This study investigated the effects of weeding frequency, slope aspect and initial tree density on the mortality and accidental cutting of planted Sugi (Cryptomeria japonica) in southern Kyushu. To predict the probability of mortality and accidental cutting,we used the multinomial logistic regression model using weeding frequency,slope aspect and initial tree density. The highest mortality on north- and south-facing slopes was found in non weeding plot (33.3%) and every-year weeding plot (16.9%), respectively. Accidental cutting of planted Sugi was found in 3 or 4 times weeding plots. Multinomial logistic regression model showed that mortality was affected by slope aspect and initial tree density. Mortality decreased with the increase of weeding frequency on northfacing slope, while mortality was invariable regardless of weeding frequency on south-facing slope. Accidental cutting rate was invariables regardless of weeding frequency on both slopes. The mortality rate and accidental cutting rate under the 1,500 trees/ha density was higher than that under the 3,000 trees/ha. These results suggest that weeding frequency should be determined according to slope aspect and initial tree density.

Objectives of this study were creating canopy shape-feature parameters using the sampled canopy point data and classifying the two species using the point data and parameters. A canopy sharpness parameter, which was standard deviation of canopy height of the points within distance of 1.5 m from the top, and a canopy top roundness parameter, which was an angle between the top point and two points at distances of 0.3 m and 1.3 m,were contrived. Larger absolute canopy sharpness values showed sharper and narrower canopies and smaller roundness parameter showed rounder canopy top. Sample trees of sugi and hinoki were selected for modeling and validation separately and a discriminant function was created (r=0.675) with accuracy of 79.2% for species classification. Classification accuracy of the validation samples was 84% and it suggested usefulness of the discriminant function for species classification of sugi and hinoki trees with similar canopy shape to the modeling samples.

We report case studies on causative factors for the establishment of hardwood or mixed forests in naturally regenerating Japanese larch stands. Field studies were conducted in two naturally regenerating larch stands, Ashoro and Ohmu. In Ashoro, the larches had died within the last few years. In Ohmu,competition existed between larch and hardwood seedlings. Most dead larches in Ashoro had scars owing to feeding by red-backed voles,whereas no hardwood had such scars. In Ohmu, the growth rates of Japanese white birch and monarch birch were faster than those of Japanese larch. These results indicated that slower growth and greater susceptibility to red-backed voles of larch than of hardwood competitors are two of the reasons why naturally regenerating larch stands develop into hardwood or mixed forests.