Cytogenetics, ploidy, and genome sizes of rose (Rosa spp.) cultivars and breeding lines

Rosa is one of the most economically important genera in ornamental horticulture. The genus' natural distribution is widespread throughout the northern hemisphere^[1]. While taxonomic classification within the genus is considered difficult with past efforts recognizing more than 300 species, Rosa is now generally considered to contain around 140 to 195 species^[2]. The currently accepted taxonomic system was devised by Rehder^[3] and revised by Wissemann^[4]. The system contains four subgenera: Hesperhodos (two species), Platyrhodon (one species), Hulthemia (one species), and Eurosa (also named Rosa; 180 species). Subgenus Eurosa is further divided into 10 sections. Despite the large number and wide distribution of species, the origin of modern garden roses (R. hybrida) is believed to have originated primarily from 10 species in the subgenus Eurosa (Table 1)^{[5, 6]}. Three more taxa believed to be involved in the origin of modern cultivars are likely hybrids, including R. damascena (R. gallica × R. moschata or R. phoenicea), R. alba (R. canina × R. gallica), and R. centifolia (R. ruba × R. moschata) × R. alba^[6]. Because of the long history of hybridization and interbreeding cultivated varieties, the cultivated rose gene pool is relatively isolated from the wild gene pool^[7].

The base monoploid chromosome number in rose is seven^[8]. Most R. hybrida-derived species are diploid (2n = 2x = 14) while R. gallica and R. foetida are tetraploid (2n = 4x = 28), and R. canina is pentaploid (2n = 5x = 35; Table 1). Although aneuploidy among garden roses may not be as rare as previously thought^{[9, 10]}, roses in the section Caninae provide the most instances of aneuploidy^[11]. The nuclear DNA 1Cx-value in subfamily Rosoideae is approximately 0.66 pg^[12], while the reported nuclear DNA 1Cx-values within Rosa range from 0.39 to 0.65 pg^[13] and most species involved in the origin of modern roses contain 1Cx-values around 0.5 to 0.6 pg (Table 1).

Meiosis in rose varies based on the ploidy as well as homology between chromosomes^[14]. Heteromorphic bivalents are often observed in cultivated roses, especially in tetraploid varieties^[10]. Both interstitial and terminal deletions alter chromosome structures, which may play a substantial role in the genetic diversity among cultivated varieties. The presence of both heteromorphic bivalents and univalents significantly affect fertility in tetraploid cultivated roses, with a higher frequency of univalents or heteromorphic bivalents being positively correlated with pollen sterility^[10].

While some modern rose cultivars are triploids or diploids, most elite cultivars are tetraploids. Ploidy differences between parents can affect both the cross-success rate and the resulting offspring's fertility. Unreduced gamete production has been observed at multiple ploidy levels in rose. Unreduced gametes in triploid genotypes led to the development of the first tetraploid Bourbon and Hybrid Tea roses^[15]. Pentaploid and hexaploid progeny has resulted from crosses between triploids and tetraploids, indicating unreduced gamete production in the triploid and tetraploid parent, respectively^[16]. Pollen size observations revealed likely unreduced gamete production in tetraploid cultivars 'Anna' and 'Sweet Promise'^[17]. Pollen observation in a different study revealed possible 2n gamete production in 20% of diploids and 4% of tetraploids^[18]. 2n pollen has also been observed in studies of dihaploid roses, potentially allowing breeders to more efficiently introgress wild, diploid germplasm into cultivated tetraploid gene pools^{[19, 20]}.

Species in the section Caninae are either tetraploid, pentaploid, or hexaploid and can exhibit an unusual form of meiosis as well as apomixis^[21]. During their unique meiosis, the chromosomes form seven bivalents, with the remainder forming univalents. During microsporogenesis, the univalents are eliminated, causing the pollen to only carry seven chromosomes originating from the bivalents (n = 7). In megasporogenesis, a combination of bivalents and univalents form the egg cell (n = 2n − 7). This results in matrilineal inheritance among offspring^[22]. Successful hybridization between Caninae species and species in the other sections depends on the direction of the cross as well as the number of chromosomes in each species^[23].

