• Hybridization and polyploidy can induce rapid genomic changes, including the gain or loss of DNA, but the magnitude and timing of such changes are not well understood. The homoploid hybrid system in Helianthus (three hybrid-derived species and their two parents) provides an opportunity to examine the link between hybridization and genome size changes in a replicated fashion.
• Flow cytometry was used to estimate the nuclear DNA content in multiple populations of three homoploid hybrid Helianthus species (Helianthus anomalus, Helianthus deserticola, and Helianthus paradoxus), the parental species (Helianthus annuus and Helianthus petiolaris), synthetic hybrids, and natural hybrid-zone populations.
• Results confirm that hybrid-derived species have 50% more nuclear DNA than the parental species. Despite multiple origins, hybrid species were largely consistent in their DNA content across populations, although H. deserticola showed significant interpopulation differences. First- and sixth-generation synthetic hybrids and hybrid-zone plants did not show an increase from parental DNA content. First-generation hybrids differed in DNA content according to the maternal parent.
• In summary, hybridization by itself does not lead to increased nuclear DNA content in Helianthus, and the evolutionary forces responsible for the repeated increases in DNA content seen in the hybrid-derived species remain mysterious.
Nuclear DNA content in angiosperms (2C-value) ranges from 0.431 pg for Arabidopsis (Schmuths et al., 2004) to 254.8 pg for Fritillaria assyriaca (Bennett & Leitch 2003). This range in C-value is not entirely dependent on changes in ploidy, nor is it correlated with organismal complexity. For example, within diploid species in Poaceae, genome sizes range from 1.11 pg in Oryza sativum to 20.90 pg in Secale cereale (Bennett & Leitch 2003), with no apparent differences in complexity between these taxa. This variation in nuclear DNA content and its lack of correlation with gene number has been termed the ‘C-value paradox’ (Thomas, 1971) or ‘C-value enigma’ (Gregory, 2001).
Adaptive explanations have been sought for differences in genome size. Differences in genome size are correlated with slope aspect in Ceratonia siliqua (Bures et al., 2004), with frost resistance in the British flora (MacGillivray & Grime, 1995), with calyx size in Silene (MeaGher & Costich, 1996) and with elevation in Patagonian species of Berberis (Bottini et al., 2000) and European species of Dactylis (Reeves et al., 1998). Studies on invasive pines have found that smaller genomes are correlated with smaller seed size and higher invasiveness (Rejmanek & Richardson, 1996; Grotkopp et al., 2004). Alternatively, nonadaptive explanations have posited that increases in genome size are the result of deleterious mutations which fix via drift in small populations (Lynch & Conery, 2003). The observation that rare and endangered species, which usually have reduced population sizes, have larger genomes than more common species within their families (Vinogradov, 2003) is in accord with this proposal.
In the past decade, research has shifted to understanding the mechanisms behind genome size changes in plants. Polyploidy has been seen as a major source of increasing genome size. However, studies of natural polyploids within Asteraceae have shown that, while total DNA content increases on average with increasing ploidy level, the DNA content of each genome (i.e. the G1 nuclear DNA content divided by the ploidy level) in the polyploid nucleus decreases (Leitch & Bennett, 2004). Differences in nuclear DNA content within diploid plants have been linked to differences in intron size (Petrov, 2001) and transposon copy number (Bennetzen, 2002). Decreases in genome size correlate with a mutational bias towards deletions over insertions (Petrov, 2001), and illegitimate recombination has been shown to eliminate retrotransposon sequences (Bennetzen, 2002; Devos et al., 2002; Ma et al., 2004).
