De novo variation in life-history traits and responses to growth conditions of resynthesized polyploid Brassica napus (Brassicaceae) †
The authors thank J. Chris Pires, Ivan Maureira, and Maria Clauss for comments on the manuscript and Liang Sun and Pablo Quijada for assistance with statistical analyses. Funding was provided by National Science Foundation Plant Genome Program (0077774), and M. E. S. was supported by a Molecular Biosciences Training Grant, the D.C. Smith Fellowship at the University of Wisconsin and continuing support from the Max Planck Gesellshaft.
Abstract
Variation that arises in generations immediately following polyploidization may be important for the establishment, adaptation, and persistence of new polyploid species. We previously showed divergence for flowering time among lines from a resynthesized Brassica napus allopolyploid lineage derived from a cross of diploid B. rapa and B. oleracea. In this study, we more fully assess phenotypic differentiation of lines from the previously studied lineage and of lines derived from an additional resynthesized B. napus lineage. Nine polyploid lines and their diploid parents were grown under four growth conditions and measured for eight life-history traits. Polyploid lines within a lineage were expected to be genetically identical because they were derived from individual, chromosome-doubled amphihaploid plants. However, significant differences were found among lines within lineages for every phenotypic trait measured and in response to different growth conditions (genotype by environment interactions). When phenotypes of each polyploid line for each trait in each environment were compared with their diploid progenitors, approximately 30% were like one or the other parent, 50% were intermediate, and 20% were transgressive. Our results demonstrate extensive de novo variation in new polyploid lineages. Such changes could contribute to the evolutionary potential in naturally occurring polyploids.
Polyploidy has long been recognized as a prominent force shaping the evolution of flowering plants (i.e., Winge, 1917; Karpchenko, 1927; Clausen et al., 1945; Stebbins, 1950). As many as 70% of angiosperms have had polyploidy events in their history (Masterson, 1994), including many “paleopolyploids” such as maize (Gaut and Doebley, 1997) and even the small-genome plant Arabidopsis thaliana (AGI, 2000; Vision et al., 2000). Both autopolyploidy (the doubling of a single genome) and allopolyploidy (the merger of two fully differentiated genomes) are still active and important mechanisms generating new species, accounting for an estimated 2–4% of speciation events in angiosperms (Otto and Whitton, 2000).
During the first generations after formation, a new polyploid species faces several hurdles before it can be successfully established and persist. For example, polyploidization can create a severe genetic bottleneck at the time the species forms (e.g., Baumel et al., 2001), although multiple polyploidy events (Soltis and Soltis, 1993) or hybridization to diploids via “triploid-bridges” (Ramsey and Schemske, 1998) can broaden genetic diversity. In sympatric speciation, polyploids must also outcompete progenitors or create niche separation to be successful (Thompson and Lumaret, 1992). Life-history traits, such as variation in flowering time and flower size, are known to differ between diploids and polyploids and to contribute to their ecological separation (Lumaret, 1988; Segraves and Thompson, 1999). Changes in response to environmental growth conditions or phenotypic plasticity may also contribute to the differentiation of polyploids from diploids (Bretagnolle and Thompson, 2001). Much of the necessary phenotypic differentiation from progenitors may result from the immediate consequences of polyploidization, such as larger cell and organ size (reviewed in Levin, 1983; Ramsey and Schemske, 2002).
Studies comparing natural polyploids to their hypothesized diploid progenitors are difficult to interpret due to uncertain parentage, multiple polyploidy events, timing of polyploidization, and subsequent hybridizations (Soltis and Soltis, 1993; Wendel and Doyle, 1998). More unambiguous comparisons can be made using resynthesized polyploid lines (reviewed in Ramsey and Shemske, 2002). For example, Bretagnolle and Lumaret (1995) studied first generation resynthesized Dactylis glomerata and found phenotypic differences between diploid parents and polyploidy progeny. Surprisingly, differences among natural polyploid and diploid lines, such as flowering time, were not found. Thus, they hypothesized that there must be “natural selection acting on the newly created polyploids to fashion some of the ecological differences frequently observed among sympatric diploid and polyploid cytotypes in natural conditions” (p. 206, Bretagnolle and Lumaret, 1995). They did not speculate about the sources of genetic variability for natural selection to act upon. Results of recent studies suggest a potential role of non-Mendelian phenomena creating de novo variation in the early generations after polyploidization (reviewed in Wendel, 2000; Liu and Wendel, 2002; Osborn et al., 2003) that could provide the raw material for natural selection to act and shape the evolutionary trajectory of the new species.
