A weeding-duration model for Larix kaempferi plantations in Hokkaido,northern Japan

In this study, a weeding-duration model for Larix kaempferi plantations was developed that employs a generalized linear model. The number of years that weeding is necessary is the response variable, and elevation, slope, maximum snow depth, annual precipitation, geological type, soil type, site index, slope aspect, and vegetation type are explanatory variables. Among the explanatory variables, geological type, soil type, slope aspect, and vegetation type are categorical data. We assumed a Poisson distribution for the response variable. The link function was log. Among the models that could be developed from these variables, we chose the model with the smallest Akaike’s information criterion (AIC). The weeding-duration model can be written as follows: years that weeding is necessary = Exp (−0.0172833 × site index + 0.0014053 × maximum snow depth (cm) + 1.7417731). The results of this study imply that weeding of Larix kaempferi plantations is needed for more years as the maximum snow depth increases and fewer years as the site index increases. This model is useful for cost–benefit analyses of afforestation or reforestation with Larix kaempferi.