, the distinction between genomes was only possible after electronic merger of the images because there was considerable cross-hybridisation of probes. The homologous group assigned to each chromosome (A–F) was determined using size, arm ratio, position of 5S and 45S rDNA, as described in Pires et al. , and by using DAPI bright bands that were revealed after denaturation in some metaphases.
Homeologous chromosome group nomenclature (A-F) follows Ownbey and McCollum  and Pires et al. . Fluorochrome colours: yellow/green (FITC, digoxigenin-labelled probes), orange/red (Cy3, biotin-labelled probes), blue (DAPI staining). Each karyotype is shown with DAPI staining, sometimes also simultaneously labelled for 45S rDNA (yellow fluorescence) or 5S rDNA (red fluorescence), and after GISH with T. dubius (green fluorescence) and T. porrifolius (red fluorescence) total genomic DNA probes. Monosomic or trisomic chromosomes are underlined. Note: (1) intergenomic translocations to chromosome Cpo/du (arrows) in individuals 2601-7 and 2601-8 and Ddu/po in individual 2602-0-3-4 (A). (2) The 45S rDNA locus on the two homologues of chromosome Adu in individuals 2602-4 and 2602-0-3-4 show different levels of decondensation, one being condensed and the other with a secondary constriction (C). (3) A large reduction in size of the 45S rDNA locus on chromosome Adu of individual 2603-33A (#) compared with other individuals (D). (4) No secondary constrictions of 45S rDNA loci on T. porrifolius origin chromosomes (Apo, Dpo) in individual 2603-33B. (D-E) Root-tip interphase nuclei of T. mirus after FISH with digoxigenin-labelled pTa71 for 45S rDNA. (D) Individual 2601-4, showing four large condensed 45S rDNA loci (arrows) and (E) Individual 2602-0-3-10 with two large condensed 45S rDNA loci (arrows). Scale bar (top right) is 10 µm.
Homeologous chromosome group nomenclature (A–F) follows Ownbey and McCollum  and Pires et al. . Fluorochrome colours and probes are as in Figure 2 except that biotinylated T. pratensis genomic DNA was used in GISH experiments. Monosomic and trisomic chromosomes are underlined. Note in (A) that there is a large intergenomic translocation, chromosome Adu/pr (arrow). Scale bar (top right) is 10 µm.
Previous cytogenetic analyses revealed that both T. mirus and T. miscellus had 24 chromosomes; aneuploidy has not been previously reported in these taxa , , . This chromosome number was also observed here in all but one individual of T. mirus, which had 23 chromosomes. It was assumed that GISH would partition the chromosome sets into 12 chromosomes of T. dubius origin and 12 chromosomes of either T. pratensis or T. porrifolius origin depending on the tetraploid being analysed. However, only four had an entirely balanced additive karyotype of that expected from the chromosomes of the diploid parents (Tables 1 and 2 and Figures 2 and 3). We were surprised to observe that five of the eight T. mirus karyotypes and two of the three T. miscellus karyoypes with 2n = 24 chromosomes had unbalanced genomic contributions. This was manifest by some chromosomes occurring in one or three copies (monosomic and trisomic, respectively; see Tables 1 and 2). In addition, four plants, including individuals of both T. mirus and T. miscellus, carried intergenomic translocations (Figures 2A, 3A, arrows).
The T. mirus individuals 2603-33A and 33B are siblings from different flower heads of the same plant. Significantly, their karyotypes exhibit substantial differences; 2603-33B has an expected karyotype assuming complete additivity of parental chromosome sets, while 2603-33A is monosomic for chromosome Fdu and trisomic for chromosome Epo.
Meiotic and mitotic aberrations in synthetic reconstruction of T. mirus
Meiosis in newly formed, synthetic allotetraploids (S0 and S1) (see Figure 1) was examined to determine if meiotic aberrations occurred in early allotetraploid generations and could be a source of genomic imbalance observed in root tip metaphases. Meiotic cells of developing anthers at diplotene were analysed (Figure 4). In many instances regular bivalent formation occurred (Figure 4A), although frequently bivalents overlapped (Figure 4C, D), making it difficult to be certain if they were in fact multivalents. However, in a number of cases resolution was sufficient to determine the presence of quadrivalents (Figure 4B). Using GISH, the genomic origin of the chromosomes could be resolved, although the distinction was less clear than at metaphase. Figures 4E–F show GISH labelling of a quadrivalent with two chromosomes of T. dubius origin and two chromosomes of T. porrifolius origin physically linked via chiasma in a large, twisted ring.
(A,B) Feulgen staining showing in (A) 12 regular bivalents and (B) one quadrivalent (arrow) and bivalents in close association with each other and a nucleolus. (C–H). Fluorochrome colours and GISH probes are as in Figure 2. (C–D) Note chromosome Adu in association with the nucleolus. Two bivalents, one from the T. dubius genome (yellow) and one from the T. porrifolius genome (orange), overlapped (arrow), perhaps occurring as a multivalent. (E–F). Note the rDNA and the bivalents carrying these genes associated with the nucleolus. A quadrivalent with two chromosomes of T. dubius origin and two of T. porrifolius origin occurring in a ring (arrow). (G–H) Note two bivalents (Adu and Dpo) with rDNA associated with the nucleolus and the Apo bivalent with condensed rDNA unassociated with the nucleolus. N = nucleolus. Scale bar (C–H) is 10 µm.
