Botanical Gazette 87(4): 478-480 (May 1929)
Eileen Whitehead Erlanson


It soon became noticeable among our living collection that nearly all the North American tetraploid species are characterized by the ability to produce flowers terminally in corymbs on the season's turions, a behavior peculiarly adapted to a prairie habitat where the aerial stems are likely to be frozen down in winter and burned off by fire in summer. They are also characterized by their long flowering period, some of them continuing to bloom after ripe fruit is developed, a performance never observed in any of the other native groups, although exhibited by R. rugosa Thunb. R. acicularis, however, will occasionally have a second blooming late in a mild autumn.

I had hoped that the ability to flower on the season's turions would prove to be a criterion for tetraploidy in American roses, and for the most part this is true. In R. virginiana, the largest of the eastern tetraploid species, this ability is only weakly developed and some plants will not produce flowering turions every year. It is very common in R. californica. R. ratonensis Erlanson in some ways resembles R. woodsii, but can immediately be distinguished from that species by its strong turions bearing terminal corymbs. It is a tetraploid.

The most unexpected exception to this generalization was the discovery of a diploid specimen of R. subserrulata, a possible hybrid origin of which has been previously considered.

Another exception to this character being confined to tetraploids is R. foliolosa Nutt., a diploid species of the Carolinae from Texas and Oklahoma. This is a low semi-herbaceous species that bears very little wood over 2 years old. It flowers in small corymbs, terminally on the annual shoots. If this species originated as a haploid from a tetraploid form and still retained the tetraploid "ever-blooming" characteristic, it would from its southern distribution be an argument in support of HURST'S theory of the loss of extra septets of chromosomes from polyploids in southern habitats. It is a very distinct species, its narrow numerous leaflets and small (sometimes curved) prickles resembling, if anything, diploid R. palustris, whose western limit is well to the east of the range of R. foliolosa  (text fig. 1), rather than any form of tetraploid R. carolina.

Our collection does not contain any mature R. nitida Willd. This species was reported as diploid by TÄCKHOLM, and it may also possess the ability to flower on the season's turions, although this is not mentioned in descriptions and can seldom be discovered from herbarium specimens. It will be of interest to discover whether the other dwarf species, R. gamella Willd. and R. nanella Rydb., also Carolinae with limited ranges in the northeastern Atlantic region, are diploid or tetraploid. R. relicta Erlanson, a dwarf form from Illinois, and the low growing R. bushii Rydb. are both tetraploid. The third exception is the case of a dwarf plant from Calgary, Alberta (no. 3371). Alter three years in the garden it is little more than a font high, a small branched woody bush resembling in habit the specimen of R. subglauca from Alberta (no. 3492). The former plant agrees with the description of R. alcea Greene. It grows very slowly and has produced no turions. The dwarf R. subglauca has produced one or two flowering turions. Both these specimens are tetraploid.

With the exception of R. virginiana, in the northeast, and of R. californica, the North American tetraploid roses are of low stature, usually under a meter. The tetraploid species belonging to the section Cinnamomeae are more specialized in habit and inflorescence type than the related diploid forms, and are perhaps of more recent origin. The setaceous stems and numerous leaflets are considered by BOULENGER (7) as relatively primitive characteristics in Rosa, but they almost always appear on young turions, and would be expected on plants with a semiherbaceous habit. Some of these forms, more particularly R. suffulta and its allies in the Mississippi region, have a markedly semiherbaceous habit, which appearing in a frutescent group is generally considered a more highly evolved condition (SINNOTT and BAILEY 56). BEWS (3) points out that the stream bank and swamp habitat is one of the primitive unchanging habitats of angiosperms, and that the grassland habitat is a relatively recent one; hence he maintains that plants adapted to the latter are derived and not primitive forms. As already stated, the North American diploid roses are characteristic of waterways (as are the hexaploids), while the tetraploids belong to the prairies and dry uplands.

From every point of view the tetraploid forms seem to be more highly evolved and of more recent origin than any other form in the group of Cinnamomeae-Carolinae in America.

Among the tetraploid Carolinae, R. virginiana stands out as the most robust and fruticose type. This species seems to be confined to the northeastern Atlantic region, although a whole series of forms can be found between it and the typical low weedy R. carolina, as well as between this latter species and the tetraploid Cinnamomeae of the R. suffulta complex. All these types flower almost synchronously when grown together. The significance of this will be considered in the discussion of phenology in Rosa. The tetraploid Carolinae have a very extensive north and south range, from the St. Lawrence to northern Mexico. It is noticeable that the glandular types of the R. carolina complex preponderate in the southern part of its range; this is also true for the western diploid R. woodsii complex, which has an even greater north and south range than the Carolinae, stretching from British Columbia to Chihuahua.