Virginia Journal of Science 1: 11-59 (1950)
Pollen Condition in Some Species and Hybrids of Rosa with a Consideration of Associated Phylogenetic Factors
Walter Samuel Flory, Jr.
The Blandy Experimental Farm, University of Virginia

Abstract: Data obtained by previous workers on pollen abortion in Rosa species are summarized. In the present study 95 species and approximately 55 botanical, horticultural and cytological variants of Rosa were analysed for percentage of normal pollen during a five year period. Only slight seasonal variation in pollen condition was encountered in most species and hybrids. A marked variation was often, but not always, observed in the pollen condition of (1) variants of the same species showing differences in chromosome number and (2) species and certain of their botanical varieties exhibiting no chromosome differences. Average percentage of normal pollen decreased with increase in chromosome number in the diploid, tetraploid, hexaploid and pentaploid groups. The single octoploid species was high in percentage of normal pollen; the triploid forms were very low. Broad differences in percentage of normal pollen according to the geographical origin of species were noted. Data on the production of normal pollen in interspecific hybrids arranged in groups based upon chromosome number conformed fairly well with observations made upon species and varieties. In several cases hybrids resulting from crosses between species believed to be quite diverse genetically and remote taxonomically produced unexpectedly high percentages of normal pollen. Hurst's differential septet theory of the evolution of Rosa species is discussed, in which most of the present day species are regarded as having descended, by loss of chromosome septets, from a now extinct hypothetical decaploid species. The author argues in favour of the view that the polyploids have originated from forms with lower chromosome numbers, i.e. by ascent rather than descent. The centre of origin of Rosa is postulated as eastern Asia; and the possible distribution and evolution of the genus are outlined. Finally, the past and present role of hybridity in speciation of Rosa is considered.

CybeRose note: I sometimes wonder how any progress is made in the sciences when critics spend their time attacking theories while studiously ignoring the basic facts that preceded the theories. If five sets of fifty traits are found in Rosa species around the world, this fact should be enough to attract attention. Hurst's "septet scheme" was consistent with what geneticists assumed to be the only way such large sets could be maintained. But to attack Hurst's "scheme" while avoiding the facts is rather silly. Furthermore, Hurst also contended that the hyphthetical decaploid would have been built up from five diploids. Such a species could, he imagined, break down, giving rise to various diploids and polyploids. This hypothesis makes more sense to me than the odd notion that a swarm of mostly tender diploids raced eastwards across the Bering Land Bridge into Canada, then moved south before assembling themselves into polyploids capable of surviving the frigid conditions they had previously endured.

By the way, Rosa praelucens is a naturally occurring decaploid rose. This is proof that decaploids are possible, but it is not the species Hurst imagined. It is interesting to note that R. roxburghii is apparently one of the ancestors. Hurst thought R. roxburghii should be removed from the genus Rosa based on certain considerations. For one thing, it is very different from most other Rosa species. On the other hand, some evidence suggests that (if we ignore the fleshy hips) it is close to the diploid Pimpinellifoliae species, along with R. minutifolia.

Hurst's Septet Scheme