Materials and Methods

Plant material.

Germplasm was collected through various means including whole plants from nurseries, cuttings from gardens and arboreta, and seeds from gardens around the world participating in Index Seminum (Table 1). The latter formed the bulk of our collection. Plants were maintained in containers or in field plots at Oregon State University and all were assigned accession numbers.

Table 1.

Source and collection information for 67 Cotoneaster accessions.

Genome sizing.

Holoploid (2C) genome sizes were determined by flow cytometry (CyFlow PA; Partec, Münster, Germany) and comparison of mean relative fluorescence of the sample against an internal standard, Pisum sativum ‘Ctirad’, with a known genome size of 8.76 pg (Greilhuber et al., 2007). Two different fluorochromes were used. A total of 67 accessions representing 65 species were sampled across the two subgenera and all 11 sections using flow cytometric analysis of nuclei stained with DAPI (CyStain ultraviolet Precise P; Partec). A subset of 17 taxa was also prepared with PI (CyStain ultraviolet Absolute P; Partec). Nuclei of each sample and our standard were concurrently prepared by chopping with a double-sided razor blade in extraction buffer (CyStain ultraviolet Precise P Nuclei Extraction Buffer; Partec) for ≈90 s before being filtered through a 50-μm nylon mesh filter (CellTrics^®; Partec) and stained with either fluorochrome. For PI-stained samples, RNase was included to ensure staining of DNA exclusively. Cotoneaster samples were prepared using 4 cm of rapidly growing terminal stem tissue including vegetative buds and we used 1 cm² of fresh pea leaf tissue. DAPI-stained samples were incubated in darkness for 5 to 10 min before analysis and PI-stained samples were incubated in darkness for at least 30 min on ice. Three replicates of each accession were prepared for both DAPI and PI. A minimum of 3000 particles were analyzed for each sample. Sample runs were rejected if the coefficient of variation (cv) was greater than 7%.

Holoploid DNA content (2C) was calculated as DNA content of standard × (mean fluorescence value of sample/mean fluorescence of standard). Then, analysis of variance and means separation by Tukey’s honestly significant difference was performed, with ploidy levels inferred from mean separation. Monoploid genome sizes were calculated by dividing each sample’s 2C genome size by inferred ploidy. Analysis of variance was then conducted on monoploid genome size by accession and then taxonomic division to test for significant differences among subgenera or sections; mean separation was performed as described above when the model was significant (α = 0.05). For the subset of 17 accessions that were examined with both DAPI and PI, genome size estimates using each fluorochrome were compared using a t test (α = 0.05) separately for each accession. Base pair composition was calculated as AT% = AT% for internal standard × {[(fluorescence sample, DAPI)/(fluorescence internal standard, PI/fluorescence sample, PI)]^{(1/binding length)}} (Godelle et al., 1993). AT% of P. sativum ‘Ctirad’ is 61.50% and has a binding length of ≈3.5 bp (Meister and Barrow, 2007).

Cytology.

Chromosomes were counted for seven species, five of which were included in the genome sizing, and two additional species Cotoneaster hebephyllus and Cotoneaster poluninii. Somatic cells were collected from actively growing root tips, which grew freely from the bottom of their containers into sand. Roots were treated with 0.003 m 8-hydroxyquinoline for 2 h at 4 °C and fixed in Carnoy’s solution [6 absolute ethanol : 3 chloroform : 1 glacial acetic acid (by volume)] overnight. Root tips were stored in 70% ethanol at 4 °C until prepared for chromosome counts (Goldblatt and Gentry, 1979). Chromosomes were examined by root tip squashes with modified carbol fuchsin, at ×63 to ×100 magnification (Axio imager.A1; Zeiss, Thornwood, NY) and images were collected using a monochromatic CCD camera (AxioCam MRm; Zeiss). A minimum of three cells were counted for each species.

