Triploid cultivars of Cymbidium act as a bridge in the formation of polyploid plants

Triploid is considered a reproductive barrier and also a bridge in the formation of polyploids. However, few reports are available in Cymbidium. In this study, diploid ‘Xiaofeng’, sexual triploid ‘Yuchan’ and ‘Huanghe’ of Cymbidium were used to evaluate hybridization compatibility of the triploids. Results showed that the sexual triploids were fertile whether they were used as male or female parents. ‘Yuchan’ produced male gametes of 1x, 1x~2x, 2x, 2x~3x, and 3x at frequencies of 8.89%, 77.78%, 6.67%, 3.33%, and 3.33%, respectively; while ‘Huanghe’ produced 3.33% 1x, 80.00% 1x~2x, 8.89% 2x, 5.56% 2x~3x, and 2.22% 3x male gametes. The cross of ‘Xiaofeng’ with ‘Yuchan’ produced progenies with a wide range of ploidy levels, including one diploid, 34 2×~3× aneuploids, 12 triploids, and one tetraploid, indicating that male gametes produced by sexual triploid were fertile and could be transmitted and fused with egg cells. On the other hand, 10 progenies obtained from the cross of ‘Yuchan’ × ‘Xiaofeng’ were all aneuploids. The cross of ‘Yuchan’ with ‘Huanghe’ produced 40 progenies including three 2×~3× aneuploids, nine 3×~4× aneuploids, 21 tetraploids, six 4×~5× aneuploids, and one pentaploid, suggesting that 2x gametes, instead of the unreduced ones played a more important role in the formation of tetraploids. The survival rates of the hybrids were all above 80.00%, with the tetraploids at 96.67%. Cytological analysis revealed that during meiosis of sexual polyploids, two chromosome sets of the 2n gamete were inclined to enter into the same daughter cell, resulting in the production of 2x gametes. Our results indicate that the triploid cymbidiums are not reproductive barrier but serve as a bridge in the formation of polyploid plants.


