Interploidy hybridization in sympatric zones: the formation of Epidendrum fulgens × E. puniceoluteum hybrids (Epidendroideae, Orchidaceae)

Interspecific hybridization is a primary cause of extensive morphological and chromosomal variation and plays an important role in plant species diversification. However, the role of interploidal hybridization in the formation of hybrid swarms is less clear. Epidendrum encompasses wide variation in chromosome number and lacks strong premating barriers, making the genus a good model for clarifying the role of chromosomes in postzygotic barriers in interploidal hybrids. In this sense, hybrids from the interploidal sympatric zone between E. fulgens (2n = 2x = 24) and E. puniceoluteum (2n = 4x = 56) were analyzed using cytogenetic techniques to elucidate the formation and establishment of interploidal hybrids. Hybrids were not a uniform group: two chromosome numbers were observed, with the variation being a consequence of severe hybrid meiotic abnormalities and backcrossing with E. puniceoluteum. The hybrids were triploids (2n = 3x = 38 and 40) and despite the occurrence of enormous meiotic problems associated with triploidy, the hybrids were able to backcross, producing successful hybrid individuals with broad ecological distributions. In spite of the nonpolyploidization of the hybrid, its formation is a long-term evolutionary process rather than a product of a recent disturbance, and considering other sympatric zones in Epidendrum, these events could be recurrent.


