【作者】 刘群秀；

【导师】 王小明；

【作者基本信息】 华东师范大学 ， 生态学， 2009， 博士

【摘要】 藏狐主要分布于青藏高原海拔3500 m以上的区域,多在地势开阔的高原草原和高山草甸生境中活动。藏狐主要捕食高原鼠兔和草原啮齿类,在控制草原鼠害、维持高原生态系统平衡方面具有十分重要的生态学作用。长期以来,藏狐的生物学信息比较匮乏,开展针对藏狐的基础生态学研究将有助于我们深入了解高原极端环境中藏狐采取的适应性生活史对策,为高原动物资源的管理和保护,高原生态环境的改善和治理奠定理论基础。2006年4月和9-10月、2007年3-5月和9-10月、2008年3-4月,我们在青海省都兰县沟里乡捕捉到5只藏狐(4雄,1雌)。通过无线电遥测技术追踪藏狐,记录其活动位点,并观察藏狐的行为。根据藏狐活动位点的分布及在该点的行为,分析藏狐昼间行为在时间和空间上的分布特征,并探讨藏狐活动指数与环境变量之间的关系;应用地理信息系统扩展软件计算藏狐的家域面积及藏狐的利用分布特征,并尝试寻找核心活动区;通过在藏狐利用分布可能性与环境变量属性间建立多元线性回归模型,探讨藏狐对环境变量的选择利用特征;根据藏狐对环境因子的喜好,开展生境评价,并计算适宜生境中藏狐的空间环境容纳量;除此之外,我们还研究了藏狐的家域功能分区及其生物学特性、藏狐的洞穴生境选择、藏狐的食性组成及其他同域分布的物种的生物学信息。1.藏狐昼间活动规律研究结果表明:藏狐主要在晨昏时段活动(χ~2检验,P<0.05),其昼间活动节律分配存在显著季节性变化(Wilcoxon Signed RanksTest,Z=-2.366,P=0.018)。藏狐在暖季的活动指数((?)=0.60,SD=0.14)高于冷季((?)=0.49,SD=0.15)。藏狐的活动高峰时段主要集中在早8:00-12:00时和傍晚16:00-20:00时,其他时段活动程度较低。藏狐行为受地形因子的影响,在空间上存在集中分布区。藏狐的活动指数与坡度因子属性呈负相关(B=-0.099),与高原鼠兔的密度呈正相关(B=0.022),鼠兔密度和坡度因子的交互作用对藏狐行为的影响更明显(B=0.073);此外,藏狐的活动指数与海拔因子属性呈负相关(B=-0.023)。2.幼龄藏狐昼间行为分配在持续时间(χ~2=124.160,df=5,P=0.000)和发生频率(χ~2=78.785,df=5,P=0.000)上均为非正态分布,以休息行为为最主要行为(持续时间百分比为55.1%,频率百分比为45.0%),其次为捕食行为(持续时间百分比为20.0%,频率百分比为19.0%)和嬉戏行为(持续时间百分比为16.7%,发生频率百分比为39.0%)。幼龄藏狐主要选择晨昏活动,早8:00-9:00时以捕食和嬉戏行为为主,9:00时以后休息行为逐渐增加并在13:00-14:00时达到最高,14:00-15:00时捕食和嬉戏行为再次增多,休息行为减少,直至15:00-16:00时,重新以休息为主。成体藏狐出现时,幼龄藏狐的行为表现出高度的活跃性(持续时间,χ~2=157.546,df=5,P=0.000;发生频率,χ~2=213.642,df=5,P=0.000),以嬉戏(持续时间百分比为50.0%,发生频率百分比为54.7%)和奔跑(持续时间百分比为16.2%,发生频率百分比为18.6%)行为为主。3.应用固定核空间法、调和平均值法和最小多边形法估算藏狐的家域面积,并应用独立区域法(AIM)划分核域。固定核空问法计算结果:暖季节藏狐家域面积从2.81 km~2到3.29 km~2不等((?)=2.95 km~2,SD=0.22),核域面积((?)=0.88km~2,SD=0.37)约占家域面积的17.3%-38.1%;冷季节家域面积从1.99 km~2到3.70 km~2不等((?)=3.36 km~2,SD=1.07),核域面积((?)=1.00 km~2,SD=0.49)占家域面积的15.7%-36.3%((?)=29.9%,SD=9.5%)。调和平均值法表明:暖季节藏狐的家域面积为0.87 km~2-2.89 km~2不等((?)=1.67 km~2,SD=1.08),核域面积((?)=0.33 km~2,SD=0.14)占家域面积的16.6%-34.5%((?)=22.7%,SD=10.2%);冷季节藏狐家域面积1.33 km~2-4.99 km~2不等((?)=3.09 km~2,SD=1.68),核域面积为0.29 km~2-1.11 km~2不等((?)=0.64 km~2,SD=0.40)。藏狐不同个体间家域及核域有不同程度的重叠:暖季节不同藏狐个体间家域重叠度指数((?)=0.24,SD=0.12)小于冷季节((?)=0.36,SD=0.19)(Mann-Whitney U,Z=-1.037,P=0.300)。在暖季节,两只雄性个体家域重叠程度较大(OI=0.38),而在寒冷季节,重叠主要发生在雌(F1)和雄(M3)个体间(OI=0.74)。核域的重叠程度显著小于家域的重叠(Mann-Whitney U,Z=-3.112,P=0.001)。核域的重叠主要表现在冷季节F1和M3个体间((OI=0.29)。4.应用无线电遥测技术和空间分析技术,将藏狐在家域范围内的空间利用分布和行为分布相结合,把藏狐的家域划分为四类功能区:核心活动区、核心非活动区、周边活动区和周边非活动区,分析家域的不同功能分区展示的生物学特性。藏狐核心活动区面积((?)=0.45km~2,SD=0.16)均小于核心非活动区((?)=0.51km~2,SD=0.16),周边活动区面积((?)=0.84km~2,SD=0.12)则大于周边非活动区((?)=0.63 km~2,SD=0.19)。藏狐家域的核心活动区(Mann-Whitney U test,Z=-6.364,P=-0.000)和周边活动区(Mann-Whitney U test,Z=-4.192,P=0.000)中高原鼠兔密度均显著高于对应的非活动区。同核心非活动区相比,藏狐的核心活动区多选择坡度较缓(Mann-Whitney U test,Z=-3.011,P=0.003),海拔较低(Mann-Whitney U test,Z=-2.570,P=0.000),中坡位或下坡位(χ~2=23.229,df=5,P=0.000)的区域;同周边非活动区相比,周边活动区多选择坡位较低(χ~2=11.257,df=5,P=0.047)的区域,对坡度(Mann-Whitney U test,Z=-1.778,P=0.075)和海拔(Mann-Whitney U test,Z=-0.881,P=0.378)无显著选择性。