We previously resynthesized allopolyploid Brassica napus (n = 19) by crossing B. rapa (n = 10) and B. oleracea (n = 9) (Song et al., 1993) and showed rapid de novo genomic change, including changes in methylation status and non-methylation related gain and loss of DNA fragments (Song et al., 1995). Changes in methylation patterns and genome structure may lead to gene expression changes, such as those observed in resynthesized Arabidopsis polyploids (Lee and Chen, 2001; Madlung et al., 2002). We also observed de novo variation in flowering time among our B. napus lines (Schranz and Osborn, 2000). Similar phenotypic changes have been reported in early generation, resynthesized polyploids of A. suecica (Comai et al., 2000), Nicotiana (Kostoff, 1938), and other species (reviewed in Ramsey and Schemske, 2002). Such de novo changes in polyploid phenotypes could have played a significant role in the evolution of polyploids and merit additional research.
In this study we examined de novo life history trait variation in early generation, resynthesized polyploid Brassica napus lines and their diploid parents in four different environments. The lines were derived from two lineages of polyploids, each of which were expected to contain genetically identical lines. However, some lines within lineages differed in flowering time, and we selected these to study several ecologically important traits associated with developmental timing, inflorescence architecture, and early reproductive potential of the plants grown under different environments. Our results suggest that de novo variation and changes in phenotypic plasticity can occur rapidly for a variety of life history traits. Such changes may play a critical role in the ecological success of polyploids and in their differentiation from progenitor taxa.
MATERIALS AND METHODS
Plant materials
Nine lines were selected from two independently derived lineages of resynthesized Brassica napus (Table 1). Lines from the first lineage, designated the Song lineage, was initially created to examine the effects of reciprocal crosses on polyploid formation and the the evolution of genome structure (Song et al., 1993, 1995). It was only after several generations that changes in flowering time were noted and selected for (Schranz and Osborn, 2000). The Song lineage was derived from reciprocal crosses between single plants of B. rapa cv. Flowering Pak Choi (KS1, genome designation Aaa) and B. oleracea CrGC3-3 (KS4, genome designation Ccc) followed by embryo rescue, colchicine treatment, and self-pollination (Song et al., 1993; Schranz and Osborn, 2000). The use of colchicine to induce chromosomal doubling ensures complete homozygosity of all duplicated homologous loci. Complete homozygosity in the first generation polyploid eliminates the confounding effects of genetic segregation and of inbreeding depression from self-pollination in later generations. All lines within the lineage should be genetically identical. One early-flowering and one late-flowering line from each reciprocal cross were advanced by bud self-pollination to the S9 generation. These early-flowering (ES1 and ES64) and late-flowering (ES6 and ES65) S9 lines were considered to come from a single lineage (same diploid parents) for this study.
The second lineage, designated the Schranz lineage, was subsequently created using different diploid genotypes as parents in order to confirm the phenotypic divergence found with the initial work with the Song lineage. The Schranz lineage was derived from a single chromosome doubled amphihaploid from a cross between B. rapa cv. Reward and B. oleracea TO1000. Thus, all polyploid lines within this lineage were expected to be genetically identical. Reward is an oilseed cultivar and TO1000 is an inbred (S5) rapid cycling plant derived from CrGC3-3. An individual amphihaploid plant (Aac) was colchicine treated to generate an S0 amphidiploid (Aaacc) by the same process as described in Song et al. (1993). The S0 plant was self-pollinated to obtain S1 seed. This S0 resynthesized B. napus plant was also used as a parent in a genetic mapping study and was found to contain complete genome complements derived from B. rapa and B. oleracea (Udall, 2003). Two S1 plants were self-pollinated to generate two pools of S2 seed. One-hundred S2 plants from each pool were transplanted to a field in Arlington, Wisconsin, USA in rows spaced 1 m apart and 0.3 m between plants within rows. The five earliest and five latest flowering plants for each of the two pools was selected and self-pollinated, giving rise to 20 lines of S3 seed. Five S3 seeds from each of the 20 selected lines were individually planted in 10-cm pots and grown in a greenhouse. Flowering time of the 100 S3 plants was measured. Selected plants were self-pollinated to generate S4 seed. Five lines of S4 seeds were used in this study: three early-flowering selections (ES70, ES91, and ES98) and two late-flowering lines (ES88 and ES105).
