Fertility relationships ofGeranium (Geraniaceae): sectt.Ruberta,Anemonifolia, Lucida andUnguiculata

Cross-pollinations were carried out among 11 briefly described species ofGeranium. Eight species pairs produced hybrids, of which five had not been reported before. The close relationship ofG. purpureum, G. robertianum andG. rubescens (sect.Ruberta) was confirmed; they form a polyploid series (diploid, tetraploid and octoploid on base x = 16). ForG. canariense (sect.Anemonifolia), another octoploid on base x = 16, the results suggest greater affinity with the former section.G. maderense andG. palmatum of sect.Anemonifolia (2n = 68) are confirmed as closely related to each other.G. maderense produced hybrids withG. robertianum (2n = 64; sect.Ruberta) and withG. cataractarum (2n = 36; sect.Unguiculata). Meiosis in the latter hybrid suggests allopolyploidy between parents with 2n = 32 and 2n = 36. Whereas all these species clearly form a very close alliance,G. lucidum (sect.Lucida) andG. macrorrhizum andG. dalmaticum (both sect.Unguiculata), appear genetically more isolated from them. One plant ofG. macrorrhizum ×G. dalmaticum was raised. — In crosses where hybrids or non-germinating seeds resulted, the reciprocal cross in the majority of cases produced a greatly inferior result or none at all. This asymmetry of response could in some cases be explained by inability of pollentubes from short-styled parents to reach the ovary of a long-styled species and in others by a modification of Hogenboom's theory of incongruity, but neither explanation works for every case. In all our asymmetric results where the ploidy level differed the diploid was the successful female, not the tetraploid, as is usually the case. — Some variation in results from year to year could be attributed to weather conditions.Dedicated to Hofrat Univ.-Prof. DrK. H. Rechinger on the occasion of his 80th birthday.