Genome-Wide Duplication of Allotetraploid Brassica napus Produces Novel Characteristics and Extensive Ploidy Variation in Self-Pollinated Progeny

Whole genome duplications (WGDs) have played a major role in angiosperm species evolution. Polyploid plants have undergone multiple cycles of ancient WGD events during their evolutionary history. However, little attention has been paid to the additional WGD of the existing allopolyploids. In this study, we explored the influences of additional WGD on the allopolyploid Brassica napus. Compared to tetraploid B. napus, octoploid B. napus (AAAACCCC, 2n = 8x =76) showed significant differences in phenotype, reproductive ability and the ploidy of self-pollinated progeny. Genome duplication also altered a key reproductive organ feature in B. napus, that is, increased the number of pollen apertures. Unlike autopolyploids produced from the diploid Brassica species, the octoploid B. napus produced from allotetraploid B. napus had a relatively stable meiotic process, high pollen viability and moderate fertility under self-pollination conditions, indicating that sub-genomic interactions may be important for the successful establishment of higher-order polyploids. Doubling the genome of B. napus provided us with an opportunity to gain insight into the flexibility of the Brassica genomes. The genome size of self-pollinated progeny of octoploid B. napus varied greatly, and was accompanied by extensive genomic instability, such as aneuploidy, mixed-ploidy and mitotic abnormality. The octoploid B. napus could go through any of genome reduction, equilibrium or expansion in the short-term, thus providing a novel karyotype library for the Brassica genus. Our results reveal the short-term evolutionary consequences of recurrent polyploidization events, and help to deepen our understanding of polyploid plant evolution.

allopolyploid Brassica napus genome instability mitotic abnormality pollen aperture polyploid evolution Polyploidy or whole genome duplication (WGD) has long been considered a prominent force driving evolution of angiosperm species (Casacuberta et al. 2016;Leitch and Leitch 2008;Soltis et al. 2009;Jiao et al. 2011). Advances in sequencing have revealed a series of ancient WGDs throughout the evolutionary history of angiosperms (Jiao 2018;Wendel et al. 2016;Jiao et al. 2011;Ren et al. 2018). All extant seed plants have experienced at least one cycle of WGD in their evolutionary history (Jiao et al. 2011). More than a third of present angiosperm plants are polyploids (Wood et al. 2009), and approximately 13% of diploid plant species populations may harbor unnamed polyploid cytotypes (Barker et al. 2016). Accumulating evidence suggests that polyploidy has played a significant role in phenotypic innovation, species diversification, facilitation of biodiversity, and adaptation to broader ecological environments (Soltis et al. 2009; Brassica oil-crops and vegetables (Wang et al. 2011). There is strong evidence that WGDs frequently occur in the evolutionary history of the Brassicaceae family (Soltis et al. 2009; Schranz et al. 2012;Jiao et al. 2011;Romanpalacios et al. 2019;Mandáková et al. 2017;Ren et al. 2018). All Brassicaceae species share a common ancient WGD event occurring around the origin of the family (Schranz and Mitchell-Olds 2006;Schranz et al. 2012). Arabidopsis thaliana has undergone at least five rounds of WGD across its evolutionary history (Jiao et al. 2011). All Brassica species share the complex evolutionary history of Arabidopsis thaliana but with the addition of a lineagespecific whole-genome triplication (Couvreur et al. 2010;Lysak et al. 2005;Wang et al. 2011;Yang et al. 2016). Diploid species Brassica rapa (AA,2n = 20) and Brassica oleracea (CC, 2n = 18) are considered mesohexaploids Liu et al. 2014;Wang et al. 2011).
Brassica napus (AACC, 2n = 4x = 38) is a recent allopolyploid species formed about 7500 years ago by hybridization between B. rapa and B. oleracea (Chalhoub et al. 2014). B. napus has undergone a predicted 72 · polyploidization since the origin of the angiosperms (Chalhoub et al. 2014). It is widely accepted that recurrent polyploidization has driven evolution of key traits, species diversity, and profound evolutionary novelty in the Brassica lineage (Schranz et al. 2012;Romanpalacios et al. 2019; Hofberger et al. 2013;Edger et al. 2015;Wang et al. 2011). B. napus is a classic example of the importance of WGD in plant evolution, and is considered an ideal system for investigation of the consequences and mechanisms of de novo and recurrent polyploidization (Jenczewski and Alix 2004;Chalhoub et al. 2014;Szadkowski et al. 2010;Ferreira de Carvalho et al. 2019;Mason and Snowdon 2016). Recent analysis suggests that recurrent WGDs may cyclically provide novel genetic resources for evolution and diversification, because duplicated sequences from ancient WGD tend to be lost in species that experienced additional recent WGD (Jiao 2018;Ren et al. 2018). However, although polyploidization remains a highly dynamic, active and ongoing process in nature (Leitch and Leitch 2008;Comai 2005;Alix et al. 2017), it is largely unclear whether recent allopolyploid species can benefit from an additional round of WGD, or if novel traits will be derived in such a scenario. Recently, we tried to double the genome of allotetraploid B. napus, and found that the octoploid B. napus has a novel function (Fu et al. 2018): when used as a pollen donor, octoploid B. napus induces tetraploid B. napus (female) to produce maternal doubled-haploid progeny with almost all genetic material coming from the female parent (Fu et al. 2018;Fu et al. 2019). This novel method allows production of homozygous B. napus lines within two years, without the necessity of using microspore culture (Fu et al. 2018;Fu et al. 2019). Nevertheless, the influences of WGD on the physiological and reproductive characteristics of B. napus are as yet largely unknown and worth exploring. Against the background of global climate change and the accelerated rate of plant species extinction (Humphreys et al. 2019;Knapp et al. 2020), doubling the genome of allopolyploid rapeseed may innovate germplasm resources for further genomic and genetic research. Here we have investigated the vegetative, reproductive, and karyotypic effects of duplicating an allotetraploid genome, and assessed the vigor, fitness and fertility of octoploid B. napus, as well as the short-term evolutionary trend of higherorder newly-formed polyploids, offering a unique perspective into the constraints and mechanisms operating in the evolution of the angiosperms.