The long developmental time of B. napus meant that the two lineages (Song and Schranz) were not able to be advanced to the same generation (S9 and S4 plants, respectively). However, we make no direct comparisons between lineages and thus this difference does not influence the analyses.
Growth conditions
The nine resynthesized B. napus lines, two B. rapa, and two B. oleracea parental lines were grown in four different environments. The growth chamber environments differed in three variables (day length, light quality [red : far red, R : FR] and temperature) in four combinations: (1) LD, long days, low R : FR, no vernalization treatment; (2) SD, short days, low R : FR, no vernalization treatment; (3) R : FR, long days, high R : FR, no vernalization treatment; (4) Vern, long days, low R : FR, 3-wk vernalization treatment.
These four growth conditions have been used to characterize flowering time of ecotypes and mutants in the related crucifer and model-plant Arabidopsis thaliana. In A. thaliana, growth of plants in long days, a low R : FR light ratio, and vernalization treatment tend to promote flowering in many ecotypes, while short days, a high R : FR ratio, and no vernalization treatment tend to inhibit flowering (e.g., Lee and Amasino, 1995; Sanda et al., 1997).
All plants were grown in PGW-132 growth chambers (Percival Scientific, Boone, Iowa, USA) under equal light intensity (550 μmol · m−2 · s−1), as measured at mid-canopy height. Long day growth was 16 h of light and 8 h of dark. Short day growth was 8 h of light and 16 h of dark. Light measurements were made using a LI-COR spectro-radiometer LI-1800 (LI-COR, Lincoln, Nebraska, USA). The ratio of R : FR was calculated from the intensity of the light from wavelengths of 655–665 nm for red light and 725–735 nm for far-red light. Growth under 28 fluorescent light bulbs (215 W) and 12 incandescent light bulbs (60 W) provided the low R : FR ratio of 1.6, and growth under only fluorescent lights gave a high R : FR ratio of 5.8. Temperature was maintained at 21°C, except during the vernalization treatment. Vernalization was 3 wk of growth at 4°C beginning 1 wk after seed germination. For each treatment, we grew five individual plants from each line in separate 10-cm2 pots in a completely randomized design using Jiffy mix soil (Jiffy Products of America, Batavia, Illinois, USA). The plants were watered with 0.5× Hoagland's solution.
Data collection
Eight life-history traits related to developmental timing, inflorescence architecture, and early reproductive potential were measured (Table 2). Three traits were measured when the first flower opened and five traits were measured 1 wk later. These traits are ecologically important and are known to differ between diploid and polyploid complexes. For example, differences have been reported for flowering time (Thompson and Lumaret, 1992; Lumaret, 1988), diameter of flowers (MacDonald et al., 1988; Segraves and Thompson, 1999), and inflorescence architecture (Husband and Schemske, 2000). Measurements of inflorescence architecture included raceme height, number of flowers on the main axis, total flowers, number of inflorescence branches, and final plant height. We measured traits 1 wk after flowering (early reproductive effort) rather than at the end of the reproductive cycle because most lines could flower indefinitely if provided enough water, nutrients, and light.
