Botanical Gazette 96(2): 197-251 (Dec 1934)
Experimental Data for a Revision of the North American Wild Roses
Eileen Whitehead Erlanson

Summary

1. Herbarium, field, and garden studies upon extensive series of North American wild roses belonging to the section Cinnamomeae have shown that many of the characteristics commonly used to distinguish between species are individual variations. These unreliable characteristics may occur combined in every possible way, sometimes in plants of a single culture. They also appear in every species and give good series of parallel variations. Because in the past rhodologists have been inclined to give these minor variations and their multitudinous combinations specific or varietal rank, nomenclatural confusion has resulted.

2. The various diagnostic characteristics are discussed and evaluated chiefly from evidence obtained from growing cultures of plants raised from the seeds of single wild individuals. As many as ten so‑called species have appeared in one culture.

3. The offspring of isolated diploid plants of R. woodsii from the Great Basin of Utah showed as much variability and pollen sterility as offspring of plants of R. blanda from northern Michigan, which grew close to roses belonging to other species groups. Hybridization is not therefore the most important source of variation.

4. Tables are presented which demonstrate the variation in minor characteristics in cultures of the groups of R. blanda, R. californica, R. pisocarpa, and R. woodsii.

5. On the evidence of data from these and other cultures, the sections Carolinae and Gymnocarpae have been merged with the section Cinnarnomeae. One hundred ten species in this section given by RYDBERG (34) and three species described by the writer have been placed in 16 Linnaean species. These 16 species are relatively stable (although individuals are highly heterozygous) and they can be distinguished from one another by general, morphological, numerical, cytological, and physiological characteristics. Ten of the sixteen are collective species with wide ranges of distribution.

6. Seven other species are suspected of being ecotypes of the various collective species but have been retained in specific rank.

7. Interspecific F1 hybrids have been raised after crossing species with ranges that overlap. Several of these hybrids are fertile and resemble roses that occur in nature. Six previously described species are considered on this evidence to be F1 hybrids.

8. These results which attribute a few highly variable Linnaean species to the genus Rosa agree with those obtained by BOULENGER from data from European rose species.

9. Sterility arises from (a) interchange, (b) internal polyploidy, (c) hybridization with polyploid species. Polyploidy is therefore an obstacle to the fusion of species:

10. Ninety‑one specific names, chiefly from RYDBERG (34), have been given as synonyms in an alphabetical list of Linneons. An analytical key is appended which includes 16 Linnaean species, seven ecotype species, and four hybrid species in the section Cinnamomeae.