Newly synthesized plant polyploids may undergo extensive genomic changes. Hybridization followed by chromosome doubling leads to loss of DNA sequences in Brassica and Aegilops (Song et al., 1995; Liu et al., 1998a,b; Ozkan et al., 2001). In addition, some polyploid lineages exhibit increased retrotransposon activity (Ozkan et al., 2001). However, polyploidization does not automatically entail genomic restructuring: studies of resynthesized cotton polyploids (Gossypium; Liu et al., 2001) and recently derived natural polyploids (Tragopogon mirus and Spartina anglica) failed to find genomic changes (Baumel et al., 2002; Soltis et al., 2004).
Several studies have explored the timing of genomic change in polyploid lineages, and it now appears that some genomic changes are initiated in first-generation diploid hybrids, whereas others are exclusive to polyploidization (e.g. Song et al., 1995; Ozkan et al., 2001; Osborn et al., 2003). In Aegilops, for example, elimination of sequences unique to one of the parental genomes but found on multiple chromosomes begins in F1 plants and is completed in just two or three generations after polyploidization (Ozkan et al., 2001; Shaked et al., 2001). In contrast, sequences unique to a single chromosome from one parental genome are maintained in diploid hybrids, but are rapidly lost following polyploidization (Ozkan et al., 2001). Similar patterns have been reported from studies of diploid hybrids. Some plant and animal hybrids show genomic changes upon hybridization, including fruit flies, wallabies, and beans (Rogers & Bendich, 1987; Petrov et al., 1995; O’Neill et al., 1998; Labrador et al., 1999), whereas others are entirely stable (Guerreiro, 1996).
Despite substantial research on the genomic consequences of hybridization and polyploid speciation, genome size changes in diploid or homoploid hybrid species remain to be explored. Sims & Price (1985) reported nuclear DNA contents for 19 diploid sunflower (Helianthus) species, and it was later shown that three of these (Helianthus anomalus, Helianthus deserticola, and Helianthus paradoxus) are diploid hybrid derivatives of the same parents, Helianthus annuus and Helianthus petiolaris (Rieseberg, 1991). Intriguingly, the three homoploid hybrid species were reported to have substantially more DNA than their parents. However, only three individuals were analyzed in each species. Also, discovery of the influence of plant secondary compounds on the estimation of DNA content by both Fuelgen densitometry and flow cytometry (Greilhuber, 1988; Price et al., 2000) has led to changes in practice and created uncertainty regarding many earlier reported C-values.
Hybrid sunflowers offer several advantages for the study of genome-size evolution in homoploid hybrid species and its potential adaptive consequences. First, two of the three species (H. anomalus and H. deserticola) appear to have arisen multiple times in nature (Schwarzbach & Rieseberg, 2002; Gross et al., 2003), so we can ascertain whether the same genomic changes have occurred independently in the wild. Secondly, numerous hybrid zones between the parental species exist naturally. These zones can be exploited to ask whether the genome size variation found in the ancient hybrid species occurs in natural hybrid zones. Finally, genetic mapping studies indicate that the genomes of the hybrid species were extensively restructured during the speciation process (Rieseberg et al., 1995, 1996, 2003). Glasshouse experiments have shown that the chromosomal changes that separate the hybrid species from the parents can be largely duplicated after just four generations of fertility selection (Rieseberg et al., 1996; Rieseberg, 2000). Thus, we can use this unique germplasm to ask whether genome size variation can be replicated in the glasshouse and whether the ‘genomic shock’ caused by hybridization can generate DNA content variation.
In this study, we examined multiple individuals from multiple populations of the three homoploid hybrid sunflower species and the two parents to confirm previous reports of a DNA-content shift and to assess variation among independently derived populations. We also examined individuals from natural hybrid zones, synthetic F1 individuals, and synthetic F6 individuals in order to understand the timing of changes in genome size in the speciation process.