Results and Discussion

Relative 2C genome sizes for 67 accessions determined by flow cytometry with DAPI ranged from 1.52 pg (Cotoneaster frigidus) to 4.71 pg (Cotoneaster kweitschoviensis) (Table 2). The 2C genome sizes showed marked divisions, which were used to assign ploidy level. Of the 67 accessions, 10 (15%) were diploids (2n = 2x = 34) with 2C values ranging from 1.52 to 1.73 pg, 5 (9%) were triploids (2n = 3x = 51) with 2C values ranging from 2.14 to 2.58 pg, 51 (76%) were tetraploids (2n = 4x = 68) with 2C values ranging from 2.88 to 3.34 pg, and 1 (1.5%) accession was hexaploid (2n = 6x = 102) with a 2C value of 4.71 pg (Table 2). Regarding relative percentage of ploidy levels, our findings generally agree with previous reports including Kroon (1975) who reported 3 diploid, 3 triploid, and 23 tetraploid species among the 28 studied. Ours is the first report of ploidy estimation for 13 of the species. However, one diploid species, Cotoneaster juratana, is not a valid species and we have been unable to confirm its identity. One assumption was a mislabeling of Cotoneaster juranus, which has only been reported as a tetraploid (Fryer and Hylmö, 2009). Unfortunately, morphology of the plant labeled as C. juratana did not match the description of C. juranus and has since been lost from our collection. We sampled two accessions of Cotoneaster adpressus and Cotoneaster acutifolius and both showed ploidy series. Cotoneaster adpressus 10-0157 was tetraploid and the other, ‘Tom Thumb’, was triploid. Previous reports for C. adpressus indicated that it was either diploid or triploid (Sax, 1954; Zeilinga, 1964), making ours the first report of tetraploidy for the species. Cotoneaseter acutifolius 09–0047 was diploid and C. acutifolius 10–0126 was triploid. Our results are consistent with Sax (1954) who reported diploids, triploids, and tetraploids for C. acutifolius. This is in contrast to Zeilinga (1964) who reported that all species in their study were diploid or tetraploid and only cultivars were found to be triploid. This provides evidence that ‘Tom Thumb’ and C. acutifolius 10-0126 are hybrids. Cotoneaster ×watereri 10-106 was a triploid in contrast to previous reports indicating this hybrid species to be diploid (Fryer and Hylmö, 2009). This species arose as a hybrid of C. frigidus × Cotoneaster salicifolius, which generally are both regarded as diploids, though Sax (1954) also reported triploidy in C. salicifolius. It is unclear how this triploid accession arose, but possibilities include a previously unknown tetraploid cytotype of one of the parent species, unreduced gamete production in one of the progenitors, an apomictic seedling, or even a self-pollinated seedling from an unreported triploid cytotype of C. salicifolius, as triploids often are regarded as facultative apomicts. Another alternative is that the parent plant from which we received seed was diploid and it was accidentally pollinated by a tetraploid, resulting in triploid progeny. There is some evidence against obligate apomixis in C. ×watereri, as Fryer and Hylmö (2009) describe five cultivars, a level of diversity that would not be expected if the species was an obligate apomict. Both Zeilinga (1964) and Kroon (1975) reported that all species were diploid or tetraploid and triploids only arise through hybridization. To determine if that assertion is supported by our study, a detailed morphological or molecular investigation of species and hybrids would be required, which is beyond the scope of the current research. In five other species, our data indicated a different cytotype than reported by Fryer and Hylmö (2009). These include Cotoneaster ganghobaensis, C. kweitschoviensis, Cotoneaster milkedandaensis, Cotoneaster niger, and Cotoneaster sikagensis. The differing cytotypes we report do not suggest consistent error in either direction (over or underestimates) and include two 4x that were reportedly 3x, one 6x previously reported as 4x, one 4x previously reported as 2x, and one 2x previously reported as 4x (Table 2). A larger screening with more accessions representing each species would likely uncover more ploidy series within other species as well as identifying more examples for which new findings would differ from previous reports. Discrepancies in ploidy between previous papers and our findings are due to testing different sources or accessions of material, some of which may have arisen from hybridization.

