Plant material.

Lilac taxa were acquired from gardens, arboreta, and nurseries. Representative taxa were obtained from five of the six series within genus Syringa including Syringa, Pubescentes, Villosae, Ligustrina, and Pinnatifoliae. Series Ligustrae, which includes genus Ligustrum nested within genus Syringa (Li et al., 2012), was not included. Included in our study were 54 total taxa including species, cultivars, and hybrids (Table 2). Species and subspecies designations are based on current taxonomy (Chen et al., 2009; Li et al., 2012). In lilac, cultivar or trademark names are rarely interchangeable with only one becoming the market name that commonly identifies a cultivar. As a reference, cultivar and trademark names are reported (Table 2), but for simplicity, only market names (cultivar or trademark) are used hereafter. A subset of hybrids was created among selected parent taxa to investigate seedling genome size variation in the following parent cytotype combinations: 3x × 2x, 2x × 3x, and 2x × 2x.

Table 2.

Taxonomic, trademark, accession, and source information for Syringa source material used in the current study.

View Table

Flow cytometry.

Flow cytometry was used to assess holoploid (2C) genome size (relative to an internal standard) for each individual taxon in the lilac collection at Oregon State University. References to genome size and ploidy follow the terminology proposed by Greilhuber et al. (2005). For each taxon, three vegetative buds or three young, fully expanded leaves were collected to represent a random sample of nuclei. Included with each taxon was a leaf sample of the internal standard of known genome size, Pisum sativum ‘Ctirad’ [2C = 8.76 pg (Bai et al., 2012; Greilhuber et al., 2007)]. Each sample was prepared by cochopping 1–2 cm² of tissue from both lilac and an internal standard (P. sativum ‘Ctirad’) with a razor in a polystyrene petri dish containing 400 μL of nuclei extraction buffer solution (Cystain Ultraviolet Precise P Nuclei Extraction Buffer; Sysmex, Görlitz, Germany). A buffer containing the chopped leaf tissue was passed through a 30-μm gauze filter (Partec Celltrics, Münster, Germany) into a 3.5-mL plastic tube (Sarstedt Ag & Co., Nümbrecht, Germany). Next, 1.6 mL of fluorochrome stain (4′,6-diamidino-2-phenylindole) was added to the nuclei suspension (Cystain Ultraviolet Precise P Staining Buffer; Partec). All samples were analyzed using a flow cytometer (CyFlow Ploidy Analyzer; Partec). A minimum of 3000 nuclei were analyzed per sample with an average CV for each fluorescence histogram under 10. Relative 2C genome size was calculated as:

View Expanded

Monoploid (1Cx) genome size was calculated using ploidy determined using root tip microscopy (described below) as:

View Expanded

The estimations of chromosome numbers in aneuploid seedlings were based on a genome size estimate of a single, theoretical chromosome (0.061 pg) calculated as:

View Expanded

Genome size variation in parent taxa and progeny resulting from different cytotype combinations were investigated using flow cytometry. Histogram figures from flow cytometry (.fcs files) were produced using open-source Cytospec software from Purdue University Cytometry Laboratories (2014).

Chromosome counts.

Chromosome counts were performed on several taxa representing four series: Syringa, Pubescentes, Villosae, and Ligustrina. An improved protocol for preparing root tips for chromosome counts (Lattier et al., 2017) was followed for lilac, with lilac root tips digested by enzyme for 2–3 h. Chromosomes were visualized (Axio imager.A1; Zeiss, Thornwood, NY) and imaged (AxioCam 105 Color; Zeiss) at different focal distances and layered to increase resolution for each photomicrograph. Focus-stacking was performed using the Auto Blend feature in Photoshop CC 2015.5.1 (Adobe Systems, San Jose, CA). A minimum of 15 resolved cells were investigated per taxa.

Pollen cytology.