Plant materials
Two octoploid rapeseed lines (Y3380 and Y3560, AAAACCCC, 2n = 8x = 76) were artificially synthesized by doubling the genome of allotetraploid B. napus ( Figure S1). The octoploid B. napus used here was the seventh self-pollinated generation of the newly resynthesized octoploid rapeseed. P3-2 (AACC, 2n = 4x = 38) was a common parent of the two octoploid B. napus lines. The conventional B. napus variety ZS11 (2n = 4x = 38) which has been widely cultivated in China was used as a control. All plant materials were grown under the same conditions at the experimental field in Wenjiang (E103.83, N30.70), Chengdu, China. The plants were self-pollinated by bagging in the field to obtain self-pollinated seeds.

Fluorescent in situ hybridization
Fluorescence in situ hybridization (FISH) analysis was carried out according to protocols detailed in (Cui et al. 2012). The probe used for in situ hybridization was BAC BoB014O06, which contains C-genome specific repeat elements (Xiong et al. 2011) and was labeled with biotin-11-dCTP. The A-genome was visualized with DAPI (4',6-diamidino-2-phenylindole) as a background stain under UV fluorescence. Photographs were taken with a fluorescence microscope (ZEISS AX10) and image processing was performed using Adobe Photoshop CS6.

Phenotypic observations
Phenotypic characteristics of the rapeseed materials were observed and photographed by SLR camera (E0S 200D, Cannon). The plant height, length of main inflorescence, primary branch number, number of effective pods per plants, pod length, number of seeds in a pod, number of ovules in a pod, 1000-seed weight and the germination rate of self-pollinated seeds of 30 octoploid B. napus plants were counted and recorded. dehydration for 15 min each, followed by incubation in isoamyl acetate for 15 min. The treated anthers were placed on a CO 2 critical point dryer for drying, and the sample was adhered to the conductive carbon film with double-sided tape, then placed on the ion sputter sample stage for about 30 sec and photographed.

In vitro pollen tube germination
The in vitro germination experiment of mature pollen grains was carried out according to the procedures detailed in Feng (2017).

