陕西黄龙山自然保护区褐马鸡（Crossoptilon mantchuricum）生境选择的研究

【作者】 李宏群；

【导师】 廉振民；

【作者基本信息】 陕西师范大学 ， 动物学， 2008， 博士

【摘要】 褐马鸡（Crossoptilon mantchuricum）为我国特有珍稀鸟类,国家一级重点保护野生动物,是世界易危鸟类之一,被我国列为濒危物种。其目前分布区很狭窄,主要分布于山西吕梁山、陕西黄龙山、河北小五台山和北京东灵山等地的局部地区。由于地理屏障或者自然植被的破坏,其分布区已被严重分割成3个区域,分别形成3个地理种群,即山西吕梁山脉的中部种群、河北与北京地区的东部种群和陕西的西部种群。陕西黄龙山林区的褐马鸡种群,其分布区位于黄土高原南缘的黄龙山腹地,涉及黄龙、宜川两县以及韩城市。由于陕西延安黄龙山褐马鸡自然保护区是刚成立,现缺乏对褐马鸡保护的一些基础资料。因此,加强对褐马鸡的栖息地研究,对保护区制定科学的管理手段尤为重要。同时,也可以弥补褐马鸡的西部种群栖息地无人研究的历史。本研究于2006年3月至2007年8月在陕西延安黄龙山褐马鸡自然保护区北寺山林区进行的。其主要结果如下:褐马鸡繁殖早期,地理性、乔木数量、灌丛种类、灌丛数量和草本盖度等5个因子的判别能力最强,由这5个变量构成方程的判别准确率达96.08%,可以较好的将觅食样方与对照样方分开;在卧息地选择方面,坡度、坡向、道路距离和乔木盖度等4个因子的判别能力最强,由这4个变量构成方程的判别准确率达75.64%,可以较好的将卧息地样方与对照样方分开。乔木密度、水源距离、灌木密度、灌木种类、乔木最大高度、海拔具有重要作用,由这6个变量构成的方程在对繁殖季节觅食地利用样方和对照样方进行区分时,正确判别率可以达到97.22%。褐马鸡春季中午卧息地多偏向针叶林、山脊和半阴半阳坡,且具有坡度小、与林间道路与居民点的距离远、灌丛高度较低、灌木密度较小、草本均高较小、乔木层植物盖度以及草本层植物盖度较大等特征。且对各种生境因子的主成分分析表明,前6个特征值的累积贡献率达到74.05%,根据载荷系数绝对值大小将褐马鸡中午卧息地生境选择影响因子分别命名为地理因子、隐蔽度因子、人类活动干扰因子和灌丛密度因子。褐马鸡对巢位具有明显的选择性:首先表现在景观尺度的选择;其次是植物结构上的选择。进一步进行逐步逻辑回归分析表明,在褐马鸡选择巢址地的过程中起主要作用的因子是:水源距离、0.5m灌丛盖度、2 m灌丛盖度和灌丛数量。褐马鸡育雏期成鸡（1.95±0.69）和幼鸡（8.45±2.28）,生境偏向于阔叶林,半阴半阳坡,下坡位和中坡位,沟底,且具有距离水源和林间小路要近,乔木层盖度和种类要大、灌丛和草本密度要小,灌木层和草本植物盖度要小,落叶要厚和裸地比例要大的特征。对各种生境因子的主成分分析,前6个特征值的累积贡献率达到74.63%,可以较好地反映其家族群觅食地生境特征,根据载荷系数绝对值大小将褐马鸡家族群觅食地选择影响因子分别命名为成鸡应急因子、小鸡应急因子、食物丰富度因子、隐蔽因子和逃跑因子。褐马鸡夏季卧息地资源选择函数为:logit（p）=-7.222+1.442×水源距离-2.132×可视度+1.862×草本盖度+2.135×草本高度,根据拟合出的资源选择函数,夏季褐马鸡对卧息地选择概率为P=e^logit（p）/(1+e^logit（p）),模型的正确预测率为94.7%。同时,根据选择系数的绝对值大小,可视度和草本高度是关键因子,草本盖度和水源距离是次要因子。采用Vanderloeg和Scavia选择指数和资源选择函数表明:春季褐马鸡取食生境偏向于低海拔、沟底、乔木密度小于5个/100m²、乔木直径大于20 cm、草本盖度小于50%的针阔混交林;夏季选择偏向于高海拔、坡度小于20°、草本高度大于20 cm和可视度小于10%的阔叶林。说明了随着北方夏春季食物数量和质量的剧烈变化,黄龙山褐马鸡对取食生境的利用对策也将发生一定程度的改变。褐马鸡秋季觅食地选择,通过主成分分析表明,影响褐马鸡生境选择的主要因子有4种,依次为:地形及草本植物因子（海拔、草本盖度、草本高度、地理性和坡位）、隐蔽因子（灌丛盖度、灌丛密度、灌丛高度和可视度）、位置因子（水源距离、人为干扰距离、植被类型和林边距离）、食物因子（乔木直径、乔木高度、乔木密度和乔木盖度）。卧息地的选择和利用资源选择函数为:logit（p）=9.162-1.605×可视度-1.187×灌丛密度-0.757×乔木盖度-0.617×坡度,根据拟合出的资源选择函数,秋季褐马鸡对卧息地选择概率为P=e^logit（p）/(1+e^logit（p）),模型的正确预测率为82.6%。同时,根据选择系数的绝对值大小,可视度和灌丛密度是关键因子,乔木盖度和坡度是次要因子。冬季褐马鸡群体平均大小为7.17±2.86只,且有固定活动范围。从逐步判别分析的结果看出,坡位、乔木盖度、灌木密度、可视度、人为干扰距离具有重要作用,由这5个变量构成的方程在对冬季栖息地利用样方和对照样方进行区分时,正确判别率可以达到81.16%。采用对整个研究区域进行系统取样的方法。采用向前筛选的逐步逻辑斯蒂回归分析,最后选择具有最小AIC_C值的回归等式为最佳的回归模型,其模型为:π（x）=e^g（x）/(1+e^g（x）),g（x）=22.107-0.004×林缘距离+13.623×乔木盖度-0.021×海拔。模型表明,距最近林边距离、乔木盖度和海拔显著影响褐马鸡群体的栖息地选择。褐马鸡群体的栖息地选择与乔木盖度正相关,与距离林边最近距离和海拔负相关。褐马鸡卧息地通过逐步判别分析表明,人为干扰距离、坡位、林边距离、海拔、可视度、灌丛盖度和乔木数量具有重要作用,由这7个变量构成的方程在对冬季卧息地利用样方和对照样方进行区分时,正确判别率可以达到93.60%。褐马鸡的卧息地选择主要与取暖条件和隐蔽性有关。褐马鸡冬季夜栖地多偏向阳坡和半阴半阳坡、中坡位、山脊和山坡、海拔高度低、接近林边、人为干扰距离较近、乔木盖度和密度较大、栖树胸径较大和草本层盖度较大的地方。主成分分析表明,冬季夜栖地生境选择影响因子分别命名为地理因子、位置和植被因子、夜栖树因子和逃避因子。我们评估冬季褐马鸡和雉鸡的生境选择及其差异,研究发现,两者生境选择存在一定的重叠性。但是它们也表现出生境选择的一些差异。分析两者共存的机制,我们发现雉鸡主要采取隐藏策略而褐马鸡采取逃跑策略来避敌和利用资源,即它们之间存在与反捕食策略相关的生境选择差异。同时,褐马鸡春夏秋冬海拔高度是发生变化的,夏季最高,次之秋季,接下来春季,最低是冬季,即夏季＞秋季＞春季＞冬季。坡位、乔木数量和灌木数量等与水源距离和食物丰富度相关的因子是其春季喜欢在低海拔区域活动的最主要原因;灌丛高度、草本高度、可视度和距林缘距离等与增加食物丰富度和隐蔽性相关的因子是促使它们在夏季倾向于在高海拔区域活动的主要因素;乔木盖度、灌丛盖度和草本盖度等与增加食物丰富度相关因子是秋季也倾向较高海拔地区活动;冬季距林缘距离、乔木盖度和可视度等与食物丰富度和隐蔽性相关因子使其向低海拔地区迁移。更多还原

