Hybrids between dog roses and species from other sections provide valuable evidence to support the view that lack of homology is not the reason why the univalent chromosomes in the Caninae do not form bivalents. We are dealing with a genetically controlled restriction of bivalent formation on the part of the chromosomes which can, and do, pair when the particular genetic constitution responsible for the restriction is broken up by hybridisation.
The Swedish geneticist Gustafsson investigated hybrids between both pentaploid and hexaploid members of the Caninae, as female parents, and the diploid R. rugosa as male. The Canina-type meiosis broke down in the hybrids; these had a considerably higher incidence of bivalents than their dog rose female parent. Thus in two pentaploid hybrids from pentaploid R. canina x R. rugosa, from 9 to 13 bivalents were present per cell, with a mean of 11.1 in one hybrid and 10.7 in the other. Even if you suppose that all 7 rugosa chromosomes had paired with canina chromosomes, this cannot give you more than 7 bivalents and the excess present above this number can only have been formed by the pairing of chromosomes both derived from the canina egg cell and which never pair in the canina parent itself.
Another example is the tetraploid variety, Carmenetta, from a cross between R. rubrifolia, a tetraploid member of the Caninae, as female parent and R. rugosa as male. This variety forms far more than 7 bivalents at meiosis, so it is again clear that chromosomes derived from the female gamete, which always remain unpaired in R. rubrifolia itself, are pairing in the hybrid.
The Canina breeding system does not, in fact, always operate with quite the precision implied in the description given of it above. Some of the results of departure from the norm are quite informative also. One example is a tetraploid seedling I discovered among a lot of otherwise pentaploid seedlings, the progeny from a pentaploid species. At the seedling stage this exceptional tetraploid really could not be distinguished morphologically from its pentaploid sibs. There can be little doubt that it arose from the development of an unfertilised egg cell. While parthenogenesis of this kind is a rare phenomenon, it is nevertheless very widespread and sporadic cases have been reported from very many genera of plants.
Meiosis in this exceptional tetraploid had a variable pattern and was obviously far more susceptible to the influence of environmental factors than is normally the case. As a rule the course of meiosis is well buffered against the impact of, for instance, sudden large variations in temperature, but not so for this tetraploid. If lack of homology were the cause of the univalents in the Caninae, the plant derived from the unfertilised egg cell and so with A B C D chromosome sets should form only univalents at meiosis28 of them. In fact, a reasonable approximation to this was found in material fixed on one occasion, when the majority of cells did have 28 univalents. But even here as many as a quarter of the cells had 1, 2 or 3 bivalents. However in material fixed 2.5 weeks earlier, during an unseasonably cold spell, the departure from 28 univalents was very marked. No cells at all had 28 univalents; most had 2, 3 or 4 bivalents and a few had a still higher degree of pairing, the mean number of bivalents per cell being nearly 3. Obviously, chromosomes are pairing which never did so in the pentaploid parent.
The destabilisation of the system in the tetraploid, which of course lacks the haploid set normally supplied by the male gamete, points to the gene or genes involved in determining the restriction of pairing in the Caninae being located, at least in part, on one or more of the chromosomes which normally form the 7 bivalents. For, in the tetraploid, these genes would be present only in single rather than their normal double dose, and so might well be operating close to the threshold level of their effect on pairing. One would then anticipate finding enhanced susceptibility to environmental fluctuations against which the normal dog rose, with its double dose of these genes, would be adequately buffered.
Recent work has demonstrated that regular bivalent formation at meiosis in the hexaploid wheats, and several other groups of polyploid plants, is based on genetic control systems which prevent pairing between chromosomes which are potentially quite capable of association at meiosis. The selection of the appropriate genetic constitution which allowed a regular meiosis to occur was a vital step in the evolution of the polyploid species. So the Caninae roses are not exceptional in possessing a genetic system which restricts chromosome pairing; but they certainly do stand apart from virtually all other plant groups in the very unusual pattern in which the genetic control is expressed.
The practical significance of this thermal variation in chromosome pairing is that the fertility of a hybrid and the crossover frequency can vary with temperature.
Nature 182, 713-715 (1958)
Genetic Control of the Cytologically Diploid Behaviour of Hexaploid Wheat
Dr. Ralph Riley and Victor Chapman