Introduction
Interspecific hybridization is frequently cited as a primary cause of the extensive morphological and chromosomal variation observed in plant genera and species complexes (Petit et al. 1999;Chapman and Abbott 2010;Souza et al. 2012;Presgraves 2013), especially in sympatric areas. The merging of two divergent genomes into a unique nucleus, caused by hybridization, can set the stage for dynamic changes in the genome, transcriptome, and phenotype of the new hybrids, what could have consequences on parental species after hybrid backcrossing species Paun et al. 2010Paun et al. , 2011aJiao et al. 2011;Buggs et al. 2012). This event is also responsible for increasing the genetic diversity within species, transferring genetic adaptations between species, breaking down or reinforcing reproductive barriers between closely related groups, which can lead to the emergence of new ecotypes or species, and playing a role in the adaptive radiation of plant species (Soltis et al. 2003;Slotte et al. 2008;Jorgensen et al. 2011).
Hybridization events have been repeatedly observed among food-deceptive orchids as Epidendrum L., the largest Orchidaceae genus in the Neotropical region, representing c. 1500 species distributed from Florida (USA) to Argentina (H agsater and Soto-Arenas 2005;Pinheiro and Cozzolino 2013). Generally, Epidendrum species lack strong premating barriers because they have high interspecific reproductive compatibility, share extensive numbers of pollinator species, and frequently have overlapping flowering periods (Almeida and Figueiredo 2003;H agsater and Soto-Arenas 2005;Pansarin and Amaral 2008). These characteristics favor the formation of hybrid swarms, such as those commonly observed in sympatric species from the subgenus Amphyglottium (Dunsterville 1979;Dressler 1989;H agsater and Soto-Arenas 2005;Pinheiro et al. 2010). Taxonomic problems within this group are probably the result of hybridization and late generation introgression among co-occurring taxa, which blur species boundaries and increase morphological variability (Pinheiro et al. 2010).
Epidendrum species present various sympatric zones throughout the distribution of the genus, and the one composed by E. fulgens (2n = 2x = 24) and E. puniceoluteum F. Pinheiro & F. Barros (2n = 4x = 52) should be highlighted (Tanaka and Kamemoto 1984;Pinheiro et al. 2009). This sympatric zone extends from the south to south-eastern coastal plains in Brazil (Fig. 1), with E. fulgens colonizing sand dunes and E. puniceoluteum swampy areas. Epidendrum fulgens and E. puniceoluteum have distinct flower color combinations (orange sepals and petals with yellow with red dots labellum in E. fulgens; red sepals, petals, and labellum with a yellow-orange callus in E. puniceoluteum), but throughout the sympatric zone, gradual color variation can be observed from one extreme to the other (Pinheiro and Barros 2006;Fig. 1), and these intermediary individuals can be observed throughout sand dunes and swampy areas (Pinheiro et al. 2010). Different butterfly and moth species are known to pollinate Epidendrum species (Pansarin and Amaral 2008), and on food-deceit systems as observed on Epidendrum, such variable flower color could be bennefical, to puzzel the pollinator. Previous analyses of nuclear and plastid markers and manual crossings have confirmed F1 and F2 hybrid formation and it backcrossing toward the polyploid specie E. puniceoluteum, but not toward the diploid E. fulgens (Pinheiro et al. 2010).
The difference in chromosome number and ploidy level between E. fulgens and E. puniceoluteum makes this sympatric zone even more challenging because hybrid fertility decreases as the difference in parental chromosome number increases (Levin 2002). The importance of interploidal hybridization for generating viable, fertile wild hybrids, and increasing population diversity and gene introgression among parental species is well documented (Stebbins 1971;Petit et al. 1999;Flatberg et al. 2006;Slotte et al. 2008;Chapman and Abbott 2010;Ricca et al. 2011). Examples in Orchidaceae are well reported in the literature, especially in the genus Dactiylohriza Neck. ex Nevski on crossings involving D. fuchsii (2n = 40) with D. purpurella and D. praetermissa (both 2n = 80; cited as Dactylorchis in Heslop-Harrison 1953, 1957 The behavior of chromosomes in a new hybrid nucleus can be analyzed through chromosome characterization techniques as chromosome banding and in situ hybridization direct localizing DNA sequences on chromosomes (Jiang and Gill 2006;Chester et al. 2010). The complete genome is useful for differentiating parental chromosomes in a cell of a hybrid individual through genomic in situ  Pinheiro et al. (2010) and morphological data from Pinheiro and Barros (2006). hybridization (GISH) techniques (Schwarzacher et al. 1989;Markova et al. 2007;Markova and Vyskot 2009). The success of GISH in demonstrating the two parental genomes depends on two main factors. First, the amount of genome differentiation that exists between the parents more related genomes are harder to differentiate in the hybrid chromosomes. Second, the age of the hybridolder hybrids suffer more complete genome turnover, diverging from the parental sequences. The divergence can be so profound that the hybrid genome does not match either parent. However, the time required for genome complete restructuring is thought to be long, for example, c. 5 million years in Nicotiana Koukalova et al. 2010).
In this sense, sympatric zones and natural hybrids formed between E. fulgens and E. puniceoluteum represent an excellent model for investigating the evolutionary significance of chromosome rearrangements and the role of postzygotic barriers, such as meiosis normality, pollen fertility, and pollen tube growth. To accomplish our objective we applied (1) meiotic analysis on both parental and hybrid individuals to verify meiotic normality, (2) pollinium analysis to estimate pollen viability and pollen tube grown, (3) karyotype analysischromosome number, chromosome banding, and in situ hybridizationto evaluate karyotype constancy throughout the sympatric zones and karyotype parental 9 hybrid relatedness and GISH analysis to determine the contribution of each parental to hybrid formation and possible causes of genome variation among hybrids.