Table 2.

Holoploid (2C) and monoploid (1Cx) genome sizes of 67 accessions of Cotoneaster determined by flow cytometric analysis of nuclei stained with 4′,6-diamidino-2-phenylindole (DAPI) using Pisum sativum ‘Ctirad’ (2C = 8.76 pg) as an internal standard, inferred ploidy levels based on flow cytometry data, previously reported ploidy, and mean 1Cx value of 10 taxonomic sections studied.

When looking across taxonomic divisions, variation in ploidy level was observed in the subgenera and within many sections (Table 2). The 1Cx genome size ranged from 0.71 pg of DNA for C. acutifolius (10-0126) to 0.87 pg of DNA for Cotoneaster dammeri. When compared across all taxa, monoploid genome sizes showed a detectable difference (P < 0.05); however, when these differences were examined by taxonomic division, they were insignificant (P ≥ 0.05). Since we did not observe differences between taxonomic groups, we infer that our mean monoploid genome size can be used to calibrate our ploidy estimations in future investigations. In addition, when compared with other genera like Magnolia, which has observable differences in monoploid genome size between taxonomic sections (Parris et al., 2010), Cotoneaster has undergone relatively little divergence in chromosome size.

DAPI genome size estimates generally were larger than PI with differences ranging from 0.03 to 0.14 pg (Table 3). Fifteen of 17 accessions were different (P < 0.05) using respective fluorochromes and four accessions had higher significance (P < 0.0001). Doležel et al. (1992) reported significant differences (P < 0.01) for five genera in four families including four genera for which DAPI overestimated genome size from 11% to 30% compared with PI and one species for which DAPI underestimated genome size by nearly 27%. Our calculations show that bp composition in Cotoneaster ranges from 58.4% to 60.8% AT (Table 3), which agrees with estimates of 59.2% to 61.1% AT for three genera from Rosaceae (Meister and Barrow, 2002). We found no relationship between ploidy level (thus genome size) and AT%, a finding consistent with previous reports in Rosaceae (Meister and Barrow, 2002). Interestingly, we observed a trend that the lower the AT%, the greater the overestimation of genome size using DAPI compared with PI. Even though Parris et al. (2010) found that DAPI underestimated genome size for Magnolia, both of our studies show the same trend: increasing AT% results in a lower DAPI estimate compared with PI. Although the trends were similar, we observed overestimation using DAPI, whereas Parris et al. (2010) observed underestimation, even though AT% was similar, albeit higher in their study. A possible source of variation in our study was using different types of tissue in our samples (young stems and vegetative buds) and pea standard (young, expanded leaves). It is possible this resulted in variation in chromatin structure, which would affect the amount of unstainable DNA (Doležel et al., 1992).

Table 3.

Monoploid (1Cx) genome sizes determined by analysis of nuclei stained with 4′,6-diamidino-2-phenylindole (DAPI) or propidium iodide (PI) using Pisum sativum ‘Ctirad’ (2C = 8.76 pg) as an internal standard, the difference between genome size estimates between fluorochromes and bp composition of 17 taxa of Cotoneaster.

Although our choice of fluorochrome influenced genome size estimates, most differences were not large enough to affect ploidy estimation. However, there could be some confusion for several species included in our analysis. For instance, Cotoneaster boisianus showed a 0.14 pg difference in monoploid genome size estimate between fluorochromes. Using the estimate from PI of 2.72, the inferred ploidy level would calculate as 3.5x. Without additional information provided from cytological analysis, accurate ploidy assignment may be challenging and sample readings could erroneously be interpreted as aneuploid. DAPI is less expensive, uses less toxic compounds, and often resulted in lower cv for mean nuclei fluorescence than PI for Cotoneaster. Overall, we consider DAPI acceptable for our purposes in an applied breeding program.