Introduction
Polyploidy plays a major role in the evolution and diversification of plants ( Thompson and Lumaret, 1992;Bretagnolle and Thompson, 1995;Kovalsky et al., 2018). In natural populations, polyploidy is formed by several different routes (Ramsey and Schemske, 1998). Among them, sexual polyploidy through unreduced gametes (2n gametes) is considered to be the main pathway (Bretagnolle and Thompson, 1995;Ramsey and Schemske, 1998;Xie et al., 2022). The union of reduced and unreduced gametes produces triploids, and the combination of two unreduced gametes forms tetraploids (Bretagnolle and Thompson, 1995;Husband, 2004). Owing to the limited chance in the fertilization between simultaneously formed unreduced male and female gametes, triploids are usually considered as the intermediate stage in the formation of stable tetraploids, and this pathway of tetraploid formation is known as the 'triploid bridge' (Ramsey and Schemske, 1998;Yamauchi et al., 2004;Jike et al., 2020).
Orchids are plants belonging to the family Orchidaceae that are prized by their ornamental and medicinal value. There are 27,801 recognized species that are globally distributed with the exception of Antarctica (Vilcherrez-Atoche et al., 2022). Orchids have been used as models for studying plant evolutionary processes and adaptability to different environmental conditions. Polyploidy plays an important role during the evolution of orchids as sequence analyses showed that wholegenome duplication (WGD) occurred widely in orchids, including Apostasia shenzhenica Zhang et al., 2021), Cymbidium ensifolium (Ai et al., 2021), Dendrobium chrysotoxum , and Phalaenopsis (Cai et al., 2015). A recent study showed that triploids clustered in an intermediate position between diploids and tetraploids in Zygopetalum mackayi (Moura et al., 2020). In the Nigritella nigra group, nuclear and plastid marker analysis showed that tetraploid N. nigra subsp. austriacais somewhat differentiated from the triploid subsp. Nigra at nuclear as well as plastid loci. The fusion of an unreduced egg cell from subsp. Nigra with a haploid microgamete from Gymnadenia conopsea gave rise to Gymnigritella runei (Hedren et al., 2018). In Phalaenopsis, diploids, triploids, pentaploids, and aneuploids were produced from the crosses of diploid × triploid or triploid × diploid. Triploids, tetraploids, octoploids, and aneuploids were identified in triploid × tetraploid crosses, while no hybrids were obtained from the cross of triploid × triploid (Zhou et al., 2009). Nevertheless, it is generally acknowledged that polyploids can be formed via polyspermy, unreduced gamete, hybridization and endopolyploidy in orchids (Okamoto et al., 2017;Vilcherrez-Atoche et al., 2022); but it is still unclear how each of these pathways contributes to the polyploidization in orchids.
Cymbidium Sw. is one of the most important orchids consisting of 74 species that are epiphytic, lithophytic, terrestrial or sometimes rarely leafless saprophytic (Ning et al., 2018;Thakur and Dutt, 2021). Among the terrestrial species, C. sinense, C. ensifolium, C. goeringi, and C. kanran are the most popular flowering ornamental plants and widely cultivated for their beauty and fragrance (Huang et al., 2012). Our previous research identified 2n gamete occurrence in cultivated cymbidiums (Zeng et al., 2020). Hybridization among selected cultivars or species produced five triploid and two tetraploid progenies. Two of five triploids were propagated through in vitro culture and evaluated in shaded greenhouse for their aesthetic value. Results showed that they had improved ornamental traits displayed by rounder flowers with wider sepal, petals, and lips compared to the diploids. The occurrence of more triploids than tetraploids was intriguing. Since orchids do not have endosperm, triploid block due to the endosperm balance could not be great concern. Besides, triploid plants can be easily propagated through in vitro culture (Zeng et al., 2020). The higher frequency in triploid occurrence, the improved ornamental traits, and little concern over the triploid block prompted us to further analyze 2n gamete occurrence in cymbidiums and the implication of triploids as a bridge in the formation of polyploid plants.
The objectives of this study were to examine microsporogenesis and microgametogenesis behaviors of two sexual triploids, determine their pollen type and fertility, evaluate their crossability with either diploids or triploids, and analyze ploidy levels and the survival rates of their progenies. Results showed that the union of 2x gametes, which were derived from the unreduced gamete, was probably the key pathway for the formation of tetraploids through 'triploid bridge'. Our studies with cymbidium demonstrated the importance of triploids in the formation of polyploid plants.

Cytological observations of microsporogenesis and microgametogenesis
Microsporogenesis and microgametogenesis were observed using the method described by Zhu et al. (2014). The pollinia of 'Yuchan', 'Huanghe', and 'Xiaofeng' at different formation and developmental stages were collected and fixed at 4°C for 12-24 hours in fresh prepared Carnoy's solution. They were then transferred to 70% ethanol and stored at 4°C. The fixed pollinia were placed on a slide; after surface dried with filter paper, two drops of improved carbolic acid fuchsin or 4,6-diamidino-2phenylindole (DAPI) [2 µg.ml −1 DAPI, 1% Triton X-100 (v/v), and 1% sucrose (w/v)] staining solution were added, they were crashed with a forceps, and stained in the dark at room temperature for 5 min. A cover glass was applied and squeezed with pencil eraser, the slide was observed and photographed under either light or UV illumination with ZEISS microscope. For observation of each microsporogensis stage per hybrid progeny, at least nine slides were observed where three slides and 100 microsporocytes as a replicate. There were three replications for each hybrid progeny. For examining each microgametogensis stage of the hybrid progenies, at least ninety pollens were observed with thirty pollens as a replicate. The observations also had three replications. The percentage of meiosis abnormalities and each male gamete type were calculated as follows: (1) the percentage of meiosis abnormalities = (the number of abnormal microspore mother cell in a replicate/100) × 100% and (2) the percentage of each male gamete type = (the number of certain male gamete type in a replicate/30) × 100%.
For the calculation of dyad and triad occurrence, ten vision fields were photographed at 40× magnitude for each slide, one slide was regarded as a replicate, and each material was replicated three times. The frequencies of dyad and triad incidence were calculated according to the formula: F Dy (%) = (Number of dyads/total microspore count observed)) ×100; F Tr (%) = (Number of triads/ total spore count observed) ×100.