Plant material
Wild specimens of E. fulgens, E. puniceoluteum, and their hybrids were collected from sympatric populations found throughout south and south-eastern Brazil ( Table 1). The same individuals were previously genotyped by Pinheiro et al. (2010); nuclear and plastid microsatellites were used to detect pure parental genotypes and hybrid individuals. Collection points were georeferenced and plotted on a map using DivaGis 7.5 (www.diga-gis.org.br; Fig. 1). The collected individuals were cultivated in the living orchid collection at the Instituto de Botânica, São Paulo, Brazil. Vouchers were deposited at the herbarium SP.
Meiosis analysis, pollen viability, and pollen tube germination To obtain pollen mother cells (PMCs) that were undergoing meiosis, young floral buds were fixed in ethanol:acetic acid (3:1, v/v) for 24 h at room temperature and stored at -20°C. To evaluate the meiotic process, pollinia were washed two times in distilled water, digested in 2% (w/v) cellulase (Serva, Heidelberg, Germany)/20% (v/v) pectinase (Sigma, St. Louis, MO) at 37°C for 5 min and squashed in a drop of 60% acetic acid. The best slides were frozen in liquid nitrogen to aid coverslip removal, stained using a solution of 2% (w/v) Giemsa (Sigma) for 5 min, and mounted in a drop of Entellan ® (Merk, Darmstadt, Germany). The slides were analyzed to evaluate the meiotic normality by the means of frequency of cells without meiotic abnormalities, for example, pairing errors, segregation errors, and presence of micronuclei or tetrads with less/more four cells.
Tetrad stainability and morphology were used to estimate pollen viability, following Alexander (1980). Three flowers from eight pure individuals of E. fulgens and E. puniceoluteum and from 15 hybrid individuals from the Ilha Comprida population, all classified by simple sequence repeats (SSRs) following Pinheiro et al. (2010), were collected at anthesis and their pollinia were removed and fixed in ethanol:acetic acid (3:1, v/v) for 24 h at room temperature and stored at À20°C. Three slides were prepared from each individual and the first 500 tetrads were classified. The average frequency of meiotic normality (cells carrying out a normal meiosis) and pollen viability among the parental species and hybrids were analyzed using a nonparametric Kruskal-Wallis test followed by a Student-Newman-Keuls test using BioEstat 5.0 (Ayres et al. 2007).

Pollen tube analyses
Hand-pollination experiments were conducted to assess the strength of postpollination barriers. To evaluate the compatibility in crossings involving hybrids and parental individuals, controlled pollination experiments were conducted in the orchid nursery of Frederico Carlos Hoehne at the Instituto de Botânica de São Paulo, São Paulo, Brazil. Flower buds in preanthesis were isolated with paper bags to exclude floral visitors. For these experiments, five flowers per plant (four E. fulgens, six E. puniceoluteum, and eight hybrid specimens) were used for each pollination treatment. The following pollination experiments were performed: (1) positive controlflowers from each parental species were cross-pollinated with pollinaria from another individual from the same species, with two crossings for each parent; (2) hybrid cross-pollinationhybrid flowers were pollinated using pollinaria from other hybrid plants in three different crossings; (3) hybrid 9 parental crosspollinations -E. puniceoluteum and E. fulgens flowers were pollinated using hybrid pollinaria, with two individuals/ parental species for each crossing; and (4) parental-hybrid cross-pollinationhybrid flowers were pollinated using E. fulgens and E. puniceoluteum pollinaria, with three different individuals/parental species for each crossing. Flowers were collected at 3, 5, 7, 9, and 12 days after pollination and stored in 70% ethanol. Pistils were stained with aniline blue to aid the observation of pollen tube growth under a fluorescence microscope (Martin 1959), using a WU filter from Olympus (Tokyo, Japan).

Pretreatment and storage
Root tips were pretreated in 8-hydroxyquinoline (0.002 mol/L) for 24 h at 10°C, fixed in ethanol:acetic acid (3:1, v/v) for 24 h at room temperature and stored at À20°C.

Chromosome counting and karyotyping
Chromosome preparations were performed by conventional staining following Guerra (1983). Fixed root tips were washed in distilled water three times for 5 min each, hydrolyzed in 5 N HCl at room temperature for 20 min and transferred to distilled water until they were squashed in a drop of 45% acetic acid. The best slides were frozen in liquid nitrogen and the coverslip was removed. The selected slides were stained using 2% (w/v) Giemsa (Sigma) for 5 min and mounted in a drop of Entellan ® (Merk).