From a previous study on cross-compatibility (Lattier and Contreras, 2017), a single seedling from an interploid cross was discovered to have a larger genome size than either parent suggesting that unreduced gametes were present in one parent. Therefore, both parents (S. vulgaris ‘President Grévy’ and S. vulgaris ‘Sensation’) as well as two other randomly selected taxa (S. vulgaris ‘Ludwig Spaeth’ and S. vulgaris ‘Miss Ellen Willmott’) were screened for unreduced pollen grains. At anthesis, fresh flowers were collected from each plant, and pollen was dusted onto microscope slides. Three slides were prepared and measured for each cultivar. To each slide, a single drop of 2% acetocarmine was added and then a cover slip was added. All slides were screened for stained pollen grains at a magnification of ×100 on a light microscope (Axio imager.A1). Single fields of view were randomly captured (AxioCam 105 Color) across each microscope slide and all stained pollen grains were measured using the line measurement tool in AxioVision SE64 4.9.1 (Zeiss). A total of 5381 pollen grains were measured in the four cultivars of S. vulgaris. Figures of reduced and unreduced pollen grains were focus-stacked to increase resolution using the Auto Blend feature of Photoshop CC 2015.5.1. To estimate unreduced pollen grains, the following equation was used:

View Expanded

As the volume of a sphere (pollen grain) doubles, the diameter increases by 26%. Therefore, any pollen grains with a diameter greater than 26% of the average for each taxa was scored as an unreduced pollen grain. Percent unreduced pollen grains were calculated as:

View Expanded

Statistical analysis.

Statistical analyses were performed using SAS Studio (version 3.6; SAS Institute, Cary, NC). Monoploid genome sizes were analyzed with the PROC GLM. Mean genome size averages for each individual taxon were separated using Tukey’s honestly significant difference test (α = 0.05). Genome size averages for each series were generated from an average of individual taxa means. Least squares means for each series were separated using a Tukey–Kramer test for unequal sample sizes (α = 0.05). Least squares means were also separated for pollen diameter measurements of four cultivars of S. vulgaris using a Tukey–Kramer test for unequal sample sizes (α = 0.05).

Genome sizes.

Holoploid 2C genome sizes ranged from 2.64 ± 0.08 pg in S. ×laciniata to 4.94 ± 0.06 pg in S. vulgaris ‘Aucubaefolia’. All 2C relative genome sizes were larger than the two previously reported genome sizes of European lilacs (Olszewska and Osiecka, 1984; Siljak-Yakovlev et al., 2010). Previous reports have shown similar variation due to different binding properties of fluorochrome stains (Lattier, 2016; Parris et al., 2010). Only three taxa of S. vulgaris, nested within series Syringa, had a 2C relative genome size larger than 4.00 pg, including S. vulgaris ‘Aucubaefolia’ (4.94 ± 0.06 pg), S. vulgaris ‘Agincourt Beauty’ (4.90 ± 0.03 pg), and S. vulgaris ‘President Grévy’ (4.85 ± 0.00 pg). Chromosome counts of S. vulgaris ‘Aucubaefolia’ revealed this group to be triploids (Fig. 1). The presence of triploids in our collection supports early reports of polyploid induction experiments and interploid hybridization (Fiala and Vrugtman, 2008) but surprisingly, no tetraploids were observed. All other root tip cells investigated were diploid, including S. ×hyacinthiflora ‘Maiden’s Blush’, S. ×hyacinthiflora ‘Old Glory’, S. ×prestoniae ‘Miss Canada’, S. reticulata, and S. pubescens Bloomerang^® Purple (Fig. 1). Chromosome counts in the current study provided no evidence for base chromosome number other than x = 23 (Fig. 1), in contrast to previous reports that varied from x = 22 to 24 (Darlington and Wylie, 1956).

Stained chromosomes in root tip cells of six accessions of Syringa. Photomicrographs viewed at ×1000 with scale bar at 1 μm: (A) Triploid (2n = 3x = 69) S. vulgaris ‘Aucubaefolia’; (B) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Maiden’s Blush’; (C) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Old Glory’; (D) diploid (2n = 2x = 46) S. ×prestoniae ‘Miss Canada’; (E) diploid (2n = 2x = 46) S. reticulata; (F) diploid (2n = 2x = 46) S. pubescens Bloomerang^® Purple.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

