【作者】 胡欣欣；

【导师】 陈平留；

【作者基本信息】 福建农林大学 ， 森林经理学， 2009， 博士

【摘要】 龙栖山国家级自然保护区以生物种类的多样性,珍稀动、植物丰富性和中亚热带森林植被的典型性而著称,是一个理想的森林生态旅游环境。然而,由于保护区经济基础薄弱,当地政府财力有限,长期以来旅游基础设施落后,旅游景点破碎化较严重,无法真正实现保护区的生态效益、社会效益和经济效益的协调发展。针对以往对于龙栖山国家级自然保护区的生态环境、景观格局等方面的研究相对薄弱的现状,本文研究分析了龙栖山国家级自然保护区景观空间格局、景观斑块特征分布规律以及外界干扰对保护区景观格局影响的关系,应用层次分析法、RBF网络、投影寻踪模型和支持向量机四种方法评价了森林景观生态质量,结合龙栖山国家级自然保护区生态旅游,分析环境容量和旅游容量,预测生态旅游规模,并给出若干建议。研究的结果有利于保护区管理者制定一个系统而有针对性的区域可持续发展战略,从而促进保护区的生态效益、社会效益和经济效益的协调发展。主要研究结果如下:1、在ArcGIS环境下将矢量图转化成栅格格式,利用Fragstats软件在斑块类型、景观类型和景观水平上计算景观指数,对龙栖山国家级自然保护区森林景观格局进行分析。1)景观格局分析结果表明,龙栖山国家级自然保护区9个景观要素的面积、周长、斑块数分布均极不均衡。阔叶林景观、松林景观和竹林景观是保护区的主要景观类型,三者的面积和周长之和分别占整个保护区面积和周长的95.3078%和85.6365%。但各景观要素斑块的面积、周长数值差异较大,同时由于自然保护区其特有的保护性质,超大斑块的数量也为数不少。2)景观要素的斑块形状特征表明,阔叶林景观要素的斑块形状最为复杂、景观连通性较好,破碎化程度和斑块分离度水平都较低,而人类活动较多的非林地景观、经济林景观则破碎化程度较高,符合人为干扰导致景观破碎化程度变化的理论。龙栖山国家级自然保护区作为国家级森林生态系统的保护区,优势度较高,景观分布较不均匀,总体多样性水平不高。2、运用正态分布、对数正态分布、Weibull分布、Gamma分布和Beta分布分别分析斑块大小、斑块面积和分维数的分布规律。结果表明景观斑块特征分布规律存在着等级相关性。研究结果表明,针对龙栖山国家级自然保护区,正态分布在分析景观斑块大小、斑块面积和分维数分布规律中的拟合效果最佳,Gama分布在面积分布规律和分维数分布规律的拟合中效果较好,而Weibull分布无论在哪个斑块特征上分布拟合效果是都是较差的。也说明了研究地的景观斑块大小、景观斑块面积和分维数大部分服从正态分布,部分服从Gama分布。3、运用现代统计学中的非线性建模方法,分析自然干扰和人为干扰对景观格局的影响。1)选择面积、蓄积、年龄、土层厚度、灌丛盖度、草本盖度、平均海拔、坡度、坡向和坡位10指标作为自然干扰因子,构建人工神经网络模型、多元自适应回归样条模型、支持向量机模型和投影寻踪回归模型分别模拟分析分维数。模拟分析结果表明当环境因子中指标值发生变化,斑块分维数也相应产生变化,从而证实并丰富了干扰对景观格局影响的论断。同时通过留一(LOO)交叉验证各模型性能,结果表明多元自适应回归样条模型用于揭示自然干扰对景观格局影响的关系,其效果较其他方法更为理想。2)人类通过改变景观类型数量导致景观格局发生变化,是人类干扰活动对景观格局影响的表现之一。选择景观类型数量作为人类干扰因子,构建人工神经网络模型、多元自适应回归样条模型、支持向量机模型和投影寻踪回归模型,分别模拟分析人为干扰对景观多样性、优势度和均匀度的影响。模拟分析结果表明人为干扰引起斑块数的变化,导致景观多样性产生较为显著的影响,而对景观优势度和均与度影响效果较弱。这一结果也证实并丰富了干扰对景观格局影响的论断。同时通过比较留一交叉验证结果各模型性能,结果表明上述模型用于模拟分析景观斑块数量对景观多样性的影响时效果均较好,其中投影寻踪回归模型性能更优。而用于模拟分析景观斑块数量对景观优势度和均匀度的影响,上述模型均未能有效的揭示两者的关系。4、从森林景观生态保护和开发利用层面出发,选择自然性、珍稀性、多样性、代表性、适宜性、稳定性、干扰性、利用性、观赏性、科研性、区位性和文化价值作为评价景观生态质量指标,构建RBF网络、支持向量机、投影寻踪评价模型和层次分析法评价模型。1)层次分析法的结果说明龙栖山国家自然保护区景观生态质量指数为0.79,其生态质量较好。2)RBF网络、支持向量机和投影寻踪评价模型的评价结果均认为龙栖山国家级自然保护区景观生态质量处于生态质量较好状态。3)在各等级范围内,随机生成10个样本数据,并用上述方法进行评价。通过Spearman等级相关性对评价结果的排序进行分析,在置信度为0.99的情况下,支持向量机、RBF网络和层次分析法高度相关,评价结果基本是一致的,而投影寻踪评价模型方法独立,评价结果差异较大。4)结合序号总和理论分析,用于评价龙栖山国家级自然保护区森林景观生态质量,首选评价方法是支持向量机,其次依次为RBF网络、层次分析法和投影寻踪评价模型。5、以景观生态学、生态经济学以及旅游经济学理论为指导,在一定规划目标驱动下,划分出不同功能区域,并适度开展生态旅游。进一步分析了环境容量和旅游容量,预测生态旅游规模,同时对自然保护区的管理提出若干建议。环境容量分析结果是全年可游天数合计约280天;近期日容量为883人次/日,远期环境日容量为1663人次/日;饱和环境容量是日环境容量的2.5倍;年环境容量近期为247240人次/年,远期为465640人次/年。旅游容量分析结果是近期生态旅游区域日游客容量为622人次/日,远期日游客容量为1159人次/日;近期生态旅游区域年游客容量为174160人次/年,远期年游客容量为324520人次/年;预测生态旅游规模近期年规模为10.11万人次,日规模360人次,而远期年规模为20.34万人次,日规模为726人次。提出应正确处理开发与保护的关系,加强领导广泛宣传,整体合作协力共建,强化内容管理,提高服务档次等若干建议。更多还原

【Abstract】 National Nature Reserve of Longqishan (NNRL) is an ideal environment for the forest eco-tourism and it is famous for the diversity of species, the richness of rare animals and plants and the typical sub-tropical forest vegetation. However, the weakness of Reserve economic base and the limitation of government financial resources have no way to throw into sufficient funds to build and perfect the infrastructure construction and scenic spot touring. All leads to hardly realize the coordinated growth of ecological, social and economic benefits.According to the weakness of studies on environment and landscape pattern of NNRL, this article focused on analyzing of Landscape pattern of NNRL, the distribution law of patch, and the relationship of outside interference pattern on the landscape impact of NNRL. Then, it also evaluated the eco-quality of forest landscape using AHP, BRF, PPE, and SVM. Last, the paper brought up and predicted the capacity of eco-tourism, and gave some advices. The results of the paper can develop a systematic and targeted regional sustainable development strategy for Reserve managers. Thus, it is useful for coordinated development of ecological, social and economic benefits.The