Pollen viability determination
Pollinia of 'Yuchan' or 'Xiaofeng' were collected from the flowers that had opened for one day and placed on a slide. After two or three drops of 0.05% of 2, 3, 5-triphenyltetrazolium chloride solution were added, the pollinia were crashed with a forceps and kept in the dark at room temperature for 2-3 h. A coverslip was applied, and the slide was observed and photographed under photomicroscope (OlympusIX71, Japan). Pollen grains with red color were regarded as viable. About 1,000 pollens were counted per slide, one slide was regarded as a replicate, and three slides per cultivar were observed. The pollen fertility was calculated according to the formula: The pollen fertility (%) = (the number of stained pollen grains/total number of pollen grains observed) × 100. et al., 2020). A total of 11 pairs of hybridization were made using the seven cultivars. The numbers of pollinated flowers and capsules produced from the pollinations were recorded, and fruit setting rates were calculated. Seeds harvested from the cross of 'Yuchan' × 'Xiaofeng', 'Xiaofeng' × 'Yuchan', and 'Yuchan' × 'Huanghe' were germinated in vitro. The protocorm like body or rhizome obtained from a seed was propagated, and test-tube seedlings were produced (Zeng et al., 2020). After the seedlings reached about 5 cm in height, they were used for identification of ploidy levels. When the seedlings were about 8 cm in height, they were removed from test tubes, rinsed with tap water, briefly air dried, and transplanted into small black plastic planting bags (100 mL). Each bag was filled with a substrate comprised of small pine bark (1 cm in length) and peat in a 3:1 ratio based on volume, one seedling per bag. Potted seedlings were grown in the aforementioned shaded greenhouse and fertigated with a Hyponex (N-P 2 O 5 -K 2 O; 20-20-20) solution every 10 d. After 6 months of growth, they were transplanted into 2.6 L bags filled with the pine bark and granite substrate and grown in another shaded greenhouse under a light intensity of 300-400 µmol·m -2 ·s -1 . A slow-release fertilizer (N-P 2 O 5 -K 2 O; 20-20-20) was applied to each bag at 3-4 g each in March and September, respectively. Meanwhile, a solution containing 0.1% KH 2 PO 4 (w/v) was sprayed monthly during growing season. Initially, a total of 90 seedlings from each hybrid were transplanted, and they were arranged as a randomized complete block design with three replicates. After 10 months of transplanting, the number of surviving seedlings were recorded, and the survival rate was calculated according to the formula: The survival rate (%) = (the number of seedlings survived in a replicate/30) × 100.

Flow cytometry analysis
The ploidy level of hybrid progenies was measured by flow cytometry Zeng et al., 2020). For each individual, young leaves, about 0.5 cm 2 , were placed in a oneoff culture dish. After adding 0.4 mL of PartecHR-A extract, the leaves were chopped quickly with a blade, following by adding 1.6 mL of Partec HR-B (DAPI, 4,6-diamidino-2-phenylindole) solution as DNA staining agent. The mixture samples were filtered through 30 mm Partec Celltrics microporous membrane, stained in darkness for 5 min and analyzed by Partec flow cytometer using CyView8.5 software (PartecGmb H, Munster, Germany). The DNA histograms of nuclei from each sample were produced.

