Proceedings of the American Society for Horticultural Science 78: 572-579. (1961)

Analysis of Nine Crosses Between Diploid Rosa Species1 (PDF)

By WALTER H. LEWIS, Stephen F. Austin Stale College, Nacogdoches, Texas,
and ROBERT E. BASYE, A. and M. College of Texas, College Station


1 Received for publication May 8, 1961. The senior author appreciates assistance from a grant-in-aid, The Society of the Sigma Xi, and from the Undergraduate Research Participation Program (NSF G-12059).

THAT many species of Rosa, even from distinct phylads of the genus, can be crossed and will produce F1 hybrids is obvious from the extensive lists reported by Ratsek et al. (10, 11). A number of these hybrids were studied for pollen viability by Flory (5), who found that 8 intrasectional hybrids (including the sections Cinnamomeae-Carolinae as one) between diploid species averaged 87.4% normal pollen, while 4 intersectional hybrids averaged only 29.1% viable pollen. The hybrids of R. multiflora x rugosa are not included in the above numbers for Flory's report of 96.0% viable pollen conflicts with the results of his other intersectional crosses and with those of Kihara (6) for the same cross. Kihara found no viable pollen. Both R. multiflora and R. rugosa are heteromorphic species and possibly genetic differences of the parental forms used by these researchers are responsible for their opposite results. Apart from the work of Flory, however, studies of diploid Rosa hybrids are very limited. Wulff (14) reported the intersectional hybrid R. wichuraiana x rugosa with only 16.0% normal pollen and Erlanson (2) recorded a high frequency of viable pollen for 3 intrasectional crosses (78.9%). Erlanson's results with R. foliolosa x rugosa are omitted for as she stated, the hybrids are scarcely hardy in Michigan and that may account for the 100% defective pollen.

Meiotic behavior of the diploid hybrids is much less known. Erlanson (2) and Lewis (9) found normal bivalent formation for the intrasectional hybrids R. rugosa x blanda and R. blanda x woodsii, respectively. Kihara (6) reported the intersectional R. multiflora x rugosa hybrids with mostly 5-7 bivalents at metaphase I followed by normal chromosome behavior throughout the course of meiosis. On the other hand, Wulff (13, 14) observed highly abnormal meiotic figures, including translocations, univalents, and multivalents, for R. wichuraiana x rugosa.

In our study of diploid species from 5 sections, data have been accumulated for: pollination attempts and per cent successful, per cent embryo development, number of F1 hybrids produced, and the meiotic behavior, defective pollen, and fruit set of the hybrids. From these results, certain suggestions regarding intersectional relationships and hybrid sterility are given.


Nine parental species with their sections (classification following Crépin (1) as modified for North America (7)), sources, and chromosome numbers are listed in Table 1. The sections Laevigatae and Microphyllae are monotypic, Bracteatae is ditypic, Synstylae is a large diploid group characteristically Asian, and Cinnamomeae, also large, is heteromorphic and with well-established polyploidy.

Table 1.—Section, source, and chromosome number of Rosa parents.

Taxon Section Source Chromosome
Number (2n)
R. bracteata Wendl. BRACTEATAE South of Millican, Brazos Co., Texas 14
R. brunonii Lindl. SYNSTYLAE Univ. of California Bot. Gard., Berkeley 14
R. foliolosa Nutt. CINNAMOMEAE Hillier & Sons, Winchester, U. K. 14
R. laevigata Michx. LAEVIGATAE 20 miles east of Opelousas, St. Landry Par., Louisiana 14
R. moschata Herrm.var. abyssinica (Lindl.) Rehd.a SYNSTYLAE Univ. of Santa Clara, California 14
R. roxburghii Tratt. MICROPHYLLAE Bobbink & Atkins, East Rutherford,N.J. 14
R. rugosa Thunb. CINNAMOMEAE T. Hilling & Co.; B. R. Cant & Sons, Colchester, U K. 14
R. setigera Michx.var. tomentosa T. & G. f. serena (Palmer & Steyerm.) Fern. a SYNSTYLAE Arnold Arboretum, Jamaica Plains, N.Y. 14
R. soulieana Crép. SYNSTYLAE Bobbink & Atkins, East Rutherford. N.J. 14
aHereafter only the binomials R. moschata and R. setigera will be used.  

