【作者】 张明海；

【导师】 马建章；

【作者基本信息】 东北林业大学 ， 野生动植物保护与利用， 2005， 博士

【摘要】 2002年11月—2005年2月,作者连续3个冬季对黑龙江省完达山林区马鹿生境破碎化及生态影响进行了较为系统地研究。在野外调查研究的基础上,基于3S技术和多元统计分析,获得了以下主要结果: 1 马鹿冬季破碎化因子分析及其空间格局特征: （1） 影响马鹿冬季生境破碎化的14个因子中的9个因子为主要生态因子,进一步的因子分析表明,人类活动这一综合因子是引起马鹿冬季生境破碎化最主要的因素,即在马鹿栖息生境内的森林采伐、农田开垦、居民点扩展和道路修筑及物资运输等一系列相关的人类经济、社会活动。 （2） 经主成分分析,结合逻辑斯缔回归分析,建立了马鹿冬季生境破碎化因子的LOGISTIC回归预测模型:log（P_i/1-P_i）=1.963+0.816PC₁-0.262PC₂-0.039PC₃,其正确预测率为71.1%,可以认为该模型基本上能够反映马鹿冬季生境破碎化因子的分布情况。 （3） 马鹿冬季生境破碎化因子具有以下主要空间格局特征:9个生境破碎化因子具有明显的空间自相关性,其显著性（p<0.05）,Moran’s,Ⅰ值达73%;动物干扰在近距离范围内表现出显著负相关特征,说明在此距离梯度下马鹿与狍、野猪等大型有蹄类存在着一定程度上的生态位竞争干扰,呈斑块状分布,斑块大小为230～920 m;居民点、农田和废弃运材路这3个因子在全部14个距离梯度下均表现出显著的正相关特征,这不仅说明它们对生境破碎化在空间格局上的影响有高度的一致性,而且说明这3个因子在马鹿冬季生境破碎化过程中起着重要的作用;马鹿在冬季破碎化生境下数量分布的半方差随空间距离梯度的变化呈“巢状结构”特点,反映出马鹿数量分布的空间格局存在着多尺度的变异;而马鹿分布数量的分维数D（1.926±1.360）接近于2,说明马鹿数量分布的空间格局变异主要发生在较小的尺度上,也就是说样方点出现马鹿足迹链数量的差异很大,较高的数量持续一段时间后往往随之以较低的数量。这充分揭示了马鹿数量分布的多尺度和等级斑块的空间分布特征和马鹿微生境的脆弱性,即一旦微生境发生变化,可能导致马鹿数量分布空间格局的较大改变。 2 马鹿冬季破碎化生境评价及景观结构分析认为: （1） 生境破碎化使马鹿冬季生境的景观适宜性面积所占比例逐渐减少,景观的异质性出现增大的趋势;斑块小而多且相互隔离,在空间分布上呈明显破碎化状态;由于景观连接度较低,降低了各生境因子在生态功能和生态过程中的联系及其作用的发挥。景观连接度进行模糊相对赋值及建立生境景观连接度评价模型,将马鹿冬季生境按由高及低的次序,划分为4个等级:适宜、次适宜、一般适宜和不适宜。结果表明:在研究地区的155.6 km²的总面积中,主要为一般适宜生境,其面积为130.05 km²,占研究地区的83.58%;适宜生境和次适宜生境的面积分别为14.81 km²和9.57 km²,仅占研究地区更多还原