View Full Size

Stained chromosomes in root tip cells of six accessions of Syringa. Photomicrographs viewed at ×1000 with scale bar at 1 μm: (A) Triploid (2n = 3x = 69) S. vulgaris ‘Aucubaefolia’; (B) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Maiden’s Blush’; (C) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Old Glory’; (D) diploid (2n = 2x = 46) S. ×prestoniae ‘Miss Canada’; (E) diploid (2n = 2x = 46) S. reticulata; (F) diploid (2n = 2x = 46) S. pubescens Bloomerang^® Purple.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Stained chromosomes in root tip cells of six accessions of Syringa. Photomicrographs viewed at ×1000 with scale bar at 1 μm: (A) Triploid (2n = 3x = 69) S. vulgaris ‘Aucubaefolia’; (B) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Maiden’s Blush’; (C) diploid (2n = 2x = 46) S. ×hyacinthiflora ‘Old Glory’; (D) diploid (2n = 2x = 46) S. ×prestoniae ‘Miss Canada’; (E) diploid (2n = 2x = 46) S. reticulata; (F) diploid (2n = 2x = 46) S. pubescens Bloomerang^® Purple.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Significant differences were found among taxa for 1Cx genome size (P < 0.0001). Values ranged from 1.32 ± 0.04 pg in S. ×laciniata to 1.78 ± 0.05 pg in S. ×hyacinthiflora ‘Old Glory’ (Table 3). Series Syringa had a significantly larger average 1Cx genome size (1.68 ± 0.02 pg) than the other four series investigated (Table 3). There were no significant differences among series Pubescentes (1.47 ± 0.01 pg), Villosae (1.55 ± 0.02 pg), Ligustrina (1.49 ± 0.05 pg), and Pinnatifoliae (1.52 ± 0.02 pg) (Table 3).

Table 3.

Ploidy and relative genome size in Syringa determined using flow cytometry analysis of DAPI-stained nuclei with Pisum sativum ‘Ctirad’ (8.76 pg/2C) as an internal standard.

View Table

Within series Syringa, S. ×laciniata had a significantly smaller genome size compared with other tested taxa in series Syringa. No reports exist on the pedigree of S. ×laciniata (Table 1), although Fiala and Vrugtman (2008) hypothesize it to a cross of the Afghan lilac, S. protolaciniata, and another unknown parent. Syringa ×laciniata has a heavily dissected leaf, much like S. pinnatifolia, while S. protolaciniata produces heterophyllous leaves with margins varying from lobed to entire (Fiala and Vrugtman, 2008; Green, 1995). In addition, the only other heavily dissected lilac, S. pinnatifolia, has also proven to be the only species successfully used in interseries crosses (Pringle, 1981). If S. ×laciniata is from an interseries hybridization, then aneuploidy concomitant with wide hybridization could explain the significant reduction in genome size compared with other taxa in series Syringa. Further chromosome counts need to be performed on this hybrid.

Most 1Cx genome sizes within series Syringa were above 1.60 pg (Table 3). Although S. oblata is native to Asia and S. vulgaris is native to southeastern Europe, their different geographical origins are not reflected in significant genome size variation. Wild-type S. oblata and the white-flowered, S. oblata var. alba, both had a 1Cx genome size of 1.73 ± 0.03 pg (Table 3). The smallest and largest monoploid genome sizes in S. vulgaris were from two white, double-flowered taxa, S. vulgaris ‘Miss Ellen Willmott’ (1.61 ± 0.01 pg) and S. vulgaris ‘Madame Lemoine’ (1.76 ± 0.05 pg) (Table 3). Taxa representing hybrids between S. oblata and S. vulgaris had a monoploid genome size range from S. ×hyacinthiflora ‘Betsy Ross’ (1.70 ± 0.02 pg) to S. ×hyacinthiflora ‘Old Glory’ (1.78 ± 0.05 pg); however, there were no differences among the four hybrid cultivars included (Table 3). One additional interspecific hybrid, S. ×chinensis (1.74 ± 0.07 pg), representing a cross between S. protolaciniata and S. vulgaris was found to have a similar 1Cx genome size to S. vulgaris (Table 3).

Within series Pubescentes, most 1Cx genome sizes were below 1.50 pg and there were no significant differences among the 12 taxa included. The smallest genome size was S. pubescens Rhythm & Bloom^® (1.43 ± 0.01 pg), whereas the largest was in S. pubescens ssp. patula ‘Miss Kim’ (1.54 ± 0.01 pg). Within series Villosae, most 1Cx genome sizes were above 1.50 pg and ranged from S. tigerstedtii (1.38 ± 0.01 pg) to S. villosa ‘Aurea’ (1.62 ± 0.03 pg) (Table 3). Syringa villosa exhibited a 1Cx genome size similar to cultivars of S. ×prestoniae, which has S. villosa along with S. komarowii in its pedigree (Table 1). Syringa tigerstedtii had a significantly smaller 1Cx genome size compared with all other taxa except S. ×prestoniae ‘Donald Wyman’ (1.50 ± 0.00 pg) and ‘Redwine’ (1.53 ± 0.02 pg) (Table 3). However, no other significant differences were found throughout series Villosae, even though this diverse series was the most species-rich in the collection.