Meiotic behavior observation
Observations of meiotic behavior were carried out according to the procedures detailed in (Li et al. 1995). Young buds (2-3 mm long) were collected in Carnoy's fixative (ethanol: glacial acetic acid, 3:1, v/v) in the morning, and fixed for 24 h at room temperature. Young anthers were dissected out, incubated in 1 M hydrochloric acid solution at 60°for 6-8 min, removed and placed on a glass slide, squeezed gently to release pollen mother cells and sporads into Carbol fuchsin solution, and then observed under an optical microscope.

Somatic chromosome counting
Young ovaries were used to count the somatic chromosome number (Zhou et al. 2016). The ovaries were treated with 0.002 mol/L 8-hydroxyquinoline solution for 3 hr in the dark, then transferred to Carnoy's fixative (ethanol: glacial acetic acid, 3:1, v/v) for more than 24 hr. Cytogenetic observation was carried out according to the procedures detailed in Li et al. (1995).
Ploidy determination by flow cytometry Flow cytometry analysis was performed to assess the plant DNA content. A total of 414 individuals were randomly selected from the self-pollinated offspring of 19 octoploid Y3380 plants, and 588 individuals derived from 30 octoploid Y3560 plants were randomly selected. Details of procedures were as referred to in (Dolezel et al. 2007), with slight modifications. Fresh young leaves were taken between 9 AM to 11 a.m, washed with distilled water and dried with filter paper. Leaves with a diameter of 0.5 cm was taken using a hole punch and placed in a pre-cooled Petri dish. Subsequently, 0.5 ml of pre-cooled LB01 cell lysis buffer (15 mM Tris, 2 mM disodium edetate, 0.5 mM spermine tetraamine, 80 mM potassium chloride, 20 mM sodium chloride, 0.1% (v/v) polyethylene glycol octylphenyl ether and 15 mM b-mercaptoethanol, pH 7.5, 0.22 mm filtered) was added. The leaves were chopped quickly into pieces with a sharp blade, filtered with a 75 mm filter, and then stained with 1 mL PI solution (5% propidium iodide and 5% RNA enzyme) for 30 min in the dark, then measured by flow cytometry (Accuri C6 Plus, BD). At least 20000 cells were collected from one sample. Data were analyzed using AccuriC software. The formula for calculating the approximate chromosome number of the samples was as follows (Dolezel et al. 2007).

Data availability
File S1 contains two sets of flow cytometry histograms of octoploid B. napus (Y3380 plants and Y3560 plants). File S2 contains two sets of flow cytometry histograms of tetraploid B. napus (ZS11 and P3-2). File S3 contains two sets of flow cytometry histograms of the self-pollinated individuals derived from octoploid B. napus. All flow cytometry fluorescence histograms were output directly from the AccuriC software. Supplemental Figures contains Figures S1-S12. Table S1 contains the fluorescence intensity values of the mitotic G1 phase of octoploid and tetraploid B. napus. Table S2 contains phenotypic data for octoploid and tetraploid B. napus. Table S3 contains the fluorescence intensity values of the mitotic G1 phase of self-pollinated individuals derived from octoploid B. napus. Supplemental material available at figshare: https://doi.org/10.25387/g3.12593627.

Chromosomal constitution of octoploid rapeseed
The octoploid rapeseed obtained by doubling the genome of tetraploid B. napus had 76 chromosomes in somatic cells (Fu et al. 2018). FISH results further confirmed that the somatic cells of the two octoploid rapeseeds contained $36 C n chromomes and $40 A n chromosomes (Figure 1), with a chromosomal composition of AAAACCCC, confirming that the octoploid rapeseed is actually octoploid B. napus. Flow cytometry analysis showed that the fluorescence intensity value of the mitotic G1 phase of octoploid B. napus was around 865,000, and that of tetraploid B. napus was about 423,000 (Table S1, File S1-S2).