An analysis of root tip metaphases in 12 allotetraploids of the subsequent S1 generation revealed one plant that was 2n = 23, the rest were 2n = 24, as expected. There were cytogenetic abnormalities in two additional plants. One of these plants (73-14) had unusually large 45S rDNA loci occurring on both copies of chromosome Adu (Figure 5 A, B). This plant also had an rDNA locus on chromosome Apo of T. porrifolius origin, but the expected site on chromosome Dpo was absent. The second plant (134-16-3) had a translocation of T. porrifolius origin to chromosome Cdu. Since there is no missing chromosome segment of T. porrifolius origin, it must be assumed that this was a non-reciprocal translocation induced during the preceding meiosis (Figure 5 C, D, E).
(A, C) Metaphase (merged images) after FISH for 45S rDNA (red) and 5S rDNA (green) and counterstained with DAPI (blue). (B) and (D) Metaphases (merged images) from (A) and (C), respectively, after reprobing using GISH with labeled genomic DNA from T. porrifolius (pink) and T. dubius (green) and counterstained with DAPI. In (B) note the particularly there are only four 45S rDNA sites, the sites on Adu. Chromosomes are particularly large (arrows) and the site expected on chromosome Dpo is missing. In (D) note the T. porrifolius translocation to a T. dubius chromosome (red arrow) and the satellites to both Adu homologues are distant from the rest of the chromosome (see connecting white lines). (E) Karyotype of (D) revealing that the T. porrifolius origin translocation is to chromosome Cdu and that no T. porrifolius chromosomes lack chromatin. Scale bar is 10 µm.
45S rDNA decondensation
Previous cytogenetic analyses of diploid Tragopogon species revealed that T. dubius and T. pratensis carry a 45S rDNA locus on chromosome Adu and Apr, respectively, while T. porrifolius has two loci on chromosomes Apo and Dpo. To determine if these sites were inherited in the respective allotetraploid species we used FISH with digoxigenin-labelled probe pTa71 (yellow fluorescence, Figures 2 and 3). In T. miscellus, 45S rDNA loci occur on chromosomes Adu and Apr (Figure 3). At metaphase both chromosomes carried secondary constrictions (Table 3). In T. mirus, 45S rDNA loci occur on chromosomes Adu, Apo, and Dpo (Figure 2). These loci were also frequently identifiable after GISH labeling, without the use of the rDNA probe pTa71, appearing with a brighter green (to T. dubius origin loci) or yellow fluorescence (Figure 2). In all cells, both mitotic and meiotic, the number of 45S loci and their parental distribution were as expected, i.e. pairing and segregation of rDNA-carrying chromosomes is probably regular. This was not the case, however, for 5S rDNA-carrying chromosomes (5S probe, red fluorescence). We would expect T. mirus to carry 5S rDNA loci on chromosomes Adu, Apo and Fpo. But here we observed one T. mirus individual (2602-03-10) trisomic for chromosome Fpo and each of these chromosomes carried a single 5S rDNA locus (Figure 2). It was a surprise that the chromosomes carrying the 45S rDNA were balanced because at diplotene they were frequently in close proximity (Figure 4 E, F), probably reflecting transcriptional activity and nucleolar function. The absence of aberrant numbers of 45S rDNA-carrying chromosomes at metaphase suggests that they paired regularly at meiosis.
An analysis of 45S rDNA-carrying chromosomes in 14 diplotene nuclei of T. mirus revealed that chromosome Adu was always associated with the nucleolus (Figure 4E–H). Often this chromosome was associated with a chromosome from the T. porrifolius genome (when identifiable it was chromosome Dpo, Figure 4G, H). Similar results were observed at metaphase (Table 3, Figure 2), where chromosome Adu and sometimes chromosome Dpo carried secondary constrictions (the exception is plant 2603-33A–below). In contrast, chromosome Apo was always condensed and unassociated with the nucleolus. Interestingly, in two individuals of population 2602 only one of the two homologues of chromosome Adu carried secondary constrictions. In plant 2603-33A, there was no secondary constriction on chromosome Adu and the 45S rDNA locus was unusually small. Instead, chromosome Dpo had a secondary constriction. Similarly in two synthetic T. mirus, plants 134-16-3 and 73-14, there were secondary constrictions on Adu and Apo, with the locus on chromosome Dpo missing in the latter individual (Figure 5 A, C).
In natural T. mirus and T. miscellus, the presence of secondary constrictions correlated well with patterns of condensation/decondensation at interphase (Figure 2D, E); for example, chromosomes Apo and Dpo have no secondary constrictions in T. mirus 2601-4 (not shown) and exhibit four sites of condensed rDNA chromatin (Figure 2D) while in T. mirus (2602-0-3-10) chromosome Apo is without a secondary constriction (Figure 2C) and at interphase there are two highly condensed rDNA loci (Figure 2E).