Genome size can give insights into genome evolution and taxonomic relationships^[24−28]. Also, genome size data can be used to estimate ploidy levels among closely related taxa when calibrated with cytogenetic work^[29−33]. Flow cytometry provides an accurate and efficient method to determine relative genome size and ploidy level of plant tissues^[34]. Previous reports on genome size in roses have focused primarily on species with little information for specific cultivars and breeding lines^{[13, 35−37]}.

Rosa hybrida is a diverse cultivated taxon with thousands of registered cultivars across dozens of horticultural classes^[1]. However, limited cytogenetic information and lack of ploidy level knowledge can hinder breeding efforts. The objectives of this study were to determine genome sizes and estimated ploidy for a broad collection of rose cultivars and breeding lines and evaluate flow cytometry as a reliable method for determining ploidy in rose.

Flow cytometry was performed on 154 rose cultivars and 20 rose breeding lines to determine relative 2C holoploid genome size, 1Cx monoploid genome size, and ploidy estimates (Table 2). Of the accessions screened, we found that 11 were diploids (2n = 2x = 14), 24 triploids (2n = 3x = 21), and 139 tetraploids (2n = 4x = 28) (Table 2 and Fig. 1).

Figure 1. 

Relative genome size for each of the three observed ploidy levels, for 174 rose species, cultivars, and breeding lines, determined by flow cytometry. Boxes display the lower and upper quartiles of the population with a horizontal line within the box showing the median. Vertical lines outside the boxes extend 1.5 times the interquartile range and data points outside of this range are considered potential outliers.

The mean 1Cx genome size for all cultivars was 0.55 ± 0.04 (SEM) pg with a range from 0.46 to 0.64 pg (Table 2). The 2C relative genome size range was 0.96–1.28 pg for diploids, 1.38–1.86 pg for triploids, and 1.87–2.50 pg for tetraploids, providing clear distinction between the genome size ranges of different euploid levels (Fig. 1). The mean relative genome sizes were 1.10 ± 0.13 pg, 1.59 ± 0.14 pg, and 2.20 ± 0.15 pg for diploids, triploids, and tetraploids, respectively.

Sixteen out of the 174 cultivars and breeding lines have had ploidy previously documented^{[38, 40]}. Of these 16 accessions, our data confirm the previously reported ploidy of 10 accessions and dispute the ploidy of the other six accessions, i.e., Knock Out®, Home Run®, Miracle on the Hudson®, Blushing Knock Out®, Double Knock Out®, and Oso Easy Paprika®. Each of these six cultivars was previously reported as triploid (D. Zlesak, Per. Comm.)^{[38, 40]}. However, the relative genome sizes determined by flow cytometry indicate that each cultivar is tetraploid (Table 2).

Chromosome counts were performed on seven of the accessions used for genome size determination including Knock Out®, Home Run®, Miracle on the Hudson®, Malvern Hills®, 'New Zealand', 'Dr. Huey', and Sunshine Daydream. Knock Out®^{[38, 40]}, Home Run®^{[38, 40]}, and Miracle on the Hudson® (D. Zlesak, Per. Communication) were chosen because of contradicting reports of ploidy between our flow cytometry estimates and previous reports. Malvern Hills® and New Zealand were chosen to calibrate relative genome size to ploidy for diploids and tetraploids, respectively (Table 2). 'Dr. Huey' and Sunshine Daydream were chosen to calibrate relative genome size to ploidy and help discern the transition point for relative genome size values for the largest triploid and the smallest tetraploid, respectively (Table 2). We found that Malvern Hills® was diploid, 'Dr. Huey' was triploid, and Knock Out®, Home Run®, Miracle on the Hudson®, 'New Zealand', and Sunshine Daydream were tetraploid (Fig. 2).

Figure 2. 

Ploidy determination by counting chromosomes of metaphase cells of seven rose cultivars. Chromosome images of (a) tetraploid Miracle on the Hudson^® (2n = 4x = 28), (b) tetraploid Home Run™ (2n = 4x = 28), (c) tetraploid New Zealand (2n = 4x = 28), (d) tetraploid Sunshine Daydream (2n = 4x = 28), (e) triploid 'Dr. Huey' (2n = 3x = 21), (f) tetraploid Knockout^® (2n = 4x = 28), and (g) diploid Malvern Hills^® (2n = 2x = 14). Bars = 10 µM.

Determination of ploidy using microscopy methods is tedious, challenging, and often difficult to obtain clear resolution of all individual chromosomes and accurate counts. Other indirect methods, such as pollen diameter and guard cell length^{[18, 40]}, can be correlated with ploidy but often lack the resolution to clearly differentiate between isoploids and anisoploids^[40]. Furthermore, guard cell length only determines the ploidy level of the L-I histogenic layer, but gives no information about the L-II layer, from which pollen and egg cells are derived^[41]. Although flow cytometry can be an effective method for estimating genome size and associated ploidy, this approach needs to be validated and calibrated among closely related species. Furthermore, variations in methodology including extraction buffers, fluorochrome stains, and internal standards can skew results from one study to another. For example, different fluorochrome stains may give slightly different estimates of genome size. Both propidium iodide (PI) and 4', 6-diamidino-2-phenylindole (DAPI) are effective and consistent for determining relative genome size and ploidy among closely related taxa^[33] when methods are standardized.

This study found that the base genome sizes (1Cx) were relatively conserved in the roses tested and that 2C genome sizes could be used to distinguish among diploid, triploid, and tetraploid cytotypes. The mean 1Cx genome size for all cultivars included in this study was 0.55 ± 0.04 pg with a range from 0.46 to 0.64 pg (Table 2). The average 1Cx genome size for these cultivars aligned well with the mean 1Cx genome size for the 10 rose species that contributed to modern cultivars estimated to be 0.57 ± 0.05 pg (Table 1) even though both Roberts et al.^[13] and Yokoya et al.^[37] used Petroselinum crispum as a calibration standard and PI as a fluorochrome stain.

The gap in genomes sizes between triploids and tetraploids was relatively small with 1.38–1.86 pg for triploids and 1.87–2.50 pg for tetraploids. Because of this small gap between genome sizes for triploids and tetraploids, chromosome counts were completed to accurately identify ploidy for cultivars near this transition point. Using microscopy, we found 'Dr. Huey' (2C = 1.86 pg) to be triploid, and Sunshine Daydream (2C = 1.91 pg) to be tetraploid. For roses with genome sizes near this transition point, chromosome counts may be required to confirm ploidy.

In cases where our ploidy estimates differed from prior reports, our reassessment of chromosome numbers using microscopy (e.g., Knock Out®, Home Run®, and Miracle on the Hudson®) substantiated our estimates derived from flow cytometry. These discrepancies may be the result of mislabeling or incorrect identification of specific clones, cytochimeras, or the inherent difficulty in clearly visualizing and counting plant chromosomes.

Accurate ploidy-level assessment is essential to breeding species with a variety of ploidy levels. A cultivar's ploidy level can affect its fertility and potential usefulness in a breeding program. For example, Hibiscus syriacus cultivars 'Minerva', 'Diane', and 'Helene' were initially reported to be sterile, triploid cultivars (allo-hexaploid) when they were released by the US National Arboretum^[42]. However, a recent flow cytometry analysis reported these cultivars found in nurseries to be fertile tetraploids^[42]. The origin of the ploidy discrepancy, whether the result of initial misidentification, propagation of a mislabeled stock plant with various ploidy levels, or the reversion of a cytochimera, is unclear^[42]. Flow cytometry is a useful, rapid tool for the determination of ploidy levels for the basis of breeding, ploidy manipulation, and uncovering past classification errors.

