Euphytica 26: 697-701 (1977)
Agriculture Canada, Ottawa Research Station. Ottawa, Ontario, Canada Received 28 February 1977


The purpose of this study was to determine at what stage of seedling development the flowering attributes in winterhardy R. rugosa hybrids could be assessed and if these attributes could be improved through breeding. The analysis of flowering was based on the duration of the flowering period and on amount of flowers produced.

Most seedlings from rugosa hybrids could be assessed at an age of 3 years. The duration of the flowering period and flower productivity were not correlated. Several seedlings were improved in both attributes over the longer flowering and more floriferous parent.


Winterhardiness and recurrent bloom are seldom found together in roses. Most garden cultivars flower recurrently but they lack winterhardiness. In Canada, cultivars of the hybrid tea, floribunda and grandiflora classes can be grown without winter coverage in relatively small areas at the coast of British Columbia and in the most southerly part of Ontario. Only the hardiest garden cultivars survive under coverage at Ottawa. Generally, species and cultivars that survive the winters at Ottawa without coverage flower during a short time in June only. However, the winterhardy R. rugosa and certain of its hybrids flower repeatedly but more or less sporadically. This study was carried out to determine whether the flowering characters of winterhardy rugosa cultivars could be improved through breeding and at what stage of seedling development selection for recurrent bloom could be carried out.


The diploid (2n = 14) R. rugosa cultivars Schneezwerg and Frau Dagmar Hartopp were used as parents. 'Schneezwerg' derived presumably from a cross of R. rugosa x Polyantha hybrid (MCFARLAND, 1969). The parentage of 'Frau Dagmar Hartopp' is not known. Both cultivars flower recurrently but 'Schneezwerg' is more florifcrous than 'Frau Dagmar Hartopp'.

Two experiments with the offspring from 'Schneezwerg' x 'Frau Dagmar Hartopp' and reciprocal will be described. The first experiment was carried out to compare the flowering characters in 2 and 3 year old seedlings. The second experiment was carried out to compare the difference in flowering characters among 11 sister seedlings and between seedlings and parents. The 11 seedlings, including the parents were evaluated for a period of 4 years, beginning with the 2nd year of growth.

The evaluation of flowering characters was based on the duration of the flowering period and on the quantity of flowers produced. The duration of the flowering period was determined by Weekly examinations during a 16 week period from June till September. In 1969 and 1970, the records were taken during 13 and 14 weeks only (Fig. 1). Flower production was estimated as of surface area of each flowering seedling covered by bloom. The codes for coverage were: 1 for 1-5%, 2 for 6-12%, 3 for 13-25%, 4 for 26 -5O%, 5 for 51-75% and 6 for 76-100%. The date for the first record in June was determined by the earliest flowering roses of the plantation. The amount of winterkill was estimated as'/ of dead wood in relation to the total wood of each shrub. The codes for % winterkill were: 1 for 0%, 2 for 1-5%, 3 for 6-12%, 4 for 13-25%, 5 for 26-50%, 6 for 51-75% and 7 for 76-1O0°. Statistical analyses were carried out with the codes for coverage and winterkill but for tabulation the data were converted into percentages (Table 2).

Fig. 1. Amount of flowers in coverage during 16 weeks from June till September for seedlings from 'Schneezwerg' x 'Frau Dagmar Hartopp'.


Relation between prolonged flowering and age of seedling. While it was possible to select for a prolonged flowering period in 2-year old seedlings, a much higher proportion of the population showed this attribute in the third year (Table 1). Selection in 2-year old seedlings would mean a loss, since 84%, of the 3-year old seedlings flowered between 7 and 16 weeks versus 62.7% of the 2-year old seedlings. The amount of non-flowering 2-year old seedlings was 5.3%. All 3-year old seedlings flowered.

Difference between reciprocal crosses. The test of heterogeneity for the duration of the flowering period of 3-year old seedlings in reciprocal crosses was not significant. The x2 test for heterogeneity was 2.53, a value which would occur in 12% of repeated trials. This test could not be carried out for flower production since 96% of the observations were clustered in one class.

Correlation of flowering attributes. The length of the flowering period was not correlated with flower production. The correlation coefficients were -- 0.0142 and 0.0453 for each of the reciprocal crosses. These values are not significant.