Chromosome counts
In order to further verify ploidy level of the hybrid progenies, the number of chromosomes in root tip cells was accounted by squash method (Zhou et al., 2009). The root tips with a length of 2~3 mm was cut from the newly formed root of in vitro grown seedlings and fixed in Carnoy's solution, which was consisted of 95% ethanol and glacial acetic acid in a 3:1 ratio based on volume, at 4°C for 12~24 h. The fixed material was washed with distilled water for 2~3 times, and then acidulated with 1 mL concentration of 1 mol·L -1 HCl in a constant temperature water bath at 60°C for about 8 minutes. The dissociated root tips were immersed in distilled water for 30 min, then stained with improved carbolic acid fuchsin staining solution, crushed with tweezers. The debris was discarded, and the sample was covered with a coverslip and observed at 100 × magnification using a photomicroscope (OlympusIX71, Japan). A digital camera system (Nikon) was used for photography. For each plantlet, at least 20 cells were observed. If more than 90% of the cells had a constant chromosome number, the chromosome number of the seedlings was confirmed. As diploid Cymbidium has somatic chromosome numbers of 40, we defined that chromosome numbers of 41-59, 61-79, and 81-99 were aneuploid of 2×~3 ×, 3×~4×, and 4×~5×, respectively.

Statistical analysis
All data were subjected to analysis of variance using Microsoft Office Excel 2019 and SPSS 26.0 (IBM Corporation, Somers, NY). When significance occurred, means were separated by Duncan's multiple range test at P< 0.05 level.

Triploid pollen viability and hybridization compatibility
The intention of making the 11 crosses (Table 1) was to assess the hybridization compatibility of two sexual triploids 'Yuchan' and 'Huanghe'. As a result, capsules were obtained in all 11 cross combinations, and all capsules obtained had seeds. The seeds collected from the crosses of 'Yuchan' × 'Xiaofeng', 'Xiaofeng' × 'Yuchan', and 'Yuchan' × 'Huanghe' were geminated normally in vitro, and the seedlings grew vigorously (Supplementary Figure S2). These results suggested that sexual triploid Cymbidium had high hybridization compatibility and could be used as male or female parent for hybridization.

Types of male gamete and pollen fertility
In order to further understand how tetraploids were formed through 'triploid bridge', the types of male gametes and their fertilities were examined in sexual triploid and diploid parents. Results showed that 'Yuchan' and 'Huanghe' produced 1x, 2x, 3x (unreduced gamete) and aneuploid male gametes (Figures 2B-L; Supplementary Figure S3). The proportion of 1x~2x aneuploid gametophytes was 77.78% and 80.00% in 'Yuchan' and 'Huanghe', respectively (Table 3), which proved that the main type of male gamete produced by triploids was aneuploidy. The occurrence of 1x and 2x male gametes with a rather high proportion in sexual triploids suggested that the unreduced gamete was inclined to enter into the same daughter cell during meiosis, thus resulting in the formation of 2x male gamete ( Figure 2J).
The viability of pollen was investigated using 2, 3, 5triphenyltetrazolium chloride staining method. Results showed that 67.88% pollen grains of 'Yuchan' were stained in red, indicating their viability ( Figures 3A2, 3, 4). Similarly, 73.32% pollen grains of 'Xiaofeng' were viable ( Figure 3A1). Besides, some pollen grains with different ploidy levels were also stained in red, suggesting that all types of male gametes were fertile or partial fertile ( Figures 3A3, 4). Moreover, there was no significant difference in the percentage of stainable pollens between the diploid and sexual triploid ( Figure 3B).