Data analysis
The SAS (2000) statistical package was used to conduct all analyses. Differences among B. napus polyploid lines (or genotypes) within lineages were examined by analysis of variance (ANOVA) of the individual traits. Type III sum of squares was used to test for significance with the general linear models procedure (GLM). Environment, genotype, and genotype by environment were considered fixed effects and repetition within genotype a random effect. To examine differences in greater detail, pairwise contrasts of means between lines in each lineage were made within each environment using the MULTTEST procedure. All contrasts were corrected using the conservative Bonferroni procedure (Milliken and Johnson, 1998).
Differences between diploid progenitors and polyploids were also examined by pairwise contrasts of means for each trait measured in each environment using the MULTTEST procedure. The mean of each trait from each B. napus line was contrasted to the mean of its B. rapa parent, the mean of its B. oleracea parent, and the mean of the two diploids (intermediate diploid value). The results of each trio of contrasts was used to classify the polyploid phenotype as B. rapa-like (not significantly different from the B. rapa mean), B. oleracea-like (not significantly different from the B. oleracea mean), intermediate (significantly different from the two diploid means but between the values of the two diploids), or transgressive (significantly different from the two diploid means and above or below the value of either diploid).
Overall phenotypic similarity among genotypes was graphically assessed by principal components analysis (PCA) of trait correlations (Sokal and Rohlf, 1995). We used each trait in each environment to calculate components. So as not to bias the contributions of any one trait to the PCA analyses, all traits were standardized by dividing the means by their standard deviations (Rosenthal et al., 2002). Finally, to visualize the patterns of phenotypic plasticity we graphed the means for phenotypic traits from each environment for both the Song and Schranz lineages (Schmalhausen, 1949).
RESULTS
Variation among B. napus lines (genotype) within each lineage was highly significant for almost every trait in both lineages (Table 3). Environment was also a highly significant factor for all traits in both lineages, except for total number of flowers in the Song lineage and flowers on the main axis in the Schranz lineage. Approximately one-half of the traits had significant genotype by environment interaction (G × E) in the Song lineage (days to flower, leaf number, number of flowering axes and flowers on the main axis), whereas G × E was significant for all traits in the Schranz lineage.
All lines within a lineage were compared by pair-wise contrasts for each trait measured in each environment (Fig. 1). In the Song lineage 25.6% and in the Schranz lineage 17.8% of all pair-wise contrasts were significant (P < 0.05). Approximately one-half of the significant contrasts were between early- and late-flowering lines, one-quarter were between early-flowering lines, and one-quarter were between late-flowering lines. Many pairs of lines were significantly different for days to flower and leaf number under all growth conditions. Only a few contrasts were significant for raceme height and diameter of flowers and only under one environment (Fig. 1).
Some traits had many significant contrasts under one growth condition but not others. For example, over one-half of all contrasts were significant for total number of flowers when grown under high R : FR conditions, but few or none were significant for other growth conditions. Similar differences were observed for plant height in the Schranz lineage when comparing growth with vernalization to other growth conditions (Fig. 1). Overall, these comparisons showed that polyploid lines from the Song lineage were most different when grown under high R : FR growth conditions (37.5% of contrasts were significant) and most similar when vernalized (only 14.6% of contrasts were significant). In the Schranz lineage, polyploid lines were most different when vernalized (25% of contrasts were significant) and most similar under short days (11.3% of contrasts were significant).
Trait means of individual polyploid and diploid lines are presented graphically in Fig. 2. Magnitude and rank of trait means varied in the different environments both between species and within the polyploid lineages. For example, in the Schranz lineage, days to flowering in B. oleracea was not affected by short days, whereas flowering time in B. rapa and all B. napus genotypes was delayed. Reaction norms also varied among B. napus lines and in comparison to their diploid progenitors. For example, the number of inflorescence axes varied from three to 16 among the B. napus lines in the Schranz lineage when vernalized; one was B. rapa-like (ES105), one was intermediate to the diploids (ES70), and the other three lines were transgressive.
Phenotypes of B. napus lines were classified as B. rapa-like (R), B. oleracea-like (O), intermediate (I), or transgressive (T) based on comparing the mean of each B. napus line for each trait in each environment to that of the diploid parents (Table 4). In both lineages approximately one-half of all B. napus phenotypes were intermediate to those of the diploids. About one-third of the B. napus phenotypes were parent-like, with twice as many B. rapa-like as B. oleracea-like, and the remainder were transgressive.