Materials and Methods
Helianthus paradoxus Heiser, H. anomalus Blake, and H. deserticola Heiser are among a handful of well-documented homoploid hybrids (Rieseberg, 1997). All three species result from hybridization between H. annuus L. and H. petiolaris Nutt. ssp. fallax Heiser (Rieseberg, 1991). Estimates of the age of origin based on microsatellite diversity of the three species range from 60 000 to 200 000 years ago (Schwarzbach & Rieseberg, 2002; Welch & Rieseberg, 2002; Gross et al., 2003). The three species occur in the south-western USA, in Arizona, New Mexico, Utah, and Nevada. H. paradoxus has a single origin, based upon all populations sharing a single chloroplast haplotype with H. annuus and high interpopulation crossability (Welch & Rieseberg, 2002). H. anomalus probably had multiple origins, as different populations have different chloroplast haplotypes that are shared with either H. annuus or H. petiolaris, and H. anomalus populations hypothesized to have separate origins have decreased fertility when crossed (Schwarzbach & Rieseberg, 2002). H. deserticola is ambiguous in its origins: patterns of microsatellite and chloroplast DNA variation can be accounted for by multiple origins or by a single origin followed by introgression with the parental species. However, partial crossing barriers imply that this taxon also arose multiple times (Gross et al., 2003).
The two parental species, H. annuus and H. petiolaris, frequently hybridize in the wild, although hybrid fertility is very low (< 1%) (Rieseberg, 2000). However, F1 and back-cross plants can be found in many locations (Rieseberg et al., 1998). Despite equal chromosome counts (2N = 34 for all five species), the three hybrid-derived species have strong crossing barriers with their parental species: F1 fertility ranges from < 1% for crosses between H. anomalus and H. annuus to c. 25% for crosses between H. deserticola and H. annuus (Rieseberg, 2000; L. H. Rieseberg, unpublished). These barriers are probably caused by several chromosomal rearrangements in each hybrid species. Fourth-generation synthetic hybrid lineages, selected for fertility, had high fertility when crossed with H. anomalus (Rieseberg, 2000), but strong barriers when crossed to the parental species, reflecting the chromosomal evolution that took place during the four generations of selection.
We selected three to five populations from each of the hybrid-derived species. Populations selected spanned the range of the species. In addition, we chose four to five populations of the parental species from the general region of interest (Table 1). Two hybrid zones were sampled, one from Colorado and one from Kansas. In addition, F1 seeds from eight H. annuus ×H. petiolaris crosses were analyzed, as well as individuals from three sixth-generation synthetic hybrid lineages; these are the same lineages analyzed for genomic composition and crossability by Rieseberg et al. (1996) and Rieseberg (2000), but extended by two generations of selfing.
Table 1. Helianthus taxa used, source information, and nuclear DNA content and standard deviation (SD) as measured by flow cytometry
Identifier and source
2C (pg) (SD)
USDA AMES 26095/LD ANO2
Little Sahara, UT
20 km N of Hanksville, UT
15 km S of Hanksville, UT
USDA PI 413024
USDA Ames 14400
USDA PI 468477
USDA PI 468607
USDA Ames 26094
Little Sahara, UT
Santa Rosa, NM
Bitterlake NWR, NM
Ft. Stockton, TX
H. petiolaris spp. fallax
USDA PI 468815
Puerto de la Luna, NM
H. petiolaris ssp. petiolaris
USDA PI 586912
USDA PI 586920
USDA PI 435809
USDA PI 435807
Dunes National Monument, CO
Kent Road, KS
H. annuus × H. petiolaris F1
Laboratory of Lisa Donovan
H. annuus × H. petiolaris F1
Laboratory of Lisa Donovan/Loren Rieseberg
H. annuus × H. petiolaris F6
Laboratory of Loren Rieseberg
We used flow cytometry to estimate nuclear DNA content, following recommendations for the use of internal standards chopped with each sample (Bennett et al., 2000; Price et al., 2000). To establish appropriate conditions for flow cytometry, juvenile leaves from glasshouse plants and/or germinating seeds were used. Tests with all species revealed that DNA content estimates were consistent between the two tissue types. Hordeum vulgare cv. ‘Sultan’ (barley) was used as an internal standard (2C = 11.12 pg) (Johnston et al., 1999). Samples and the internal standard were chopped together in a buffer containing 50 mm Tris, 1 mm MgCl2, 0.1% Triton X-100, 0.1% polyvinylpyrrolidone (PVP: average molecular weight 40 000), and 0.5 mm spermine at pH 7.2, then filtered through two layers of Miracloth (CalBiochem, Pasadena, CA, USA). The filtrate was then stained with 50 ppm propidium iodide and 1 mg ml−1 RNase. Samples were kept on ice and run on a Becton-Dickinson FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA). At least 5000 nuclei were counted for each sample.