Southern blot analysis of 5S rDNA loci
Given the imbalance observed in the inheritance of individual chromosomes, we would expect the copy number of 5S rDNA units to vary among plants. For example, individual 2602-3-10, showing trisomy of the Fpo chromosome, had seven 5S rDNA signals instead of the expected six (Figure 2C). On the other hand, if there was Fpo monosomy, we might expect a reduction in gene copy numbers. Using Southern blot hybridisation, we determined the parental T. porrifolius/T. dubius 5S rRNA gene copy number ratio in the progeny of a single T. mirus plant, 2602-0-3. Several progeny were studied, including those analysed by GISH (plants 2602-0-3-10 and 2602-0-3-4). To ascertain the parental origin of the 5S rDNA units, we digested genomic DNA with TaqI restriction enzyme. This enzyme takes advantage of a restriction site present in T. porrifolius units that is not found in T. dubius units. Consequently, after digestion the T. porrifolius 5S rDNA was digested mostly to monomers while there was little digestion of T. dubius units that remained at high molecular weight (Figure 6). In T. mirus, the 5S rDNA probe used for Southern hybridisation revealed signals from both parental gene families. However, the distribution of signal intensities between genes of T. porrifolius and genes of T. dubius origin differed between progeny. For example, the trisomic individual 2602-0-3-10 had a T. porrifolius/T. dubius 5S rDNA unit ratio close to 1.0, while the “normal” disomic plant 2602-0-3-4 had a ratio that was 0.82. Thus, trisomy of chromosome Fpo resulted in an approximately 10% increase in the number of 5S genes in the genome. This value is in a good theoretical agreement with the expected increase in number of 5S arrays from six to seven (per diploid cell). Three other plants (2602-0-3-5, -7 and -8) had a 5S rDNA gene ratio corresponding to a trisomic genotype, suggesting meiotic aberrations were frequent in this lineage. On the other hand, two plants (2602-0-3-2 and -3) had a decreased number of T. porrifolius genes compared with the expectation (plant 2602-0-3-4, with balanced chromosomal sets). Perhaps these two individuals are monosomic for Fpo.
(A) Southern blot hybridisation of 5S rDNA probe to TaqI-digested genomic DNA of the progeny of 2602-0-3 individual and diploid parental accessions. (B) Quantitative representation of the parental 5S gene families determined by a phosphorimager scanning of radioactivity signals on blots (three independent experiments).
Effectiveness of GISH
Phylogenetic analyses of internal (ITS) and external transcribed spacer (ETS) sequences of nuclear 45S rDNA suggest that the diploid progenitors of both T. mirus and T. miscellus are distantly related, with T. dubius in one major clade of Tragopogon and both T. pratensis and the populations of T. porrifolius that served as parents in another major clade , . Earlier allozyme studies also suggested that the three diploid progenitors were well differentiated genetically 16, 28. Likewise, comparisons of cDNA-AFLP genetic markers between the diploid species reveal that T. pratensis and T. dubius share only between 30-40% of markers , again suggesting considerable genomic divergence.
Recently, Markova et al.  used GISH with genomic DNA probes from diploid species of Silene onto metaphases of related diploids to show that labeling strength was inversely correlated to genetic distance, i.e. there was strongest labeling to the most closely related species. Given these data and an apparently large genetic distance between Tragopogon diploids, we might expect GISH to work effectively on the derived allopolyploids. Instead, we observed that it worked rather weakly, with much cross-hybridisation of probes, and the genomic distinction was only resolvable after electronic merging of images. Indeed previously we had reported that GISH had not worked at all in Tragopogon . Our success here is due to improved quality of the electronics on the microscope's camera and improved facilities to electronically merge images using Openlab software, but it remains an enigma why GISH does not work more effectively in T. mirus and T. miscellus.
rDNA inheritance and expression patterns
We expect T. miscellus to inherit two 45S rDNA loci (carried on chromosome Adu from T. dubius and Apr from T. pratensis), while T. mirus inherits three loci (carried on chromosomes Adu from T. dubius and Apo and Dpo from T. porrifolius). The observations of only two 45S rDNA in synthetic T. mirus 73-14 (Figure 5A, B) can be explained in one of two ways. First, the enlargement of the rDNA locus on chromosome Adu and the locus loss on Dpo occurred through an rDNA translocation/fusion involving the loci on Dpo and Adu. However, given that the karyotype is balanced and the material is the S1 generation, this scenario seems unlikely because it would suggest the union of two gametes carrying the same abnormality, or arising via restitution of segregating chromosomes in meiosis II. The second and alternative explanation is that the rDNA locus number may reflect the situation in the T. porrifolius parent used for the cross, although our past analyses of T. porrifolius never revealed such a polymophism . Unfortunately the precise T. porrifolius parent used in the cross is now deceased (the plants are annuals or biennials), preventing us from determining which of the competing hypotheses is correct.
Previously we showed that natural Tragopogon allopolyploids also do not have fixed patterns of 45S rDNA inheritance, with some individuals showing a balanced distribution of rDNA sequences originating from each of the parental diploids and some showing biased inheritance, typically with the number of rDNA units from T. dubius being underrepresented . However, from molecular studies it was not clear whether non-Mendelian rDNA inheritance was caused by a decrease in copy number (elimination) or unit replacement (e.g. via homogenisation mechanisms). In one individual of T. mirus, 2603-33 (referred to here as 2603-33A), there was a considerable reduction in the number of T. dubius units (to ~100 copies/diploid) from an expectation of ~700 units (typical number for any given tetraploid population). An analysis of this individual's karyotype (Figure 2) revealed that the loss in copy number can be accounted for by a reduction in the size of the rDNA locus on chromosome Adu. The sibling to this plant, 2603-33B, does not show this rDNA copy number deletion. Therefore, the locus size reduction in 2603-33A may have arisen from a deletion event occurring as a consequence of meiotic instability in the parent.