【Abstract】 Brown-eared pheasant（Crossoptilon mantchuricum） is a world vulnerable and nation threatened species of pheasant that is native to China,and is listed in the first grade nationally protected animals of China.It is distributed in the Xiaowutai Mountain of Hebei,Lvliang Mountain of Shanxi, Huanglong Mountain of Shaanxi Provinces and Dongling Mountain of Beijing municipality.They are divided 3 isolated geogrpahical population which are respectively called by Middle population only including ones of Shanxi,east population including ones of Hebei and Beijing,and west including ones of Shaanxi because of geography barrier and destroy of nature vegetation.The brown-eared pheasant from Huanglong Mountain of Shaanxi Provinces lies only in Huanglong Mountain region involved in Huanglong county,Yichuan county and Hancheng city.Because the Yanan Reserve of the Huanglong Mountains（109°38′-110°12′E,35°28′-36°02′N） in Shaanxi Province has been bulit before long,lots of basic data were needed for pretection of brown-eared pheasant.Therefore,the further understanding of feeding-sites,noon-resting sites,nest-sites selection,roosting-sites and flocking behavior is essential before any managemen efforts are initiated by the resevre.And these may make up the blankness of habitat of west population.We have been carrying out on habitat selection of brown-eared pheasant with transect line methods in Beisi Mountain of Huanglong region for March 2006 to January 2007.There are some results as follow:From March to April,geographical,density of trees,type of shrub,density of shrub and cover of grasses were critically factors to discriminate the feeding-sites and random plots using discriminant analysis,the overall classification model developed from those five variables was successful to distinguish the feeding-sites from random ones at probability of 96.08%;Slope degree, slope aspect,distance to road and cover of trees were critical factors to discriminate the noon-resting sites and random plots using discriminant analysis,the overall classification model developed from those four variables was successful to distinguish the feeding-sites from random ones at probability of 75.64%.From April to June,density of trees,distance to water source,density of shrub,type of shrub,upmost height of trees and altitude were critical factors to discriminate the feeding-sites and random plots using discriminant analysis.The overall classification model developed from those six variables was successful to distinguish the feeding-sites from random ones at probability of 97.22%. Feeding-site selection of brown-eared pheasant was mainly related to food,concealment and water source.Furthermore,the noon-resting sites of brown-eared pheasant preferred conifer forest,ridge, half sunny and half shady slope,smaller slope degree,more far distance to trail and village,lower average height and density of shrub,lower average height of grasses,and bigger coverage of trees and grasses.And the results of principal component analysis showed that the first 6 principal components explained 74.05%of the total variance among all habitat variables.According to the absolute value of coefficient,the 6 components were classified as geography factor,sheltering class factor,human-disturbance factor and density of shrub factor.The nest-sites selection of brown-eared pheasant had a significant tendency by sampling method.Firstly,this tendency was represented on the scale of landscape.Secondly,it depended on vegetation structure.By Univariate analysis of logistic regression,the cover degree of shrub at height of 0.5m and 2m,density of shrub and distance to water source are the most important factors in the nest-sites selection of brown-eared pheasant. The brood size averaged 10.44±2.60,including adult pheasant（1.95±0.69） and infants（8.45±2.28）. Brood habitat of brown-eared pheasant are