WGD-driven changes in phenotypes and physiology
Although the appearance of octoploid B. napus was roughly similar to that of tetraploid B. napus (Figure 2), WGD caused phenotypic and physiological differences between the two B. napus types. The octoploid B. napus was shorter ( Figure 2A), with plant height ranging from 62 to 189 cm (average 146 cm), while the average plant height of tetraploid B. napus was 183 cm ( Figure 3A, Table S2). The average length of the main inflorescence of octoploid B. napus decreased by 36% relative to the tetraploid ( Figure 3B, Table S2), and the average number of primary branches of octoploid B. napus showed a reduction of about 20% compared to that of the tetraploid ( Figure 3C, Table S2). Additionally, compared with tetraploid B. napus, the octoploid B. napus grew slower at the seedling stage ( Figure S2), the leaves were thicker and deformed ( Figure S3), and the floral organs were slightly larger ( Figure 2B-D). Novel phenotypic variation was also prevalent in the octoploid B. napus population ( Figure S3-S4).

Ploidy-induced variation in male reproductive organs
Pollen viability of the octoploid B. napus was generally extremely high, up to 90%, only slightly lower than that of the tetraploids ( Figure 4A). Compared to tetraploid B. napus, octoploid B. napus had larger anthers, and each anther contained more pollen grains ( Figure  4B). Scanning electron microscopy observation of pollen grains Sample chromosome No: ¼ reference chromosome No: · mean position of the G1 sample peak mean position of the G1 reference peak revealed that more than 98% of the octoploid B. napus pollen grains had mutated exine patterns, with an increase in the number of pollen apertures to four ( Figure 4C) (natural Brassica plants have only three pollen apertures per pollen grain). Some octoploid B. napus pollen had apertures asymmetrically distributed on the pollen wall ( Figure  4C). In vitro pollen germination assays showed that the pollen grains of octoploid B. napus might grow faster than those of tetraploid B. napus ( Figure S5).

Meiotic process of octoploid B. napus
In order to explore the mechanisms underlying high pollen viability in octoploid B. napus, the meiotic process of the pollen mother cells was studied. Many abnormal chromosomal behaviors were detected, including multivalent chromosome pairing, irregular chromosome arrangements during metaphase, lagging chromosomes, chromosome bridges and chromosome breakages at anaphase, and unequal chromosome distribution at telophase ( Figure 5). However, some cells still had relatively normal meiotic processes ( Figure 5I). It is worth noting that after telophase II, more than 99% of pollen mother cells were divided into four parts, eventually forming normal tetrads. The tetrads finally formed regular pollen grains with four apertures, and a few pollen grains had five or more apertures ( Figure  5K). Overall, there were certain abnormalities during meiosis in octoploid B. napus, but the general process was relatively normal, which may ultimately promote the formation of viable pollen grains.

