荒漠绿洲交错区景观稳定性与维持机制

景观稳定性与维持机制是景观生态学理论的重要基础。由于人类活动干扰和景观斑块自然属性及资源条件约束，荒漠绿洲交错区景观稳定性研究成为干旱区生态学研究热点。选取描述景观斑块大小（MPA）、数量（PD）、大斑块优势度（LPI）和斑块空间关系（ENN）等4个指数，采用统计学中的变异系数和景观稳定性指数分析了荒漠绿洲交错区不同景观水平稳定性。从景观指数稳定性来看，无论是景观或斑块类型水平，斑块间平均距离指数稳定性最高，平均斑块面积指数次之，斑块密度指数居第三，大斑块优势度指数最低。景观与斑块类型水平比较而言，景观尺度同类指数稳定性都高于斑块类型尺度。从景观稳定性时间尺度依赖特征来看，线性趋势分析表明景观水平尺度依赖特征达到了显著水平（P=0.003），低于斑块类型水平的沙质荒漠（P<0.001）与砾质荒漠（P<0.001）。从景观稳定性维持机制来看，Pearson相关分析显示人类活动导致的绿洲化和地下水埋深是影响荒漠绿洲交错区景观稳定性的主要原因（双侧检查都为P<0.01）；沙漠化过程（0.01 < P < 0.05）和相邻斑块属性等自然现状是次要原因。

Landscape stability and maintaining mechanism in desert-oasis ecotone

It is one of theoretical fundament to identify landscape stability and maintaining mechanism in landscape ecology research. Due to restriction of anthropogenic activities disturbance, patch natural attributes and natural resource support, it is a focus area in desert-oasis ecotone of arid zone to study landscape stability. To best understand landscape stability in different scales, we selected landscape indices such as mean patch area (MAP), patch density (PD), large patch index (LPI) and Eucidean nearest-neighbor distance (ENN) to evaluate the stability by employing Coefficient of Variance (CV) and Stability Index. Results showed that there was obviously difference to landscape stability assessed by CV analysis for four landscape indicators that indicate and the stability values were in turn ENN, MPA, PD, and LPI both of landscape and patch class level. However, the stability index was higher in landscape level than in patch class level for all selected indices. In terms of temporal scale dependence for the stability, significant F check of linear trend analysis indicated that sandy desert (P<0.001) and gravel desert area (P<0.001) were higher than landscape level (P=0.003). In terms of landscape maintaining mechanism, Pearson Correlation Analysis illustrated that oasisization and underground water table were main factor to impact on landscape stability that induced by anthropogenic activities disturbance (two tailed significant check both of P<0.01), and desertification (two tailed significant check P<0.01) and neighbor patches natural attribute were secondary factor.