5.基于多元线性回归模型对藏狐的生境选择进行初步分析。模型筛选中,高原鼠兔密度及其平方项均为藏狐生境选择重要的影响因子,地形起伏度、坡度和海拔亦重要的回归变量。在温暖季节,最佳简约模型包含高原鼠兔密度及其平方项、地形起伏度、坡度、海拔、鼠兔密度与地形起伏度和坡度的交互作用项等(k=8),模型的AICc权重值为AICcw=0.998;在寒冷季节,最佳简约模型中包含高原鼠兔密度及其平方项、地形起伏度、坡度、海拔、鼠兔密度与坡度的交互作用项等(k=7,AICcw=0.374)。藏狐在冷季和暖季均对高原鼠兔密度因子呈现非线性选择,其活动范围主要集中于鼠兔密度中等的区域;藏狐对地形起伏度呈现正选择,而对地形起伏与鼠兔密度的交互项呈负选择;藏狐回避高海拔区域。藏狐对坡度的选择在不同季节呈现相反的选择趋势。藏狐活动点与河流因子的距离((?)=806.4m,SD=301.9)显著小于随机点((?)=917.9m,SD=467.6)(ANOVA,F=7.876,P=0.005),同时对道路呈现正选择性利用(ANOVA,F=10.556,P=0.001)。藏狐对人类活动无明显回避行为,距离较近,但没有呈现显著选择(ANOVA,F=2.830,P=0.093)。6.将藏狐洞穴划分为繁殖洞穴和休息洞穴,洞穴的生境选择利用存在以下特征:繁殖洞穴位点对坡向没有明显的选择性(Rayleigh Test,P>0.05),但休息洞穴主要位于阳坡((?)=43.0°,SD=6.9)和半阳坡((?)=98.2°,SD=18.0)区域(Rayleigh Test,P<0.05)。地形起伏较大的区域常被藏狐利用作为洞穴生境(Mann-Whitney U Test,P<0.05),并且藏狐洞穴位点的视野范围显著大于随机点(Mann-Whitney U Test,P<0.05)。藏狐的休息洞穴多位于陡坡、低鼠兔密度、与其他洞穴距离较近的区域。藏狐洞穴位点并未呈现出对人类活动的回避,海拔、光照指数、与水源和公路的距离等对藏狐的洞穴生境影响亦不明显(Mann-Whitney U Test,P>0.05)。7.应用粪便分析法初步研究了藏狐的食性组成。分析检测到的食物条目包括高原鼠兔、啮齿类、草原旱獭,牦牛、鸟类、两栖类、昆虫和植被等。藏狐对不同食物的条目具有选择性(9-10月份:χ~2=360.61,df=9,P=0.000,3-5月份:χ~2=358.17,df=10,P=0.000)。食物组成中,高原鼠兔出现比例最高((?)=84.8%,SD=10.6),其次为啮齿类((?)=17.3%,SD=13.4)。在鼠兔密度较低的3-5月份,藏狐食物组分中草原旱獭出现频率较高。牦牛、藏原羚、岩羊等仅在少数样本中发现。9-10月份高原鼠兔的密度((?)=4.15,95%置信区间为3.42-4.88)显著高于3-5月份((?)=2.67,95%置信区间2.31-3.03)(U=2376,Z=-3.961,P<0.001),但藏狐食性组成无季节性的显著变化(χ~2=9.14,df=11,P=0.609)。藏狐全年的食物生态位宽度为1.32,Shannon-Weaver多样性指数和均匀性指数分别为2.29和0.66。8.依据藏狐的“出现点”数据,基于海拔、坡度、坡位和地形起伏度等4个地形因子对藏狐的生境质量进行初步评价,并基于藏狐最小空间需求估算研究地区该物种的空间环境容纳量。通过比较藏狐对各因子的利用及其可获得性建立单个因子的评价准则,并依据各因子的分布特征进行综合生境评价。藏狐最适宜的海拔高度为4050-4300 m,坡度为5-20°,坡位为上坡位和下坡位,对地形起伏度因子没有明显的喜好。根据藏狐对不同生境因子的利用,及不同个体对生境利用的一致性,建立综合评价准则:(?)。研究地区适宜性区域面积约为17.4 km~2,占总研究地区面积的38.8%,其中最适宜生境仅0.4 km~2。藏狐的最小空间需求为2.09-3.55 km~2,研究地区藏狐的空间环境容纳量为7-12只。9.通过指数法调查高原鼠兔的空间密度分布特征。暖季节高原鼠兔种群密度显著高于冷季节(U=2376,Z=-3.961,P<0.001),并在空间分布上呈现季节性变动。高原鼠兔的密度分布与海拔因子呈非线性相关(y=-20.484 x~2+20.501x,r~2=0.324),随着海拔的升高,鼠兔密度呈下降趋势,在海拔高度达到4300 m以上鼠兔密度降到最低,并逐渐趋于稳定;高原鼠兔对坡度因子呈负选择(y=-0.784 x~2+0.737 x,r~2=0.218),对地形起伏度的选择利用主要选择较平坦和较起伏的区域(y=1.415 x~2-1.455 x,r~2=0.246)。更多还原

【Abstract】 The Tibetan fox distributed in Qinghai-Tibet plateau with elevation higher than 3500 m, and prefer broad alpine steppe and meadow areas. Tibetan fox prey mainly on Plateau pika and alpine rodents, which is necessary to maintain grassland quality and keep balance of alpine ecosystem. For a long time, the biological information of Tibetan fox is scarce and little is known about this species. Therefore, research works on fundamental ecology about Tibetan fox is helpful to understand the living pattern adapted by this species in extreme environment, and provide theorietical basis for specis conservation and habitat enhancement.We captured 5 Tibetan foxes ( 4 M, 1 F ) in Gouli township, Dulan county, Qinghai Province in April 2006, September-October 