A recent paper by Crespel, L; Gudin, S.; Meynet, J.; and Zang, D.; Theor Appl Genet, vol. 104, pages 451-456 (2002) reports on the results of crossing 2 dihaploid R. hybridia L. clones, called H61 and H3 with normal diploids R. rugosa and R. wichuraiana. The dihalpoids were created by in situ parthenogenesis induced by using irradiated pollen and embryo rescue of a rose identified as FJV6 and Sweet Promise.
The results are a little surprising (one would of expected that all of the seedlings would of been diploids like the parents). The actual results for H61 X r. wichuraiana are 8 diploids, 19 triploids and 5 pentapoids (32 total seedlings); for H61 X r. rugosa - 6 diploids, 28 triploids, 2 tetraploids, and 2 pentaploids (38 total seedlings); and for the reverse cross r. winchuraiana X H3 - 19 diploids, 9 triploids (28 total seedlings).
Thus, we see that the division does not always follow the classical expected division but “often follows abnormal (or absence of)cytokinesis”. Please consult the original paper for a more technical discussion.