Within the tree lilacs (series Ligustrina) only two species, S. reticulata and S. pekinensis, and few cultivars exist. Green and Chang (1995) previously reported only one species, S. reticulata, with other species circumscribed to the rank of subspecies. The 1Cx genome sizes of S. pekinensis China Snow^® and Summer Charm^® were not significantly different. However, a significant difference was detected between S. pekinensis China Snow^® (1.41 ± 0.02 pg) and S. reticulata (1.59 ± 0.03 pg) (Table 3). Within the monotypic series Pinnatifoliae, S. pinnatifolia var. alashanensis had a 1Cx genome size of 1.52 ± 0.02 (Table 3).

Hybrid genome sizes.

Based on genome size estimates of parent taxa, hybrids from a previous cross-compatibility study on lilacs (Lattier and Contreras, 2017) were evaluated for genome size variation. As a seed parent, more than 800 flowers of the triploid S. vulgaris ‘President Grévy’ were pollinated in intraspecific crosses with four different diploids: S. vulgaris ‘Tiny Dancer’, S. vulgaris ‘Angel White’, S. vulgaris ‘President Lincoln’, and S. vulgaris ‘Sensation’ (Lattier and Contreras, 2017). Of the four pollen parents, only crosses with the picotee-flowered S. vulgaris ‘Sensation’ produced seed. From 240 pollinations, 107 seeds were obtained; however, only one seed germinated and produced a viable seedling (H2013-150-01). Flow cytometry revealed H2013-150-01 to be a polyploid/aneuploid with a 2C relative genome size of 5.65 ± 0.02 pg (Table 4). This genome size was significantly larger than any other seedling and is currently the largest reported in any lilac (Table 4) but is lower than expected for a tetraploid. In addition, the genome size of H2013-150-01 was larger than both its seed parent [S. vulgaris ‘President Grévy’ (2C = 4.85 ± 0.00 pg)] and its pollen parent [S. vulgaris ‘Sensation’ (2C = 3.33 ± 0.04 pg)]. In a combined analysis on the flow cytometer, histograms for each parent as well as the hybrid were clearly distinguishable (Fig. 2).

Table 4.

Comparison of hybrid genome size from interploid and intraploid crosses in Syringa.

View Table

Flow cytometry histogram of three taxa of Syringa vulgaris with an internal standard: (A) ‘Sensation’ (2C relative genome size = 3.33 pg); (B) ‘President Grévy’ (2C relative genome size = 4.85 pg); (C) hybrid (H2013–150–01) ‘President Grévy’ × ‘Sensation’ (2C relative genome size = 5.65 pg); (D) internal standard Pisum sativum ‘Ctirad’ (2C genome size = 8.76 pg).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

View Full Size

Flow cytometry histogram of three taxa of Syringa vulgaris with an internal standard: (A) ‘Sensation’ (2C relative genome size = 3.33 pg); (B) ‘President Grévy’ (2C relative genome size = 4.85 pg); (C) hybrid (H2013–150–01) ‘President Grévy’ × ‘Sensation’ (2C relative genome size = 5.65 pg); (D) internal standard Pisum sativum ‘Ctirad’ (2C genome size = 8.76 pg).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Flow cytometry histogram of three taxa of Syringa vulgaris with an internal standard: (A) ‘Sensation’ (2C relative genome size = 3.33 pg); (B) ‘President Grévy’ (2C relative genome size = 4.85 pg); (C) hybrid (H2013–150–01) ‘President Grévy’ × ‘Sensation’ (2C relative genome size = 5.65 pg); (D) internal standard Pisum sativum ‘Ctirad’ (2C genome size = 8.76 pg).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Other studies have yielded similar results in crosses between diploids and triploids. Seedlings with genomes sizes surpassing their parents have been reported for diploid-triploid crosses in other woody plants including Pyrus (Phillips et al., 2016), Ulmus (Santamour, 1971), and Populus (Harder et al., 1976; Wang et al., 2010). Similar results have been found in herbaceous taxa including Miscanthus sinensis (Rounsaville et al., 2011) Rudbeckia (Palmer et al., 2009), Lilium (Lim et al., 2003; Marasek-Ciolakowska et al., 2014), Musa (Osuji et al., 1997), and Cucumis sativus (Diao et al., 2009). Most studies proposed sexual polyploidization via the union of unreduced gametes from one or both parents as the likely cause of the large seedling genomes. In the current study, H2013-150-01 resulted from an unreduced gamete from the diploid S. vulgaris ‘Sensation’ (discussed below) and an aneuploid gamete from the triploid S. vulgaris ‘President Grévy’ that was 0.11 pg below the expected haploid (1.5x) value. However, S. vulgaris ‘Sensation’ used in 2x × 2x reciprocal crosses with S. vulgaris ‘Tiny Dancer’ failed to contribute unreduced gametes to produce triploid seedlings (Table 4).