Meiotic abnormalities during microsporogenesis
Microsporogenesis of 'Yuchan', 'Huanghe', and 'Xiaofeng' were observed in order to gain a better understanding of cytological mechanisms behind the formation of different types of male gametes (Figure 4; Table 4 and Supplementary  Figures S4, S5). The results indicated that meiotic abnormalities included meiosis asynchrony, lagging chromosomes, chromosome bridges, and abnormal orientation of spindles during the microsporogenesis. Univalents, bivalents, trivalents, and multivalents were observed at diakinesis of 'Yuchan' (Figure 4A). At metaphase I, there were 33.0% and 24.5% microspore mother cells of 'Yuchan' and 'Xiaofeng', respectively at either diakinesis or pachytene stage ( Figure 4B; Table 4). Those microspore mother cells probably missed the meiosis I ( Figure 4C) but proceeded with normal meiosis II ( Figure 4D), which resulted in the formation of dyads ( Figure 4E). Lagging chromosomes were noticed at every stage of meiosis from metaphase I to telophase II ( Figures 4H, I, K, L). Chromosome bridges were found   Figure 4M). The abnormal orientation of spindles was observed during metaphase II, including tripolar spindle ( Figure 4N) and fusion spindle ( Figure 4O), which resulted in the formation of triads ( Figure 4P) and dyads ( Figure 4E). The percentage of lagging chromosomes and chromosome bridges in sexual triploids was significantly higher than that in diploid. As a result, a higher percentage of aneuploid male gametes occurred in sexual triploids (Table 4).

Statistical analysis of survival rate of plantlets with different ploidy levels
To determine the survivability of plantlets, the survival rates of plantlets with different ploidy levels grown in the shaded greenhouse were evaluated. The survival rate of tetraploid plants was 96.67%, which was significantly higher than that of triploids. Interestingly, the survival rates of 2×~3× and 3×~4× aneuploid plants were also significantly higher than that of triploids. There was no significant difference in the survival rate between 4×~5× aneuploid and triploid plants ( Figure 5).

Discussion
Polyploidization is considered as an important evolutionary force. The most common mechanism of polyploid origin is believed to be through production of unreduced gametes (Clo et al., 2022). There are two main models explaining the pathways of polyploid formation in diploid populations: (i) frequencydependent minority cytotype exclusion (Levin, 1975;Husband, 2000) and (ii) the 'triploid bridge' hypothesis (Ramsey and Schemske, 1998;Burton and Husband, 2001;Husband, 2004;Peckert and Chrtek, 2006;Schinkel et al., 2017). According to the 'triploid bridge' hypothesis, triploids are first formed via the union of reduced and unreduced gametes. Subsequently, backcrosses of triploids to diploids or crosses between triploids can generate tetraploids (Bretagnolle and Thompson, 1995;Ramsey and Schemske, 1998). In the present study, we documented that sexual triploid cymbidiums produced functional 1x, 2x, 3x, and aneuploid gametes after backcross to diploids or through the cross between triploids. We further showed the triploid cross resulted in the formation of tetraploids at a high percentage and also pentaploids. Our results demonstrate that the sexual triploids act as a bridge for efficiently producing tetraploid and even polyploid with higher ploidy level in Cymbidium.

