American Rose Annual 48: 188-193 (1963)
A Self-Pollination Mechanism And Other Items In Rose Species
Dr. Eileen W. Erlanson Macfarlane

Cytologist, Cancer Center, Ohio State University
Columbus, Ohio

WHEN THE final report on my extensive wild rose collection at the Botanical Gardens of the University of Michigan was prepared for the 1961 American Rose Annual (7), several interesting features were noticed. One was a possible self-pollination mechanism which was absent in the more primitive species. Other floral characteristics are now reported, some of which may be due to a single gene (Mendelian unit of heredity) and therefore of value in studies of heredity.


     In many wild roses the petals are past their prime by the third day and the flowers no longer close at night. The stamens may then rise up and their delicate filaments curve over to bring the pollen-bearing anthers onto the receptive stigmatic heads of the ovaries. It was found in one of my earliest studies (1) that the pollen cells mature before the embryo sac and egg cells. Most of the pollen is shed when the petals first open and insects will carry some of it to flowers opening for the second day, which may then be ready for cross-fertilization. However, a few pollen grains remain on the shriveled anthers on the third day and the stigmas should still be receptive. The rising and incurving of the stamens could very well provide self-fertilization for any remaining eggs not already cross-fertilized. My experiments showed that the roses are self-fertile.
     This feature may have been observed elsewhere, but if so the report has not come to my attention. In Table I are listed 35 species from all over the world. Among those with strongly rising, incurved stamens are Rosa setigera (the Prairie Rose), and others that are considered by botanists to be more specialized on the whole, than the species in the third column, whose stamens show no self-propulsion. The latter, negative group, includes the far northern species Rosa acicularis and Rosa pimpinellifolia and related forms, as well as the primitive Rosa roxburghii and some dog roses (Caninae section). Seven species in which the character is weakly developed are listed in the second column.

Self-pollinating mechanism in Rosa—35 species & 17 hybrids

Selfing-strongly + Selfing-weakly ± Non-selfing (negative)
foliolosa arkansana acicularis
michiganensis*      blanda x arkansana      x californica
     x blanda carolina      suffulta x acicularis
multiflora cinnamomea alpina
nitida      alpina x cinnamomea amblyotis
palustris (most) housei* blanda
rubiginosa** (some) jundzillii**      x arkansana
Rubrifolia (some) nutkana (few)      x woodsii
rudiuscula      tomentosa x nutkana canina
rugosa      californica x nutkana centifolia
setigera++ palustris (few) dumetorum**
virginiana (most)      x blanda engelmannii
     x foliolosa      x nutkana
     x virginiana hugonis
rubiginosa (most) macrophylla
rubrifolia (most) nutkana group
suffulta      x spaldingii
     x carolina      villosa x nutkana
virginiana (few) omeiensis var. pteracantha
     blanda x virginiana pimpinellifolia**
ratonensis *
woodsii group
(fendleri, Macounii et al.)

*Erlanson, Eileen W., 1928. Ten new American species and varieties of Rosa. Rhodora 30: 109-121.
**Meikle, Katherine, 1958. Modern Roses V.
R. dumetorum = canina var. dumetorum
R. jundzillii = marginata (Caninae)
R. pimpinellifolia = spinosissima var. pimpinellifolia
R. rubiginosa = eglanteria (Caninae) Most botanists use the specific names given in the Table.

CybeRose note: Heslop Harrison (1921) wrote, "I therefore got up earlier, at 4 a.m. (GMT), before any insects were at work, when I found that even then every newly expanded R. pimpinellifolia had its stigmas powdered with pollen from its own overarching stamens." Perhaps the self-pollination mechanism is influenced by environmental conditions.