Despite the relatively small size (perhaps only 100 copies/diploid of the 45S rDNA unit) of the T. dubius 45S rDNA locus, it usually dominates rRNA expression in leaf material of T. mirus . Even in the 2603-33A individual, with extremely reduced numbers of 45S rRNA genes of T. dubius origin, the locus accounts for 97% of the rRNA transcripts (see also Table 3). Nevertheless, in root-tip metaphases a secondary constriction is observed on chromosome Dpo, suggesting some transcription of the T. porrifolius locus occurs in that tissue, perhaps to compensate for the reduced 45S rDNA gene copy number when there is a high demand for ribosomes in metabolically active meristematic cells. In plant 2603-33B, with substantially higher numbers of T dubius derived 45S rRNA genes, the units of T. dubius origin are transcribed, and those rDNA units of T. porrifolius origin are silent . This is reflected in the lack of a secondary constriction at the rDNA locus on chromosomes Apo and Dpo in this individual (Figure 2, Table 3). The presence of secondary constrictions correlates strongly with levels of decondensation at interphase and almost certainly reflects transcriptional activity at the preceding interphase. Two individuals of T. mirus (2602-4 and 2602-0-3-4) show different decondensation of the two Adu homologues, probably reflecting genetic or epigenetic differentiation between the two homologues. In population 2605 of T. miscellus, all individuals investigated show only partial dominance in the expression of rDNA units of T. dubius origin over those of T. pratensis origin . This is also seen in the occurrence of secondary constrictions on both rDNA-carrying chromosomes Adu and Apr (Figure 3, Table 3).
An analysis of 45S rDNA evolution in Nicotiana polyploids indicates that parental loci are initially maintained in young polyploids, although the sequences within a locus may be subject to concerted evolution, and over time frames of >1 million years individual loci are lost . In Tragopogon polyploids we do not know which of the different karyotype variants will survive selection and become fixed. Perhaps over longer evolutionary timescales interlocus homogenisation and new rDNA variants will occur and spread across all rDNA loci.
Our study shows that more than one pathway can lead to non-Mendelian inheritance of rDNA units in allotetraploids: (i) elimination or amplification of repeats within an array can occur without changes in locus number (as in the case of 45S rDNA locus in 2601-33A); (ii) a change in the number of rDNA-bearing chromosomes (and loci) without any material change in the number of genes at each locus (as with the 5S rDNA locus–Figure 6); or (iii) a combination of both, although this situation was not detected in this study.
The chromosome multiplication step of polyploidy is thought to establish species isolation barriers between the newly formed polyploid and its diploid parents, whilst providing a homologous partner for each of the chromosomes. However, analyses of newly synthesised allopolyploids reveal that early-generation individuals are often infertile, or have highly reduced fertility, due to problems with meiosis including irregular pairing of homologous chromosomes , . Selection for fecundity in synthetic polyploids is associated with generation-by-generation increased fertility . Nevertheless even after many thousands of years of evolution, meiotic irregularities can still occur, as observed in Triticum aestivum (wheat), an allohexaploid where meiotic misdivision has been exploited in the formation of wheat aneuploid lines , .
An analysis of meiosis in the newly synthesised Tragopogon allotetraploids revealed the frequent occurrence of multivalents (Figure 4). Such aberrant pairing patterns may result in imbalanced chromosome contribution in subsequent generations as well as intergenomic translocations. Both abnormalities were observed in several of 12 synthetic (S1 generation) T. mirus plants analysed at metaphase. One plant was 2n = 23 and lacked a T. dubius chromosome and one plant carried a non-reciprocal translocation from a T. porrifolius chromosome to chromosome Cdu (Figure 5E)
Multivalents were also observed at a low frequency in natural populations of T. mirus and T. miscellus  and in the few F2 plants resulting from diploid F1 hybrids of Tragopogon . Meiotic abnormalities can cause unequal segregation of homeologous chromosomes and are a likely driver of the chromosomal imbalance between genomes observed in both Tragopogon allotetraploids. The two sibling plants of T. mirus 2603-33 had different karyoypes: one had a balanced karyotype of 12 chromosomes from each parental genome (2603-33B), while the other (2603-33A) was monosomic for chromosome Fdu and trisomic for chromosome Epo. Given these data, it is likely that the parent plant had a balanced karyotype, and meiotic irregularities, probably arising through multivalent formation, gave rise to the imbalanced karyotype of plant 2603-33A.
Multivalent pairing can arise either through (1) synapsis and recombination between homeologous chromosomes in meiosis I, or (2) synapsis between chromosomes carrying intergenomic translocations. Four out of 12 mitotic karyotypes in plants from natural populations had intergenomic translocations visible at mitotic metaphase following GISH (Tables 2 and 3), and additional smaller translocations, not resolvable by GISH, may also be present. Nevertheless, the quadrivalent indicated in Figure 4F (see arrow) does not show intergenomic translocations at the resolution obtained using GISH. Our analyses of plants from different populations, although sample sizes were small, showed that aberrant chromosome numbers were more prevalent for certain chromosome types. For example, chromosome A was not involved in any of the aneuploidy evetns detected, whereas chromosome F accounted for almost 40% of all subgenomic chromosome imbalances. There might be several explanations for these results. First the homology between F-type chromosomes may be larger than between other chromosomes. It is likely that the greater sequence and morphological similarities between homeologous chromosomes, the more likely there will be homeologous and multivalent pairing. In support of this, Nicolas et al.  observed in Brassica napus haploid hybrids that chromosomes with the highest synteny had the highest frequency of homeologous pairing. Nevertheless, in T. mirus, the presence of a 5S rDNA locus on Fpo but not on Fdu indicates some divergence between the homeologues. Secondly, all chromosomes may be equally likely to form multivalents, but aberration in copy number of some chromosomes, e.g. homeologous group A, may not be favored by selection.