All parents have 2n = 14 chromosomes and, except for R. rugosa and R. setigera, have rarely been used in experimental hybridizing.

The usual plant breeding techniques were followed to obtain the F1: removal of anthers from immature buds of the female, pollen from the carefully brushed on the stigmas, and bagging of hypanthia to prevent contamination. After harvesting, the achenes were stratified in sand for about 6 weeks outside, washed and placed in petri dishes containing bacto-agar at 0° ± 3°C. As the seeds germinated they were removed from the agar plates to 2" pots in cold frames and later placed under standard field conditions at College Station, Texas. Using this procedure, percentages of germination were high and seeds germinated rapidly, not only seeds from diploid hybrids, but also from polyploid hybrids, Rosa species, and many commercial varieties studied simultaneously.

Achenes containing embryos were estimated using the water-sink test, i.e., sinking achenes contain embryos, those floating lack embryos. Exceptions are R. foliolosa, R. rugosa, and their hybrids (cf. R. palustris Marsh. 4) whose achenes invariably float.

During 1960, flower buds from the F1 hybrids were fixed in 4 parts chloroform, 3 parts absolute ethyl alcohol, and 1 part glacial acetic acid, and stored for varying lengths of time, up to 6 months, at 0 3C. When studied, the buds were hydrolyzed in N HCl at 60C for 5-8 minutes, returned to fixative for 5 minutes, and the anthers were squashed in 1% acetic-orcein. Meiotic figures were drawn with the aid of a camera lucida at X2300. At the same time pollen was removed and stained in 1% acetic-orcein. Normal appearing pollen was considered viable; shriveled, non-staining pollen, defective. Sampling at random, observations, and records were made on 500-1000 pollen grains from each hybrid with pollen being secured from 2 plants of each respective cross.

Herbarium specimens of parents and hybrids are filed in the Stephen F. Austin State College Herbarium (ASTC).


Numbers of pollinations and those successful for 9 crosses are listed in Table 2. Frequencies of success ranged from 60% to 100% with an average of 82%. No significant difference was noted between the number of hips formed by the intrasectional and the intersectional crosses. The production of achenes with embryos was also high excepting those of R. bracteata x rugosa (15%). The number of hybrids raised to maturity is also given in Table 2. The proportion of germinations and seedlings produced was good, but because of limitations in space and time, all plants were not grown to maturity. This success did not materialize, however, for the F1 of R. bracteata x laevigata, even though 80% of the achenes from this cross possessed embryos and germination was good. Of the 500-600 seedlings transplanted to pots over a 2 year period from a total of 40 hips, few developed beyond the second node and all perished within 2 months of transplanting. Since the plants were placed in cold frames with seedlings from other crosses, a high proportion of which always survived, various diseases and environmental factors do not appear to have been responsible for their complete loss.

Table 2.—Pollinations made between diploid Rosa species, together with
data on resulting hips, achenes with embryos, and F1 hybrids.

Hybrids Pollinations Per cent
achenes with
F1 hybridsa
Number Per cent
Intrasectional crosses
foliolosa x rugosa 14 79 b 6
setigera x brunonii 30 89 75 70
Intersectional crosses
moschata x rugosa 2 100 60 2
bracteata x foliolosa 24 71 60 10
bracteata x roxburghii 10 60 60 25
bracteata x rugosa 31 71 15 8
bracteata x laevigata 48 83 80 0
soulieana x bracteata 19 100 90 6
soulieana x laevigata 13 85 80 8
aOnly those plants grown to maturity and maintained for 5 years.
bR. foliolosa and R. rugosa do not respond to the "sink test" for achenes.