In addition to validating the application of flow cytometry for ploidy determination in roses, the results of this study provide an expanded database on the genome sizes and ploidy of 174 diverse rose cultivars and breeding lines. This research further contributes to the cytogenetic understanding of roses and a broader foundation for future breeding and improvement.

One hundred and fifty four rose cultivars and 20 rose breeding lines were included in this study (Table 2). Sixteen accessions with previously reported ploidy levels were included in this study to help calibrate genome size to ploidy level and to determine validity of previous reports^[38−40]. Emerging flower bud tissue of these cultivars and breeding lines were collected from rose gardens, public breeding programs, and nurseries (Table 2). If emerging floral bud tissue was not available, emerging new leaf tissue was collected. Leaf tissue was collected from P. sativum 'Ctirad' grown in the greenhouse and used as the internal standard (2C DNA content = 8.76 pg^[43]).

Flow cytometry

Relative genome size was determined using flow cytometry. Tissue samples were chopped using a double-sided razor blade in a 50 mm petri dish with 200 μL of extraction buffer (CyStain PI Absolute P; Sysmex Partec, Munster, Germany). P. sativum 'Ctirad' was chopped with Rosa tissue. The finely chopped tissues and buffer were poured through a 50 μm nylon-mesh filter into a test tube and stained with 800 μL 4',6-diamidino-2-phenylindole (DAPI) fluorochrome (CyStain ultraviolet Precise P Staining Buffer; Sysmex Partec). DAPI was used due to its speed, low toxicity, precision (low sample coefficient of variance), and repeatable results for determining ploidy^[37]. Samples were tested in a completely randomized design using a flow cytometer (Quantum P; Quantum Analysis, Munster, Germany). 2C DNA content was calculated using the formula: DNA content of the internal standard × (mean fluorescence of sample / mean fluorescence of the internal standard). At least two replicates were used for each cultivar. The relationship between ploidy and genomic size was calibrated using roses in the same section with known ploidy as a reference^[13].

Chromosome counts

Chromosome counts were conducted to confirm the relationship between genome size and ploidy^[44]. Roots or young shoot tips of 0.5 cm in length were collected and placed in the pre-fixative solution containing 2 mM 8-hydroxyquinoline and 70 mg·L⁻¹ cycloheximide in microcentrifuge vials. Vials were stored in the dark for 3 h at ambient room temperature and subsequently placed in a dark refrigerator at approximately 4 °C for an additional hour. Following the pre-fixative treatment, tissue was removed and rinsed thoroughly in deionized water. The tissue was then placed in Farmer's fixative solution containing 95% ethanol and glacial acetic acid (3:1 v/v) and left at 4 °C overnight or until tissues could be further processed.

Fixed tissues were removed from the fixative solution and rinsed with deionized water for 1 min per time in triplicate. After washing, the youngest, most tender tissues were placed on a microscope slide with 30 µL of a cell wall digestive enzyme mixture of 6% pectinase (Sigma Chemical; St. Louis, MO, USA) and 6% cellulase (Onozuka R-10; Yakult Honsha, Tokyo, Japan) in 75 mM KCl buffer (pH 4.0). Prepared slides with the cell wall digestive enzyme buffer were placed in an enclosed container with moistened paper towels and incubated at 37 °C between 3 h and overnight depending on the cultivar. The next day, the buffer was diluted with 1−3 drops deionized distilled water and any excess buffer was removed with a Kimwipe®. Then, two drops of a modified fixative buffer containing 95% methanol and glacial acetic acid (3:1, v/v) were added to the sample on each slide. Tissue was broken up using a small spatula, spreading the macerated tissue evenly on the slides in a circular pattern. Additional drops of the modified fixative buffer were added when needed.

Once the tissue was spread out, the extra fixative buffer was burned off and the slides were incubated at 37 °C overnight to be fully dried, then stored at room temperature. Dried slides were stained using Vectashield® Antifade Mounting Medium with (Vector Laboratories; Burlingame, CA, USA) with DAPI. Chromosomes of metaphase cells were counted using a compound microscope (Axio Imager.A2, Zeiss; Oberkochen, Germany).