characterized by the broadleaf forest,half sunny and half shady slope,lower and mid slope,gully,closer to water resources and trail,bigger coverage and type of trees,lower density and cover of shrub and grasses,heavier leaf-fallen layer and greater proportion of bareness.And the results of principal component analysis showed that the first 6 principal components explained 74.6%of the total variance among all habitat variables.According to the absolute value of coefficient,the 6 components were classified separately as an emergency factor of adult pheasant,an emergency factor of young pheasant,food abundance factor,sheltering class factor and escaping factor.During July to August in 2006,the habitat selection of brown-eared pheasant was also investigated.The Resource selection functions of the noon-resting sites is a log-linear model: Logit（p）=-7.222+1.442×DW-2.132×VC+1.862×CG+2.135×HG（DW:distance to water source; VC:visibility class;CG:coverage of grasses;HG:average height of grasses）.From this model,we can estimate the noon-resting selection probability at P=e^logit（p）/(1+ e^logit（p）),the overall classification model developed from those four variables was successful to distinguish the noon-resting sites from random ones at probability of 94.7%.The model also indicated that the key factors influenced on the noon-resting selection were visibility class and average height of grasses, the secondary factors were the coverage of grasses and distance to water source.Then we compared with feeding samples between sping and summer using Vanderloeg & Scavia Index and Resource Selection Functions,the results indicated that the key factors in sping were lower altitude,gully, lower density of trees（＜5 individuals/100 m²）,bigger diameter of trees（＞20cm） and coverage of grasses（＜50%）,while in summer the key factors were higher altitude,slope degree（＜20°）,higher hight of grasses（＞20 cm） and lower visibility class（＜10%）.These showed that in the northern temperate region,the quantity and quality of feeding species have experienced great changes between summer and spring,hence the strategy of feeding habitats used by brown-eared pheasant varied with the season.During September to October in 2006,the habitat selection of brown-eared pheasant was also investigated.The results of principal component analysis showed that the first 5 principal components explained 74.069%of the total variance among all habitat variables.According to the absolute value of coefficient,the 5 components were classified as geography and grasses-vegetation factor（altitude,coverof grasses,average height of grasses,geographical and slope location）, sheltering class factor（cover of shrub,density of shrub,average height of shrub and visibility class）, position factor（distance to water source,distance to edge of woods,distance to human disturbance and vegetation type） and food factor（average diameter of trees,average height of trees,density of trees and cover of trees）.The Resource selection functions of the noon-resting sites is a log-linear model:Logit（p）=9.162-1.605×VC-1.187×DS-0.757×CT-0.617×SD（VC:visibility class;DS:density of shrub;CT:coverage of trees;SD:slope degree）.From this model,we can estimate the noon-resting selection probability at P=e^logit（p）/(1+ e^logit（p）),the overall classification model developed from those four variables was successful to distinguish the noon-resting sites from random ones at probability of 