2006, March-May 2007, September-October 2007 and March-April 2008. Radio telemetry technique was used to locate all collared foxes, and the behavior happened at each location was recorded. The temporal and spatial distribution characteristics of the diurnal behavior of Tibetan fox were analyzed, and the relationship between activity index and habitat variables was studied by regression; Based on locations, the home range and home range core of Tibetan fox were estimated through extensions of GIS software; Multi-linear regression model was set to explore the relationship between the probability of utilization distribution and environmental variable attributes; Habitat quality was evaluated according to animal preference (or avoidance) to certain habitat, and the spatial carrying capacity was calculated for Tibetan fox in suitable habitat; We also sepetated fox home range into different biological function areas, studied the denning habitat selection of Tibetan fox, analyzed the food habits of Tibetan fox and provide biological information of some species sympatric with Tibetan fox.1. Diurnal activity characteristics: Tibetan fox moves mainly at dawn and dusk (x~2 test, P < 0.05 ) , and there is significant seasonal difference in behavior rhythm ( Wilcoxon Signed Ranks Test, Z =-2.366, P = 0.018 ) . The fox activity index in warm season ((x|-) =0.60, SD = 0.14) is higher than that in cold season ((x|-)= 0.49, SD = 0.15) . In both warm and cold season, foxes were more active at 8:00-12:00 and 16:00-20:00 and less active at other periods. The spatial distributon of fox activities are influenced by topographic factors. There is minus correlation between fox activity index and slope attributes (B = -0.099), and the activity index is positively correlated with pika density (B - 0.022) . The interaction between pika density and slope affects fox activity index positively (B = 0.073). Foxes will be less active with the increasing of elevation (B = -0.023) .2. The diurnal behavior rhythm of Tibetan fox pups: The diurnal behavior rhythm of young pups was not normally distributed (χ~2 test, P < 0.05) , and mainly composed by resting (Time percentage is 55.1 %; Frequency percentage is 45.0 %) , hunting (Time percentage is 20.0 %; Frequency percentage is 19.0 %) and gamboling behavior (Time percentage is 16.7 %; Frequency percentage is 39.0 %) . The pups were active at dawn and dusk; they are active at 8:00-9:00 and 15:00-16:00, and less active 13:00-14:00. Young foxes were much active when adult fox appeared(χ~2 test, P < 0.05), and their behavior were mainly composed by gamboling (Time percentage is 50.0 %; Frequency percentage is 54.7 %) and running behavior (Time percentage is 16.2 %; Frequency percentage is 18.6 %) .3. Fixed kernel, Harmonic mean