Fertility of octoploid B. napus
The number of self-pollinated siliques harvested from octoploid B. napus ranged from 34 to 409, with an average of 216, while tetraploid B. napus produced an average of 692 siliques per plant ( Figure 3D).  The relatively small number of effective siliques (with normal seeds inside) on an octoploid B. napus plant was partly due to the fact that some flowers could not develop into effective siliques ( Figure 6A). The siliques derived from self-pollination of octoploid B. napus were significantly (P , 0.05) shorter than those of tetraploid B. napus: the average length of the former was about half that of the latter ( Figure  3E, Figure 6). Additionally, the average number of seeds in a silique derived from self-pollination of octoploid B. napus was between 5.3 and 10.8, with a median of 8.6, while that of tetraploid B. napus was about 25.9 ( Figure 3F, Figure 6). In addition to the small number of ovules in a pistil ( Figure 3G) compared with tetraploid B. napus, the reasons for the reduced fertility of octoploid B. napus included the following factors: some pollen grains could not germinate on the stigma ( Figure S6); pollen tubes could not elongate properly ( Figure  S6); the number of pollen tubes entering the style was reduced ( Figure   S6); some pollen tubes could not accurately identify the ovule and release sperm ( Figure S6); and some embryos aborted after fertilization during the post-zygotic development phase ( Figure S7). Although the seeds from self-pollination of octoploid B. napus varied in size and shape ( Figure S8), and their average thousand-seed weight ranged from 3.8 to 6.9 g ( Figure 3H, Table S2), they had a high seed germination rate (average 93%) ( Figure 3I). Overall, the octoploid B. napus had an intermediate level of reproductive capability.
DNA content of self-pollinated progeny of octoploid B. napus Flow cytometry analysis showed that the DNA content of the selfpollinated progeny of octoploid B. napus varied greatly and was normally distributed (Figure 7). Of the 558 self-pollinated offspring of octoploid Y3560, about 237 individuals had DNA content close to that of octoploid B. napus (with a confidence level of 99%), 210 individuals had lower DNA content than the octoploid parent, and the genome size of the remaining 141 individuals was larger than that of the octoploids ( Figure 7A, Table S3). Among the 414 self-pollinated progeny of octoploid Y3380, approximately 70 individuals had DNA content close to that of octoploid B. napus (with a confidence level of 99%), while 335 individuals had DNA content lower than that of the octoploid parent, and only 9 individuals had a genome size larger than that of the octoploid parent ( Figure 7B, Table S3). In the Y3560 offspring population, the maximum DNA content was 2.7 times the minimum DNA content; and in the Y3380 offspring population, this maximum DNA content was 7.3 times the minimum DNA content (Table S3). Due to the linear relationship between the DNA content and the number of somatic chromosomes ( Figure S9), we speculated that the somatic chromosome number of the self-pollinated offspring derived from octoploid B. napus could be up to 100 (Table S3). Additionally, FISH analysis showed that the C sub-genome chromosomes from octoploid B. napus might be more vulnerable to loss in the self-pollinated progeny than the A sub-genome chromosomes ( Figure S10). The self-pollinated progeny generation of octoploid B. napus had an extensive range of ploidy variation, where ploidy was not only maintained around the octoploid level in some individuals, but also reduced or expanded in others.
Abnormal mitosis in the self-pollinated progeny of octoploid B. napus Some anomalies, such as mixed-ploidy, were found in the selfpollinated progeny generation of octoploid B. napus. The frequency of occurrence of mixed-ploidy individuals in the self-pollinated progeny population of octoploid B. napus was between 1% and 3%. Several different types of mixed-ploidy plants were detected by flow cytometry analysis ( Figure S11). The mixed-ploidy plants were mostly aneuploids, as confirmed by somatic chromosome counting (Figures 8, S12). There was no obvious integer relationship between the chromosome numbers of different cytotypes (Figures 8, S12). Due to the slight asymmetry and partial overlap of the flow cytometry peaks ( Figure S11), the results of flow cytometry could not fully reflect the true chromosome numbers of the mixed-ploidy plants: the results obtained by somatic chromosome counting were far more complicated than the corresponding flow cytometry results (Figures 8, S12). Moreover, there were a variety of mitotic abnormalities in the cells of mixed-ploidy plants, including lagging chromosomes during mitotic metaphase, unequal chromosomal distribution to daughter cells, abnormal chromosome arrangements caused by multipolar spindles, and chromosome loss ( Figure  9). Therefore, mitotic instability appears to further exacerbate the genomic instability of the mixed-ploidy plants.

