【作者】 赵安玖；

【导师】 胡庭兴；

【作者基本信息】 四川农业大学 ， 森林培育学， 2007， 博士

【摘要】 空间异质性是生态学研究的一个极为重要的理论问题，也是生态学家研究不同尺度的生态系统功能和过程中最感兴趣的问题。由于森林群落在结构、功能上的异质性及在时空上的动态变化，加之表达的多样化，使得群落异质性度量变得非常困难，所以本研究以四川省雅安天全县白沙河景区天然亚热带山地常绿落叶阔叶林为研究对象，采用样地调查方法，设立1个100m×100m样地，采用传统的罗盘仪对树体的空间位置进行准确定位，通过不同的分析方法，从林分局部和整体出发，开展了常绿落叶阔叶林空间异质性研究。其结果如下：(1)运用两参数Weibull方程对林分进行分层模拟，能更好地分析林分树体大小(tree sizes)的结构；H≥10m层对林分树体大小的结构起主导作用；运用Weibull方程对林分密度进行模拟，反映天然林在一定地域空间范围内也呈连续分布状态。(2)在0-75m范围内不同层的空间格局(全林、H＜10m层、H≥10m层)呈规则分布且趋于随机分布，不同树种的空间点格局亦呈规则分布且趋于随机分布；通过Ripley’s K函数分析，不同层、不同树种在每个空间格局下都存在多个尺度转折点，表明与林分的结构和外部环境有密切的关系。(3)通过半方差函数分析，表明林分的DBH、BA的空间自相关范围狭窄，在45°方向上空间变异性表现明显。(4)各树种的林分断面积增长量的时间序列以AR(2)为主，且林分林木断面积增长量在1986-1991年，1986-1996年，1986-2001年，1986-2006年，在45°方向上空间变异较为明显；林木生物量有较强的空间变异，且在45°和97°方向最为明显。(5)冠层下散射的光通量密度(PPFDDiffuseU)与空隙度(Gapfraction)、开度(Openness)和叶面积指数(LAI)有极强的线性相关关系；通过半方差分析，冠层下散射的光通量密度(PPFDDiffuseU)、叶面积指数(LAI)、丛生指数(ClumpFact)具有较高的异质性，而林分冠层结构指数(空隙度、开度)的空间变异较差。同时，结果证明了光环境对林分冠层结构空间依赖的重要性，也强调了单次数据的局限性，光能利用的长时间观测和实验手段对林分结构研究是必须的。(6)山地常绿落叶阔叶林、常绿落叶阔叶林次生林和人工林(柳杉、杉木林)之间的土壤有机质含量差异性明显。山地常绿落叶阔叶林土壤有机质具有较大的空间自相关距，土壤有机质含量分布趋于大块状变异，在45°和90°这两个方向上的变异最大。(7)在0-100m的尺度下，H、E具有负空间自相关性，但这种空间自相关性不明显；H、E在各个方向上异质性表现不明显，在整个尺度上具有恒定的变异。(8)在全尺度下，通过主轴邻距法对林分环境异质性的空间分析是可行的。PCNM的分析表明，土壤有机质、胸高断面积、林分密度、空隙度和开度与PCNM在小尺度上是十分显著的；多样性指数、土壤有机质、断面积、林分密度、空隙度和开度在中大尺度呈显著状态。(9)相关分析表明，林分结构因子和冠层光因子相关性明显，它们与PCNM变量相互影响，相关性较高。(10)林分在时间和空间表现为连续变化的状态，这也表明了林分结构因子在一定尺度内可简化为区域连续变量；同时证明了林分结构因子为连续变量的假定是成立的。更多还原

