Abstract
Estimating forest tree seeding before seedfall is important because annual fluctuations in seeding affect forest regeneration and wildlife behavior. Visual surveys for estimating tree seeding can be conducted easily, but cannot distinguish sound and unsound seeds, and cannot directly measure sound seedfall density, which closely correlates to seedling density and the food availability for wildlife. To establish an effective model for estimating sound seedfall density of Fagus crenata from a visual survey, seeding intensity at the tree level was visually rated into five classes in four stands for 7 years. Seedfall density of the visually surveyed trees was also measured by seed traps. Seeding intensity was strongly correlated with sound seedfall density at the tree level, and sound seedfall density differed among the five classes. However, a model that used only seeding intensity as an explanatory variable was not as effective at estimating sound seedfall density as more complex models that used the stand-level seeding index, calculated from the composition ratio of seeding intensity, in the current and previous years. More complex models improved the estimates because the proportion of sound-to-total seeds at the tree level depended on stand-level total seed production resulting from pollination efficiency and insect predator satiation. This method should be useful for forest management and wildlife conservation because it is rapid, easy, and highly accurate.
Acknowledgments
I thank the anonymous referees and editor for their valuable comments. I am grateful to Toyama District Forest Office and Nanto City for permission to use the stands, to my colleagues at Toyama Prefectural Government for their helpful support, and to Yukimi Kayamori and Natsuko Aoki for their assistance with counting seeds in the laboratory. This study was funded by Toyama Prefecture, Japan.