American Rose Annual 44: 113-116 (1959)
Genetics of Doubleness In Roses
Dennison Morey*
Pleasanton, Calif.

*Director of research, Jackson & Perkins Co. of California.

Floral parts of the Rosaceae family are arranged in whorls of five. Normally, wild members have only this one set of five petals. There are, however, a number of wild semi-double species. In general, these extra petals (called petaloids) result from the failure of stamen initials to develop properly.

This is also true in cultivated roses. For example, if we select a seedling which has five petals and from 1-10 petaloids,subsequent flowers will generally have anywhere from 5-10 petals, depending upon external conditions (climate, etc.), the plant itself and the position of the flowers on it. Petaloids vary rather markedly in their degree of perfection. So, it is quite difficult to demonstrate on a morphological basis whether or not one is dealing with true petals, or with very perfect petaloids. This is important, because the concept of "dedoublement" has been a classical hypothesis to explain increases in petal number. Dedoublement means that instead of one whorl of petals being formed, there are two, three or more.

If a Hybrid Tea rose seedling with five petals and one or two petaloids is budded and observed for several years, we see that those flowers with few petaloids have but five perfect petals. However, when seven or eight petaloids are present, several of them may be perfect petals. If enough petaloids are formed, five of them might be perfect. This would create an impression of dedoublement. But, previous observation of the variety has emperically established the fact that there are but five true petals.

Thus, "false" petals in the form of perfect petaloids vary in number. If there are five of these "false" petals, in the absence of prior observation one might conclude that he is dealing with dedoublement rather than petaloidy (also called staminody).

If dedoublement is the basic cause of doubleness in roses, a polymodal (or at least a bimodal) frequency curve for petal numbers would be expected in a self-population of a semi-double rose.

We grew the rose Golden Scepter and a population of its self-seedlings in order to determine the frequency pattern of petal number segregation. A strongly bimodal curve is evident in the petal frequencies of the seedlings (Figure l). At first glance, this curve seems to establish dedoublement as a basic feature of doubleness in roses. Thus, the more plausible hypothesis that doubleness in roses is the result of the action of genes causing petaloids must be examined very critically.

From data taken from the parent, it is obvious that petal number is considerably influenced both by the position of the flower on the plant and by the overall environmental situation. Terminal and Spring flowers tend to have both more petals and petaloids than lateral and Summer flowers (Figures 2 and 3). These phenomena undoubtedly account for the bimodality of the seedling petal frequency curve where the distinction between lateral or terminal and Spring or Summer flowers could not be conveniently made.

It is evident that the basic pattern in Golden Scepter lateral flowers is for 19 petals and 4 petaloids (Figure 2 B). Moreover, the number of petals and petaloids depends very much upon the total number of floral parts initiated by the floral primordium.

It appears that Golden Scepter is genetically designed to have an average of 22 perianth segments, but unless there are an average number of total floral parts initiated, the number of petals will be lesser or greater, depending upon how much the total flower part number varies from the mean. Thus, the degree of "doubleness" is esentially the proportional division of the number of floral primordia initiated in the particular bud into petals, petaloids, stamens and carpels.

It is well-established in hybrid roses that the alternation of vegetative impulses to reproductive is a thing of very delicate balance, and that, under certain conditions, no floral primordia of higher order than petals are possible, so that all parts of the flowers are petalaceous or vegetative. This results in flowers that are excessively double through what is probably heterotically-induced physiological imbalance.

The distribution observed in petal number frequencies in Golden Scepter seedlings, and in roses in general, can then be best-explained on the basis of five true petals plus the independently-controlled production of petaloids, the number of which are subject to wide limits of externally-induced variation. Moreover, the absolute petal numberdepends upon the total number of flower parts initiated in any given flower, and upon independently-regulated tendencies toward hypervegetative flowers.