Abstract: The reproductive biology of Loropetalum subcordatum was studied. The floral phenology, pollen histochemistry, pollen-ovule ratio (P/O), pollen viability and floral visitors were investigated and determined. Assisted pollination experiments were carried out to examine the breeding system of L. subcordatum. Scanning electron microscopy (SEM) and fluorescence microscopy (FM) were employed to check the pollen germination and the growth of pollen tubes. Results were obtained as follows: (1) The flowering period of L. subcordatum was from September to the next February with a peak at September-October; the longevity of a single flower was 4-6 days. (2) L. subcordatum was protogynous, and pollen grains could be found on self-stigmata when the anthers opened 12-24 h after petals unrolling. (3)The pollen viability (MTT test) maintained for ca. 26 hours; the pollen was starchless; the P/O was 8420±720.86 (n=10); no nectar secretion was observed. (4) Thrips (Thrips sp.) were the only floral visitors observed but which seldom moved among inflorescences, thus played limited role in pollination. (5) Fruit sets of untreated bagged (5.30±1.83%) and hand assisted cross pollinated flowers (6.67±1.91%) were not significantly different from that of open flowers (4.79±1.45%). (6) SEM and FM observations proved that pollen germinated on self-stigmata and pollen tubes growed in self-styles. The results indicated that L. subcordatum was facultatively autogamous and without apomixes. The possibility of outcrossing and the protogyny might indicate that the species was originally a crosser. Flowers pollinated in September-October usually start ovule growth in next summer, and set mature fruits in next October (after the next flowering peak), suggesting the occurrence of retard embryo development. The possibility that autogamy in Hamamelidaceae could have been developed from fly pollination was also discussed.

Key words: autogamy, Hamamelidaceae., Loropetalum subcordatum