**The self-pollinating mechanism and precocious petals are well shown in Roses by Eric Bois, Nelson, 1962. Plate 7. ('American Pillar,' a setigera hybrid)
In some instances there is variation among individuals of the same species; these are then listed in each of two columns. The variation involves: strong and weak selfing or weak and non-selfing (negative) never strong and negative. There are plants in the Ann Arbor garden of Rosa blanda, R. palustris, R. woodsii and other species from all parts of their geographic ranges (2). The variation in the selfing mechanism may be due to hybridization which occurs in nature (3) e.g., R. palustris (+) with R. blanda (-), R. virginiana (+) with R. carolina (weak). Some of my species hybrids are listed in Table I and the selfing character seems to be weakly dominant. My R. michiganensis (1) which is positive, crossed with negative R. blanda gave offspring that were both positive and negative for selfing mechanism.**


     Mendelian inheritance in roses has been reported by Dr. Lammerts (4, 5), who has also irradiated buds in an attempt to unlock genetic secrets quickly (6).
     The opportunity to observe many rose species growing under the same conditions, as at Ann Arbor, is infrequent. The following observations are offered in the hope that they may help a little to solve some of the problems of the hereditary make-up of Rosa.


     1. Tiered. In wild roses the number of stamens ranges from a few (50-125) to many (150 to about 250) and is an important characteristic in classification (3). Usually the outer stamens have the longest filaments and the innermost very short filaments. The rest are gradually reduced in length from outside inwards, a feature that may be designated as "tiered stamens." It is well developed in the many stamens of Rosa rugosa, also among the few stamens of R. blanda, carolina, cinnamomea, nutkana, suffulta and woodsii. Exceptionally the filaments are all of about equal length, or "non-tiered," the anthers then form a crown or corona effect which is very attractive. This is well shown in our plants of two Chinese species of the section Cinnamomeae: R. amblyotis C. A. Meyer and R. macrophylla Lindley, which have many stamens. Few stamens, non-tiered, were observed in plants of R. acicularis from Alberta and from Northern Ontario.
     2. Color. Stamen filaments are usually nearly colorless but occasionally pink. In a R. pimpinellifolia from Scotland, both filaments and the disc to which they are attached, are reddish pink.


     1. Color. The fruits of a rose are the nutlets enclosed in the hollow stem tip which softens and turns red as the hip ("berry"). The nutlets are usually straw colored except in the R. spinosissima-pimpinellifolia group where they vary from pink (in a cultivated escapee with double flowers) to deep maroon wine red in plants from Lake Baikal, U.S.S.R. R. hugonis also has dark maroon or mahogany red nutlets. The related Chinese R. omeiensis var. pteracantha was the only other species in which pink nutlets were found.
     The hips and their short peduncles (stems) are black or mahogany red in the species that have colored nutlets.
     2. Size. Nutlets are relatively large (4-5 mm.), also very hairy in R. acicularis and R. canina. Few in the former, in the base of the hip and many in the latter, attached all over the inner walls of the hip. The small hips of R. multiflora contain a few large, almost naked nutlets. There are many small nutlets in the specialized American species R. palustris and R. setigera.
     3. Exposed or Naked Nutlets-Gymnocarpic. The delicate species of the forests of western North America, R. gymnocarpa, is so named because after flowering it sheds the sepals along with the stigmas on their styles and the contracted tip of the hip. This character was observed to be weakly developed in a R. nutkana var. muriculata from Seattle and in its hybrid with R. acicularis; also in two other species hybrids involving plants from the West: R. engelmannii x acicularis and R. californica var. myriantha x nutkana var. spaldingii. These species were listed 1934 (3).


     Color. The receptive stigmas form a cushion in the center of most roses and, like the filaments and disc, are usually colorless. These are pink in a few plants of R. acicularis, palustris and virginiana, also in my species R. michiganensis (1). Some of the offspring from palustris x blanda and a virginiana x carolina plant had pink stigmatic cushions, indicating that the pigmentation may be dominant.