and Minimum convex polygon estimators were used to calculate the home range of Tibetan fox, and the home range core was determined by Area of independent method (AIM). Fixed kernel estimator: In warm season, the home range size of Tibetan fox was 2.81 km~2-3.29 km~2 ((?) = 2.95 km~2, SD = 0.22) , the home range core ( (?) = 0.88 km~2, SD = 0.37) take a percentage of 17.3 %-38.1 % of the total home range; in cold season, the home range size is 1.99 km(?)-3.70 km(?) ((?) = 3.36 km~2, SD = 1.07), and the home range core size ((?)= 1.00 km~2, SD = 0.49) is 15.7 %-36.3 % ((?) = 29.9%, SD = 9.5%) of the total home range. Harmonic mean estimator: In warm season, the home range size is 0.87 km~2-2.89 km~2 ((?) = 1.67 km~2, SD= 1.08), and the home range core ((?) = 0.33 km~2, SD = 0.14) take a percentage of 16.6 %-34.5 % ((?) =22.7 %, SD = 10.2 %); In cold season, the home range size of Tibetan fox is 1.33 km~2-4.99 km~2 ((?) = 3.09 km~2, SD = 1.68) , and the home range core size is 0.29 km~2-1.11 km~2 ((?)= 0.64 km~2, SD = 0.40)。There is overlap between home range and core area of different fox individuals. The home range overlap in warm season ((?) =0.24, SD = 0.12) is significantly lower than that in cold season ((?) = 0.36, SD = 0.19) (Mann-Whitney U, Z = -1.037, P = 0.300) . The overlap between home range cores is significantly lower than that between home ranges (Mann-Whitney U, Z = -3.112, P = 0.001) .4. Based on the behaviors and spatial distributions of resource use, we partitioned home ranges into 4 main functional regions: core active region, core inactive region, peripheral active region, and peripheral inactive region. We measured biological characteristics of each region. Core active regions ((?) - 0.45km2, SD = 0.16) were smaller than core inactive regions ((?) = 0.51 km2, SD = 0.16), and the peripheral active regions ((?) = 0.84km2, SD = 0.12) were larger than peripheral inactive regions ((?) = 0.63 km2, SD = 0.19). Pika (Ochotona curzoniae) densities in both the core active region (Mann-Whitney U test, Z=-9.310, P = 0.000) and peripheral active region (Mann-Whitney U test, Z=-4.762, P = 0.000) were significantly higher than those in counterpart inactive regions. Compared with core inactive regions, core active regions were more likely to be located in areas with gentle slopes (Mann-Whitney U test, Z =-3.011, P = 0.003) , lower elevations(Mann-Whitney U test, Z =-2.570, P = 0.000) and lower positions on slopes(χ~2 =23.229, df- 5, P=0.000) . Compared with peripheral inactive regions, lowerslope positions (χ~2 =11.257, df- 5, P= 0.047) were preferred by Tibetan foxes forperipheral active areas, whereas slope (Mann-Whitney U test, Z =-1.778, P = 0.075)and elevation (Mann-Whitney U test, Z =-0.881, P = 0.378) did not differ.5. Habitat