Overall relationships among traits and among diploid and polyploid genotypes were determined by principal components analysis (PCA) for both the Song (Fig. 3a and b) and Schranz (Fig. 3c and d) lineages. The PCA of traits (Fig. 3a and c) illustrate the relative contribution of each trait to the separation of genotypes. The greater the contribution by traits, the closer the value is to the circles edge (Fig. 3b and d). The proximity of traits to one another (Fig. 3a and c) indicates trait correlations and generally agreed with calculations of correlation coefficients (data not shown). For many traits, measurements from the four environments were highly correlated (e.g., leaf number in the Song lineage, Fig. 3a). However, for other traits, the measurements varied in different environments (e.g., days to flower in the Schranz lineage, Fig. 3c). Some of the different traits were highly correlated with each other in all environments, such as leaf number with raceme height in the Song lineage (Fig. 3a).
The PCA of genotypes (Fig. 3b and d) illustrates the divergence of polyploids from each other and from their diploid parents. Polyploids in the Song lineage had phenotypes that were more intermediate to the two diploid parents compared to polyploids and parents in the Schranz lineage. In both lineages, the late-flowering polyploids clustered together and were more divergent from the intermediate parental phenotype than were the early-flowering polyploids. The early-flowering polyploids also tended to cluster together, although ES70 was more divergent from the other early-flowering polyploids in the Schranz lineage (Fig. 3d).
DISCUSSION
Our analysis of two resynthesized B. napus lineages revealed extensive variation for eight life-history traits and their developmental interactions with the environment. These differences are remarkable given that the lineages were young, they were expected to be genetically homozygous, and they had undergone limited selection for just one trait. Such changes occurring in early generations after polyploidization could play a major role in the establishment, adaptation, persistence, and ultimate success of polyploid species. Next, we discuss three major findings from analyses of these new polyploids.
De novo variation of life history traits of polyploid B. napus
One major finding was that significant de novo variation exists within both polyploid lineages for all eight life-history traits, not just the one for which we had selected differences among lines (Fig. 1 and Table 3). Differences in multiple traits may be due to genetic correlations. It is known that the timing of the transition from vegetative to reproductive growth affects many life-history characteristics (e.g., Mitchell-Olds, 1996; Ungerer et al., 2002), and we found that many of our traits were correlated (Fig. 3a and 3c). This could be due to pleiotropic effects of regulatory loci or to linkage disequilibrium of the loci causing de novo variation. However, not all traits were correlated, and many significant contrasts were detected between early-flowering lines or between late-flowering lines within lineages. Thus, our selection for early- and late-flowering lines could not account for all the de novo variation observed, suggesting that genetic changes other than those involved in the selected flowering time differences may have occurred in these lines and contributed to the phenotypic differentiation among lines.
Variation in responses to environmental growth conditions
Our second major finding was that polyploid lines varied in their response to environmental growth conditions. The differences in G × E provide information on pathways and genes that may have been modified in the two lineages. For example, in the Song population, there were few significant contrasts among lines when plants were vernalized (Fig. 1), which suggests possible modification of genes that are involved in the vernalization response pathway (Finnegan, 2002). Another example of differences in G × E can be seen in the Schranz population in which one line (ES105) was greatly delayed in flowering under short-day conditions. This could be due to modifications of the short-day flowering pathway, including alterations of the gibberellin (GA) signaling pathway (Levy and Dean, 1998). Gibberellin content is known to be highly correlated with plant height in B. napus (Rood et al., 1990, Zanewich et al., 1990); thus it was not surprising that ES105 is the shortest plant when grown under short day conditions.