results listed below.1. It analyzed of forest landscape pattern of NNRL with landscape indexes, which were computed under patch-type-level, class-type-level, and landscape-type-level using FRAGSTATS software with grid picture converted under ARCGIS environment.1) The results of analysis of forest landscape pattern indicate that there are nine type forest landscape, and their areas, perimeters, and the number of patch distributes very unevenly. The results further present that broadleaf forest, pine forest, and bamboo forest are mainly landscape styles, and that the sum of their area is 95.3078 percent of the NNRL, while 85.6365 percent of the perimeter of NNRL. But, they also suggest that the numerical area and perimeter varies sharply discrimination. Besides, because the nature reserve has its unique protection character, the number of the bigger patches is not big enough.2) The results of analysis of shape features landscape elements show that the patches shape of the broadleaf forest landscape is the most complicated, and that landscape connectivity is good. They further indicate the degree of fragmentation and patches separation degree are fairly low, while the degree of fragmentation in non-wood landscape and economic landscape are high which accords with the theory that landscapes with human interference suffer from high degree of fragmentation. But, as a reserve of national-level forest ecosystem, NNRL enjoys a high predominance, uneven landscape distribution and a low variety on the whole.2. Normal, Lognormal, Weibull, Gamma and Beta distribution respectively are used to analyze the distribution law in the size, the area, and the fractal indices .The analysis results of Landscape patches characteristic distribution law proves a level relativity exists among landscapes patches. The results also show normal distribution fitness is the best, and that Gama distribution has a better effect on the area distribution and fractal distribution fitness, while the Weibull distribution fitness has the worst of all. They further present that the size, the area, and the fractal indices submit to normal distribution mostly, while the parties of indices submit to gamma distribution.3. The article uses nonlinear modeling of modern statistical to simulate and analyze the effects of landscape pattern by Natural disturbance and human disturbance.1). It chooses 10 indices as a natural disturbance factors such as size, volume, age, soil thickness, shrub cover, herbaceous cover, average altitude, slope, aspect and slope position to construct of artificial neural network model, multi-model adaptive regression splines, Support vector machine model and project pursuit regression model to analog analyze of fractal index separately. That the value of fractal index is changing when the value of one index changes, proves and riches the conclusion on interference affecting on landscape pattern. Through leave-one-out cross-validation for the simulation of natural disturbance on the impact of landscape pattern, multiple adaptive regression splines model has the most ability of revealing the relationship between interference and impression of landscape pattern.2). Human beings cause changes in landscape pattern by changing the number of landscape types and it is one of the performances of human Interference activities on the landscape pattern affect. It chooses the number of landscape types as human interference factors to construct artificial neural network model, multi-model adaptive regression splines, support vector machine model and projection pursuit regression model to respectively analog analysis human disturbance on the impact of the landscape diversity, dominance and evenness. The results show that the change of the