Pollen mother cells (PMC's) of hybrids of the remaining 8 crosses were examined at meiosis (Table 3). One or two typical plants were used in each case and the results were so consistent that additional analysis seemed unnecessary. All phases from diakinesis to anaphase II were observed with high frequencies of normal meiotic figures (Figs. 1-2, 4-5, 7-8). Regular meiosis averaged 98.3% for the 2 intrasectional hybrids and only slightly lower, 93.6%, for the 6 intersectional crosses. From a total of 345 PMC's studies, only 19 or 5.5% exhibited irregular meiosis (Figs. 3, 6). In these cells, the abnormalities were all of a simple type, viz., univalent formation with rarely less than 6 bivalents or more than 2 univalents, and at meiosis II, 6 and 8 chromosomal divisions. Since 13 PMC's were noted at diakinesis and at metaphase I, cells of this tpe could exhibit normal behavior at later stages giving the expected number of 7 chromosomes at each pole. Conceivably, therefore, the 5.5% meiotic irregularity noted for the intersectional hybrids might be decreased by as much as one-half if these meioses could have been traced to their conclusion.

Table 3

Hybrids Number
of plants
Per cent
Intrasectional crosses
foliolosa x rugosa 1 28 0
setigera x brunonii 2 29 3.4
Intersectional crosses
moschata x rugosa 2 87 3.4
bracteata x foliolosa 2 63 4.8
bracteata x roxburghii 1 46 6.5
bracteata x rugosa 2 49 10.2
soulieana x bracteata 2 13 0
soulieana x laevigata 1 30 13.3
Totals 13 345  


Fig. 1-8. Meiotic chromosomes of Rosa hybrids. Fig. 1. R. setigera var. tomentosa f. serena x brunonii, 7+7 (Lewis 5477). Fig. 2-3 R. moschata var. abyssinica x rugosa, 711, 611 + 21 (Lewis 5472). Fig. 4. R. bracteata x foliolosa, 7+7 (Lewis 5476). Fig. 5. R. bracteata x rugosa, 611 + 21 (Lewis 5471). Fig. 7. R. soulieana x bracteata, 711 (Lewis 5225). Fig. 8. R. soulieana x laevigata, 7+7 (Lewis 5473).

Table 4.—Defective pollen and fruit set for F1 Rosa hybrids.

Hybrids Per cent defective
pollen, 1960
of plants
Fruit sets
Intrasectional crosses      
foliolosa x rugosa a 16.7 6 equal parents
setigera x brunonii 96.0 70 numerous
Intersectional crosses      
moschata x rugosa 71.1 2 few, 1 yr.
bracteata x foliolosa 94.1 10 numerous
bracteata x roxburghii 99.8 5 5 hips, 1 yr.
bracteata x rugosa 78.5 8 none
soulieana x bracteata 91 .0 6 1 achene
soulieana x laevigata 59.9 8 1 achene
a lncluding the reciprocal cross.

In contrast to the low percentage of irregular meiosis and to normal tetrad formation was the high proportion of defective pollen for the intersectional hybrids (Table 4). The average percentage of defective pollen, when these 6 hybrids were considered collectively, was 82.4. However, the amount of defective pollen for the intrasectional hybrids of R. foliolosa x rugosa was only 16.7% which is comparable to the results of Erlanson (2) and Flory (5). That no rule can be made regarding all species in the same section is apparent from the results of R. setigera x brunonii hybrids (Synstylae) for defective pollen averaged 96%. However, the female parent in this cross is male sterile, a characteristic of many R. setigera individuals (8), and it appears that the gene(s) controlling this sterility has been inherited by the F1 as a dominant factor.

The number of hybrids maintained for a 5 year period and their fruit sets from open pollinations are recorded in Table 4. Of the 6 intersectional crosses only the hybrids R. bracteata x foliolosa had any degree of successful fruit production. Attempts are now being made to obtain the F2, but the paucity of achenes makes this difficult. As anticipated from the morphological grouping of the parents and the viability of the hybrid pollen, those individuals of R. foliolosa x rugosa had fruit sets about equal to the parents. The R. setigera x brunonii hybrids also had successful fruit sets, somewhat less than the parents, as expected on the basis of taxonomic position, but not from the results of high pollen sterility. This adds evidence, however, to the control of male sterility by one or more genes, which is not a barrier to fertility when cross pollination is possible.