82.6%.The model also indicated that the key factors influenced on the noon-resting selection were visibility class and density of shrub,the secondary factors were the coverage of trees and slope degree.From November to December 2006 and January 2007,the winter habitat of brown-eared pheasant was also investigated.Population size of brown-eared pheasan averaged 7.17±2.86,and activitied in a stable home range.Slope location,coverage of trees,density of shrub,visibility class and distance to human disturbance were critically factors to discriminate the feeding-sites and random plots by using discriminant analysis.The overall classification model developed from those five variables was successful to distinguish the feeding-sites from random ones at probability of 81.16%.By using Line transects of systematic sampling,the wintering habitat model of brown-eared pheasan could be formally expressed as:π（x）=e^g（x）/1+e^g（x）,g（x）= 22.107-0.004×DW+13.623×CT -0.021×AH（DW:distance to nearest edge of woods;CT:cover of trees;AH:altitude）.The model suggested that wintering habitat selection of brown-eared pheasant was negatively related to distance to nearest edge of woods and altitude,positively related to cover of trees.Distance to human disturbance,slope location,distance to edge of woods,altitude,visibility class,coverage of shrub and density of trees were critically factors to discriminate the noon-resting sites and random plots by using discriminant analysis.The overall classification model developed from those seven variables was successful to distinguish the feeding-sites from random ones at probability of 93.60%.The noon-resting sites of brown-eared pheasant is related to sunlight and concealment.The roosting-sites of brown-eared pheasant were characterized by sunny slope,and half sunny and shady slope,mid slope,slope side and ridge,lower altitude,closer to edge of woods and human disturbance,bigger density and coverage of trees.And the results of principal component analysis showed that the first 4 principal components explained 73.542%of the total variance among all habitat variables. According to the absolute value of coefficient,the 4 components were classified as geography factor, position and vegetation factor,roosting-tree factor and escaping factor.And we studied wintering habitat selection of brown-eared pheasant and common pheasant.Habitat selection of brown-eared pheasant and common pheasant showed a certain extent of overlap,while showed a certain difference.We suggest that the two species coexist in the Yanan Reserve of the Huanglong Mountains as a result of differential habitat selection associated with their species-specific anti-predator strategies.Common pheasant avoid predators by concealment whereas brown-eared pheasant evade predators by running as their escape strategy.And the altitude of their home range varied with seasons:highest in summer and higher in autumn,lower in spring and lowest in winter.The most likely reason that brown-eared pheasant preferred to live at lower altitude in winter is to have access to the abundance of food and shelter.In summer,they tended to appear at highest altitude,correlated with the distance to human disturbance and shelter.In spring,they tended to appear at higher altitude,correlated with the distance to water source and the abundance of food.In autumn,they tended to appear at higher altitude,correlated with the abundance of food.更多还原