number of landscape types because of human beings interference is remarkable to diversity, but weak to dominance and evenness. It also proves and riches the conclusion on interference affecting on landscape pattern. By comparing the results of leave-one-out cross-validation for the simulation analysis of the number of landscape patches on the landscape diversity, the analytical capacity of the above-mentioned models are stronger than the generalization ability of each model. Among them, the projection pursuit regression model is the optimal performance. While according to the analysis used to simulate the number of landscape patches dominance on the landscape and the impact of uniformity, the above-mentioned models have not been able to obtain better performance.4. From the landscape ecological protection and the development and utilization level,it chooses the characteristics of Nature, value, diversity, representativeness, suitability, stability, interference, and usability, and ornamental, research, and regional and cultural values as indicators to construct RBF networks, support vector machines, Projection Pursuit Evaluation Model and AHP Evaluation Model to evaluate the eco-quality of forest landscape.1). The result of AHP shows the index of eco-quality of forest landscape of NNRL is 0.79, which means it lies in a better state of eco-quality.2). The results of RBF model, SVM model, and PPE model also show the eco-quality of forest landscape of NNRL lies in a better state of eco-quality.3). Within the scope of all levels, it randomly generated 10 samples data and used these methods to evaluate. In the case of 0.99 Credibility and highly concerning Support Vector Machine, RBF network and analytic hierarchy process, the results of evaluation are consistent. On the contrary, Projection Pursuit evaluation model is an independent method, so the results of evaluation are different.4)Combined with the total numbers and theoretical analysis, it is used in evaluating the eco-quality of landscape. The methods which ranged from the best to the poor are Support Vector Machine, RBF network, the level analysis and projection pursuit evaluation model in turns.5. We plan for ecotourism according to different function areas, estimate and predict the capacity of environment and the capacity of tourism based-on bionomics, eco economics, and the tourism economics theory driven by some planning aims. The results of estimation of the capacity of environment show that it is about 280 days for tourism in one year. They also present that the daily capacity is 883 persons each time per day for near future and is 1663 persons each time per day for long-date. They further indicate that the annual capacity is 247240 persons each time per year for near future and is 465640 persons each time per year for long-date. The results of estimation of the capacity of eco-tourism show that the daily capacity is 622 persons each time per day for near future and is 1159 persons each time per day for long-date. They also indicate that the annual capacity is 174160 persons each time per year for near future and is 324520 persons each time per year for long-date. The results of prediction on the size of eco-tourism show that the daily size is 360 persons each time per day for near future and is 726 persons each time per day for long-date. They also indicate that the annual size is 101.1 thousand persons each time per year for near future and is 203.4 thousand persons each time per year for long-date. While in the process of carrying on ecotourism, we should properly deal with the relationship between development and protection, strengthen and lead extensive propaganda, work together and intensify contents management to update the service level.更多还原