DISCUSSION
Genome duplication can affect plant phenotypes and physiology in various ways (reviewed by Doyle and Coate 2019), but the effects may vary significantly by taxa and ploidy level. Although polyploidy generally leads to increased organ size, with larger, more vigorous plants (Ramsey and Ramsey 2014), this is not always the case. We observed a decrease in vegetative organ size in our autoallooctoploids relative to the allotetraploid B. napus. Similar results were also found in chromosome-doubled B. napus (Olsson 1960b) and B. juncea (Olsson 1960a), where general plant height and size were reduced relative to the allotetraploid parents. Abel and Becker (2007) had similar findings in synthetic autotetraploid B. rapa and B. oleracea, and attributed this effect to inbreeding depression caused by increased homozygosity. Hence, crossing between different autooctopoloid lines may be of interest in restoring "hybrid vigour": (Gowers 1977) also had some success with this approach in tetraploid turnip breeding, where crosses between two inbred autotetraploid lines showed increased biomass in many cases relative to the homozygous parent autotetraploids. In addition, the octoploid B. napus plants are relatively short, with strong lodging resistance, large flowers and abundant pollen, and thus could have good ornamental potential.
Our results indicated that additional WGD significantly altered the structure of the male reproductive organs, specifically resulting in an increase in the number of pollen apertures in octoploid B. napus relative to tetraploid B. napus. This finding supports the hypothesis that a ploidy-sensitive mechanism affects the establishment of pollen apertures (Reeder et al. 2016). The germination apertures are where water and air enter and exit the pollen grain, and where the pollen tube extends out (Edlund et al. 2004). Pollen grains with three germination apertures are ubiquitous in dicotyledonous plants, including the Brassica genus (Reeder et al. 2016). Changes in the aperture number may affect the ability of pollen grains to regulate volume changes and to be involved in reproduction (Albert et al. 2018). Pollen tubes of octoploid B. napus pollen with four apertures could grow faster than those of tetraploid B. napus pollen with three  apertures, but encountered more difficulties during in vivo fertilization, which was in line with the view that an increase in the number of germination apertures could be conducive to competition of pollen on the stigma but not conducive to fertilization (Prieu et al. 2016). Thus, four or more apertures may affect the chance of successful fertilization in higher-order dicotyledonous polyploids. We confirm that recurrent polyploidization can drive novel traits in Brassica evolution.
Polyploidy is a problem for the normal completion of mitosis and meiosis due to the complex chromosome pairing that can occur when more than two copies of each chromosome are present (Comai 2005). Meiotic stability is a key hurdle that must be overcome following WGD (Pelé et al. 2018) and is one of the hallmarks of an adapted polyploid (Baduel et al. 2018). Meiotic instability in newly synthesized B. napus has been extensively documented, and found to be due to frequent homeologous chromosome pairing between the A and C sub-genomes (Gaeta et al. 2007;Xiong et al. 2011;Szadkowski et al. 2011). The meiotic response to additional WGD of an established allopolyploid has not previously been reported. In this study, it was found that WGD of allopolyploid B. napus did not majorly disrupt meiotic stability and in particular pollen viability. Despite many meiotically abnormal behaviors, pollen mother cells of octoploid B. napus (AAAACCCC) mostly formed tetrads, the majority of which successfully developed into normal pollen grains, with pollen viability per plant up to 90%. Similar results of high pollen fertility were observed by (Olsson 1960a) and (Olsson 1960b) in genomedoubled B. juncea and B. napus, supporting our finding. Resynthesized Brassica hexaploids (AACCCC (Li et al. 2013) andAAAACC (McNaughton 1973)) resulting from crosses between B. rapa and B. napus also possessed fairly stable meiotic behavior in most pollen mother cells, with pollen fertility of approximately 90% (Li et al. 2013;McNaughton 1973). However, in autooctoploids formed from the diploid species, including autooctoploid B. rapa (AAAAAAAA, 2n = 8x = 80), autooctoploid B. nigra (BBBBBBBBBB, 2n = 8x = 64), and autooctoploid B. oleracea (CCCCCCCC, 2n = 8x = 72), the meiotic process was extremely disordered, pollen grains were severely . DNA content was detected by flow cytometry, and expressed as fluorescence intensity of the mitotic G1 phase. The green columns represent plants with DNA content lower than that of their octoploid parent, that is, the genome size of these progeny has shrunk. The orange columns represent individuals with DNA content close to that of the octoploid B. napus at a 99% confidence level (genome equilibrium). The pink columns represent plants with expanded genome size compared to the octoploid parent. The blue solid line indicates the normal distribution curve. deformed, and pollen viability was reduced to 45.4%, 2.1% and 36.8%, respectively (Liu 2014). More seriously, the majority of pollen grains of autooctoploid A. thaliana were dead at anther dehiscence (Wang et al. 2010). It appears that higher-order allopolyploids have a stronger ability to regulate meiotic stability and to form normal pollen than higher-order autopolyploids. This suggests that higherorder allopolyploids have greater reproductive capacity and adaptability than higher-order autopolyploids, supporting the view that allopolyploidy results in higher adaptive potential than autopolyploidy ( Van de Peer et al. 2017). Interactions between sub-genomes may also play an important role in maintaining meiotic stability and pollen viability in higher-order polyploids. In resynthesized Brassica allohexaploids (AABBCC), meiotic stability and fertility vary based on parental species origin and genotype (reviewed by Gaebelein and Mason 2018). For instance, some allohexaploid genotypes from the cross B. carinata · B. rapa are immediately meiotically stable (Gupta et al. 2016), while the majority are unstable (Tian et al. 2010), while striking differences in fertility and stability have been observed between allohexaploids derived from the cross B. juncea · B. oleracea (Mwathi et al. 2020;Zhou et al. 2016), B. napus · B. nigra (Gaebelein et al. 2019a) and (B. napus · B. carinata) · B. juncea (Gaebelein et al. 2019b;Mwathi et al. 2017). The meiotic stability of allohexaploid AABBCC may be affected by both interactions between the A, B and C sub-genomes, as well as specific genetic loci (Gaebelein et al. 2019b;Zhou et al. 2016). The regulation of meiotic stability of octoploid B. napus, which contains two sub-genomes in two copies each, may fall somewhere between that of a true autopolyploid and a true allopolyploid.