selection of Tibetan fox was studied based on multi-linear regression model. Pika density, terrain ruggedness, slope and elevation are all important impacts on habitat selection by Tibetan fox. In warm season, the most parsimonious model contains pika density, the quadratic term of pika density, terrain ruggedness, slope, elevation, the interaction between pika density and terrain ruggedness, the interaction between pika density and slope (k = 8), the AICcw value is 0.998; In cold season, the most parsimonious model contains pika density, the quadratic term of pika density, terrain ruggedness, slope, elevation, the interaction between pika density and slope (k = 7, AICcw = 0.374) . There is no-linear relationship between pika density and utilization distribution, and Tibetan fox more prefer moderate pika density areas; Tibetan fox keep avoidance to higher elevations, use terrain ruggedness with no-linear selection and showed reverse selection on slope in different season. The distances of fox locations from river ((?) = 806.4 m, SD = 301.9) are significantly shorter than those of random points((?) = 917.9 m, SD = 467.6) (ANOVA, F= 7.876, P = 0.005). Comapred with random points, Tibetan fox shows positive selection to road (ANOVA, F = 10.556, P = 0.001), and did not avoid human activities (ANOVA, F = 2.830, P = 0.093) .6. Tibetan fox dens were classified into natal dens and resting burrows. Tibetan foxes did not use special slope aspects for natal dens (Rayleigh Test, P > 0.05) , but use sunny ((?) =43.0°,SD =6.9) and half-sunny ((?)=98.2°,SD=18.0) slope aspects for resting burrows (Rayleigh Test, P < 0.05) . Tibetan fox choose more rugged area for both natal dens and resting burrows (Mann-Whitney U Test, P < 0.05) . The visible viewshed gird number of natal den and resting burrow sites were significantly larger than those of random points (Mann-Whitney U Test, P < 0.05) . Resting burrows located at areas with steeper slopes, lower pika abundance, smaller distance to other dens. Locations of dens and burrows were not significantly affected by elevation, sun index, distances to river and road. Tibetan foxes did not avoid manmade features for their resting natal dens and resting burrows (Mann-Whitney U Test, P> 0.05) .7. The diet composition of Tibetan fox was analyzed by frequency of occurrence method based on scat-analysis in periods of September-October (high pika density) and March-May (low pika density) . The food items contain plateau pika, rodents, marmmot, yak, birds, reptile, insect and plant. Tibetan fox did not prey food item randomly ( September-October,χ~2 =360.61, df = 9, P = 0.000; March-April,χ~2 =358.17, df = 10, P = 0.000) . Black lipped pika ((?) =84.8%, SD=10.6) and rodents ((?)