Initial trials revealed that the G1 nuclear DNA content of the hybrid species was very close to the barley standard, making DNA quantification difficult. For these species, French green lentils (Lens culinaris ssp. culinaris) (2C = 9.0 pg; calibrated using H. vulgare cv. ‘Sultan’) were used as an alternate internal standard, with samples of the parental species run with both standards to verify consistency.
Sampling design and analysis
We quantified nuclear DNA content from at least four individuals per population. We tested for differences between populations within each species using analysis of variance (anova) (sas proc glm; SAS Institute, Cary, NC, USA). Differences between species were tested using nested ANOVA, with population within each species treated as a random effect (sas proc mixed; SAS Institute). Pre-planned contrasts tested for differences in means for synthetic F1 lineages with H. annuus or H. petiolaris as the maternal parent, and for the synthetic hybrids compared to the three homoploid hybrid species. All other posthoc pairwise comparisons of means were corrected for multiple tests.
Parents and hybrid species
Homoploid hybrid species showed a consistent pattern of increased nuclear DNA content (Fig. 1). After adjustment for multiple tests, the three hybrid species were each significantly different from their parents, H. annuus (7.23 pg; P < 0.0001) and H. petiolaris (6.68 pg; P < 0.0001). The three homoploid hybrid species all showed increases compared with the parental species, but differed from each other: H. anomalus had a higher 2C-value (11.46 pg) than either H. deserticola (10.79 pg; t = 4.36, P = 0.002) or H. paradoxus (10.78 pg; t = 3.90, P = 0.009), after adjustment for multiple contrasts.
Synthetic hybrids and hybrid-zone plants
The DNA contents of natural and synthetic hybrids did not show increases similar to those found in the hybrid species (Fig. 2). Correspondingly, the three hybrid species were each significantly different from all synthetic and natural hybrid populations (P < 0.0001 in all cases). F1 plants derived from H. annuus maternal plants had a mean 2C-value of 7.26 pg, while F1 seeds from the reciprocal cross had a mean 2C-value of 6.91. In an analysis treating each cross as a random factor nested within the maternal species, this difference was significant (P = 0.008). The synthetic F6 lineages, derived from H. annuus maternal plants in their initial cross but back-crossed to both H. annuus and H. petiolaris, had the same mean nuclear DNA content as the H. annuus maternal F1 plants (7.26 pg), indicating no genomic size changes in the first six generations following hybridization. Plants from natural hybrid zones had values near that of H. petiolaris (6.86 pg).
Within each species, populations did not differ significantly in C-value for H. annuus, H. paradoxus, and H. anomalus (Table 1). However, interpopulation differences were found for the two remaining species. For H. petiolaris (F6,24 = 4.51, P = 0.003; Table 1), the 2C-value of the Monahans, Texas population of ssp. fallax was lower than that of any other population, and significantly lower than the values of four of the other six populations. The two H. petiolaris subspecies did not differ consistently: the Montana H. petiolaris ssp. petiolaris population had the second-lowest C-value (Table 1). Populations also differed significantly in H. deserticola (F5,17 = 49.84, P = 0.0001); two populations with significantly lower C-values were located near Glen Canyon in southern Utah (Fig. 3).