As a pollen parent, the triploid S. vulgaris ‘President Grévy’ was used in crosses with S. vulgaris Blue Skies^®, S. vulgaris ‘President Lincoln’, and S. vulgaris ‘Sensation’ totaling 459 pollinations (Lattier and Contreras, 2017). Crosses with S. vulgaris Blue Skies^® produced the only viable seed with 12 seedlings recovered from 135 pollinations (Lattier and Contreras, 2017). With the exception of one seedling (H2014-033-11), all hybrid seedlings varied in 2C genome size between the two parents, S. vulgaris Blue Skies^® (3.44 ± 0.03 pg) and S. vulgaris ‘President Grévy’ (4.85 ± 0.00 pg) (Table 4). Relative 2C genome sizes of seedlings varied significantly from 3.41 ± 0.03 pg (H2014-033-11) to 4.35 ± 0.03 pg (H2014-033-01) (Table 4), the majority representing aneuploid genome sizes.

The estimations of chromosome numbers across aneuploid populations can be performed with knowledge of holoploid 2C genome sizes using Eq. [1]. Previous studies have produced models based on a holoploid genome size of a theoretical average, single chromosome based on parent genome sizes, and chromosome counts. Although some estimates of aneuploid chromosomes have been based solely on hypothetical chromosome size (Palmer et al., 2009), several studies have tested this model with root squashes and found most of their predictions to be concurrent with the true chromosome number or accurate within two to three chromosomes in Primula (Hayashi et al., 2009), Lilium (Lim et al., 2003), and Calluna (Behrend et al., 2015). Considering these previous studies and the relatively uniform chromosome size observed in lilac (Fig. 1), a simple linear model was used to predict chromosome number in the 2x × 3x aneuploid population (Fig. 3).

Linear model of lilac 2x × 3x aneuploid progeny with predicted chromosome number based on theoretical chromosome size of 0.061 pg [(4.85 − 3.45 pg)/23 chromosomes]. Parent taxa of aneuploid progeny: diploid female parent Syringa vulgaris ‘Blue Skies’ (2n = 2x = 46) and triploid male parent S. vulgaris ‘President Grévy’ (2n = 3x = 69). Linear model follows the formula: y = 16.224x − 9.7743.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

View Full Size

Linear model of lilac 2x × 3x aneuploid progeny with predicted chromosome number based on theoretical chromosome size of 0.061 pg [(4.85 − 3.45 pg)/23 chromosomes]. Parent taxa of aneuploid progeny: diploid female parent Syringa vulgaris ‘Blue Skies’ (2n = 2x = 46) and triploid male parent S. vulgaris ‘President Grévy’ (2n = 3x = 69). Linear model follows the formula: y = 16.224x − 9.7743.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Linear model of lilac 2x × 3x aneuploid progeny with predicted chromosome number based on theoretical chromosome size of 0.061 pg [(4.85 − 3.45 pg)/23 chromosomes]. Parent taxa of aneuploid progeny: diploid female parent Syringa vulgaris ‘Blue Skies’ (2n = 2x = 46) and triploid male parent S. vulgaris ‘President Grévy’ (2n = 3x = 69). Linear model follows the formula: y = 16.224x − 9.7743.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Based on a linear model with an average chromosome size of 0.06 pg (Fig. 3), the chromosome numbers in our aneuploid seedlings from S. vulgaris Blue Skies^® × S. vulgaris ‘President Grévy’ varied from 46 to 61 with an average of 54.3 ± 1.4 chromosomes. Chromosome numbers of triploid gametes can be deduced by subtracting the euploid chromosome number from the seedlings somatic chromosome numbers (Iorizzo et al., 2012). Assuming that S. vulgaris Blue Skies^® consistently contributed haploid gametes with 23 chromosomes, S. vulgaris ‘President Grévy’ produced a range of aneuploid pollen from 23 to 38 chromosomes to progeny from this cross. Previous research in other crops has shown that triploids produce a higher percentage unreduced and/or aneuploid gametes than their diploid or tetraploid counterparts (Burton and Husband, 2001; Herben et al., 2016; Ramsey and Schemske, 1998). Viable aneuploid gametes have been described in plants such as C. sativus (Diao et al., 2009), Brassica (Brassica; Lu and Kato, 2001), and Tulipa (Marasek-Ciolakowska et al., 2014), yet other plants only tolerate euploid gametes as in Vaccinium corymbosum (Vorsa and Ballington, 1991). The resulting aneuploids from the 2x × 3x lilac crosses were skewed slightly to the diploid cytotype compared with a theoretical bimodal distribution with an average of 57.5 chromosomes (Fig. 3). In the 3x × 2x cross, S. vulgaris ‘President Grévy’ × S. vulgaris ‘Sensation’, the triploid parent contributed 39 chromosomes (2n = 7) as a seed parent. It is unclear if this slightly higher contribution from the triploid is due to combining with an unreduced gamete, the direction of the cross, or chance. Brandham (1982) reported a greater prevalence of aneuploidy over the range between diploid and triploid when the latter are females—presumably associated with endosperm balance number (discussed in later sections). However, with only a single seedling it is impossible to draw conclusions.