=17.3%, SD=13.4) served as the main food components of Tibetan fox. Marmot, Blue sheep, Tibetan gazelle were found in fox scats, and more frequently detected in March-May season. In September-October season, the pika density ((?)= 4.15, 95% confidence interval is 3.42 - 4.88) is significantly higher than that in March-May season ((?) = 2.67, 95% confidence interval is 2.31 -3.03) (U= 2376, Z= -3.961, P < 0.001), but the diet composition of Tibetan fox did not significantly changed (χ~2 = 9.14, df=11,P = 0.609) . The food niche breadth of Tibetan fox is 1.32, the Shannon-Weaver diversity index and evenness index are separately 2.29 and 0.66. 8. Habitat quality evaluation: Based on "Presence data", we assessed the habitat quality of Tibetan fox in Gouli Township, Dulan County, Qinghai Province, China. We also estimated the spatial carrying capacity of Tibetan fox according to their minimum space requirements. A total of 4 topographical factors, elevation, slope, terrain ruggedness and slope position, were used to conduct habitat assessment for Tibetan fox. The suitability of habitat factors were determined by comparing the utilization by Tibetan fox and its availability based on 95% Bonferroni confidence interval. A comprehensive assessment criterion was set up for all topographic factors. For Tibetan fox, the elevations of 4050-4300m, the slopes of 5-20°, the upper and lower slope positions were suitable habitat. Tibetan fox use terrain ruggedness non-selectively. According to the utilization of Tibetan fox on different topographic factors and the consistence of habitat utilization by different fox individuals, we setthe comprehensive criterion as: (?). The area of suitable habitat is about 17.4 km~2, which take a percentage of 38.8% of research area. The area of most suitable habitat is only 0.4 km~2. The home range of Tibetan fox ranges from 2.53 km~2 to 4.99 km~2, and the overlap index (OI) of different individuals were 0.16-0.66. The minimum space requirements of Tibetan fox are 2.09-3.55 km~2, and the spatial carrying capacity of Tibetan fox in suitable habitat is 7-12.9. Index of pika abundance was used to investigate the spatial density distribution characteristics. The pika density in warm season is significantly higher than that in cold season (U = 2376, Z = -3.961, P < 0.001) , and the spatial distribution changed. The pika density distribution was correlated with elevation attributes (y = -20.484 x~2 + 20.501 x,r~2= 0.324), the pika density will decrease with the enhancement of elevation, when the elevation attributes reached 4300 m, the pika density will not decrease any more. The regression relationship between pika density and slope was: y = -0.784 x~2 + 0.737 x, r~2 = 0.218, and plateau pika prefer flat or most rugged areas (y= 1.415 x~2-1.455 x, r~2=0.246) .更多还原