Assessment of provincial-scale forest fragmentation in Chinese mainland | 中国大陆省级尺度森林破碎化特征评价

chinois. 评价中国大陆省级尺度森林破碎化模式，分离导致森林破碎化的人为干扰及自然干扰贡献份额并探究破碎化的社会经济驱动力。提取2006年中国遥感影像地图中的道路信息并整合进全球最新300 m分辨率土地覆盖数据库Globcover；在3×3像元分析窗口下，采用森林破碎化分析模型，将中国各省（市、区）森林归并为内部、孔洞、边界、斑块、过渡及未确定6种破碎化成分，同时识别人为干扰空间位置及烈度，以此建立省级尺度的具有明确空间意义的森林破碎化及干扰模式地图。另外，计算景观类型间的空间关联指数，以刻画森林与其他土地利用类型的空间交互效应。福建、内蒙古、江西、浙江及西藏5省区森林的破碎化程度较低，而河北、北京、山东、江苏、上海及青海森林的破碎化程度较高；新疆、内蒙古、宁夏、青海、山东及江苏等省（区）森林的人为干扰广泛存在且干扰烈度较高；江苏、重庆、青海、山东、上海等省（市）的森林与农业土地利用的空间关联度较高，黑龙江、吉林、内蒙古、辽宁等地的森林与自然干扰土地利用类型的关联度较高，各省份森林与城市类型连接的可能性均较低。（1）中国森林破碎化的主要驱动力在于巨大人口压力驱使下农业土地利用的无序扩展；（2）由森林破碎化模型所导出的具有明确空间含义的森林破碎化地图，是科学制定土地利用决策、生物多样性保护规划和林业可持续经营战略发展的基础，具有传统景观指数分析方法不可比拟的优势；（3）森林破碎化是一个长期的、复杂的社会问题，需要整个社会采取涉及政策、经济及教育等领域的综合措施，以实现对现存森林的最小化干扰，降低对森林生态系统的潜在威胁，保障国家的生态安全。

anglais. This study was to assess forest fragmentation patterns at the provincial scale in Chinese mainland to separate anthropogenic disturbances from natural disturbances responsible for forest fragmentation, and to probe the socioeconomic drivers for forest fragmentation. Roads were extracted from the 2006 Remotely Sensed Map for China and integrated into the latest global land cover database, Globcover v2.2, with a spatial resolution of 300 m. Then a forest fragmentation model was applied to the revised Globcover to classify each forest pixel into one of the six fragmentation components, including interior, perforated, edge, patch, transitional and undetermined at the analytical window of 3×3 pixels. Those areas having high amount of anthropogenic disturbances were located by interpreting anthropogenic disturbance values derived from the model. Ultimately, a suite of forest fragmentation maps with spatially explicit implications was produced to present the fragmentation patterns. Additionally, spatial association index was computed to characterize the spatial interactions between forest class and other target land use types of interest. A low forest fragmentation intensity in Fujian, Inner Mongolia, Jiangxi, Zhejiang and Tibet followed by a high fragmentation in Hebei, Beijing, Shandong, Jiangsu and Shanghai, as well as Qinghai was observed. Furthermore, forests in Xinjiang, Inner Mongolia, Ningxia, Qinghai, Shandong and Jiangsu were heavily disturbed by far-reaching anthropogenic events at a high severity. Additionally, a high spatial interaction between forest and agricultural land use was identified in Jiangsu, Chongqing, Qinghai, Shandong and Shanghai, while forests in Heilongjiang, Jilin, Inner Mongolia, Liaoning were highly associated with natural land-use types. Contributions of urban land uses to forest fragmentation were observed at a low rate in contemporary China. Forest fragmentation in China is primarily attributed to the widespread anthropogenic disturbances, particularly, the expansion of agricultural land uses driven by a huge population. The spatially explicit forest fragmentation maps derived from implementing the fragmentation model are the information basis for governments and conservation communities to develop reasonable land use decisions and biodiversity conservation practices, and to formulate ecologically sustainable forest management strategies, implying that the fragmentation model has an incomparable strength compared to the conventional landscape indices when dealing with forest fragmentation issues. Meanwhile, we need to recognize that forest fragmentation is a long-term, complex social issue, which requires a comprehensive measure that is closely associated with political, economic and educational process to be taken to minimize the impacts of human events on the existing forests, to mitigate the potential threats of forest fragmentation to forest ecosystems to ensure the ecological security of nation.