FIGURE 5
Survival rates of hybrid seedlings with different ploidy levels after being transplanted to plastic bags containing a substrate and grown in a shaded greenhouse. Bars represent standard error, and different letters on the top of bars indicate significant difference in survival rates among hybrid progenies analyzed by Duncan's multiple range test at P< 0.05 level. Li et al. 10.3389/fpls.2022.1029915 Frontiers in Plant Science frontiersin.org between sexual triploids could be a principal way of producing tetraploid through 'triploid bridge' in Cymbidium.
In theory, triploids are sterility due to the unbalanced meiotic chromosome segregation, which resulted in the production of aneuploid gametes (Kohler et al., 2010;Wang et al., 2017;. But in practice, a lot of triploids can produce functional euploid gametes, especially for x and 2x gametes in different proportions, which can be used as male or female parent in cross breeding programs (Lim et al., 2003;Hayashi et al., 2009;Zhou et al., 2009;Nakato and Masuyama, 2021). Why does a triploid produce euploid gametes and why is it regarded as a bridge in polyploid evolution? Thus far, little information is available to the questions. Here we propose a hypothesis of coordinate actions of unreduced gamete to address the questions: During meiosis of a sexual polyploid, two chromosome sets of the 2n gamete are inclined to be assorted to a daughter cell, resulting in the production of 2x gamete, such a chromosome behavior during meiosis mainly depends on the genetic relationship of the parents who provide the chromosome set. When the genetic relationship is very close, such as sexual autopolyploid, the main chromosome pairing configuration at diakinesis is trivalent (sexual autotriploid) or quadrivalent (sexual autotetraploid); when the genetic relationship is far different, such as sexual allopolyploid, the main chromosome pairing configuration is a univalent and a bivalent (sexual allotriploid) or two bivalents (sexual allotetraploid). In fact, the meiotic configuration 8I+8II+2III was the most common in two natural triploid populations of Campuloclinium macrocephalum, and their pollen fertilities were 44.74 and 52.69%, respectively (Farco and Dematteis, 2014). Similar results were obtained in allotriploid P. alba × P. berolinensis 'Yinzhong' (Wang et al., 2017). Lavia et al. (2011) reported that the main chromosome paring configuration in sexual autotriploid Arachis pintoi was trivalent, and the pollen grain viability was 42.47%.  reported that most Alstroemeria interspecific F 1 hybrids of Chilean-Brazilian species simultaneously produced 2n male and female gametes; and all the F 2 progeny plants, which were resulted from self-pollination of the F 1 hybrids, were typical allotetraploids. Additionally, most of them formed 16 bivalents and a small proportion formed multivalents during metaphase I stages of meiosis. Triploids that originated through the fusion of 2n × n gametes of the same taxon showed more regular meiotic behavior and higher fertility than triploids from the contact zone of diploids and tetraploids or triploids of hybrid origin (Kovalsky et al., 2018). Natural Dactylis polyploids exhibited successful chromosome pairing during meiosis, whereas artificial polyploids did not, suggesting that there was a selection for sexual fertility in order to stabilize meiosis in natural polyploids (Lumaret and Retired, 1988). Our results indicated that both sexual triploids 'Yuchan' and 'Huanghe' produced x and 2x male gamete ( Figure 2J) with 2x gamete frequencies at 6.67% and 8.89%, respectively, and the percentage of viable pollen was 67.88% in 'Yuchan'. These results further proved that two chromosome sets of the 2n gamete were inclined to be assorted to a daughter cell. However, due to the sophisticated origin of 2n gamete and sexual triploid, the chromosome paring configuration was not typical and the occurrence percentage of 2x and x gametes was not high. Nevertheless, due to the occurrence in 2n gametes, triploids play an important role in polyploid evolution.
Our results indicated that in the cross of 2× × 3×, 25% hybrids were triploids, which was similar to the results in Tulip (Marasek-Ciolakowska et al., 2014) and Populus (Wang et al., 2017). These triploids were probably formed by the fusion of a 2x male gamete produced by the triploid with a haploid female gamete from the diploid or originated from the fusion of a haploid male gamete from the triploid with unreduced female gamete from the diploid. A tetraploid was also obtained in the cross of 2× × 3×, which was likely formed by the fusion of an unreduced male gamete (3x) produced by the triploid with a haploid female gamete from the diploid or originated from the fusion of 2x male gamete from the triploid with unreduced female gamete from the diploid. In the cross of 3× × 3×, 52.5% progenies were tetraploids, and these tetraploids probably originated from the fusion of an unreduced gamete produced by one parent with a haploid gamete from the other or the fusion of two 2x gamete from the parents. Similar results were reported in H. echioides where 3× × 3× produced 60% tetraploid (Peckert and Chrtek, 2006). Because of the low percentage of occurrence in euploid gametes in triploid, the aneuploid gametes might play an important role in the production of tetraploids. The possible pathways of producing tetraploids through 'triploid bridge' in Cymbidium are illustrated in Figure 6.
A long-standing problem in polyploid breeding is triploid block, which is a reproductive barrier caused by malfunction of endosperm. The endosperm supports the development of the embryo by providing nutrients for its growth, and the genetic relationship between the endosperm and the embryo is important for higher plants evolution (Johnston et al., 1980;Koehler et al., 2010;Stoute et al., 2012;Huc et al., 2022). In the majority of plants, the maternal and paternal chromosome dosages in the endosperm are considered to be critical for seed development and fertility (Vinkenoog et al., 2003;Koehler et al., 2010;Schinkel et al., 2017). However, orchid seeds have rudimentary embryo and lack of endosperm (Yeung, 2017;Chen et al., 2018). Seed germination and seedling establishment depend on the successful interaction between protocorms and mycorrhizal fungi either in vitro or ex vitro, thus endosperm is not a limiting factor affecting seed germination (Chugh et al., 2009;Xu et al., 2011;Yeung, 2017;Yeh et al., 2019;Bhatti and Thakur, 2022). Our study showed that hybrid seeds derived from the crosses of 3× × 2×, 2× × 3×, and 3× × 3× were well developed and able to germinate normally on culture medium. Plantlets with different ploidy levels grew healthy in greenhouse. Thus, triploid block is not a concern in Cymbidium as sexual triploids can produce seeds through hybridization with appropriate parents and the seeds can readily germinate.
Cymbidium is a renowned genus in the orchid family that distributed in tropical and subtropical areas of Asia, Papua New Guinea, and Australia (Ai et al., 2021). It exhibits distinctive ecological diversification and occurs in terrestrial, epiphytic, lithophytic, and saprophytic life forms (Yukawa and Stern, 2002;Ning et al., 2018). The cross compatibility between different species was reported to be high, and even an increasing number of interspecific and intergeneric hybrids has been obtained through artificial pollination Ogura-Tsujita et al., 2014;Joffard et al., 2022). These interspecific hybrids have a high percentage in occurrence of 2n gametes and have been successfully employed to create sexual polyploids in breeding programs (Zeng et al., 2020;Kondo et al., 2022). However, because of reproductive isolation caused by geographical, temporal, and spatial isolations, the interspecific hybrids are rare in nature. The triploid crossability and 2n gamete formation demonstrated in the present study may represent a viable way for creating new polyploid cymbidium through hybridization.

