【作者】 雷雅凯；

【导师】 吴波；

【作者基本信息】 中国林业科学研究院 ， 水土保持与荒漠化防治， 2012， 博士

【摘要】 毛乌素沙地是我国四大沙地之一，位于中国北方半干旱区的农牧交错带。由于气候及人为干扰因素的影响，该地区出现了固定沙丘活化、草场生产力下降、物种多样性丧失和耕地退化等问题，是我国荒漠化发展较快、危害比较严重的地区之一。油蒿为旱生沙生植物，分布于暖温带的干草原和荒漠草原带，喜生于固定沙地和覆沙土壤上，是草原区沙地半灌木群落的重要建群植物，为优良的固沙植物。本文基于野外调查及遥感影像数据，应用点格局、地统计、多元统计、区组方差分析等方法研究了毛乌素沙地油蒿种群及其组分的空间格局，结果表明：（1）在小尺度域范围内，随着沙地固定年限的增加，油蒿种群密度逐渐变大，而油蒿的冠幅呈减小趋势，其空间变异程度也逐渐减小。油蒿种群的空间自相关性都随着间隔距离的增加而迅速减小，且都在0.6m的空间尺度上呈现最强的空间自相关性；随着沙地固定年限的增加，油蒿种群的空间自相关性随着间隔距离的增加而减弱的速度更快，空间异质性程度增加的更加快速，而油蒿种群的半方差模型的变程则显著减小，表明油蒿种群空间自相关的尺度逐渐减小。在油蒿种群盖度小于30%时，一般具有2-3次斑块分化现象，个体水平上的分化现象出现的区组尺度为2-2.5m，中区组尺度为6m附近，而较大区组尺度为11-15.5m，且大尺度上的分化现象要比小尺度上的强烈。当油蒿盖度大于35%时，油蒿种群出现2-3次斑块分化现象，且个体水平上分化的区组尺度为1m-1.5m，中尺度群体水平上分化现象出现在4m，大尺度群体水平分化出现在7.5m-15.5m，且个体水平上的分化现象比群体水平上的分化现象强烈。油蒿种群盖度越小，其个体水平分化现象出现的区组尺度越大。（2）毛乌素沙地中尺度域油蒿种群的空间格局分为三种类型，第一种为上升式，第二种为波浪式，第三种为平稳式。上升式类型样地中油蒿种群空间自相关的距离较大，并且随着尺度的增加，其空间异质性程度增加的速率相对较小，一般在很大的尺度上才能达到平稳状态，其斑块分化现象只出现在较大区组尺度上，与之对应的沙地景观空间格局也呈现出同步的特征。波浪式类型样地中油蒿种群半方差函数都具有比较小的变程值及较大的分维数值，其空间异质性程度随着尺度的增加而呈现波浪式特征，并且函数曲线的波浪式特征逐渐减小，趋于平稳；油蒿种群呈现出明显的等级分布特征；油蒿种群在较小及较大的区组尺度上都具有斑块分化现象，而一般在中间区组尺度的分化强度较小，与之对应的沙地景观的空间格局特征具有相似特征。平稳式类型样地油蒿种群的半方差函数变程值以及分维数值介于上升式及波浪式类型之间，具有较高的拟合优度，半方差函数曲线比较平稳，油蒿种群的斑块分化现象一般出现中间区组尺度上，与之对应的沙地景观的空间格局特征具有相似的特征。毛乌素沙地中尺度域上沙地发育的空间格局先后顺序为上升式——波浪式——平稳式。（3）开花油蒿在较小尺度上表现为非聚集分布，在较大的尺度上为聚集分布，且随着尺度的增加其聚集程度逐渐减小并趋向于随机分布。未开花油蒿在小尺度上为聚集分布，大于6m的空间尺度上更趋向于随机分布。死亡油蒿的L(r)函数在较小尺度上空一般为随机或均匀分布，而在较大尺度上呈聚集或随机分布，死亡油蒿的g(r)函数一般在较小尺度上表现为聚集分布或随机分布，而在较大尺度上呈随机分布。随着沙地环境的改善，开花油蒿在较小尺度上逐渐由随机分布转变为均匀分布，而较大尺度上聚集分布的强度逐渐变小，具有转变为随机分布的趋势。未开花油蒿随着沙地环境的改善，其L(r)函数在极小尺度上出现了随机分布，其在大尺度上的聚集程度呈现出先增大后减小的趋势，g(r)函数在较大尺度上从随机分布转变为聚集分布而后又转为随机分布。死亡油蒿随着沙地环境改善，其较大尺度上经历了聚集——随机——聚集的变化过程。存活油蒿与开花油蒿相似，在较小尺度上由随机分布逐渐变为均匀分布，其较大尺度上聚集分布的强度有减小的趋势，个别样地在大尺度上已经表现为随机分布。更多还原