Although our seedling cytotypes varied from a random distribution of aneuploid cytotypes, lilac aneuploid segregation conflicts with the limited number of similar studies on 2x x 3x crosses by being less concentrated at either euploid level (diploid or triploid). In lilies, these crosses resulted in all triploid or near-triploid seedlings derived from viable 2n gametes from triploid male parents (Marasek-Ciolakowska et al., 2014). In Tulipa, 2x × 3x crosses yielded a majority of diploid and near-diploid progeny with a small percent of near-triploids (Mizuochi et al., 2009). This same study found that the reciprocal cross in Tulipa yielded a binomial distribution of aneuploids, with the female triploid parent producing a wide range of fertile aneuploids (Mizuochi et al., 2009). Similar to lilies, 2x × 3x crosses in Allium schoenoprasum (Levan, 1936) and C. sativus (Diao et al., 2009) resulted in diploids or near-diploids with a small percent of near-triploids; the reciprocal crosses yielded a wider range of aneuploids. Brandham (1982) reviewed a number of studies on interploid crosses and found that with very few exceptions the triploid parent generally produced gametes that were either haploid or diploid based on the ploidy of the other parent. Populus was a notable exception from other examples given, namely a considerably higher chromosome number than other taxa discussed. Populus (x = 19) is similar to Syringa (x = 23) in chromosome number and both are almost certainly of polyploid origin. This high chromosome number indicates that there is redundancy present that likely allows the survival of aneuploid gametes produced by triploids that are inviable in taxa with fewer base chromosomes.

The importance of embryo and endosperm cytotypes often plays a role in seedling cytotype segregation and has been studied at length in diploid and triploid crosses of potato (Solanum). A 2:1 maternal to paternal endosperm balance ratio must be maintained for successful hybridization in potato, preventing 2x x 3x crosses and yielding progeny from 3x x 2x that are skewed to near-triploid cytotypes (Carputo, 1999; Iorizzo et al., 2012). The origin of S. vulgaris ‘President Grévy’ could play a role in its fertility as a triploid parent. Allotriploids are rarely used in breeding because of their difficulty in chromosome pairing during meiosis; however, autotriploids can overcome problems with meiotic pairing to produce haploid to triploid gametes (Brandham, 1982; Hayashi et al., 2009; Kato et al., 2001). The history of wide hybridization and polyploid induction in lilac leaves the question open to the origins of triploid cytotypes. In addition, it remains unclear if meiotic abnormalities in gamete formation, preferential fertilization, or preferential embryo/endosperm survival skewed the distribution of aneuploid cytotypes. Our results may simply be due to the small sampling population of aneuploid seedlings resulting from the 2x × 3x and 3x × 2x crosses.

