Cone and seed development was observed from 1993 to 1998 in wind-pollinated Keteleeria davidlana (Franchet) Beissner var. formosana Hayata trees grown at Taipei Botanical Garden. Seed efficiencies per cone (percent filled seed per cone divided by seed potential) were persistently less than 1% during the investigation period. Sections of large seeds at different stages during the period from polyembryos to cotyledonary embryos revealed that 80% of them were successfully fertilized initially, and then almost all degenerated before developing into mature embryos. Cone length growth was evidently affected by the number of large seeds inside the cone, and there was a very significant (p<0.001) linear regression between these 2 factors. Before fertilization occurred, conelets and ovules reached the same sizes as did mature cones and seeds. Pollen grains had a high germination percentage and were found in mycropyles of ovules, indicating no pollination barrier between catkins and conelets. Unpollinated cones containing small seeds were not shed. All these facts indicate that ovules grew after they obtained pollen, subsequently stimulating cone development. The great variations of large seeds, numbering from 50 to 114 per cone, and cone length, differing from 6.7 to 9.0 cm, in 6 years came from the pollination degree of ovules. According to the year with the best large-seed production, the reasons for empty seeds could be estimated as follows: Seed potential per cone calculated as 2 times the number of fertile ovuliferous scales was about 206 seeds. Percentages of seed losses resulting from pre-and post-pollination ovule abortion, prezygotic factors, embryo degeneration, and efficient seed were 33%, 27%, 39%, and<1%, respectively. Even in the year with good pollination effectiveness, embryo degeneration was so great as to cause most fertilized seeds to become empty.
Cone and seed development was observed from 1993 to 1998 in wind-pollinated Keteleeria davidlana (Franchet) Beissner var. formosana Hayata trees grown at Taipei Botanical Garden. Seed efficiencies per cone (percent filled seed per cone divided by seed potential) were persistently less than 1% during the investigation period. Sections of large seeds at different stages during the period from polyembryos to cotyledonary embryos revealed that 80% of them were successfully fertilized initially, and then almost all degenerated before developing into mature embryos. Cone length growth was evidently affected by the number of large seeds inside the cone, and there was a very significant (p<0.001) linear regression between these 2 factors. Before fertilization occurred, conelets and ovules reached the same sizes as did mature cones and seeds. Pollen grains had a high germination percentage and were found in mycropyles of ovules, indicating no pollination barrier between catkins and conelets. Unpollinated cones containing small seeds were not shed. All these facts indicate that ovules grew after they obtained pollen, subsequently stimulating cone development. The great variations of large seeds, numbering from 50 to 114 per cone, and cone length, differing from 6.7 to 9.0 cm, in 6 years came from the pollination degree of ovules. According to the year with the best large-seed production, the reasons for empty seeds could be estimated as follows: Seed potential per cone calculated as 2 times the number of fertile ovuliferous scales was about 206 seeds. Percentages of seed losses resulting from pre-and post-pollination ovule abortion, prezygotic factors, embryo degeneration, and efficient seed were 33%, 27%, 39%, and<1%, respectively. Even in the year with good pollination effectiveness, embryo degeneration was so great as to cause most fertilized seeds to become empty.