Newly formed polyploids may be particularly common in angiosperms, but most polyploid plants are predicted to go extinct shortly after formation, such that few successfully establish as a new species (Leitch and Leitch 2008;Soltis et al. 2014;De Storme and Mason 2014;Wang et al. 2010). Although the gametophytes of our octoploid B. napus were vigorous, they still encountered certain difficulties during the reproductive process. Fertility-related traits (silique number, seed number, silique length) of the octoploid B. napus were significantly lower than those of tetraploid B. napus, but the octoploid B. napus could successfully set hundreds of self-pollinated seeds with an average of eight seeds per silique, which was enough to propagate the next generation. This phenomenon appeared consistent over several generations. Our resynthesized higher-order octoploids were propagated and maintained for 7 generations of self-pollination, indicating that the post-pollination reproductive barriers were not extremely serious during this self-pollination process. We hence propose based on these results that the octoploid B. napus has achieved reproductive success over a short evolutionary period. Similarly, previous studies found relatively good fertility in hexaploid Brassica composed of the A and C sub-genomes (AACCCC (Li et al. 2013) and AAAACC (McNaughton 1973)) in the first three generations. Meanwhile, some hexaploid Brassica AAAACC plants also failed to set seed (McNaughton 1973), similar to observations in autooctoploid Brassica (Liu 2014). Some resynthesized Brassica allohexaploids (AABBCC) are also highly fertile, but with some sterile plants within each generation (Mwathi et al. 2017;Mwathi et al. 2020;Gaebelein et al. 2019b). Whether the fertility in octoploid B. napus in our study is related to sub-genomic interactions or to particular genetic combinations is unclear.
Although several rounds of genome doubling have occurred in the evolutionary history of the Brassica genus, the highest ploidy level observed in natural Brassica species is only tetraploid, indicating that Brassica lineages have the tendency and ability to downsize their duplicated genomes over a long period of time (Hohmann et al. 2015). It has been proposed that three strategies of genome evolution (shrinkage, expansion, or equilibrium) are possible for organisms to find an optimal balance between genomic stability and plasticity (Schubert and Vu 2016); of these strategies, genome contraction appears to be the most common (Hufton and Panopoulou 2009;Mun et al. 2009;Wendel 2015;Leitch and Leitch 2008;Yang et al. 2011). However, genome size variation and changes in the early generations of newly formed polyploids can be complicated even within a single genus. The genome sizes of self-pollinated offspring of the resynthesized octoploid B. napus were normally distributed, with the maximum being several times the minimum, and each of genome expansion, contraction and equilibrium being observed. However, there was a general trend toward loss of chromosomes over gain of chromosome copies in the early generations of allohexaploid Brassica (AABBCC) (Zhou et al. 2016;Gaebelein et al. 2019b). The chromosome numbers of offspring of newly formed tetraploid B. napus were maintained at or near 38 (Xiong et al. 2011). Our results indicate that additional WGD may increase the plasticity and flexibility of the polyploid genomes, thereby effectively buffering against the unbalanced numbers of chromosomes and gene copies, and thus allowing plants that deviated from euploidy to survive. This finding is in line with the view that the success of newly formed angiosperm polyploids is partly attributable to their highly plastic genome structure (Leitch and Leitch 2008). Moreover, the chromosome loss in the self-progeny of octoploid B. napus seemed to be more likely to occur in the C subgenome, which was consistent with the previous finding that the A sub-genomes are more stable than the C sub-genomes (Zhou et al. 2016). Besides, the octoploid B. napus appears to have the ability to rapidly change genome size, which may promote the acquisition of novel traits and aid in species diversification (Puttick et al. 2015), and which enabled us to observe some highly divergent lineages.
Chromosomal instability is common after WGD in both natural and synthetic lines, and extensive aneuploidy is commonly observed in early generation neopolyploids (De Storme and Mason 2014; Chester et al. 2012;Zhou et al. 2016;Xiong et al. 2011). Possibly due to factors such as tetrasomic sub-genome dosage, homeologous chromosome pairing and chaotic segregation of chromosomes, a large number of aneuploids with an extremely wide range of ploidy variations were involved in the early generations of synthesized octoploid B. napus and accrued across multiple generations. This observation is in full agreement with previous studies and further strengthens the view that polyploidy and aneuploidy have played an important role in the evolution of the Brassicaceae (Warwick and Alshehbaz 2006). Additionally, chromosome instability may also result from the inability of cells to handle aberrant behaviors during mitosis (Hufton and Panopoulou 2009;Comai 2005). Although genomic instability such as multiple spindles has been reported in plant meiosis (Risso-Pascotto et al. 2005), abnormal mitosis in polyploids has rarely been directly studied. We found clear evidence of genomic instability during mitosis in Brassica polyploid plants. We found severe cases of unequal segregation of chromosomes and multipolar spindles in mitosis cells of self-pollinated progeny of octoploid B. napus, possibly due to the higher number of chromosomes passing through the threshold for error-free segregation in mitotic cell division (Comai 2005). This phenomenon could also be due to sub-genomic chromosome imbalance; similar situations have been observed in aneuploid and polyploid human cancer cells (Ganem et al. 2009). These mitotic abnormalities increased the chance of chromosome mis-segregation and the formation of cytotypes with a series of chromosomes, consistent with the discovery of mixed-ploidy descendants of octoploid B. napus. Mixed-ploidy plants may play a bridging role in the formation of plants with different ploidy levels (Varela-Álvarez et al. 2018). However, mixed-ploidy plants have seldom been reported in Brassica, such that our findings of multipolar spindles in mitosis and mixed-ploidy plants in the selfpollinated offspring of higher-order octoploid B. napus provide insights into the diversity and complexity of the Brassica genome and the role of ploidy variation in adaptation.