Chromosome banding
Fixed root tips were washed in distilled water and digested in a 2% (w/v) cellulase (Serva)/20% (v/v) pectinase (Sigma)/1% macerozyme (Sigma) solution at 37°C for 30 min. The meristem was squashed in a drop of 45% acetic acid and the coverslip was later removed in liquid nitrogen. After 3 days, the preparations were stained with chromomycin A 3 (CMA; 0.5 mg mL À1 ) for 1 h and counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1 lg mL À1 ) for 30 min. After the analysis, the best slides were destained and stored for fluorescence in situ hybridization (FISH) and GISH.
For GISH experiments, E. fulgens and E. puniceoluteum genomic DNA was obtained following Ferreira and Grattapaglia (1998) and labeled, respectively, with biotin-14-dUTP and digoxigenin-11-dUTP by nick-translation (Roche Biochemicals). Initially, both probes were used simultaneously on Epidendrum hybrid metaphases, the E. fulgens probe was detected using avidina-FITC (Sigma) and the E. puniceluteum probe was detected using antidigoxigenin-Rodamina (Roche Biochemicals). A second experiment was performed using biotin-labeled E. fulgens DNA as a probe and unlabeled E. puniceluteum genomic DNA as a block at five different concentrations: 109, 209, 309, 609, and 909, following Moraes and Guerra (2010).

Analysis and editing
All of the slides that were prepared using nonfluorescent stains (Giemsa and Alexander) were examined under an Olympus BX 50 microscope coupled with an Evolution TM MP camera (Media Cybernetics, Rockville, MD) and analyzed using the program Image ProPlus v6 (Media Cybernetics). Slides from CMA/DAPI banding, FISH and GISH, which used fluorescent stains, were examined using a DMRA2 epifluorescence microscope (Leica, London, UK), photographed with a Leica DCF365 FX camera and analyzed using the program LAS 3.0 (Leica). Pistil slides that were stained with aniline blue were examined with an Olympus BX 50 fluorescent microscope with a WU filter, photographed with an Olympus DP73 camera, and analyzed using the cellSens Entry software (Olympus). All images were uniformly processed for color balance, contrast, and brightness using Adobe Photoshop CS5 (Adobe Systems, San Jose, CA).

Meiotic analysis
The meiotic analysis of parental E. fulgens and E. puniceoluteum found high levels of meiotic normalityin average 99.60% and 96.06%, respectivelywith chromosomes pairing as bivalents and following meiotic division forming four equal cells. The chromosome number could be determined in both species: n = 12 in E. fulgens ( Fig. 2A and B) and n = 28 in E. puniceoluteum (Fig. 2C). A few abnormalities were observed in the parental slides, including unpaired chromosomes at prophase I/metaphase I and/or early disjunction at metaphase I in E. fulgens (0.37% and 0.31%, respectively) and unpaired chromosomes at prophase I/metaphase I and anaphase I bridge in E. puniceoluteum (0.58% and 0.48%, respectively; Fig. 2D and E; Table S1). An additional curious abnormality was observed in E. puniceoluteum: metaphase I with the metaphase plate divided in two (0.26%; Fig. 2F). The meiotic analyses of hybrids (13 individuals from Canan eia and Ilha Comprida/SP) found lower levels of normality compared to the parental species (H = 24.019, P < 0.0001), with only 9.35%, in average, of cells carrying out normal meiosis. Nevertheless, two of 13 individuals had 88.8% and 87% normal meiotic cells, respectively, both presenting n = 19. The remaining 11 individuals had a mean normality of 0.12%. Unpaired chromosomes at metaphase I and anaphase I bridges were frequently observed; a consistent number of late chromosomes was observed, including 13-14 bivalents on the plate and 10-12 univalents; in anaphase I c. 13 chromosomes could be observed on poles and more than 12 chromosomes lagged on the metaphase plate ( Fig. 2G-I; Table S1). However, 91.64% of the analyzed meiocytes presented complex abnormalities that were difficult to classify because they involved metaphase clumping/sticking of chromosomes with unpaired chromosomes.