【Abstract】 Abstracts: As 0patial heterogeneity becomes a major theme in a wide range of ecological studies, the concepts of scale and hierarchy become increasingly important in ecology in general. This confirms, and is indicative of, the rapidly rising awareness of the importance of scale and hierarchy by ecologists. Since structures of forest communities, functional heterogeneity and its spatially and temporally dynamic varieties which expresses by many approaches, it is difficult to quantity community heterogeneity. In this research, one sampling in sizes of 100 m×100 m were established in subtropical hilly evergreen-deciduous broad-leaved forest in Baisha River of southwestem Sichuan province, and expanded the researches of forest spatial heterogeneity by many statistics methods. The results were as following:（1） To simulate stand structures of different forest layers using two-parameter Weibull function, it can prefer to elucidate the structures of tree sizes. And the results demonstrated that H≥10m layer was important layer that controlled the forest stand structure, in simultaneity, we obtained better simulated forest stem density by Weibull function, and this reflected continuous distributions about nature forests at certain spatial ranges.（2） Spatial patterns of different layers which included whole stand, H＜10m layer and H≥10m layer represented Regularity distribution tending to randomness, and the same as different tree species at the distance interval 75m. By analysis of Ripley’s K function, inflexions in multi-scales were presented in spatial patterns of different forest layers and tree species that indicated the affinities between forest structures and exterior conditions.（3） By analysis of semivariogram function, diameter at breast height and basal area of forest stand show spatial autocorrelations in strict ranges, further more, they represented spatial varieties only in orientations of 45 degree.（4） Time series equations of basal area increment of each tree species were established, and AR （2） was suit for many tree species. By analysis of semivariogram function, basal area increment of each tree species showed spatial varieties only in orientations of 45 degree during 1986-1991, 1986-1996, 1986-2001, 1986-2006. In addition, trees biomass represented spatial varieties only in orientations of 45 and 97 degree.（5） There were robust linear relationships between diffuse of Photosythetically active Flux Density under canopies （PPFDDiffuseU） and gapfraction, openness as well as leaf area index （LAI） （R²=0.9836, P＜0.00001）, moreover, PPFDDiffuseU had positive linear correlations with gapfraction, openness, and had negative linear correlation with the LAI. By analysis of semivariance PPFDDiffuseU and LAI had obvious spatial heterogeneities, however gapfraction, openness had weak spatial variability. Our results confirm importance of examining spatial dependence of light availability in studies of forest canopy, but they also underscore the limitations of a single period of data collection. Long-term studies as well as experimental manipulations of light availability are needed to establish causal relationships between light availability and stand-level patterns.（6） Organic matters contents which were in hilly evergreen-deciduous broad-leaved forest, second forest of hilly evergreen-deciduous broad-leaved forest and plantation forest such as Japan cedar and fir forest, showed evidently different. Organic matters contents in hilly evergreen and deciduous broad-leaved forest represented spatial autocorrelations in large ranges, and tended to distributions of patchiness varieties, at same time, represented spatial varieties only in orientations of 45 and 97 degree.（7） Linear models for isotropy were optimal models by simulating Shannon-Wiener diversity indices （H） , Pielou evenness indices （E） ; and for anisotropy linear model was optimal model to H, as well as Gaussian model to E. moreover, the ratio of C/（C₀+C） showed that anisotropic heterogeneities was not obvious, namely isotropy of H and E was effective to spatial pattern of tree species diversity, and appeared steady variances in whole scales; in sampling scales, H and E represented negatively spatial autocorrelation, and its spatial autocorrelation presented not remarkable in regional scales （-0.6386、-1.0902＜-1.96）（8） Under all scales, principal coordinates of neighbour matrices （PCNM） analysis was best approach to spatial analysis of ecologically environment data, by means of PCNM varieties in fine scale of OM contents, DBH, stem densities, gapfraction and openness were very outstanding, Moreover, diversity indices, OM contents, BA, stem densities, gapfraction and openness in middle scales showed very outstanding.（9） By analysis of canonical correlation, the results showed that they had obvious correlations among forest structures, canopy structures and light environment factors, additionally, they represented high affinities with PCNM variables.（10） From above analysis, we know that forest stand represent continuous change situations in space and time, and this confirms that forest stand structures factors can simplify regional continuous variables, and this proved that previous hypothesises were correct.更多还原