From our chromosome counts in seven species, we found one diploid accession and six tetraploid accessions (4x Cotoneaster vandelarii not shown; Fig. 1). These ploidy estimates matched our results from flow cytometry for the five species that were examined by both methods. Overall, when our results were compared with literature for both cytology and flow cytometry, most were in agreement. Where there is conflict in the reports, the conflict may be from the way that the ploidy reports have been generated. Sax (1954) conducted estimations via chromosome counting in pollen mother cells and Zeilinga (1964) found several conflicting reports when root tips were examined. Zeilinga suggested that with polyploids, the chromosome pairing observed in pollen mother cells was crowded and led to confusion in counting. With a relatively high base chromosome number and common occurrence of polyploidy (102 chromosomes in hexaploids), it is possible that previous chromosome counts included errors. Determining ploidy level by counting chromosomes in Cotoneaster was time consuming and difficult, while flow cytometry proved to be much faster and accurate.

• Download Figure
• Download figure as PowerPoint slide

Photomicrographs of six Cotoneaster accessions including (A) a diploid (2n = 2x = 34) Cotoneaster henryanus, and (B) five tetraploids (2n = 4x = 68) Cotoneaster buxifolius, (C) Cotoneaster vandelarii, (D) Cotoneaster hebephyllus, (E) Cotoneaster lidjiangensis, and (F) Cotoneaster poluninii. Cells were prepared from root tips and chromosomes were stained using modified carbol fuchsin.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 141, 5; 10.21273/JASHS03776-16

Thus far, taxonomic organization in Cotoneaster has largely failed to incorporate molecular data and has relied on morphology and species provenance, although a report by Bartish et al. (2001) using randomly amplified polymorphic DNA supported the recognition of subgenera. Because of the number of species within the genus and the difficulty in organizing Cotoneaster, we hope fundamental information on genome size, ploidy level, and bp composition may give others insight to the relationship among the species. However, the results of this study do not show a relationship of chromosome size to current taxonomic organization, as monoploid genome sizes did not appear to be linked to taxonomic division.

This work may be useful to breeders for predicting success of interspecific hybridization and fertility of F₁ populations. Along with other factors, similarity in chromosome size contributes to functional meiosis and bivalent pairing between genomes. In this study, monoploid genome sizes varied up to 23% among species. In Rudbeckia, a hybrid was recovered when there was a difference in genome size of >300% (Palmer et al., 2009). The much smaller range in Cotoneaster suggests variation in monoploid genome size, thus chromosome size, is not expected to hinder interspecific hybridization. Furthermore, we have successfully performed several intersubgeneric and interploidy crosses including Cotoneaster ×suecicus ‘Coral Beauty’ (2x, 2C = 1.53 pg, subgenus Chaenopetalum) × Cotoneaster splendens (4x, 2C = 3.03 pg, subgenus Cotoneaster) that resulted in a triploid hybrid, which was confirmed using flow cytometry (2C = 2.42 pg), thus supporting broad compatibility in the genus. Breeding programs should conduct ploidy analysis for each accession included in a germplasm collection without relying on previous reports. We expect the broader the survey among species from various sources, the more examples of ploidy series will be found. Nevertheless, we have demonstrated the utility of apomictic polyploids for breeding when used as pollen parents.

Taxonomy of Cotoneaster is challenging due to morphological similarities, the propensity for hybridization, and the presence of apomixis. The tangled issue of separating and correctly identifying these species is emphasized by Dirr (2009) who stated, “Cotoneaster identification is not easy with 400 species, many possibly the result of hybridization, subsequent apomixis, which leads to microspecies that essentially reproduce vegetatively via seed.” Our study relied heavily on material obtained through Index Seminum and here we present the material labeled as we received it. Because of the presence of apomixis, particularly among tetraploids, we have strong confidence in the identification of most species presented. It is worth noting that materials in our study were not wild collected as one may expect in a classical floristic study but our findings should be quite relevant to applied plant breeders or others studying cultivated material of Cotoneaster or Maloideae.