【Abstract】 As one of the four huge sandlands in china, Mu Us sandland is in the ecotone betweensemi-arid and sub-humid zone in northern China. Due to climate and human disturbancefactors, some problem came out, such as fixed dunes activation, pasture decline in productvity,biodiversity loss and degradation of agriculture lands. The desertification developed rapidly inMu Us sandland.Artemisia ordosica was desert plant, which distributed in warm temperate drygrassland and desert steppe belt, born in the fixed sandy soil and overlying sand, and it wasconstructive species and excellent sand-fixation plants in steppe sandy shrub communities.Based on field survey and remote sensing data, spatial distribution pattern of the Artemisiaordosica population and its components in Mu Us sandland was studied with the methods aspoint pattern analysis,geostatistics,multivatiate statistics, and block variance. Results showed:(1) In the scope of small-scale domain, with the improvement of the sandy habitats,Artemisia ordosica population density gradually became larger, and the crown showed adecreasing trend, so was its extent of spatial heterogeneity. With the increasing of intervaldistance, the extent of spatial autocorrelation of Artemisia ordosica population decreasedrapidly, and it showed high extent of spatial autocorrelation in the scale of0.6m. With theincrease of Artemisia ordosica coverage, the extent of spatial autocorrelation weakened morerapidly. With the interval distance, and extent of spatial heterogeneity increased more rapidly,while the range of semi-variance model of Artemisia ordosica population significantly reduced.As the coverage of Artemisia ordosica population <30%, there was2-3patch differentiation,while the differentiation occured in the scale of2-2.5m in individual level,6m in small grouplevel, and11-15.5m in large group level. As the coverage of Artemisia ordosica population>35%, there was2-3patch differentiation, while the differentiation occured in the scale of1-1.5m,4m in the small group level, and7.5-15.5m in large group level. The differentiation extent of individual level was higher than group level. The block scale of differentiation waslarger in individual level when the coverage of Artemisia ordosica population was smaller.(2) Spatial pattern of Artemisia ordosica populations in mesoscale domain divided intothree types, first the rise style, second wave style, and third for the smooth style. Spatialautocorrelation range of Artemisia ordosica populations was large in the plots of rise style.Spatial heterogeneity of Artemisia ordosica populations increased more slowly with theincrease of scale, and the phenomenon of patch differentiation just occured in large block scale,while the spatial pattern of sandy landscape showed the similar characteristics. Thesemivariance function of Artemisia ordosica populations of wave style had relatively smallerranges and the larger fractal dimension, while its spatial heterogeneity showed wave-likecharacteristics with increasing scale, and the wave function curve became stabilized gradually.Artemisia ordosica populations pattern showed significant level distributioncharacteristics. Artemisia ordosica populations had patch differentiation in various block scales,and the differentiation extent in intermediate block scale was small. The rang and fractaldimension of semi-variance function in smooth style plots was between the rise andwave-style,and it had a high goodness of fit, its semi-variance function curve was relativelystable. The phenomenon of patch differentiation occured in intermediate block scale. Spatialpattern of the sandy landscape had the similar characteristics as Artemisia ordosica populationsand sandy landscape in mesoscale domain was rise style–wave style–smooth style.(3) The fowering Artemisia ordosica showed non-aggregeted distribution in small scalesand aggregated. Distribution in large scales. The extent of aggregation decreased graduallywith increasing scale and tends to be random distribution. The non-flowering Artemisiaordosica showed aggregated distribution in small scales and tend to be randomly distributed asthe scales>6m. The L(r) function of dead Artemisia ordosica showed non-aggregateddistribution in small scales and aggregated or random distribution in large scales. The g(r)function of dead Artemisia ordosica showed aggregated distribution in small scales andrandom distribution in large scales. With the improvement of the sandy environment, spatial point pattern of floweringArtemisia ordosica turned from random distribution into uniform distribution in the small scale,while the extent of aggregated distribution decreased gratually, with a trend into randomdistribution. With the improvement of the sandy enviroment, the L(r) function showed that theextent of aggregation of non-flowering Artemisia ordosica firstly increased and thendecareased. The g(r) function showed that spatial point pattern of non-flowering Artemisiaordosica turned from random distribution into aggregated distribution and then randomdistribution. Dead Artemisia ordosica experienced the process of aggregation–random–aggregation. The spatial point pattern of living Artemisia ordosica was as similar as floweringArtemisia ordosica.更多还原