Despite the numerous pollinations and few resulting seedlings, all aneuploid lilacs appear to be healthy and vigorous after their first 2 years of growth (J.D. Lattier, personal observation). This conflicts with some studies which reported that aneuploid seedlings from diploid-triploid crosses were nonviable past initial germination exhibiting abnormal, stunted growth (Behrend et al., 2015; Osuji et al., 1997), and sometimes reverting to euploids after more than a year (Behrend et al., 2015). Although our aneuploids appear to grow as vigorously as their diploid counterparts, female fertility and pollen viability in the aneuploid population has not been investigated as the plants have yet to reach maturity during this study. However, flow cytometry was performed on the aneuploid population for more than 2 years after germination. While it cannot be assured that this aneuploid series will not stabilize at a euploid level (diploid or triploid), the fact that these plants have maintained aneuploid chromosome compliments for more than 2 years contrasts with previous studies and suggests that they may be stable.

Unreduced pollen.

Stained pollen grains from four taxa of S. vulgaris were scored as viable and were measured for variability in diameter. Unstained pollen grains were negligible in all taxa, and pollen germination was not investigated. There were significant differences among taxa for pollen diameter (P < 0.0001) and every pairwise comparison between taxa was significant (P < 0.01). The largest average pollen grains were detected in S. vulgaris ‘Sensation’ (35.74 ± 0.16 μm), a measure likely overinflated by the presence of 8.5% unreduced pollen grains identified because of their increased volume (Fig. 4). This is the first report of unreduced (2n) pollen in lilac, but is not the first report in the Oleaceae. B-chromosomes and unreduced pollen have been reported in cultivars of Olea europaea (Sheidai et al., 2008). Syringa vulgaris ‘Sensation’ was also the only diploid observed to produce unreduced pollen, as S. vulgaris ‘Ludwig Spaeth’ and S. vulgaris ‘Miss Ellen Willmott’ produced only 1n pollen (Fig. 4).

Frequency distribution of viable pollen grain diameters of four cultivars of Syringa vulgaris. Regions to the right of asterisks were measured to be 26% larger than the mean and indicate unreduced gametes: (A) ‘Ludwig Spaeth’ (0% unreduced gametes); (B) ‘Sensation’ (8.5% unreduced gametes) [insert: unreduced (left) and reduced (right) pollen grains stained with 2% acetocarmine and viewed at ×630 magnification (scale bar = 10 μm)]; (C) ‘Miss Ellen Willmott’ (0% unreduced gametes); (D) ‘President Grévy’ (0.6% unreduced gametes).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

View Full Size

Frequency distribution of viable pollen grain diameters of four cultivars of Syringa vulgaris. Regions to the right of asterisks were measured to be 26% larger than the mean and indicate unreduced gametes: (A) ‘Ludwig Spaeth’ (0% unreduced gametes); (B) ‘Sensation’ (8.5% unreduced gametes) [insert: unreduced (left) and reduced (right) pollen grains stained with 2% acetocarmine and viewed at ×630 magnification (scale bar = 10 μm)]; (C) ‘Miss Ellen Willmott’ (0% unreduced gametes); (D) ‘President Grévy’ (0.6% unreduced gametes).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Frequency distribution of viable pollen grain diameters of four cultivars of Syringa vulgaris. Regions to the right of asterisks were measured to be 26% larger than the mean and indicate unreduced gametes: (A) ‘Ludwig Spaeth’ (0% unreduced gametes); (B) ‘Sensation’ (8.5% unreduced gametes) [insert: unreduced (left) and reduced (right) pollen grains stained with 2% acetocarmine and viewed at ×630 magnification (scale bar = 10 μm)]; (C) ‘Miss Ellen Willmott’ (0% unreduced gametes); (D) ‘President Grévy’ (0.6% unreduced gametes).

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 142, 5; 10.21273/JASHS04189-17

• Download Figure

Because of its low fertility in crosses with the triploid seed parent S. vulgaris ‘President Grévy’ and the presence of an inflated genome size in the single viable seedling recovered (H2013-150-01), an unreduced pollen grain from S. vulgaris ‘Sensation’ likely contributed to the production of this single polyploid/aneuploid progeny. It remains unclear whether the rare picotee flower mutation is in some way related to the production of an unreduced pollen grain, or if other diploids of S. vulgaris that were not included in the current study produce unreduced pollen at a similar rate as S. vulgaris ‘Sensation’. Since an unreduced pollen grain resulted in the only viable seedling between these two cultivars, this may indicate the utility of an increased ploidy level for improving cross-compatibility in lilac.