Flowering period. The length of the flowering period was improved through breeding. The 11 seedlings from the cross 'Schneezwerg' x 'Frau Dagmar Hartopp' showed significant differences among seedlings and between seedlings and parents (Table 2). The average flowering period for 'Schneezwerg' was 13.5 weeks and for 'Frau Dagmar Hartopp' it was 10.0 weeks. Multiple comparison showed an overlap between parental means (Table 2) but comparison between parental means only showed a significant difference at P = 0.05. Three seedlings flowered longer than 'Schneezwerg', the parent with the longer flowering period (Table 2).

Table I. Flowering period, from June till September, of seedlings from
'Schneezwerg' x 'Frau Dagmar Hartopp' and reciprocal during the second and third year of growth.

of weeks
2 years old seedlings 3 years old seedlings
  number % number %
0 4 5.3 0 0.0
1-6 24 32.0 12 16.0
7-11 36 48.0 37 49.3
12-16 11 14.7 26 34.7
Total 7 75 100.0 75 100.0

Table 2. Flowering characters and winterhardiness levels of F1 seedlings from 'Schneezwerg' x 'Frau Dagmar Hartopp'

Seedling Flowering period in weeks
from June till Sept.
Amount of flowers in 
% coverage
Winterkill in %
  mean range1 test mean range1 test mean range1 test
1 14.8 a 16 a 4 a
11 14.5 a 18 a 10 a
7 14.0 a 15 ab 4 a
3 13.8 ab 17 a 8 a
6 13.0 ab il bcd 8 a
2 12.8 ab 9 def 4 a
5 12.5 ab 10 de 8 a
9 12.5 ab 8 efg 2 a
4 12.0 ab 9 def 2 a
8 11.8 ab 7 fg 7 a
10 7.3 c 6 g 4 a
'Schneezwerg' 13.5 ab 13 abc 8 a
'Frau Dagmar Hartopp' 10.0 bc 10 de 3 a

 1 Means followed by the same letter(s) are not significantly different at P <= 0.05 as determined by Duncan's Multiple Range Test.

These findings are not in disagreement with previous reports that recurrent bloom in roses is a recessive character with simple Mendelian inheritance (HURST, 1941; SEMENIUK, 1971). In spite of the large variation in the duration of the flowering period, non-recurrent types were not obtained. This would be expected since both parents flower recurrently. Presumably, the heritable variation of the flowering period is due to modifying factors.

Flower productivity. Flower productivity was improved through breeding. Significant differences were found between means of the li sister seedlings and between seedlings and at least one parent (Table 2). The parental means for percent coverage was 13.3 for 'Schneezwerg' and 9.9 for 'Frau Dagmar Hartopp'. The difference between parental means is significant at P = 0.05. Even so, the correlation coefficients between flowering attributes were not significant, the 3 seedlings with the longest flowering period produced also more flowers (Table 2).

The variation in flower production between seedlings No. I and No. 5 in the period from 1967 until 1971 is shown in Fig. 1. Seedling No. 1 produced several, in some years overlapping, peaks of flowers. Seedling No. 5 produced one peak of flowers. In the spring of 1969 the plants were moved to a different location. This was the one year in which the two seedlings did not differ significantly in flower production. The gaps in 1969 and 1970 are due to missing records.

Hardiness level. The average amount of winterkill between the 11 seedlings varied from 2 to 10%. These differences were not significant (Table 2). This was expected because the offspring from winterhardy parents is also hardy.

Rating. The system of rating, described under Methods, permits the differentiation of relatively small amounts of winter injuries and flower production. It will be noted that the same system is used for both attributes except that no winter injury is noted as 1 but a rating of 1 for flower production stands for 1 to 5% coverage. In our weekly examinations we simply leave a blank for non-flowering seedlings. This system of rating is a valuable aid in plant selection (SVEJDA, 1971). The hardiest seedlings show little variation in the yearly fluctuations of winterkill. Individuals with two or more peaks of bloom have a higher mean for flower production over the entire flowering period than individuals with one peak of bloom. Breeding efforts should be directed towards the production of well adapted plants with several, preferably overlapping, peaks of bloom during a prolonged flowering season.