Our results reveal three phenomena. First, one of the hybrid taxa (H. deserticola) had populations that significantly differed in DNA content. Secondly, the nuclear DNA content of F1 hybrids between H. annuus and H. petiolaris differed depending on the direction of the cross, suggesting nuclear–cytoplasmic interactions. Thirdly, the three hybrid-derived species showed a marked increase in DNA content compared with their parents, while synthetic hybrids showed no increase. We discuss each of these patterns in turn.
Intraspecific variation in genome size
Before 2000, intraspecific genome size variation was reported for many taxa (Price et al., 1983; Price & Johnston, 1996). However, the observed variation declined in many species with the use of internal standards (Baranyi & Greilhuber, 1999; Schmuths et al., 2004). Using internal standards, we found a 10% difference in nuclear DNA content within H. deserticola. The two populations with decreased genome size occur at the southern-most part of the range in Utah, and appear to be basal in the phylogeny of the species (Gross et al., 2003). Other data suggest that these populations may have had an independent origin from the rest of the species (Gross et al., 2003).
Nuclear–cytoplasmic interactions in hybrid DNA content
We found an effect of maternal species on the C-values of hybrid plants. To our knowledge, this is the first case in which DNA content has been found to depend on the identity of the maternal parent in hybrid plants, although related phenomena have been reported in other contexts. Variation from the mid-parent genome size that depended upon the identity of the parental plants was seen in Dasypyrum villosum (Caceres et al., 1998). Work on synthesized wheat polyploids has shown that one genome may be stable while the other undergoes rearrangement (Levy & Feldman, 2004), but this is not dependent on the direction of the cross. Synthesized Brassica polyploids show a maternal effect on genome change (Song et al., 1995). The ease of crossing H. annuus and H. petiolaris, the repeatability of the apparent nuclear–cytoplasmic interaction in determining nuclear DNA content, and the availability of natural hybrid zones make this an excellent study system for unraveling the mechanism behind this pattern.
Nuclear DNA content of hybrid species
We found that Helianthus species derived from interspecific hybridization had significantly higher nuclear DNA contents than their parent species. The three hybrid taxa are derived from the same parental species, but evidence points to multiple origins within two of the three hybrid taxa. Therefore, the increase in nuclear DNA content occurred independently and repeatedly.
Unfortunately, we cannot assess how common this pattern is for homoploid hybrid species. Of the handful of well-documented cases (Rieseberg, 1997), only those in Helianthus have published C-values for both the parental and descendant hybrid taxa. We are currently seeking seed sources to establish whether the increases seen in the Helianthus hybrid species are exceptional. Evidence from some animal hybridization studies indicates that large changes in genome size can occur in a single generation through transposon replication (O’Neill et al., 1998; Labrador et al., 1999), although other animal and most plant studies have failed to find similar patterns (Guerreiro, 1996; Rayburn et al., 1993; Williams et al., 2002). Given the repeated independent emergence of larger genomes in hybrid sunflower taxa, we expected to see frequent increases in genome size in F1 hybrids. We did not: F6 hybrids did not differ from the F1 hybrids, and naturally occurring hybrid-zone plants showed a nonsignificant decrease in genome size compared with the parental plants. These results suggest several hypotheses: (1) DNA increases occur much later than the F6 generation; (2) DNA increases occur rapidly following hybridization, perhaps as early as the F6 generation, but occur infrequently or depend on particular genotypes not included in this study; and (3) DNA increases occur in the field in response to environmental conditions which were lacking in the glasshouse environment.
Potential evolutionary forces underlying genome size increases in Helianthus
The repeated large increases in genome size may be attributable to a ratchet effect. Hybrid speciation probably proceeds through a tight bottleneck. If a rare, large-genome plant existed during the bottleneck, larger genome size may persist if it is selectively neutral, or even if it is disadvantageous (Lynch & Conery, 2003). While several mechanisms, including unequal and illegitimate recombination, can eliminate genes, it is unclear whether selection would favor genomes that were 0.0001% smaller as a result (Bennetzen, 2002). Repeated evolution of large genome size across multiple origins argues against this mechanism.