     1. Color. Petal color and its inheritance is very important for the commercial rose breeder (5, 6) but has not been studied extensively in simple diploid species with only fourteen chromosomes (4). Among the American rose species there are diploids, tetraploids and hexaploids, most of which show the magenta petal shade, that becomes bluer as it fades. Magenta has been reported as genetically dominant by Dr. Lammerts who named the controlling gene M (5). The recessive form, a nonfading currant red (gene m), has not so far been found in our wild roses. We have several examples of clear rose reds and pinks in the dog roses (Caninae): R. canina, rubiginosa, rubrifolia; also in R. kurdestana, received from the Arnold Arboretum, which seems to be a form of R. canina var. dumetorum. Rose red and clear pinks are found in the American species: R. carolina, nitida, virginiana and setigera.
     The attractive salmon or coral pink tint (light scarlet of The British Horticultural Colour Chart) (6) is characteristic of many plants of R. arkansana including R. suffulta. It is also found in dog roses. R. suffulta var. valida Erlanson (1) crossed with R. carolina gave a brilliant light scarlet color. Roses with non-magenta petals are listed in Table II, also some minor petal characters.

Petal Characteristics—Rosa

Color - Rose pinks (R)
           Light Scarlet (S)
Precocious opening
alpina S      canina (few)
arkansana S      multiflora
     blanda x arkansana S      rubiginosa (50%)
canina R-S      rubrifolia
carolina R      setigera
housei S      alpina x cinnamomea (some)
kurdestana R Cordate (broader than long)
nitida R      nutkana
rubrifolia R Narrow
setigera R      rubiginosa
suffulta S Toothed (small point top center)
     x acicularis S      michiganensis
     x arkansana S      rubiginosa
     x carolina S      blanda x virginiana
virginiana R      tomentosa x nutkana
     blanda x virginiana R Open flat (in sun)
webbiana S      arkansana
Pure White      ratonensis
     dumetorum      nutkana
     foliolosa var. alba  
     blanda var. hispida
          (from Manitoba)

     2. Size. In some dog roses the petals unfold when they are very small and increase at least twice in size during the flowering period. This character I have called "precocious petals."
     3. Doubling. In our collection several to many petals are present only in some plants of R. gallica, cinnamomea, pimpinellifolia and rugosa. Petaloids, or small extra petals are rare in American species. One plant of R. acicularis (no. 13321) from Edmonton, Alberta, and a Rosa blanda from Grand Marais, Michigan, each has from one to four large petaloids on many of the flowers to give the effect of from six to ten petals. Only three of my many species hybrids showed petaloids (7).


     1. Stamens that rise and curve over the stigmas and could provide self-pollination, are reported in several rose species. In others the mechanism is weak, while it is absent in many of the more primitive wild roses.
     2. Other minor characteristics of stamens, stigmas and nutlets are described which may have simple inheritance.
     3. The occurrence of rose pink and salmon pink in species is recorded, as well as other colors and minor petal characteristics.
     Post Script. As reported in 1961 the majority of the 530 wild rose plants (7) were moved to the new Botanical Gardens of the University of Michigan. They were visited early in June 1962 and the above observations were reconfirmed. Seventy plants succumbed to the winter. There remain 150 accessions labelled with registration numbers, with a total of 275 plants. This includes 29 hybrid cultures with 66 hybrid plants.


  1. Erlanson, Eileen W., 1928. Ten new American species and varieties of Rosa. Rhodora 30: 109-121
  2. ——————— 1929. Cytological conditions and evidences for hybridity in North American wild roses. Bot. Gazette 87: 443-506
  3. ——————— 1934. Experimental data for a revision of the North American wild roses. Bot. Gazette 96: 197-259
  4. Lammerts, Walter E., 1945. The scientific basis of rose breeding. American Rose Annual 1945, 30: 71-79
  5. —————————1960. Inheritance of magenta red color in roses. American Rose Annual 1960, 45: 119-125
  6. —————————1961. Neutron-induced variation of roses. J. Amer. Scientific Affiliation
  7. Macfarlane, E. W. Erlanson, 1961. Report on hybrids of American wild rose species at the Botanical Gardens at Ann Arbor, Michigan. American Rose Annual 1961, 46: 107-116