【Abstract】 Fragmentation habitat and its influence on red deer population distribution were studied systematically in the Wandashan Mountains in Heilongjiang Province during three successive winters from November of 2001 to February of 2005. Based on field survey, the main results by geomatics technique （3S） and multiple statistics analysis were revealed as followings:1 Analysis of habitat fragmentation factors of red deer in winter and its characteristics of spatial pattern:（1） The nine factors were main ecological ones among the fourteen that had an infuence on habitat fragmentation of red deer in winter. Further analysis showed that human activity, a synthetical factor, played a main role in habitat fragmentation of red deer in winter, including forest logging, agricultural reclamation, settlement extension, road construction and material transportation.（2） Logistic regression model was established: log （P_i/1-P_i） =-1.963 + 0.816PC₁ — 0.262PC₂ — 0.039PC₃, which classification rate correctly was 71.1%. Hence it was believed that the model may reflect the distribution of habitat fragmentation factors of red deer during winter by way of PCA （principal component analysis） and binary logistic analysis.（3） Spatial autocorrelation analysis indicated that main spatial pattern characteristics of habitat fragmentation factors of red deer in winter were as following: a) under fourteen distance grading （230³220 m）, nine habitat fragmentation factors had spatial autocorrelation significantly （p < 0.05） and Moran’s I reached at 73%; b) ungulate disturbance appeared negative correlation significantly during the nearer disturbance grading, which revealed that there was niche competition among red deer, roe deer and wild pig to some extent, and this disturbance distributed fragmentally and the sizes of these patches fluctuated from 230 to 920m; c) settlements, abandoned logging roads and farmlands appeared correlated significantly during all fourteen distance grading which indicated not only higher consistency in influecing on the spatial pattern of habitat fragmentation but also their important role in the process of habitat fragmentation of red deer in winter; d) semi-variance analysis and analysis on fractals showed that "nest structure" characteristics of the changes of semivariance of population distribution with the changes of spatial distance degrading, which pointed out that there was a multiple scale variation in spatial pattern of red deer distribution; e) however, fractal number-D of red deer distribution approached 2, which indicated the variation of spatial pattern of red deer distribution mainly occurred at smaller scale, i.e. differences of track number of red deer occurred in sample sites were very large, and small number came after a period time of large number. The above revealed that multiple scale of red deer distribution, grading patches of spatial distribution and vulnerability of red deer miro-habitat, i.e. once microhabitat changed, spatial pattern of red deer distribution may change greatly.2 Analyses of fragmented habitat evaluation and landscape structure（1） By analysis of the importance of landscape factors, landscape connectivity, assessment of suitability and fuzzy value of connectivity, and establishment of a landscape connectivity model that applied landscape ecology and GIS technology, the connectivity level, suitability assessment and fragmentation status of all types of suitability areas were studied. The habitat of red deer was divided into four classes: suitable, marginally suitable, less suitable and unsuitable from high quality to the low. The result indicated that suitable habitats were 14.81 km2 （only 9.52%）, marginally suitable habitats were 9.57 km2 （6.15% ）, less suitable habitats were 130.05 km2 （83.58%）, unsuitable habitats were 1.17km2 （0.75%） in the total area of 155.6 km". In addition, among 51 patches in study area, the number of suitable habitat patches （23） was the most, accounting for 45.10%, but the numbers of less suitable habitats （5 patches） and unsuitable habitat （6 patches） were the least, accounting for 9.80% and 11.76% respectively.（2） Three landscape indices of describing the landscape pattern and change used presently, Mean PARA, Mean FRAC and Mean CONTAG, were introduced to estimate the landscape characteristic of the above four classes. The result indicated that the Mean PARA of suitable habitat （332.43） was the highest, but its Mean FRAC （1.050） and Mean CONTAG （0.749） were the lowest; while, the Mean PARA of unsuitable habitat （186.16） was the lowest, its Mean FRAC （1.059） was higher and its Mean CONTAG （0.837） was the highest, which indicated fully that the suitable habitat was more fragmented more isolated and heterogeneous than unsuitable habitat.3 Influence of forest logging on fragmented habitat selection and use of red deer in winter（1） Red deer heavily used clear-cutting land （SI=0.29） and mixed broadleaf forest （SI=0.08） which were selected positively, while selected negatively mixed coniferous and broadleaf stands, mixed poplar broadleaf forest, planted larchen forest, brush land and bare land, and did not use farmland completely （SI=0）. Habitat with low arbor crown density was heavily and positively selected by red deer （SI=o.62）, and with middle arbor crown density was positively selected （SI=0.16）, but with high arbor crown density was selected negatively.（2） Red deer selected positively the habitat range from 600 m to 1 km to open road, while hardly appeared in the ranger of 100 m to open road.（3） Settlements due to occuring to frequent human activity around it formed a buffer zone to red deer with radium 2.74 km, where no activity of red deer had been observed.（4） Red deer used higher the habitat that was logged at the middle or high frequency （logged one to five times during the recent 15 years）, while used lower tht habitat of being logged at the low frequency （no logging in the recent 15 years）.（5） With the extension of logging interval, the clear-cutting land experienced succession of different vegetation, which led to difference in abundance and availability of food. The forest that was logged in 7 years showed the higher suitable to red deer, where abundance and availability of food was comparatively higher. On the contrary, red deer used lowly the habitatthat was logged 7 or more than 7 years ago.（6） At present, clear-cutting and selection-cutting were two main logging way in the study area. Clear-cutting had a great change on landscape including component and structure, while selection-cutting had little influence. Red deer had no use planted forest （0.00%） after clear-cutting, had low use rate of natural secondary forest （4.40%） after clear-cutting, while had high use rate of newly clear-cutting and selection-cutting habitat （54.40% and 41.21% respectively）.4 Effect of fragmented habitat edge on population number and distribution of red deer in winter（1） Forest-shrub edge and farmland-forest edge had different edge effects on population number and distribution of red deer in winter significantly, and significant difference existed them by analysis of Mann-Whitney U （U=3.000, Z =-3.130, P=0.002）. Forest-shrub edge had a positive effect on activity and distribution, which meant the activity and distribution density of red deer was higher than those in bordered on interior of shrub or forest. Farmland-forest edge had a negative effect, namely the activity of red deer mainly were in a distance range of 200-680 m from forest interior to the edge, and little in the edge or farmland.（2） Red deer had a much higher use of forest-shrub edge （Hi=0.332） than farmland-forest edge （Hi=0.032）, due to high forage abundance an coverage of forest-shrub edge.（3） By the comparison of effects between forest-shrub and farmland-forest edge, the habitat factors of forest-shrub edge that had a positive edge effect on activity, population number and distribution of red deer were shrub coverage and arbor crown density to the edge, not arbor density and forage abundance. While the habitat factors of farmland-forest edge that had a negative edge effect were nearby human disturbance activity and foraging sites of red deer were mainly located the distance of 600- 800 m from forest interior to farmland.更多还原