【作者】 李远发；

【导师】 惠刚盈；

【作者基本信息】 中国林业科学研究院 ， 森林培育， 2013， 博士

【摘要】 空间结构对于描述林分及其状态的改变有特别重要的意义。量化不同林分之间的结构关系可以帮助简化测量、了解和管理森林结构的整个过程。林分尺度上可量化的结构要素很多，各要素之间常常保持着紧密的联系。许多描述林分结构的指标或多或少都与相邻木有关，它们中的一些指标描述的是林分空间结构，而另一些指标描述的却是非空间结构，并且各指标之间的理论基础和表达方式不同，缺乏必然联系。这些方法的共同特征是它们只能刻画单方面或整体的结构特征(宏观结构)，而无法同时提供两个或多个方面的林分结构信息。为了解决这一难题，本研究提出一种新的基于最近4株相邻木空间结构关系的微观结构分析方法——结构参数二元分布分析，它包含3种不同形式的结构组合，即混交度和大小比数的二元分布、混交度和角尺度的二元分布以及大小比数和角尺度的二元分布。为检验这种新方法分析不同水平层次林木结构的能力以及它在森林经营中的可能应用，在中国东北地区典型的植被类型红松阔叶林建立了6块面积大小均为100m×100m的标准固定样地，其中3块(a、b、c)作为结构化森林经营示范，另外3块(d、e、f)作为对照样地。同时在西北秦岭北坡的天然松栎混交林中建立了2块面积均为70m×70m的标准固定样地(h、i)，1块作为结构化森林经营示范(h)，而另1块则作为照样地(i)。采用结构参数二元分布分析了红松阔叶林林分、种群以及种群内不同直径林木的空间结构特征，还分析了两类经营林分中第一次经营时获得的采伐木的空间结构特征。此外，还将它扩展应用到结构化经营中采伐木的优先性选择中。主要结果如下：(1)林分水平上，天然红松阔叶林的大部分林木同时处于强度混交和随机分布状态；不同优劣态势的林木多数与其他树种伴生；随机分布的林木其周围一般都是其他树种；中庸状态的林木在林分中的分布格局为随机分布。(2)种群水平上，水核林的3个主要种群水曲柳(Fraxinus mandshurica Rupr.)、核桃楸(Juglans mandshurica Maxim.)和暴马丁香(Syringa amurensis Rupr.)多数个体同时处于高度混交和随机分布状态，只有少数处于团状或均匀分布；同一种群中，处于团状或均匀分布的林木数量基本相等，并且具有相似的混交度和优劣态势；暴马丁香种群在林分中处于中庸或劣势状态，而大部分水曲柳和核桃楸处于优势状态，但核桃楸的优势更明显。(3)种群内，水曲柳的小树和大树常被其他树种包围，小树多为随机分布的非优势木，而大树为偏规则分布的优势木；核桃楸的小树和大树的周围通常是其他树种并且这些相邻木随机地围绕在它们的周围，但小树多为亚优势木，而大树呈典型的优势木；暴马丁香的小树和大树也常被其他树种包围，它的小树为随机分布的劣势木，而大树中仅有少部分为优势木且整体偏向规则分布。(4)红松阔叶林的采伐木广泛地分布在林分各空间层次中，既有被压木、中庸木，也有优势木，但整体属于中小径木，多数个体同时处于高度混交和随机分布状态。相对而言，松栎混交林的采伐木分布较为集中，几乎均匀地分布在各径阶中，多数采伐木属于高度混交或随机分布的优势木；伐后两类林分的结构特征向原始林更加靠近。(5) M–U疏伐优先性指数表明样地a中优先采伐的是不同混交的劣势木和绝对劣势木，样地b中优先采伐的是高度混交的优势木和劣势木，样地c中优先采伐的是不同优势的高混交林木，而样地h中优先采伐的是中度混交的亚优势木和高度混交的优势木；M–W疏伐优先性指数表明4块经营样地中优先采伐的是同时处于高度混交和随机分布状态的林木；U–W疏伐优先性指数表明样地a中优先采伐的是处于随机分布的劣势木，样地c中优先采伐的主要对象是不同优势的随机分布林木，而样地b和h中优先采伐的是随机分布的优势木和亚优势木。结构参数二元分布充分利用了角尺度、混交度和大小比数在表达结构上的频率优势，同时从两方面刻画不同水平层次林木的空间结构特征，进一步细化了它们的空间结构特征，可提供比结构参数一元分布或传统单一指标更加丰富、直观的有效信息。这些信息可能对恒续林经营中择伐木的选择和林分模拟与重建具有指导作用，亦可能有利于生物多样性的发掘和保护。更多还原