Pollen from the double-flowering, triploid S. vulgaris ‘President Grévy’ proved difficult to obtain as many flowers simply did not produce viable anthers. Unreduced pollen grains (0.6%) were detected in S. vulgaris ‘President Grévy’ out of 1689 grains measured. Although at a much lower percentage, S. vulgaris ‘President Grévy’ was the only parent besides S. vulgaris ‘Sensation’ to produce unreduced pollen in the current study, indicating some level of meiotic irregularities. Triploids are more likely to undergo irregularities during meiosis such as irregular chromosome pairing, supernumerary B chromosomes, laggard chromosomes, chromatin bridges, cytomixis, and out of plate chromosomes during metaphase I (Farco and Dematteis, 2014; Lavia et al., 2011). Triploids and resulting aneuploid progeny may prove to be sterile, yielding cultivars with reduced weediness and extended bloom times. Irregular meiosis during microspore development likely contributed to the poor performance of S. vulgaris ‘President Grévy’ as a seed parent and the subsequent aneuploidy seen in its viable seedlings (Fig. 3). Based on its aneuploid offspring when used as a male parent, S. vulgaris ‘President Grévy’ likely produces a range of aneuploid pollen. The average pollen grain diameter of S. vulgaris ‘President Grévy’ (35.28 ± 0.07 μm) was significantly larger than the two diploids that exhibited normal meiosis, S. vulgaris ‘Ludwig Spaeth’ (33.96 ± 0.05 μm) and S. vulgaris ‘Miss Ellen Willmott’ (34.32 ± 0.06 μm) from which no unreduced gametes were observed in a combined 2866 pollen grains (Fig. 4).

This study provides valuable information for future lilac breeding and informs a previous study on cross-compatibility among elite cultivars of lilac (Lattier and Contreras, 2017). In addition, this study contributes genome size and ploidy information to the growing database of angiosperm genome sizes, recommended by Galbraith et al. (2011). The discovery of three triploid lilacs, S. vulgaris ‘Aucubaefolia’, S. vulgaris ‘Agincourt Beauty’, and S. vulgaris ‘President Grévy’ lends evidence to previous reports of artificial tetraploid development and subsequent hybridization (Fiala and Vrugtman, 2008). However, no tetraploids were discovered among the research population. The discovery of high levels of aneuploidy in interploid hybrids indicates meiotic irregularities in pollen development of polyploid lilacs. Further cytological studies of pollen mother cells and meiotic analyses could contribute to understanding the complexities within developing gametes of taxa in the heavily hybridized series Syringa.

The development of an aneuploid series in 2x × 3x crosses provides an avenue to develop a model for cytotype prediction in seedlings of interploid lilac hybrids. Future efforts to confirm initial predictions of aneuploid chromosome numbers will include chromosome counts on parent taxa, chromosome counts on a subset of aneuploids, and repollination of the parent genotypes to increase the number of seedlings in the aneuploid population. Further, aneuploids can be highly variable in morphology, including reduced vigor and can have greatly reduced fertility. This may be a detriment in breeding most crops, but could be an avenue for ornamental breeders to recover more compact, longer-blooming, and sterile cultivars. As the aneuploid population matures, plants will be compared for differences in gross morphology, and flowers will be compared for pollen viability and female fertility. Reanalyzing the genome sizes of this population will be necessary over subsequent years in light of previous reports of euploidization of woody aneuploids (Behrend et al., 2015).

The discovery of unreduced pollen in S. vulgaris ‘Sensation’ and subsequent production of a seedling from a 3x × 2x cross with a larger genome than either parent, indicate that unreduced gametes or polyploidy may contribute to cross-compatibility in wide hybridization of lilac. Future work using high-throughput pollen screening by flow cytometry may reveal other cultivars with high levels of unreduced gametes. Wide hybridization with polyploids may reduce the impact of chromosome loss, which has been reported in previous cytological studies on lilac (Taylor, 1945) and the smallest genome recorded in the current study was a dissected-leaved, interspecific hybrid, S. ×laciniata. Identifying parents with unreduced pollen or generating autopolyploids in each lilac series may prove a valuable method for recovering viable progeny from wide hybridization in lilac. Interseries hybrids continue to be the most elusive quarry for lilac breeders, with only the pinnately compound S. pinnatifolia in the monophyletic series Pinnatifoliae proven to be a successful parent in crosses with S. oblata var. giraldii, S. vulgaris, S. ×laciniata, and S. ×hyacinthiflora (Pringle, 1981). Using S. pinnatifolia as well as an induced autopolyploids of cultivars proven to produce fruit and seed in interseries crosses (Lattier and Contreras, 2017) may spark a renaissance in the storied history of lilac breeding.