Abstract
We investigated how height growth and suppression by weed of planted sugi trees (Cryptomeria japonica) differed with topographic variables along a slope (distance from the ridge and slope angle). We also assessed different statistical models to detect these relationships. We developed three types of models (without random effects, with random effects and with spatial autocorrelation) to predict three and four years old sugi tree height, conditions of suppression by weed and topographic variables measured at 16 subplots along a slope. Sugi trees further from the ridge were taller. The probability of being suppressed by weed was higher for sugi trees that were shorter, or further from the ridge, or on a steeper slope. These effects of topographic variables were reduced in the spatial auto-correlation model, with corresponding increases in credible intervals. We concluded that the reduction in weeding frequency within a stand along a slope may be one possible option in low-cost weeding management, but sampling and statistical approaches methods must take care to evaluate topographic effects with spatial information.
