The system of the Saxifragaceae sensu lato is discussed with special reference to the comparative morphology of pollens. The pollens treated here cover about 108 species belonging to 38 genera of our family. These pollens are classified into 18 Types, and their relationship is shown in Fig. 6. The results obtained by this study will be summarised as follows: 1. By this study no positive evidence is available as to which is most natural among the systems by ENGLER's syllabus, HUTCHINSON, HUBER, and TAKHTAJAN. Two centers of affinity seem to be in this family. These two are Tiarella Type and Deutzia Type, though they have no vital discrepancy to each other. It is not elucidated by this palynological study whether these two Types show a natural group or the resemblance caused by the parallel evolution. 2. The ligneous Ribes and Bauera seem to have affinities to the herbaceous genera, and the herbaceous Parnassia and Deinanthe to the ligneous ones. Several genera of Saxifragoideae and several ones of Hydrangeoideae are similar in the pollen morphology to each other, though the former is herbaceous and the latter mainly ligneous. Therefore, the systems of HUTCHINSON, HUBER, and TAKHTAJAN, which consist of two phylogenetic groups-the herbaceous and the ligneous-should further be revised. 3. Each of the genera Ribes, Bauera, Parnassia, Itea, Polyosma, Deinanthe is suggested to have a weak affinity to the other genera: the first 3 genera are supported to be separated in subfamily, or even in family, rank, the last two, included in Escallonioideae (Escalloniaceae) and Hydrangeoideae (Hydrangeaceae), respectively, should further be investigated in detail. 4. The genus Itea seems to have no close relationship to any of our genera, and is worthy to place in a distinct subfamily or a family. The HUTCHINSON's system including it in the Escalloniaceae may better be rejected. 5. The systematic position of Kirengeshoma, Cardiandra, and Deinanthe seem to have no close relationship in view of the pollen morphology, and the Tribe Kirengeshomeae in the sense of HUTCHINSON should be revised. 6. There is no positive evidence to show that the genus Penthorum should be separated as an independent family. This genus may be included in the Saxifragaceae rather than in the Crassulaceae.
View full abstract