Pollen viability and pollen tube growth
The pollen grains were delivered in tetrads inside a coherent mass, the massulae. Pollen grain stainability (strong/ light purple stained) and morphology (normal/wrinkling) were examined to classify the grains as viable or unviable ( Fig. 2L and M). The parental and hybrid individuals showed high percentages of normal strong stained pollen grains inside the tetrads, but hybrid viability (98.99%) was lower than observed in E. fulgens (99.78%; H = 15.1; P < 0.05; Table 1).
Pistils showed pollen tube growth only 12 days after hand-pollination in all crossings, except those involving E. fulgens and hybrid flowers, which were aborted soon after pollination and no flowers could be collected. Pollen tube growth was observed in all other crossings, with pollen tubes reaching to the ovules, but with differing intensities (Fig. 3). Positive control treatments showed intensive pollen tube growth, as was observed in Epidendrum hybrid (♀) 9 E. puniceoluteum (♂) crosses ( Fig. 3A and B). However, when hybrids were used as pollen donors (E. puniceoluteum (♀) 9 Epidendrum hybrid (♂) and Epidendrum hybrid (♀) 9 Epidendrum hybrid (♂)), pollen tube germination was weak ( Fig. 3C and F).

Karyotype analysis -chromosome number, chromosome banding, and in situ hybridization
In addition to counting the number of chromosomes through meiosis, it was also determined through mitosis, with 2n = 24 in E. fulgens and 2n = 56 in E. puniceoluteum. The hybrids chromosome numbers were defined here for the first time and presented an aneuploid variation of 2n = 38 and 2n = 40 (Fig. 4) (Table 2).

CMA/DAPI banding
The CMA/DAPI banding revealed a diverse band pattern, with DAPI + band variability (Table 2). Epidendrum fulgens had eight (sometimes up to 10) terminal DAPI + bands and two CMA + bands (Fig. 4A), while E. puniceoluteum had four DAPI + bands (two terminals and two subterminals) and three CMA + band (Fig. 4C). Among the hybrids, two (one terminal and one subterminal) and three (one terminal and two subterminal) DAPI + bands were observed in individuals with 2n = 38 and 2n = 40, respectively ( Fig. 4E and I). The CMA + band pattern was more conservative, with two terminal bands in E. fulgens and hybrids, and three terminal bands in E. puniceoluteum, in addition to tiny proximal bands on all chromosomes and hybrids (Fig. 4A, C, E, and I).

5S and 45S rDNA
Two proximal 5S rDNA and two terminal 45S rDNA sites were observed in E. fulgens (Fig. 4B), while in E. puniceoluteum, three proximal 5S rDNA and three terminal 45S rDNA sites were observed (Fig. 4D). In both species, 45S rDNA sites were colocalized with terminal CMA + bands. Among the hybrids, all individuals (with 2n = 38 and 2n = 40) had two proximal 5S rDNA and two terminal 45S rDNA sites that were colocalized with CMA + terminal bands ( Fig. 4F and J).

GISH
GISH on hybrid mitotic metaphases using E. fulgens-and E. puniceoluteum-labeled genomic DNA as probes, without blocking DNA, failed to differentiate the two subgenomes; all of the chromosomes were labeled uniformly (data not shown). However, when E. fulgens-labeled DNA was used as a probe and E. puniceoluteum-unlabeled genomic DNA was used for blocking (1:30), differential labeling was observed on hybrid metaphases. In 2n = 38 individuals, 12 completely labeled chromosomes, 14 partially labeled chromosome, and 12 unlabeled chromosomes were observed ( Fig. 4G and H). However, in 2n = 40 individuals, 12 completely labeled chromosomes, 14 partially labeled chromosome, and 14 unlabeled chromosomes were observed ( Fig. 4K and L).

Discussion
The study of hybrid zones between flowering plant species have provided important results regarding the evolution of reproductive barriers (Widmer et al. 2009), the role of habitat selection in diverging lineages (Johnston et al. 2001;Marques et al. 2010), and the process of speciation (Abbott et al. 2013). As differences in chromosome numbers and ploidy levels are historically considered strong isolating barriers among species (Stebbins 1971;Coyne and Orr 2004), few studies have investigated putative hybrid zones between species showing such differences. However, recent studies have provided data showing extensive hybrid zones presenting high levels of interploidal gene exchange (Ramsey and Schemske 1998;Petit et al. 1999;Chapman and Abbott 2010). The results of this study support the findings of Pinheiro et al. (2010), which characterized an extensive hybrid zones between two species with different ploidy levels: E. fulgens  (2n = 2x = 24) and E. puniceoluteum (2n = 4x = 56).
Here, we observed that despite an almost completely abnormal meiosis and lower pollen viability, hybrids still present few pollen tube growths when used as pollen donors in controlled crossing experiments. However, the introgression was unidirectional toward just one parental, E. puniceoluteum. The karyotype analyses indicate a reduction on amount of repetitive DNA (heterochromatic blocks and rDNA sites) on hybrid genomes, in agreement to expect diploidization event. Chromosome doubling is assumed to be one of the first events during the diploidization process, ensuring the fertility restoration (Soltis et al. 2003), but until now it has not occurred in these hybrids, and the hybrid are still triploids. These unexpected finding highlight the importance of studies investigating hybrid zones formed by species with different ploidy, challenging the widely held view of 'instant isolation' among species of different ploidy (Coyne and Orr 2004).
The chromosome number observed in E. fulgens (n = 12/2n = 2x = 24) agreed with previous reports (Blumenschein 1960;Tanaka andKamemoto 1984, Pinheiro et al. 2009), but the chromosome number observed in E. puniceoluteum (n = 28/2n = 4x = 56) were not the same of earlier reports of 2n = 52 (Pinheiro et al. 2009). The most likely explanation for variation in chromosome number of E. puniceoluteum is the occurrence of two cytotypes. This possibility has direct consequences for the hybrid swarm, especially considering the observation of two chromosome numbers among hybrids: 2n = 38 (supposedly 12 E. fulgens chromosomes and 26 E. puniceoluteum chromosomes) and 2n = 40 (supposedly 12 E. fulgens chromosomes and 28 E. puniceoluteum chromosomes). In fact, the variation of the chromosome number in hybrids and GISH results suggest the additional chromosome pair in 2n = 40 individuals came from the E. puniceoluteum because the 12 well labeled chromosomes that came from E. fulgens were constant between the 2n = 38 and 40 individuals.
Another possibility for this aneuploid variation is meiotic errors in E. puniceoluteum, despite the high meiotic normality index (97%; Table 1). A closely related polyploid species, E. cinnabarinum Salzm., had a reduced meiotic normality index (66% in meiosis II) and low tetrad viability (8.2%), suggesting an unbalanced gamete formation (Da Conceic ßão et al. 2006). However, considering that no aneuploid gametes were observed in E. puniceoluteum and that crossings involving hybrids as the pollen donor did not form seeds (Pinheiro et al. 2010), hybrid aneuploid variation could be a combined consequence of the effect of hybrid meiosis abnormalities on ovule formation and E. puniceoluteum backcrossing.
If interploidal hybrids were completely sterile, they could represent a genetic dead end of little evolutionary relevance (Mayr 1942), but even with low fitness, hybrids may act as a conduit for genetic exchange and could have substantial impacts on the population (Husband et al. 2008;Arnold et al. 2012). The 98.99% of viable tetrads in hybrids is probably an overestimation, consequence of staining technique employment to estimate pollen viability, but the data from hybrid pollen tube germination is a stronger indicative of low hybrid pollen viability and the strong postpollination barrier between hybrid and E. fulgens, but not so strong between hybrid and E. puniceoluteum. In fact, previous studies by Pinheiro et al. (2010) have demonstrated asymmetric introgression toward E. puniceoluteum, but always with this species as the pollen donor.
