Scale-dependent wintering habitat selection by brown-eared pheasant in Luyashan Nature Reserve of Shanxi, China
山西芦芽山褐马鸡越冬栖息地选择的多尺度研究

The selection of different spatial scales is critical when investigating ecological processes, because different patterns emerge in spatial data at different scales. The wintering habitat selection of Brown-eared Pheasant Crossoptilon mantchuricum was studied on four spatial scales (10m, 100m, 300m and distance scale) in Luyashan Nature Reserve of Shanxi from 1998 to 2000. At the 300m scale, the area of coniferous forests around selected sites was larger than that of unselected sites (F=(-3.116,)P=0.002), and the area of coniferous-deciduous mixed forests was less than that of unselected sites (F=-2.255,P=0.024), though the percent area of mixed forests was lower around the two kinds of sites. In addition, there was no difference for shrubs and grass. The results at the 300m scale show that the pheasants prefer to use coniferous forests rather than use mixed forests that cannot supply better shield in winter. At the 100m scale, there were no shrubs, and the area of conifer forests around the selected sites was also larger than that of unselected sites (F=-2.931,P=0.003), but for mixed forests and grass, there was no difference between the two kinds of sites. The results suggest that the pheasants occur in coniferous forests, and also freely use mixed forests and grass at the 100m scale. So it is concluded that after the better shields are satisfied on larger scale, Brown-eared Pheasant can take more food by using all available habitats on smaller scale. The distance scale applied showed some variables, such as the distance to villages and roads, had significant differences between the selected and unselected sites (F=15.621; 6.048, P=0.000; 0.018). Furthermore, the model based on data collected at four different spatial scales was developed by stepwise logistic regression. Presence of Brown-eared Pheasant was best predicted by the distance to shrubs, distance to villages, area of coniferous forests of 100m around the selected sites, height of trees and food richness. We tested the model's ability to predict the presence of Brown-eared Pheasant by an independent data set collected at another study area in this nature reserve. In general, the model has a good ability to predict the presence of Brown-eared Pheasant.