In addition to finding G × E differences among polyploids, differences between progenitor species were also observed. Overall, the rapid-cycling lines of B. oleracea were phenotypically invariant for many traits across environments, whereas both B. rapa and many B. napus lines were more plastic (Fig. 3). The environmental modification of genotypic expression, or phenotypic plasticity, has been widely cited as an important element in the adaptive repertoire of plants (Bradshaw, 1965; Schlichting, 1986; Schlichting and Pigliucci, 1998). Polyploids have been hypothesized to have less phenotypic plasticity than progenitors due to greater genotypic buffering from increased gene dosage and heterozygosity (Bretagnolle and Thompson, 2001). We found no clear trend of reduced plasticity for allopolyploids relative to diploids (Fig. 3). These results agree with other recent studies that have failed to identify a clear link between polyploidy and reduced plasticity (Garbutt and Bazzaz, 1983; MacDonald et al., 1988; Bretagnolle and Lumaret, 1995; Bretagnolle and Thompson, 2001; Murren et al., 2002).
Comparisons of polyploids to diploid progenitors
Our third major finding was that polyploids had a range of phenotypes compared to diploid progenitors (Table 4). Traditionally, chromosome doubling itself has been thought to cause transgressive segregation (phenotypes that are extreme or novel relative to parental lines) (Levin, 1983), whereas hybridization in allopolyploids leads to intermediate phenotypes (e.g., Clausen et al., 1945). However, a recent study of hybrids concluded that transgressive segregation was very prevalent (Rieseberg et al., 1999). We compared polyploid and diploid phenotypes for each trait in each environment and found that about half of all B. napus phenotypes were intermediate to those of the diploids (Table 4). The large number of intermediate phenotypes could be due to additive effects between homologous loci. About 15% of phenotypes were transgressive (Table 4), including days to flower under many growth conditions. Transgressive characteristics could be due to new intergenomic epistatic interactions and/or to the direct or epistatic effects of de novo allelic variation. Many of the remaining polyploid traits were similar to the B. rapa parent, with fewer traits being similar to the B. oleracea parent (Table 4). This bias towards B. rapa phenotype was unexpected. However, it is noteworthy that the B. rapa ribosomal genes show nucleolar dominance over B. oleracea loci in B. napus polyploids (Chen and Pikaard, 1997). Perhaps the mechanisms favoring the expression of B. rapa ribosomal genes in B. napus similarly favor gene expression of some subset of regulatory loci.
Comparisons of our contrasts between diploids and polyploids generally agreed with results of other studies. For example, polyploids often flower later than their diploid parents (Thompson and Lumaret, 1992). We found that most of our polyploid lines had significantly later flowering times than their respective diploid parents (Fig. 3). However, the early-flowering polyploid genotypes ES6 and ES64 in the Song lineage flowered as quickly as the diploids (Fig. 3). Later flowering of polyploids is generally attributed to slower growth rates of polyploids than diploids. The intermediacy of leaf number but transgressiveness of days to flower (Fig. 3) support the hypothesis that slower growth rates likely contributed to differences in flowering times. Levin (1983) concluded that it was difficult to predict whether the number of flowers will be fewer or greater in polyploids than diploids. Likewise, we found differences in inflorescence architecture measurements between taxa; however, there were no clear patterns.
Overall, there were differences in the life histories of the B. rapa and B. oleracea parents in both lineages (Figs. 2 and 3). Brassica oleracea had fewer leaves, larger flowers, greater raceme heights, and produced fewer total flowers on fewer inflorescence axes than B. rapa parents of both lineages. Although not measured in this study, B. oleracea also produces larger but fewer seeds. The overall phenotypes of the polyploid lines were a mixture of these life history strategies (Fig. 2). However, some polyploid lines have become more similar to one parent than the other. For example, the late flowering lines (ES6 and ES65) in the Song lineage have adopted many more B. rapa-like characteristics. Also, line ES70 in the Schranz lineage has become more like B. oleracea than B. rapa. The differences in overall life-history strategies, both between species and within B. napus, suggest that these phenotypes are not canalized or rigidly fixed by developmental constraints. Rather, there can be shifts in “the evolutionary gestalt of polyploid lineages” (Levin, 1983).