The surprisingly high incidence of trisomy and monosomy in highly fertile plants with 2n = 24 suggests that “compensating trisomy” may be operating in natural populations of both Tragopogon allotetraploids. Compensating trisomy was first described by Blakeslee  to refer to a situation in which the loss of a normal chromosome is compensated by the presence of the two arms in new translocated associations (secondary chromosomes). That concept was extended to include the replacement of the primary chromosome by two tertiary chromosomes, or by a secondary and a tertiary chromosome . Compensatory trisomy has been reported in Datura stramonium from progeny of a plant exposed to radium , , and from crops, including from Poales ,  and tomato ; all were produced experimentally . The putative compensatory trisomy observed here may be an example from natural plants.
The allotetraploids T. mirus and T. miscellus formed from their diploid progenitors within the last 80 years (perhaps even within the last 60 years) and given their biennial habit, the number of generations to present is likely to be less than 40 . The long-term outcome of meiotic irregularities in these species is not easily predicted. On the one hand, the genomic imbalance will reduce fitness and perpetuate cycles of meiotic irregularities. This may lead to a cascade of reduced fitness, generation upon generation. Such a phenomenon was observed in synthetic Brassica allopolyploids maintained by single seed descent . In the wild, cryptic karyotypic instability manifest by aberrant ratios of homeologous chromosomes might ultimately lead to a slow reduction in fitness and ultimately extinction. Perhaps this accounts for the loss of some local populations of both T. mirus and T. miscellus  and of some recently formed Senecio polyploid populations . Nonetheless, it is noteworthy that despite meiotic irregularities in the initial synthetic S0 plants and subsequent S1 generations, pollen fertility and seed set were generally high in the synthetic Tragopogon polyploid lines (Tate et al., in prep.). In addition, selection will favour the most fertile individuals, likely to be those with the most regular chromosome pairing. Thus, if the population can expand through early bottlenecks of reduced fertility, the derived populations are likely to be more fertile with regular bivalent pairing. Certainly in well-established allotetraploids (104–105 myrs old), e.g. of Nicotiana and Triticum, no major imbalances in chromosome numbers or the distribution of chromosomes to subgenomes are normally observed , .
The angiosperm genome is characterised by its plasticity to genetic change, including large-scale chromosome number changes, aneuploidy and polyploidy . However, it may be significant that in Tragopogon polyploids, all the imbalances in parental chromosome dosages between individuals occurred within a near-regular karyotype of 2n = 24 (one exception at 2n = 23). Given the frequency of plants showing genomic imbalance we might expect to find more plants with unexpected chromosome numbers. Perhaps there is selection against plants that deviate from 2n = 24 and against those that are nullisomic for a particular chromosome, or perhaps our sample size was too small to find a representative range of abnormalities.
GISH data revealed that in two out of three individuals of T. miscellus and four of nine individuals of T. mirus, there is an imbalance in the parental contribution of the chromosomes. However, the imbalance observed resulted from monosomy or trisomy, and no individual was nullisomic for a particular chromosome. An analysis of 10 genes in T. miscellus using genomic and cDNA CAPS revealed that 65% of individuals displayed losses of one of the two homeologues and a further 5% of individuals showed silencing of one of the two homeologues in leaves . The missing alleles were interpreted as sequences that had been stochastically eliminated from the T. miscellus genome. The results here may suggest that chromosome loss or non-reciprocal translocations may contribute to the loss of alleles, although we did not find any example of a homozygous deletion for a chromosome segment. The differential expression of rDNA on the Adu chromosomes of two T. mirus individuals (Figure 2) points to epigenetic or genetic heterozygosity between the homologous chromosomes.
Large-scale genetic changes caused by parental genome imbalance will influence the inheritance of genetic markers, which will in turn influence the transcriptome, proteome and metabolome. Clearly genetic analyses of young or synthetic polyploids require in-depth cytogenetic studies to assess the contribution that chromosomal changes play in the inheritance of genetic markers. Unfortunately in recent years that work has seldom been standard practice. Such cytogenetic data are clearly needed even if chromosome counts appear regular [see also 43], since a deeper analysis of the genome and chromosome substructure can reveal substantial chromosome dosage deviation from expectation.
Materials and Methods
Seeds of Tragopogon were collected from natural populations in Idaho (ID) and Washington (WA) (USA) (Table 1) and planted either in a greenhouse at the Department of Botany, University of Florida, or in field plots at the Institute of Biophysics, Academy of Sciences of the Czech Republic, Brno. Root tips from young, healthy, vigorously growing plants were harvested and placed in ice cold and saturated aqueous 2mM 8-hydroxyquinoline (Sigma-Aldrich Company Ltd, Poole, Dorset, UK). After 60 min incubation on ice, the roots were fixed in ethanol-acetic acid (3:1) at room temperature overnight (several washes) and stored in the same solution at −20°C until use. Developing anthers of 1 mm length or less, that contain diplotene nuclei, were excised from young buds of length 1 cm or less. Root tip and meiotic material were fixed in freshly prepared 3:1 ethanol: glacial acetic acid for two days. Root tip material was then transferred to 90% ethanol at −20°C for long-term storage.