Hybrid fertility is expected to be a function of the percentage of meiotic paired chromosomes -the higher the number of univalents and multivalents, the lower the fertility (Levin 2002)plus allelic interactions among fertility genes (Buerkle et al. 2000;Bomblies et al. 2007;Chase 2007;Rieseberg and Blackman 2010;He et al. 2011). No multivalents were observed in the hybrids examined here, but univalents were currently detected (c. 10-12). Despite the fact that the hybrid is triploid and the presence of univalents, some chromosome pairing was detected, ensuring some gamete production, probably unbalanced ones, with a portion of the univalents going together to one side or the other during anaphase I or even forming a separate gamete. As such triploid gamete production is expected to be very low, we can consider the possibility that E. puniceoluteum pollen grains (n = 28) have fertilized different unbalanced hybrid gametes (n = 10 and n = 12), which gave rise to new hybrid backcrosses with 2n = 38 and 2n = 40. Similar patterns were suggested from analyses of triploids formed by Dactylohriza fuchsia (2n) 9 D. praetermissa (4n) and D. fuchsia (2n) 9 D. purpurella (4n), with the conclusion being that a consistent number of 20 univalents occurred during hybrid metaphase I, which would tend to be grouped in an aneuploid gamete (Heslop-Harrison 1953). However, Lord and Richards (1977) examined hybrids formed by D. fuchsia (2n) 9 D. purpurella (4n) and found a variable univalent number, between 9 and 12, and 38% aneuploid hybrids.
It is accepted that backcrossing occurs more often between a triploid hybrid and its diploid parental (Ramsey and Schemske 1998), and that such interspecific crosses are more fertile when the female gamete contains more chromosomes than the male gamete (Stebbins 1971;Soltis et al. 2003;Slotte et al. 2008;Erilova et al. 2009;Jorgensen et al. 2011). This is thought to be related to the maternal:paternal (2:1) genome balance in the endosperm, which plays an important role in hybrid viability (Johnston et al. 1980;Erilova et al. 2009). However, because Orchidaceae has no endosperm, this could facilitate crossing in both directions. As observed here, a triploid hybrid that formed from diploid D. incarnata and tetraploid D. lapponica could only backcross with the tetraploid parental (but authors did not indicate the direction of crossing; Aagaard et al. 2005), contradicting the previous triploid x diploid preferential crossing (Ramsey and Schemske 1998). Also introgression between D. fuchsii (2n) and D. maculata (4n) was observed in both directions (St ahlberg 2009). Moreover, the pollen grain conformation as a cohesive mass of tens of tetrads, the pollinium, can have a direct influence on hybrid reproductive success (Harder and Johnson 2008) by minimizing the loss of gametes. Because a pollinium is deposited on a stigma as an entire unit, it reduces pollen loss, and the highly abnormal hybrid meiosis (which should produce only a few viable female gametes) could be compensated for by a massive E. puniceoluteum pollen deposition, which should pollinate any viable female hybrid gametes.
Three hybrid individuals (two with 2n = 40 and one with 2n = 38) of the 23 analyzed here were initially identified by nuclear SSR loci as E. puniceoluteum individuals (Pinheiro et al. 2010), indicating that a high incorporation of parental genome sequences can occur into late hybrids as a consequence of recurrent backcrossing. The same pattern was observed in a sympatric population of Serapias (Orchidaceae), which was analyzed by amplified fragment length polymorphism (AFLP), in which many plants classified as parental species were actually introgressed hybrids, suggesting that the co-occurring Serapias vomeracea and S. cordigera undergo extensive interspecific gene flow and hybrid backcrossing (Bellusci et al. 2010). Such recurrent longterm hybridization and introgression events could contribute to increasing biodiversity, as observed in bromeliads species (Palma-Silva et al. 2011).
