The three forms of chromosome grouping in tetraploids:
Autotetraploids (eg. Gloire de Dijon) may form upto 7 quadrivalents (groups of four). Typically there will be some irregularity: univalents, trivalents and pairs among the quadrivalents.
Allotetraploids (eg. R. damascena L.) form 14 bivalents (pairs)
Amphitetraploids (eg. Queen Elizabeth) may have one or more quadrivalents, with the remaining chromosomes forming bivalents.
Quadrivalents give an interesting pattern of segregation. During meiosis, the chromosomes double, then the cell divides into 4 pollen grains or ova. Thus, a parent carrying only one copy of a recessive allele and 3 copies of the dominant may give gametes with 2 copies of the recessive. If the 4 chromosomes are freely interchangeable, then for every 28 gametes there will be one with 2 copies of the recessive. If the parent is self-pollinated, one of 784 offspring will carry four copies of the recessive.
This inheritance can become even more complicated:
There is some evidence (Cytology of Two Fertile Triploid Roses: Dr H. D. Wulff) that environmental conditions - temperature or photoperiod - may influence the formation of unexpected chromosome groupings. He found that one of the triploids formed 7 bivalents + 7 univalents during warm weather, but tended to form 3-4 trivalents (+ 4-3 bivalents + 4-3 univalents) in cooler weather. This agrees well with Hurst's observation of 3-4 quadrivalents among some HTs.
Lammerts found that roses with much pelargonidin tended to have some lack of vigor when compared with siblings of other colors. Something similar has been found when some native American species are found to segregate for the pigment peonin. Apparently as the imperfectly matched chromosomes (from different septets) exchange the genes that are being studied, other genes are carried along in the transfer.
For instance, pelargonidin was first noted in Polyanthas (eg. Paul Crampel), and from them was transferred to Floribundas (eg. Baby Chateau) and HTs (eg. Tropicana). So long as the chromosomes are behaving in the normal bivalent manner, a tetraploid could not have more than two copies of the gene that allows pelargonidin to be formed. If there 3 copies, we may assume that an A-septet chromosome has exchanged a segment of DNA with a C-septet chromosome. Any genes on the C-chromosome segment that was exchanged may be lacking in the plant with enhanced pelargonidin production.
In the hypothetical case of a tetraploid derived from a cross of a peonin-bearing R. rugosa X pelargonidin-bearing Paul Crampel, our plant would carry two copies of the pelargonidin-allowing gene in the AA chromosomes, and two copies of the peonin-producing gene in the CC chromosomes.
Any seedling that produced an excess of peonin would probably also be deficient in certain genes normally carried in the A-chromosome, while having extra copies of the genes located near the peonin-gene. And any seedling with an excess of pelargonidin, on the other hand, will have extra A-genes and a deficiency of C-genes.
The solution to the deficiency problem will be to backcross to plants with the "normal" chromosomes and hope for further crossovers that will restore the missing genes.
The pattern of quadrivalent segregation is more complex and interesting than the simple Mendelian inheritance of diploids. This is possible because all the chromosomes are duplicated at meiosis, allowing a gamete to carry two copies of a gene even though the parent plant has only one. In the extreme case, a plant with one copy of a gene can produce a self-seedling with 4 copies.
In this example the plant carries one copy of a gene which confers glossy leaves, and three for dull leaves: G1,g2,g3,g4. At meiosis each is duplicated giving G1,g2,g3,g4,G1,g2,g3,g4. Since these 8 genes segregate randomly, it is possible for a gamete to carry both copies of G1.
"Sumarizing all the gametes having dominant G or GG and those lacking it, we have 1 GG plus 12 Gg or 13 G-bearing gametes to 15 gg gametes. Accordingly, on crossing a glossy plant of the above genetic constitution to dull, gggg, which forms only gg gametes, we expect to get a ratio of 15 dull to 12 glossy to 1 very glossy." Lammerts, Scientific Breeding.