One of the most significant results from the data given was the inability of the seedlings from the R. bracteata x laevigata cross to develop to maturity. The death of all during the second month suggests the absence of some factor, perhaps an enzyme system, required by the hybrids after reaching a definite stage of development. This is a striking example of incompatibility not yet described in Rosa.

The hybrids of R. foliolosa x rugosa grow vigorously in Texas, and the difference between our results of only 16.7% defective pollen and the 100% sterility noted by Erlanson (2) in Michigan, where the hybrid is not hardy, is as she noted one of environment, although genetic differences of the parents cannot be overruled.

Otherwise the small amounts of defective pollen for intrasectional hybrids given by Erlanson and by Flory (5) agree with our results. The cross of R. setigera x brunonii is exceptional with 96% defective pollen, but, as explained, male sterility is inherited from the female parent. Amounts of defective pollen for the intersectional crosses by Kihara (6), Flory (5), except R. multiflora x rugosa, and Wulff (14) are all in accord with our high percentages for 6 crosses.

Normal meiotic behavior for the intrasectional crosses reported by Erlanson (2) and by Lewis (9) is supported by observations, in the present work, of almost complete regularity of the PMC meiotic divisions in the hybrids R. foliolosa x rugosa and R. setigera x brunonii. High percentages of normal meiotic figures, totaling 93.6%, were also found for the microsporogenesis of the 6 F1 intersectional crosses. These data are similar to those of Kihara (6) for R. multiflora x rugosa, but are unlike the extensive irregularities noted by Wulff (14) for R. wichuraiana x rugosa. Both intersectional crosses have in common the male parent (R. rugosa) and the sections Synstylae and Cinnamomeae as does our cross of R. moschata x rugosa. Wulff concluded that pollen defectiveness is directly correlated with the meiotic abnormalities of his hybrid, but such a conclusion is impossible for Kihara's and our results since most microspores obtain the normal compliment of 7 chromosomes. Disintegration is probably caused by incomplete genotypes due to crossing-over with "incompletely" homologous chromosomes during prophase I. The amount of defective pollen would depend on the significance of the genic deletions affecting metabolic imbalances in the microspore. This would vary according to the frequency of crossing-over, the distinctiveness of the parental genotypes, and the importance to cellular metabolism of the genes omitted. Very possibly other factors Introduced by crossing-over, such as the gene sequence and the presence of genes incompatible to the microspore, might affect their normal development. This may also be true for other hybrids from distantly related taxa that exhibit near normalcy at meiosis but produce only small fractions of viable gametophytes

It is probably safe to assume for our hybrids that megasporogenesis was equally regular, producing high proportions of megaspores with 7 chromosomes. For the intrasectional hybrids, these remained functional and produced successful fruit sets, but because of incomplete or imbalanced genotypes, the normal development of the intersectional female gametophytes was prevented in most cases and few fruit developed. Relatively successful fruit set for R. bracteata x foliolosa hybrids suggests genetic affinities between these species not known to the other intersectional hybrids.

Data from these crosses indicate other relationships, qualified, of course, by the extent of our results. 1. The small per cent of defective pollen and excellent fruit formation for the hybrids R. foliolosa x rugosa add evidence for the treatment of the sections Cinnamomeae (R. rugosa) and Carolinae (R. foliolosa) as one, a position arrived at on morphological grounds by Lewis (7) and others. 2. Mature hybrids of R. bracteata x laevigata were not formed, yet each parent crossed with R. soulieana (Synstylae) to produce adult hybrids, suggesting that the Bracteatae and Laevigatae are more distantly related to each other than they are to R. soulieana. Even though R. soulieana x laevigata hybrids have higher percentages of irregular meioses than do those of R. soulieana x bracteata, the reverse is true for the amount of defective pollen. From a total of 14 hybrids, however, only 2 achenes were produced, so these hybrids are about equally sterile. In other words, R. soulieana representing the Synstylae has about the same affinity to the Bracteatae as it does to the Laevigatae. 3. Rosa rugosa (Cinnamomeae) is about equally related to R. moschata (Synstylae) and to the Bracteatae, but not closely, because of the high pollen sterility and small fruit set by their hybrids. This relationship is as near as R. soulieana is to the Bracteatae and Laevigatae. Based on a more successful fruit set, a second Cinnamomeae species, R. foliolosa, is more compatible with the Bracteatae than is R. rugosa. Since R. foliolosa and R. rugosa cross readily, there is obviously a great range of sterility vs. fertility in the Cinnamomeae as would be anticipated in this large and heteromorphic group. 4. The Microphyllae-Bracteatae hybrids gave similar results for meiosis, pollen sterility, and fruit set, as did other intersectional hybrids. In the most current classification of Rosa (12), R. roxburghii (Microphyllae) is distinguished as subgenus Platyrhodon and has variously been considered as a section or as a separate genus. Our evidence supports the sectional rank for this taxon and not that of subgenus or genus, which clearly express a more distant relationship for a more distinct phylad.


