I wonder if trying wide crosses of other tribes within the Rosaceae with roses would have a higher probability of success if the other tribes chosen were those with similar pollen such as the Dryadeae, Kerrieae, and Rubeae of subfamily Rosoideae.
Title: Studies on pollen morphology of Rosaceae
Authors: Hebda, Richard J.; Chinnappa, C. C.
Authors affiliation: Botany Earth History, Royal British Columbia Museum, Victoria, BC V8V 1X4, Canada.
Published in: Acta Botanica Gallica, volumn 141, pages 183-194,(1994).
Abstract: " Light and scanning electron microscope studies of pollen representing genera from all tribes of the Rosaceae reveal a variety of form and sculpturing. All genera examined produce radially symmetric isopolar monads. Most genera in subfamilies Maloideae, Prunoideae, and Spiraeoideae produce tricolporate striate grains with large perforations in valleys between ridges. These tectate perforate grains have a chambered pore covered by arching ektexinal pore flaps. The ridge-and-valley pattern can vary from 1-long ridges parallel to the colpus, to 2-medium to long ridges tooping near the poles, to 3-short weaving and crossing ridges. Striate perforate pollen occurs in tribes Dryadeae, Kerrieae, Roseae (operculate) and Rubeae of subfamily Rosoideae. in some Rubus species perforate and verrucate sculpturing occurs. Tribe Potentilleae (= Fragarieae) produces pollen with microperforations rather than typical perforations and most genera have an operculum. Coluria, Fallugia, Geum, Orthurus, and Waldsteinia of Dryadeae produce striate microperforate pollen, suggesting that they may belong in the Potantilleae. Filipendula (Ulmarieae) is prominently verrucate. Tuberculate perforate sculpturing occurs in Cercocarpus, Cowania, and Purshia suggesting a natural group distinct from the rest of the family. The diverse Poterieae (= Sanguisorbeae) has mainly tricolporate and some hexacolporate (Sanguisorba) grains all with an operculum. Some genera (Agrimonia group) have striate pollen, but most have microverrucae and perforations. Within the tribe, a distinctive group of mainly south hemispheric genera (Acaena, Cliffortia, Cowania, Hagenia, Leucosidea, Margyricarpus, Polylepis, Tetraglochin) has tricorporate perforate pollen, often with a short colpus, sculpturing of macroverrucae and rugulae covered by microverrucae. These pollen characteristics suggest a distinct evolutionary lineage."
Another possible route is to utilize members from the Sanguisorbeae and/or Potentilleae clades since their DNA is closest to Rosa.
The diagram below is from the following web site (the 2 roses studied are Rosa Persica and Rosa Majalis):
The abstract of their 2003 paper (which contains the same diagram on page 205 of the published manuscript) is at: