【目的】以扬州市为例,分析城市景观破碎化的空间格局,定量研究社会经济因素对景观破碎化的影响及该影响的相对重要性。【方法】利用空间分辨率30 m的Landsat卫星影像、乡镇层次的扬州统计年鉴等数据,运用景观格局分析法、增强回归树模型(boosted regression trees, BRT)等方法,开展景观破碎化空间格局及社会经济驱动的研究。【结果】①扬州的土地利用类型以耕地为主,占全市总面积的54.9%,其次为水体(22.0%)和建设用地(21.9%),林地和草地很少(1.1%)。距离市中心越远,景观破碎化程度越高,在城区边缘达到峰值,而后随着距离的继续增大,破碎化程度逐渐减小。②随着城市化水平的增加,景观破碎化程度呈先增加后减小的趋势,其在城市化水平35%~45%时达到峰值。③在景观水平上,社会因子(人口数量和人口密度)对景观破碎化的影响以抑制为主,而经济因子(人均GDP、人均财政收入、第一产业产值占比和第三产业产值占比)的影响以促进为主。与景观水平相比,社会经济因子在类型水平上对景观破碎化的影响更大,而且它们与破碎化不同特征(斑块密度、边缘密度、分离度和聚合度)的作用关系也存在明显差异,且主要体现在分离度和聚合度上,例如人均财政收入在景观水平上促进分离度的增加,在类型水平上却表现为抑制作用。【结论】社会经济发展对城市景观破碎化产生较大影响,不仅体现在不同的尺度(景观和类型水平)上,还体现在社会经济因子对破碎化特征不同的驱动作用(抑制或促进)上。因此,需要综合考虑社会经济发展对景观破碎化影响的尺度效应及作用强度与方向,有效缓解城市景观的破碎化,保障城市的生态安全和可持续发展。

【Objective】 We analyzed spatial patterns of urban landscape fragmentation by taking Yangzhou City as a case study. Here, we further quantitatively explore the effects of socioeconomic factors on landscape fragmentation.【Method】 Multi data sources (e.g., Landsat-5 TM imagery with 30 m spatial resolution, and the Statistical Yearbook of Yangzhou City) and several data analysis methods (e.g., landscape pattern analysis and the boosted regression trees machine learning technique) were employed. 【Result】 ① Arable land was the dominant land use, covering 54.9% of the total area of the city in 2010, followed by water (22.0%), built-up land (21.9%), and forestland and grassland (1.1%). As the distance from the city center increased, the degree of landscape fragmentation first increased and then decreased, whereby the turning point was at the edge of the city. ② The degree of landscape fragmentation increased with increasing level of urbanization, and reached an extremum at 35%-45% before declining gradually. ③ At the landscape level, social factors (i.e., total population and population density) generally had an inhibiting role on landscape fragmentation, while economic factors (i.e., per capita GDP, per capita revenue, primary and tertiary industries) usually promoted landscape fragmentation. In comparison to landscape level, socioeconomic factors had greater impacts at the class-level, and their relationships with class-level fragmentation indices (patch density, edge density, landscape division index and aggregation index) showed distinct differences. For example, per capita revenue promoted the division of landscape at the landscape level, but inhibited it at class level.【Conclusion】 Socioeconomic development played an important role in landscape fragmentation, not only at different levels (landscape and class levels), but also in terms of different driving roles (inhibition or promotion) of fragmentation indices. Therefore, it is necessary to consider the scale effect, impact intensity, and impact direction of socioeconomic factors in order to effectively alleviate urban landscape fragmentation.