Nine Rosa species representing 5 sections of the genus were crossed. The study of their hybrids revealed: the presence of a dominant gene(s) producing sterility in R. setigera and its hybrids; the absence of a certain factor(s), possibly an enzyme(s), causing the death of hybrids between R. bracteata x laevigata after limited growth; and the high sterility of most intersectional hybrids not directly controlled by meiotic abnormalities. Rather, genic deletions and possibly genic positions and incompatibilities introduced by crossing-over producing metabolic imbalances are the probable causes of sterility. The results also show that the sections Bracteatae and Laevigatae are more closely (and equally) allied to the Synstylae R. soulieana than to one another. The Cinnamomeae R. rugosa is about as related to R. moschata (Synstylae) and the Bracteatae as the Synstylae is to the Bracteatae and the Laevigatae, although the Cinnamomeae R. foliolosa has greater affinities to the Bracteatae. Data for a consectional treatment of the Cinnamomeae and the Carolinae and for maintaining the Microphyllae at sectional rank are also given.


  1. CRÉPIN, F. 1889. Sketch of a new classification of roses. Jour. Roy. Hort. Soc. 11:217-228.
  2. ERLANSON, E. W. 1931. Sterility in wild roses and in some species hybrids. Genetics. 16:75-96.
  3. —————. 1934. Pollen analysis for rose-breeders. Amer. Rose Ann. 19:63-68.
  4. ERLANSON MACFARLANE, E. W. 1957. American rose species and their hybrids at the botanical gardens of the University of Michigan. Amer. Rose Ann. 42:90-98.
  5. FLORY, JR., W. S. 1950. Pollen conditions in some species and hybrids of Rosa with a consideration of associated phylogenetic factors. Va. Jour. Sci. 1 (NS):11-59.
  6. KIHARA, H. 1944. Chromosome conjugation in a rose hybrid, Rosa multiflora x R. rugosa. Jap. Jour. Genet. 20:55-58.
  7. LEWIS, W. H. 1957. An introduction to the genus Rosa with special reference to R. acicularis. Va. Jour. Sci. 8 (NS):197-202.
  8. —————. 1959. A monograph of the genus Rosa in North America. III. R. setigera. Southwest. Nat. 3(1958):154-l74.
  9. —————. 1962. Monograph of Rosa in North America. IV. R. x dulcissima. Brittonia 14 (in press).
  10. RATSEK, J. C.. S. H. YARNELL, and W. S. FLORY, JR. 1939. Crossing relations of some diploid species of roses. Proc. Amer. Soc. Hort. Sci. 37:983-992.
  11. RATSEK, J. C., W. S. FLORY, JR., and S. H. YARNELL. 1940. Crossing relations of some diploid and polyploid species of roses. Proc. Amer. Soc. Hort. Sci. 38:637-654.
  12. REHDER, A. 1940. Manual of Cultivated Trees and Shrubs. MacMillan Co., New York.
  13. WULFF, H. D. 1951. Rosa kordesii, eine neue amphidiploide Rose. Züchter 21:123-132.
  14. —————. 1953. Max Graf and its progeny, with special reference to Rosa kordesii. Amer. Rose Ann. 38:111-122.