Possible sources and mechanisms for de novo variation
De novo variation could be due to only a few changes of large effect in particular genes or genomic regions, or there could be a myriad of smaller, perhaps random, changes throughout the genome. Recent research supports the importance of few genes of large effect. For example, flowering time in Brassica species appears to be controlled by homologs of the A. thaliana regulatory genes FLC and CO (Axelsson et al., 2001; Schranz et al., 2002). Quantitative trait loci (QTL) of life-history traits and agronomic yield are pleiotropic with genomic regions containing these genes in A. thaliana (Ungerer et al., 2002) and in resynthesized and natural B. napus (Udall, 2003). Using a population derived from the same S1 resynthesized B. napus plant as the Schranz lineage, Udall (2003) found clusters of QTL associated with agronomic performance, including flowering time, plant height, and seed yield. In selecting for flowering time, we may have been selecting for de novo variants in particular regulatory genes; this could be the cause of many of the phenotypic changes observed.
However, not all of our traits were correlated and some traits had different responses to growth conditions. These results suggest that genetic changes may have occurred at multiple loci and that their effects could be fixed in lines due to the action of genetic drift, rather than our selection for flowering time differences.
There are several mechanisms that could generate de novo variation in new polyploids (Osborn et al., 2003). The use of cholchicine for genome doubling could be a potential source of variation by causing anueploidy. However, cytological studies (Song et al., 1993) and genomic mapping (Udall, 2003) found no evidence for aneuploidy in our newly resynthesized polyploids. Translocations could be an important mechanism creating de novo variation. Both reciprocal and nonreciprocal translocations have been observed in natural B. napus polyploids and in progeny from crosses of natural and resynthesized B. napus (Parkin et al., 1995; Sharpe et al., 1995; Udall, 2003). Translocations involving genomic regions containing important genes could arise in resynthesized lineages and generate de novo variation directly. Alternatively, epigenetic changes, such as alterations in chromatin structure and methylation patterns, could cause changes in gene expression (Osborn et al., 2003). These changes could occur at specific loci, although results from other studies of new polyploids indicate that epigenetic changes occur throughout the genome (Song et al., 1995; Shaked et al., 2001; Madlung et al., 2002). Epigenetic changes are a plausible cause of de novo variation considering the many emerging connections between chromatin structure, flowering time, and other phenotypic changes (Sung et al., 2003). Future comparisons of genome structure, methylation patterns, and gene expression between these early- and late-flowering lines could elucidate the mechanisms underlying de novo phenotypic variation in new polyploids.
Summary of pairwise contrasts of resynthesized Brassica napus polyploid lines for eight traits measured in four environments (LD = long days, SD = short days, R : FR = high red : far red light, Vern = vernalization). Contrasts were done between each polyploid line and all other lines within the Song (a) and Schranz (b) lineages for days to flower (dtf), leaf number (ln), raceme height (raceme), diameter of flowers (diam flwr), height to top (ht to top), number of inflorescence axes (inflor), number of flowers on the main axis (flwrs main), and number of total flowers (flwrs total) measured under four growth conditions. Results graphed by the percentage of significant pairwise contrasts (P < 0.05) for each trait in each environment
Plots of phenotypic trait means for eight life-history traits of individual polyploid (B. napus) and diploid (B. rapa and B. oleracea) lines grown in four environments (LD = long day, SD = short day, R : FR = high red : far red light ratio, Vern = vernalization). The variation in the magnitude and rank of trait means in the different environments reflects genotype by environment interactions (G × E) and phenotypic plasticities between species and between polyploid lines
Principal components analysis (PCA) of trait data for genotypes within the Song (a and b) and Schranz (c and d) lineages. All trait measurements were standardized by dividing the means by their standard deviations. The first two principle components were calculated for each trait in each environment and plotted to illustrate the relationships among traits (a and c). The first two principal components were calculated for each diploid and polyploid genotype and plotted to illustrate overall phenotypic relationships among diploid and polyploids within each lineage (b and d). The distance of each trait or genotype from the origin along the x- and y-axes is proportional to the contribution to the first and second components, respectively. The percentage of total variation explained by each component is shown. Lines that had been selected for early- and late-flowering phenotypes are indicated by E and L, respectively, after the line designations (b and d)