【Abstract】 Spatial structure is important in describing forest stand structure and change. Quantifyingrelationships among different forst structures simplifies the process of measuring,understanding, and managing forest structure. At the stand scale, there are a great number ofstructural elements that often maintain close links to each other can be quantified. Manystructural indicators described forest stand are more or less related to the adjacent wood, someof them describe the spatial attributes of forest stand, while the others depict the non-spatialcharacteristics. Commonly, there is no necessary connection among such indexes that aretypically built on variously theoretical backgrouds and have different expression ways at thesame time. The common feature of these methods is that they can only be portrayed unilateralor whole characteristic of tree structure (i.e, macroscopic structure). That is to say, they cannotsimultaneously provide any information about two or more aspects of stand structures. In orderto solve such difficult problem, this contribution proposed a mircroscopic structure analysismethod based on the spatial structural relationships of the nearest four adjacent woods—theanalysis of bivariate distribution of structural parameters. It contained three different forms ofstructural combination, i.e., bivariate distribution of mingling-dominance, bivariate distributionof mingling-uniform angle index and bivariate distribution of dominane-uniform angle index.In order to examine its ability in analyzing different levels of tree structure and itspossible application in forest manangement, six permanent fixed standard plots with length100m×100m were established in Northeast China PR where are commonly covered by thezonal climax vegetation the Korean pine broad-leaved forest. Each plot was only lettered by aArbica character. Three of them (a, b, c) have been treated as the demonstration ofStructure-based forest management (a advanced kind of near to natural forest management)after the plots being built, while the residual (d, e, f) was treated as control. During the sameperiod, two fixed standard samples (h, i) whose areas are70×70m2were also set up in the natural oak-pine mixed forest on northern face of Qinling Mountains, Northwest China PR. Asthe way of plots (a-f) treated in the Korean pine broad-leaved forest, one of them (h) was alsotreated as the demonstration of Structure-based forest management in Qingling area, the otherone (i) was treated as still (CK). With the purpose to exame the ability of the bivariatedistribution of the structural parameters in analyzing forest structure on different levels, i.e.,individual, population and stand (community), the spatial characteristics of the Korean pinebroad-leaved forest stands (a-f) were explored. The spatial attributes of the harvested trees inthe four managed stands which were operated for the first time were also disclosed by thismethod. What is more, the bivariate distribution of the structural parameters was extended tothe priority selection of wood harvesting in natural forests. The main results are as follows:(1) On the stand level, most trees in the natural Korean pine broad-leaved forest werehighly mixed by species and randomly distributed. Trees with different advantages weretypically surrounded by other species; trees within stochastic distribution patterns were usuallysurrounded by different species; and medium-sized trees were randomly distributed.(2) On the population level, three major populations Manchurian ash (Fraxinusmandshurica Rupr.), Manchurian walnut (Juglans mandshurica Maxim.) and Syringa reticulataHong (Syringa amurensis Rupr.) in the mixed forest of Manchurian ash and Manchurianwalnut were usually mixed with other species and distributed randomly. Only a few individualwas in clump or uniform distribution. Trees in clump or regular distribution had a similarnumber of individuals when they belonged to the same species, and they also had a very closemixture or dominance. Normally, Syringa reticulata population were in medium or inferiorstatus, however, most of Manchurian ash and Manchurian walnut had a dominant state whileManchurian walnut population won more obvious advantages thant that of Manchurian ash.(3) Intraspecies, both small (23cm<DBH≥5cm) and big trees (DBH≥23cm) ofManchurian ash were often surrounded by other species. The small ones within random patternwere non-dominant, while the large ones departing from regular dispersion were dominant.Most of the time, Manchurian walnut population‘s small and big trees were randomly encircled by plants that were belong to other species. However, the small Manchurian walnuts weredominant trees while the individuals with bigger DBH were the typically advantage woods. AsSyringa reticulata‘s small trees (8cm<DBH≥5cm) to do, its large trees were often surroundedby other species. The small trees, with random pattern, were in the status of disadvantage. Thebigger ones (DBH≥8cm) tended to be regular on the whole and only a small part of them weredominant plants.(4) The harvested woods were widely distributed on different vertical levels of the Koreanpine broad-leaved forest, including trees in the state of compression, medium and dominance.Most of them belonged to the small diameter category and were highly mixed and randomlydistributed, simultaneously. On the contrary, trees harvested in the oak-pine mixed forestdisplayed a relatively single characteristic. They had a attribute of even distribution among theall diameter classes. Most of them were dominant, and were in the status of high mixture orrandom distribution pattern. After managed, both of natural forests were closer to originalforest in spatial structural features.(5) Even depicting the same plot, different thinning priority indexes may exhibit acompletely different tendency of frequency distribution on the5×5=25combinations. M-Uthinning priority index suggested that trees preferred to be cut in plot a were the ones haddifferent degrees of mixture and in the state of disadvantage or completely disadvantage; Inplot b, the woods cut firstly were highly mixed and suppressed by their nearest neighbors orwere completely dominanct; Trees with different advantages and high mixtures werepreferentially cut in plot c; Nevertheless, plants in plot h had high value of thinning prioritywere the ones in the status of dominance. Some of them had a medium mixture, and otherswere highly mixed by other plant species. M-W thinning priority index indicated that fouroperated samples had a similar frequency distribution on different conbinations. That is to say,most trees preferred to be cut were highly mixed and randomly distributed at the same time. Inplot a, U-W thinning priority index revealed that most trees cut were dominant and were inrandom distribution pattern; In plot c, those different dominant trees that were randomly surrounded by their nearest four neighbors were liked to be cut firstly; But in plot b and h,dominant woods that were random surrounded by their nearest four neighbors in the structuregroup of five were become the selective hotspot of harvesting.The bivariate distribution of the structural parameters takes full advantage of thefrequency distribution of species mingling, uniform angel index and dominance when they areused to express the spatial characters of forest stand. They simultaneously quantify two aspectsof stand structures from different points of view. As a result, it further refines the spatialstructure of the stand, and provide more direct and useful information about the heterogeneityof spatial structure than can univariate distributions or other conventional stand descriptors.This could be helpful for selective thinning in continuous cover forest management and inmodelling and restoring forests. It may also be conducive to the excavation and protection ofbiodiversity.更多还原