Micromorphological features of the seed surface of the Schisandraceae are reported for the first time. One hundred and seventeen seed samples from 92 populations, representing 24 species, seven varieties and one form of the Schisandraceae, were examined and photographed under scanning electron microscope (SEM). Micromorphological features of seed surfaces of the Schisandraceae are little affected by the habitats in which plants grow, and are quite constant within species, therefore they can be used as reliable diagnostic characters to distinguish species. The following taxonomic
treatments are not supported by micromorphological features observed: ( 1 ) To reduce Schisandra wilsoniana A. C. Smith to S. henryi Clarke; (2) To reduce S. neglecta A. C. Smith, S. arisanensis Hayata, S. viridis A. C. Smith, S. sphenanthena Rehd. & Wils., S. gracilis A. C. Smith, S. micrantha A. C. Smith and S. lancifolia var. polycarpa Z. He to S. elongata ( Bl. ) Baill.; (3) To reduce S. henryi var. longipes (Merr. & Chun) A. C. Smith, S. tomentella A. C. Smith and S. pubescens var. pubinervis (Rehd. & Wils.)A. C. Smith to S. pubescens Hemsl. & Wils.; (4) To reduce S. rubriflora(Franch.)Rehd. & Wils., S. flaccidiramosa C. R. Sun, S. incarnata Stapf, S. sphaerandra Stapf, S. sphaerandra f. pallida A. C. Smith and S. glaucescens Diels to S. grandiflora ( Wall. ) Hook. f. & Thoms.; (5) To reduce Schisandra wilsoniana to S. bicolor Cheng; (6) To reduce S. lancifolia var. polycarpa to S. neglecta; (7) To raise S. henryi var. longipes to S. longipes(Merr. & Chun)R. M. K. Saunders; (8) To reduce Kadsura polysperma Yang to K. heteroclita(Roxb. )Craib. The relationship between Schisan-
dra Michx. and Kadsura Kaempf. ex Juss. revealed by the micromorphological features of seed surface is very similar to that revealed by pollen morphology, namely these two genera might have originated from a common ancestor and then evolved along two different routes. However the probability of a more complicated relationship between the two genera than we have known hitherto can not be excluded. The general evolutionary level of Kadsura seems to be higher than that of Schisandra. Therefore, the viewpoint that Kadsura is more primitive than Schisandra is not supported.