Making synthetic Tragopogon allotetraploids
Descriptions of the methods used to generate synthetic polyploid lines are given in detail in a separate paper reporting the formation and availability of these lines for research (Tate et al. in prep). Briefly, numerous repeated T. dubius×T. porrifolius crosses were made, and seeds from successful crosses germinated on moist filter paper. The chromosome number was doubled to resynthesise polyploids that closely resemble T. mirus by placing seedlings with fully emerged cotyledons in 0.1% or 0.25% colchicine solution overnight. After washing with water for two-three days, the seedlings were transferred to 2.5” pots with soil (and grown in the glasshouse at the University of Florida under standard conditions). Plants coded N197-3.132-1 and N197-4.98-1 were used for meiotic analyses (S0 generation) and root tips of selfed progeny (S1 generation) analysed in root tip mitosis, metaphases of plants coded 134-16-3 and 73-14 are shown.
Feulgen staining of meiotic cells
For meiotic squashes, small developing inflorescences (1 cm or less in length) were collected from the greenhouse and fixed in 3:1 (as above, for root tips). Individual anthers were then removed and macerated in 60% glacial acetic acid, stained in aceto-orcein , spread under a coverslip by warming over a naked flame, and observed using a Zeiss Photomicroscope III. At least 15 meiotic cells per plant were scored.
Preparing cell spreads for in situ hybridisation
Chromosome preparations for FISH, using either cloned probes or total genomic DNA probes (called GISH) were made using modifications of established methods , . Briefly, root tips or developing anthers were digested in 0.3% (w/v) cellulase Onozuka R-10 (Apollo Scientific Ltd, Stockport, Cheshire, UK), 0.3% (w/v) pectolyase Y23 (MP Biomedicals, Solon, Ohio, USA) and 0.3% (w/v) drieselase (Sigma-Aldrich Company Ltd., Poole, Dorset, UK) for 28 min and transferred to 1% citrate buffer for 2 h. The meristematic cells behind the root cap were isolated in a drop of 60% acetic acid and squashed onto a glass slide. For meiotic preparation, fixed anthers were dissected and meiotic cells gently dispersed into a drop of 60% acetic acid, a coverslip applied and gently warmed. Coverslips were removed following freezing with dry ice.
In situ hybridisation
Fluorescence in situ hybridisation followed standard protocols, Figures 2–4, Leitch et al. , Figure 5, Telgmann-Rauber et al. . Genomic DNA from T. dubius, T. pratensis, and T. porrifolius for labelling by GISH was extracted using DNeasy Plant mini kit (Qiagen Ltd, Crawley, West Sussex, UK) following manufacturer's instructions. Genomic DNA was labelled with biotin-16 dUTP (Sigma-Aldrich Company Ltd., Poole, Dorset, UK) or digoxigenin-11-dUTP (GE Healthcare, Chalfont St Giles, Buckinghamshire, UK) using nick translation following standard protocols. The probe against 5S ribosomal DNA (rDNA) was prepared by amplifying the gene using primers described in Fulnecek et al.  and biotin-16-dUTP-labelling protocol as described in Leitch et al. . The probe against 45S rDNA was the clone pTa71, which includes the 18- 26S rDNA subunit isolated from Triticum aestivum, which was labelled with digoxignenin-11-dUTP as described in Leitch et al. . Briefly, slides were denatured in 70% (v/v) formamide in 2× SSC (0.3 M sodium chloride, 0.03 M sodium citrate) at 70°C for 2 min and the hybridisation mix added (4 µg.ml−1 labeled probes and 50% (v/v) formamide, 10% (w/v) dextran sulphate, 0.1% (w/v) sodium dodecyl sulphate in 2× SSC). In situ hybridisation was carried out overnight at 37°C, after which the slides were given a stringent wash (20% (v/v) formamide in 0.1× SSC at 42°C). Sites of probe hybridisation were detected using 20 µg.ml−1 fluorescein-conjugated anti-digoxigenin IgG (GE Healthcare, Chalfont St Giles, Buckinghamshire, UK) and 5 µg.ml−1 Cy3-conjugated avidin (Roche Pharmaceuticals, Lewes, East Sussex, UK) in 4× SSC containing 0.2% (v/v) Tween 20 and 5% (w/v) bovine serum albumin. Chromosomes were counterstained with 2 µg/ml DAPI (4′,6-diamidino-2-phenylindole, Sigma Aldrich Company Ltd., in 4× SSC) and stabilised in Vectashield medium (Vector Laboratories Ltd, Peterborough, UK) prior to data acquisition using either: 1) Leica DMRA2 epifluorescent microscope fitted with an Orca ER camera and Open Lab software® (Improvision, Coventry, UK) (Figures 2–4) or; 2) Olympus BX61 epiflurescent microscope using Olympus Microsuite 5 software® (Olympus America Inc, Center Valley, PA, USA) (Figure 5). The images were analysed with Adobe Photoshop® version 7 and treated for colour contrast and uniform brightness only. At least 5 mitotic or meiotic cells per plant were scored with each probe used.
TaqI restriction enzyme digestion was carried out on genomic DNA extracted from leaves using standard protocols  with modifications as in . After fractionation in 1% agarose by gel electrophoresis, the DNA was transferred to a Hybond N+ membrane (GE Healthcare, Chalfont St Giles, Buckinghamshire, UK). The 32P-labelled 5S rDNA probe was from a 120-bp XbaI/EcoRI fragment of 5S rDNA cloned from N. tabacum . Probe hybridisation was conducted under high-stringency conditions in the Church-Gilbert hybridisation buffer at 65 °C overnight. The radioactivity signals were quantified by phosphorimager scanning (Storm, GE Healthcare, UK).
Conceived and designed the experiments: DS PS AK AL. Performed the experiments: KYL DS JT RM HS ZX AL. Analyzed the data: KYL PS RM HS AK CP ZX AL. Contributed reagents/materials/analysis tools: DS PS AK CP AL. Wrote the paper: DS PS JT AK CP AL.
- 1. Blanc G, Wolfe KH (2004) 2004. Widespread paleopolyploidy in model plant species inferred from age distributions of duplicate genes. Plant Cell 16: 1667–1678.
- 2. Bowers JE, Chapman BA, Rong R, Paterson AH (2003) Unravelling angiosperm genome evolution by phylogenetic analysis of chromosomal duplication events. Nature 422: 433–438.
- 3. Cui LY, Wall PK, Leebens-Mack JH, Lindsay BG, Soltis DE, et al. (2006) Widespread genome duplications throughout the history of flowering plants. Genome Research 16: 738–749.
- 4. Leitch AR, Leitch IJ (2008) Genomic plasticity and the diversity of polyploid plants. Science 320: 481–483.
- 5. Ainouche ML, Baumel A, Salmon A (2004) Spartina anglica C. E. Hubbard: a natural model system for analysing early evolutionary changes that affect allopolyploid genomes. Biological Journal of the Linnean Society 82: 475–484.
- 6. Abbott RJ, Lowe AJ (2004) Origins, establishment and evolution of two new polyploid species: Senecio cambrensis and S. eboracensis in the British Isles. Biological Journal of the Linnean Society 82: 467–474.
- 7. Urbanska KM, Hurka H, Landolt E, Neuffer B, Mummenhoff K (1997) Hybridization and evolution in Cardamine (Brassicaceae) at Urnerboden, central Switzerland: Biosystematic and molecular evidence. Plant Systematics and Evolution 204: 233–256.
- 8. Soltis DE, Soltis PS, Pires JC, Kovarik A, Tate JA, et al. (2004) Recent and recurrent polyploidy in Tragopogon (Asteraceae): cytogenetic, genomic, genomic and genetic comparisons. Biological Journal of the Linnean Society 82: 485–501.
- 9. Salmon A, Ainouche ML, Wendel JF (2005) Genetic and epigenetic consequences of recent hybridization and polyploidy in Spartina (Poaceae). Molecular Ecology 14: 1163–1175.
- 10. Abbott RJ, Ireland HE, Rogers HJ (2007) Population decline despite high genetic diversity in the new allopolyploid species Senecio cambrensis (Asteraceae). Molecular Ecology 16: 1023–1033.
- 11. Hegarty MJ, Barker GL, Wilson ID, Abbott RJ, Edwards KJ, et al. (2006) Transcriptome shock after interspecific hybridization in Senecio is ameliorated by genome duplication. Current Biology 16: 1652–1659.
- 12. Tate JA, Ni ZF, Scheen AC, Koh J, Gilbert CA, et al. (2006) Evolution and expression of homeologous loci in Tragopogon miscellus (Asteraceae), a recent and reciprocally formed allopolyploid. Genetics 173: 1599–1611.
- 13. Ownbey M (1950) Natural hybridisation and amphidiploidy in the genus Tragopogon. American Journal of Botany 37: 487–499.
- 14. Buggs RJA, Soltis PS, Mavrodiev EV, Symonds VV, Soltis DE (2008) Does phylogenetic distance between parental genomes govern the success of polyploids? Castanea 73: 74–93.
- 15. Soltis DE, Mavrodiev EV, Doyle JJ, Rauscher J, Soltis PS (2008) ITS and ETS sequence data and phylogenetic reconstruction in allopolyploids and hybrids. Systematic Botany 33: 7–20.
- 16. Soltis PS, Plunkett GM, Novak SJ, Soltis DE (1995) Genetic variation in Tragopogon species - additional origins of the allotetraploids T. mirus and T. miscellus (Compositae). American Journal of Botany 82: 1329–1341.
- 17. Novak SJ, Soltis DE, Soltis PS (1991) Ownbey's Tragopogons: Forty years later. American Journal of Botany 78: 1586–1600.
- 18. Cook LM, Soltis PS, Brunsfeld SJ, Soltis DE (1998) Multiple independent formations of Tragopogon tetraploids (Asteraceae): evidence from RAPD markers. Molecular Ecology 7: 1293–1302.
- 19. Soltis DE, Soltis PS (1989) Allopolyploid speciation in Tragopogon - Insights from chloroplast DNA. American Journal of Botany 76: 1119–1124.
- 20. Soltis PS, Soltis DE (1991) Multiple origins of the allotetraploid Tragopogon mirus (Compositae) - rDNA evidence. Systematic Botany 16: 407–413.
- 21. Soltis DE, Soltis PS (1995) The dynamic nature of polyploid genomes. Proceedings of the National Academy of Sciences of the United States of America 92: 8089–8091.
- 22. Kovarik A, Pires JC, Leitch AR, Lim K, Sherwood A, et al. (2005) Rapid concerted evolution of nuclear ribosomal DNA in two allopolyploids of recent and recurrent origin. Genetics 169: 931–944.
