景観生態学研究における階層線形モデルの有用性の検討

概要

論文の詳細を見る
Landscape ecological data often have spatial hierarchical structure. Let us suppose that avian species richness is surveyed in patchy woods and several sampling stations are located in each surveyed patch. In this situation each sampling station forms the lower level (Level 1) survey unit while each patch forms the upper level (Level 2) unit. This mixture of survey units of different spatial level is not appropriate for most traditional methods of data analysis. If the upper level units are ignored, the analyst will suffer pseudoreplication problem with variables obtained for patches (the upper level units). If the lower level units are ignored,information on local heterogeneity (i.e., heterogeneity among the sampling stations within the same patch) will be lost. Hierarchical Linear Models can handle such data set with spatial hierarchical structure. First, a linear model is constructed for each lower level unit considering a dependent variable and the variables measured for the lower level units. Then, each of the intercept and the coefficients of the linear model is represented as a linear combination of variables obtained for the upper level units, a random factor and an error term. In this paper we present the usability of Hierarchical Linear Models by showing an example of application of this method.
社団法人　日本造園学会の論文