All hybrids presented 2n = 38 or 40, showing a nonrandom distribution of chromosome number and suggesting that certain somatic chromosome combinations are more viable than others, possibly reflecting nonrandom hybrid meiotic products (Lord and Richards 1977). In the hybrids, the DAPI + bands were reduced compared to the expected number of bands, with just two (2n = 38) and three (2n = 40) bands being observed (for total parental DAPI + bands and possible gametes see Fig. 5). Two sites for each of 5S and 45S rDNA were observed in all hybrids, but considering the E. puniceoluteum karyotype, one could even expect three sites (for expected patterns of 5S and 45S rDNA, see Fig. 5). Hybridization events are usually followed by a diploidization process, in which repetitive DNA suffers a dynamic reorganization by expansion/reduction of sequences, which could explain the reduction in the repetitive DNA sequences (heterochromatic blocks and rDNA sites) in hybrids (Clarkson et al. 2005;Renny-Byfield et al. 2013), in addition to alterations in the DNA methylation pattern (Paun et al. 2009(Paun et al. , 2010Flatscher et al. 2012). The time required to complete the diploidization process can be determined from patterns of epigenetic variation, as observed in Dactylohriza (Paun et al. 2011b), but also GISH patterns (Lim et al. 2004;Renny-Byfield et al. 2013). The hybrid formation examined here is not thought to be a recent event (Pinheiro et al. 2010), but not old enough to GISH fail and hybrid/parental complete isolation (Clarkson et al. 2005;Koukalova et al. 2010). Estimating time in generations, complete hybrid/ parental separation requires c. 60 generations (Rieseberg and Willis 2007).
Note that the hybrid individuals examined here have not duplicated their chromosomes. However, hybrid speciation that starts from interploidal crossing is particularly prone to resulting in allopolyploids (Chapman and Burke 2007;Buggs et al. 2009;Paun et al. 2009Paun et al. , 2011b. Some hybrids are formed between closely related species with the same ploidy level and represent introgressed ecological forms that colonize diverse habitats and lack postzygotic isolation from their parents, as in the hybrid examined here. However, the hybrid presented here is a triploid formed by interploidal crossing, but curiously it is able to backcross with one parental. Even without duplicating their chromosomes and having low fertility, the Epidendrum hybrids largely colonize both parental habitats (the sand dunes of E. fulgens and the swampy regions of E. puniceoluteum, both along the restinga vegetation on the Brasilian seashore; Pinheiro et al. 2010), suggesting that hybridization might increase ecological flexibility or colonization ability (Petit et al. 1999). In the present hybrids, their vegetative reproduction and perennial habits should have contributed to their persistence in the habitats of both parental species (Pinheiro et al. 2010).
The hybrid GISH pattern was able to separate the chromosomes into three groups: well labeled (originated from E. fulgens), poorly labeled, and nonlabeled (both originated from E. puniceoluteum). The two latter groups indicated that E. puniceoluteum chromosomes share some E. fulgens sequences (poorly labeled chromosomes). During the diploidization period, retrotransposons and DNA satellites jump from one genome to another and begin hybrid genome homogenization, what could explain these 14 constant E. puniceoluteum chromosomes that were poorly labeled by the E. fulgens genome probe. An alternative explanation could be a hybrid origin for E. puniceoluteum, with E. fulgens as one of its parents. Throughout its distribution, E. fulgens overlap with other species (i.e., sympatric zones with E. denticulatum and E. secundum that contain putative hybrids; Pinheiro et al. 2009), and E. puniceoluteum could have originated from one of these E. fulgens sympatric zones.
The pattern of introgression between E. fulgens and E. puniceoluteum challenges the widely held view of hybrid 'instant isolation' and the polyploidy tendency after divergent parental crossing (Coyne and Orr 2004;Chapman and Burke 2007;Buggs et al. 2009;Paun et al. 2011b), and it opens interesting research possibilities throughout the Epidendrum sympatric zones for investigating the evolutionary potential of interploidal hybridization. In addition, both species and their hybrids are distributed throughout the restinga vegetation, an harsh environment in which plants face flooding, drought, constant wind, high salinity, a lack of nutrients, and a broad ecological amplitude (Scarano 2002). Considering these conditions, the adaptive component of these extensive hybridization zones is of particular note. Next-generation sequencing techniques could improve the detection of specific genomic regions that will be useful for improving our understanding of genomic reorganization after hybridization (Chester et al. 2010Buggs et al. 2012), especially after this intriguing interploidal crossing resulted in diploid hybrids, and for localizing sequences associated with selection for divergent habitats, which also occurs in hybrid genomes (Twyford and Ennos 2012) and may be associated with the broad ecological amplitude and high frequency of hybrids observed in natural populations.