- 23. Matyasek R, Tate JA, Lim KY, Srubarova H, Leitch AR, et al. (2007) Concerted evolution in recently formed Tragopogon allotetraploids is typically associated with an inverse correlation between gene copy number and expression. Molecular Biology and Evolution 176: 2509–2519.
- 24. Pires CJ, Lim KY, Kovarik A, Matyasek R, Boyd A, et al. (2004) Genome size and distribution of tandem repetitive DNA in allopolyploid Tragopogon (Asteraceae). American Journal of Botany 91: 1022–1035.
- 25. Ownbey M, McCollum GD (1954) The chromosomes of Tragopogon. Rhodora 56: 7–21.
- 26. Mavrodiev EV (2007) Polyphyly of Tragopogon porrifolius L. (asteraceae), a European native with intercontinental disjuncts. International Journal of Plant Sciences 168: 889–904.
- 27. Mavrodiev EV, Tancig M, Sherwood AM, Gitzendanner MA, Rocca J, et al. (2005) Phylogeny of Tragopogon L. (Asteraceae) based on internal and external transcribed spacer sequence data. International Journal of Plant Sciences 166: 117–133.
- 28. Roose ML, Gottlieb LD (1976) Genetic and biochemical consequences of polyploidy in Tragopogon. Evolution 30: 818–830.
- 29. Markova M, Michu E, Vyskot B, Janousek B, Zluvova J (2007) An interspecific hybrid as a tool to study phylogenetic relationships in plants using the GISH technique. Chromosome Research 15: 1051–1059.
- 30. Kovarik A, Dadejova M, Lim K, Chase M, JJ C, et al. (2008) Evolution of rDNA in Nicotiana allopolyploids: A potential link between rDNA homogenization and epigentics. Annals of Botany 101: 815–823.
- 31. Goodspeed TH (1954) The genus Nicotiana. Massachusetts, USA: Chronica Botanica Company.
- 32. Burk LG (1973) Partial self-fertility in a theoretical amphiploid progenitor of N. tabacum. Journal of Heredity 64: 348–350.
- 33. Riley R (1960) The diploidisation of polyploid wheat. Heredity 15: 407–429.
- 34. Sears E (1954) The aneuploids of common wheat. Research Bulletin of Missouri Agricultural Experimental Station 572: 1–58.
- 35. Ownbey M, McCollum GD (1953) Cytoplasmic inheritance and reciprocal amphiploidy in Tragopogon. American Journal of Botany 40: 788–796.
- 36. Nicolas SD, Le Mignon G, Eber F, Coriton O, Monod H, et al. (2007) Homeologous recombination plays a major role in chromosome rearrangements that occur during meiosis of Brassica napus haploids. Genetics 175: 487–503.
- 37. Blakeslee AF (1927) Genetics in Datura. Berlin: Vererbungswiss. pp. 117–130.
- 38. Avery AG, Satina S, Reitsema J (1959) The Genus Datura. New York: Ronald Press.
- 39. Sears E (1953) Nullisomic analysis in common wheat. The American Naturalist 87: 245–252.
- 40. Saini RS, Minocha JL (1981) A compensating trisomic in pearl millet. Journal of Heredity 72: 354–355.
- 41. Kush GS, Rick CM (1967) Novel compensating trisomics of the tomato: Cytogenetics, monosomic analysis, and other applications. Genetics 56: 297–307.
- 42. Singh RJ (2002) Plant Cytogenetics. New York: CRC Press.
- 43. Gaeta RT, Pires JC, Iniguez-Luy F, Leon E, Osborn TC (2007) Genomic changes in resynthesized Brassica napus and their effect on gene expression and phenotype. Plant Cell 19: 3403–3417.
- 44. Lim KY, Matyasek R, Kovarik A, Leitch AR (2004) Genome evolution in allotetraploid Nicotiana. Biological Journal of the Linnean Society 82: 599–606.
- 45. Liu Z, Li DY, Zhang XY (2007) Genetic relationships among five basic genomes St, E, A, B and D in triticeae revealed by genomic southern and in situ hybridization. Journal of Integrative Plant Biology 49: 1080–1086.
- 46. Jackson RC (1973) Chromosomal evolution in Haplopappus gracilis: A centric transposition race. Evolution 27: 243–256.
- 47. Telgmann-Rauber A, Jamsari A, Kinney MS, Pires JC, Jung C (2007) Genetic and physical maps around the sex-determining M-locus of the dioecious plant asparagus. Molecular Genetics and Genomics 278: 221–234.
- 48. Leitch AR, Lim KY, Webb DR, McFadden GI (2001) In situ hybridisation. In: Hawes C, Satiat-Jeunemaitre B, editors. Plant Cell Biology, a practical approach. Oxford: Oxford University Press. pp. 267–293.
- 49. Fulnecek J, Lim KY, Leitch AR, Kovarik A, Matyasek R (2002) Evolution and structure of 5S rDNA loci in allotetraploid Nicotiana tabacum and its putative parental species. Heredity 88: 19–25.
- 50. Saghai-Maroof MA, Soliman KM, Jorgensen RA, Allard RW (1984) Ribosomal DNA spacer-length polymorphisms in barley - Mendelian inheritance, chromosomal location, and population-dynamics. Proceedings of the National Academy of Sciences, USA 81: 8014–8018.
- 51. Kovarik A, Fajkus J, Koukalova B, Bezdek M (1996) Species-specific evolution of telomeric and rDNA repeats in the tobacco composite genome. Theoretical and Applied Genetics 92: 1108–1111.