CONCLUSIONS
In this study, we assessed the effects of additional WGD on extant allotetraploid B. napus. WGD decreased the size of vegetative organs but slightly increased the size of reproductive organs in allotetraploid B. napus. Despite some abnormal chromosomal behaviors, the octoploid B. napus exhibited fairly high meiotic stability and produced about 90% viable pollen. The extra WGD significantly changed the appearance of the male reproductive organs of B. napus: in particular the number of pollen apertures of octoploid B. napus increased from three to four. The octoploid B. napus demonstrated generally moderate seed fertility, and could set several hundreds of seeds per plant on average. Higher ploidy is not necessarily a barrier to reproductive success in Brassica, or possibly in induced autopolyploids of allopolyploid species in general. The genome size of self-pollinated progeny of the octoploid B. napus varied greatly, and was accompanied by extensive genomic instability, such as aneuploidy, mixed-ploidy and mitotic abnormality. The additional WGD may have amplified the tolerance of B. napus to genome instability. The octoploid B. napus could go through any of genome reduction, equilibrium or expansion in the short-term, thus providing a novel library of karyotypes for the Brassica genus. Our results provide novel insights into the early stages of recurrent polyploidization events and the short-term evolutionary fate of newly-synthesized octoploid rapeseed, as well as supporting an extremely high degree of plasticity in Brassica genomes. Additional WGD in present allopolyploid species could drive novel traits or features, and in future, related studies in other genera may help to improve our understanding of the evolution of polyploid angiosperms.