中文摘要：

2007年12月至2008年1月，在贺兰山地区，利用痕迹检验法和直接观察法对马鹿阿拉善亚种的冬季取食和卧息生境选择进行研究。通过在选定的15条沟段里进行调查，共测定了72个取食利用样方，59个卧息利用样方和131个对照样方的18种生态因子。结果表明，马鹿在冬季偏好的取食地为平滑起伏坡，以酸枣、柳为优势乔木、混合型树林或空地，乔木高大稀疏、间距较远，灌木密度大，草本盖度较高，位于〈15。的半阴半阳坡的下坡位，距裸岩远，隐蔽度高；偏好的卧息地在山地疏林草原带的平滑起伏坡上，以酸枣、柳为优势乔木、混合型树林或空地，乔木高大稀疏，灌木矮小、稀疏且距离较远，草本盖度较高，位于〈15。阳坡的下坡位，远离裸岩，隐蔽程度高。马鹿冬季的取食和卧息生境在草本盖度和隐蔽度上差异极显著。相对于卧息生境，马鹿冬季的取食生境对草本盖度和隐蔽度要求更低一些。马鹿冬季取食地的资源选择函数为1．155—0．149×乔木高度-0．066×草本盖度＋0．190×坡度，模型的正确判别率为86．8％；马鹿冬季卧息地的资源选择函数为-30．936＋0．494×乔木高度＋0．257×坡度-0．002×海拔高度＋0．387X隐蔽度，模型的正确判别率为95．8％。食物、隐蔽条件以及贺兰山的独特地形特征是影响马鹿冬季取食和卧息生境选择的主要因素。

英文摘要：

Winter feeding and bedding habitat selection by red deer （ Cervus elaphus alxaicus ） were studied from December 2007 to January 2008 in the Helan Mountains of Ningxia and Inner Mongolia. We identified 72 feeding sites and 59 bedding sites by observation of fresh sign and direct observation of animals; for comparison, we identified 131 sites from systematically placed transects on which we observed no recent evidence of deer use. At each plot, we measured 18 topographic and biological variables. Indices of selectivity for feeding were positive for sites with smooth undulating slopes, overstories characterized by Ziziphus jujube, Salix spp. , mixed species, and unforested. Additional variables yielding positive selectivity indices at feeding sites included tree heights 〉2.5 m, shrub density 〉9 trees/100 m2, vegetation coverage 〉20%, tree density 〈 2/100 m2, slopes 〈 15°, aspects with both sun and shade, distance from nearest tree 〉 6 m, distance to nea- rest bare rock 〉 20 m, and hiding cover 〈 86%. Selectivity indices for bedding sites were positive for smooth undulating slopes, elevations of 1 600 -2 000 m, woodlands of 〈 2/100 m2 characterized by Ziziphusjujuba, Salix spp. , Juniperus rigida, mixed species, and unforested, slopes 〈 15° on southerly aspects, shrub density 〈 2/100 m2, shrub height 〈0. 2 m, distance from nearest shrub 〉 10 m, distance from nearest bare rock 〉20 m, vegetation coverage 〉20%, and hiding cover 〈 86%. Feeding habitat selected by red deer had lower vegetation coverage and less hiding cover than bed- ding habitat. We fitted a logistic regression model describing feeding habitat selection： 1. 155 - O. 149 （ tree density） - 0. 066 （ vegetation coverage） ＋ 0. 190 （slope）. The model correctly predicted 86. 8% of feeding sites. The model of bedding habitat was best fit by - 30. 936 ＋ 0. 494 （ tree density） ＋ 0. 257 （slope） - 0. 002 （elevation） ＋ 0. 387 （ hiding cover）, correctly predicting 95.8% of bedding sites. Food