Alternatively, large genomes may have been directly favored in the transition of the hybrid sunflowers to their new habitats. The three hybrid species inhabit extreme environments relative to their parents. H. deserticola inhabits the desert floor, H. anomalus is found in sand dunes, and H. paradoxus occurs in salt marshes. Perhaps these distinct harsh environments all favor larger genomes, and so selected for increased genome size. Supporting the plausibility of this hypothesis, larger genomes are found in more variable environments in C. siliqua (Bures et al., 2004). However, the opposite pattern is more frequent, with plants in more stressful environments exhibiting lower rather than higher DNA contents (e.g. Price, 1988; Castrojimenez et al., 1989).
As a third alternative, we propose that increased genome size in the Helianthus ancient hybrids might have been indirectly favored by selection. This might have occurred in two ways. H. annuus and H. petiolaris differ by numerous chromosomal rearrangements (Burke et al., 2004). As a consequence, recombination in F1 hybrids generates mostly inviable gametes because of the presence of large chromosomal deletions (Chandler et al., 1986). Genomic redundancy provides a simple way of reducing the initial fitness costs of these rearrangements. Thus, there might have been strong selection for early-generation hybrids that carried large-scale chromosomal duplications.
Increased genome size might also have been indirectly favored if it increased recombination between homeologous chromosomes. Selection for increased fertility in the glasshouse leads to rapid genomic restructuring and recovery of fertility (Rieseberg et al., 1996). We also know that the successful generation of the novel phenotypes that characterizes each of the hybrid species requires recombination of parental quantitative trait loci (Rieseberg et al., 2003). If an increase in genome size (e.g. the amplification of repetitive elements) increased recombination and chromosome pairing, then selection might favor larger genomes, not because of the favorable effects of larger genomes per se, but because of the favorable recombinations or increased fertility that resulted from increased chromosome pairing. The ecological differentiation and establishment of homoploid hybrids may have required the fortuitous involvement of individual lineages that were predisposed to repetitive element increases, allowing increased recombination and thus the formation of new, recombinant genotypes. This proposal is supported by work in other systems linking the quantity of repeated elements to recombination rates. Disease resistance clusters have been shown to have repetitive elements that may favor recombination (Richter & Ronald, 2000). In allopolyploid wheat, deletions of repeated elements may decrease homoeologous pairing (Ozkan et al., 2001).
Here, we have shown that genomic organization in three Helianthus hybrid species differs from that of their parents and early-generation hybrid lineages (both natural and synthetic). Genome size increases are therefore not automatic in the first six generations following hybridization. Future work could aim to distinguish between the alternative possibilities that genome size increase occurs early but infrequently (requiring large-scale sampling of early generation hybrids), occurs later in the process (requiring maintenance of hybrid lineages for many generations), or only occurs under particular ecological conditions, for example in stressful environments.
Whether or not the timing of genome size changes is ultimately resolvable, we show here that hybrid sunflowers also offer the opportunity experimentally to examine the mechanisms and potential adaptive roles of genome size change. Synthetic hybrid lineages have high fertility when crossed with the natural hybrid species (Rieseberg, 2000), yet differ in genome size by 50%. By crossing these synthetic lineages with the hybrid species, it should be possible to generate a range of genome sizes, and then to examine the effect of genome size on both recombination rates and ecological performance.
Thanks are due to the US Department of Agriculture – Agricultural Research Service National Plant Germplasm Resource Center as well as Lisa Donovan, Briana Gross, Nolan Kane, David Rosenthal, and Mark Ungerer for providing seeds used in this study; to Angela Omilian and two anonymous reviewers for helpful comments on the manuscript; and to the Center for Genomic Biology and the Drosophila Genome Resource Center at Indiana University for access to the flow cytometer.