Explanation of land use in mountainous area, China: from field to village level

Abstract

Few studies of land use change were particularly considered the hierarchical data structure originating from different scales and levels. Using interviewing data collected from 107 villages, 1,050 households and 4,780 fields between November 2003 and August 2005, the objective of this paper is to predict the occurrence of land use from field to village level in mountainous area, China, and to improve our understanding of the causes of land use. Household’s behavior in the choice of land use type is guided by multiple, often confiding, household objectives, subject to the available resources, possible productive activities, and external economic and biophysical constraints. For rice model, the household level variables cannot be substituted by village level aggregates. Aggregated variables at village level do not capture any of the variability at the household level. Village level variables can virtually be explained in virtue of the variables of field and household level. The households and the villages show significant clustering of the occurrence of rice, and they explain the 11.3 and 4.5% variance, respectively. For corn model, corn as dependent variable does not show any significant variance component. The variables of household and village level have lower effects on the occurrence of corn. There is a significant relation between slope of a field and the choice to cultivate corn and a significant random effect at the village level. However, cropland size, input–output, transportation cost, even family income in household level and road density and food market development in village level, at some extent, are controlled by slope. These variables do not influence corn cultivation significantly and that those are predominantly determined by slope. In a word, the household level can be crucial in explaining land use at the field level. Multilevel analysis can be applied to statistically model the occurrence of land use, and to explore a number of cross-scale propositions.