Conclusion
As far as is known, this is the first investigation of the crossability of sexual triploids in cymbidiums. Our study documented that both triploid 'Yuchan' and 'Huanghe' were fertile and able to be used as male or female parents in cross breeding. Sexual triploid cymbidiums produced fertile male gametes of 1x, 1x~2x, 2x, 2x~3x, and 3x. The production of 2x male gametes could be resulted from the probability that two chromosome sets of the 2n gamete were inclined to enter into the same daughter cell during meiosis. The cross of diploid × triploid generated diploid, triploid, and tetraploid with frequencies at 2.1%, 25.0%, and 2.1%, respectively. The cross of triploid × triploid produced tetraploid and pentaploid hybrids with proportions of 52.5% and 2.5%, respectively. The survival rate of tetraploid was significantly higher than that of triploid. Our results indicate that the triploid cymbidiums are not reproductive barrier but act as a bridge in the formation of polyploid plants. A hypothesis of coordinate actions of unreduced gamete was proposed to explain why sexual triploids produce 1x and 2x euploid gametes. Further research with more sexual polyploids including auto and allopolyploids, along with the use of genomic in situ hybridization (GISH) should be conducted to test this hypothesis.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

Author contributions
Z-SZ and R-ZZ designed the study. M-ML, R-ZZ, LX, H-RG, and J-RZ performed the experiment. M-ML, R-ZZ, and LX carried out and analyzed the data. M-ML and Q-LS wrote a first draft of the manuscript that was further critically reviewed and edited by JC, R-ZZ, LX, QW, H-RG, and J-RZ. All authors contributed to the article and approved the submitted version.