生态系统服务（Ecosystem service, ES）是指人类从自然生态系统中直接或间接获得的收益，作为连接生物物理系统和社会经济系统的桥梁，ES显著地促进了人类福祉。近年来，将ES的框架纳入国家和地区发展战略的主流，已经引起了世界范围内政府和公众的高度重视。其中，生态系统服务权衡和协同效应已成为ES研究的重要主题，更好地理解权衡的时空特征和跨尺度的驱动机制，有利于利益相关者平衡ES的供给和需求，也是进行可持续生态系统规划和管理的先决条件。 内蒙古草原处于干旱半干旱脆弱地带，对气候变化和人类活动非常敏感。该区域提供了丰富且多样的生态系统服务，是中国北方的重要生态屏障。自2000年退耕还林还草等生态恢复工程实施以来，该地区生态系统服务的格局以及服务之间的关系均发生了巨大的变化。 本研究选取内蒙古锡林浩特市作为研究区，应用GIS和RS平台，运用InVEST等评估模型和多源数据，对研究区产草量（GY）、产水量（WY）、水源涵养（WR）、土壤保持（SR）、防风固沙（SF）、碳固持（CS）、栖息地质量（HQ）和游憩潜力（RP）8项重要的生态系统服务进行定量模拟。在分析生态系统服务时空分布特征的基础上，在不同时间节点（2000、2010和2017年）和不同空间尺度（像元、格网和子流域尺度）上，应用空间自相关分析、Spearman秩相关分析、主成分分析（PCA）与自组织映射网络（SOM）分别检测了单项生态系统服务的分布模式、成对生态系统服务的相互关系和生态系统服务簇三个方面的动态变化，探讨了生态系统服务的权衡与协同效应的时间变化和空间尺度依赖。研究结果表明： （1）2000-2017年，研究区的产草量、产水量、水源涵养、防风固沙在空间上均呈现从东南到西北递减的分布趋势；而土壤保持、碳固持、栖息地质量和游憩潜力在空间上没有明显的分布趋势。从时间变化来看，除栖息地质量呈现逐渐下降之外，产草量、土壤保持、防风固沙呈现先减小后增加的趋势，而产水量、水源涵养、碳固持和游憩潜力呈现逐渐增加的趋势。 （2）在2000、2010和2017年三个时间点下，以像元尺度ES数据为例，全局Moran's I指数显示8项服务均在三个年份全都能维持空间上具备显著性的高度集聚，但每项服务的空间集聚性强弱程度发生了变化，LISA显示每项服务局部关联模式也随时间发生了面积和分布位置上的差异。相关分析显示成对生态系统服务的相互关系也会通过时间改变，主要体现在相关系数数值的变化。研究区服务簇的服务项构成和各类服务簇的面积占比也随着时间发生了改变，能揭示多项生态系统服务之间潜在的权衡协同作用的主要因素所起到的重要性程度也随时间发生了变化；与此同时不同土地利用类型上生态系统服务的组合情况是随着时间不断变化的，草地的产草量、产水量、水源涵养和碳固持在提高。权衡与协同效应的时间动态要求管理者更新特定生态系统服务的保护与修复边界以适应冷热点区域的动态变化；有助于回答在什么时间以及在哪里交付旨在改善的一系列生态系统服务的问题；综合考虑恢复水分条件和植被类型而进行半干旱区生态恢复活动从而弱化关键生态系统服务间的权衡。 （3）在像元、格网和子流域三个空间尺度下，以2010年ES数据为例，生态系统服务的空间集聚性有随着尺度变大而逐渐减弱的趋势，在子流域尺度下，栖息地质量和游憩潜力呈现随机分布。某些生态系统服务可能在不同尺度的分析单元中具有不同的局部关联模式，特别是冷热点区域。能够维持显著相关性的成对生态系统服务越来越少，并在相关关系的正负类型和相关系数的数值两方面表现出差异。呈现正相关的成对生态系统服务之间的相关性有逐渐增强的趋势，而呈现负相关的成对生态系统服务之间的相关性逐渐减弱，并且升尺度的过程会造成两两服务由负相关转变为不相关或正相关。大尺度下只需要较少的因素就能解释服务簇中ES项组成的变化，驱动服务簇发生分离的主导因素发生了改变，各类服务簇的面积占比和位置均随尺度发生了较大变化。权衡与协同效应的空间依赖要求管理者制定自上而下的决策框架以进行跨尺度生态系统保护优先级的设置；需要在不同的空间单元边界进行不同的生态系统服务组合设计；基于子流域尺度进行多功能景观的识别可以得到更具空间连续性的结果。 以上结果表明：单一时空尺度的分析容易对生态系统服务的权衡和协同效应产生片面的结果，进行多尺度分析有助于深入理解半干旱草原地区多种生态系统服务之间复杂的相互作用，从而为草原景观空间规划提供理论基础，以促进区域的可持续发展。

Ecosystem service (ES) refers to the direct or indirect gains from natural ecosystems. As a bridge connecting biophysical and socio-economic systems, ES significantly promotes human well-beings. In recent years, the mainstream of the ES framework into national and regional development strategies has attracted great attention from governments and the public around the world. Among them, the trade-offs and synergies of ecosystem services have become an important theme of ES research. A better understanding of the spatial and temporal characteristics of trade-offs and the cross-scale driving mechanism is conducive to balancing the supply and demand of ES among stakeholders, and is also a prerequisite for sustainable ecosystem planning and management. Inner Mongolia grassland is located in arid and semi-arid fragile zone, which is very sensitive to climate change and human activities. The region provides rich and diverse ecosystem services and is an important ecological barrier in northern China. Since the implementation of ecological restoration projects such as returning farmland to forestry and grassland in 2000, the pattern of ecosystem services and the relationship between services in the region have undergone tremendous changes. In this study, Xilinhot City, Inner Mongolia, was selected as the research area. Eight important ecosystem services were quantitatively simulated on GIS and RS platforms by using InVEST and other evaluation models with multi-source data, including grass yield (GY), water yield (WY), water retention (WR), soil retention (SR), sand fixation (SF), carbon sequestration (CS), habitat quality (HQ) and recreational potential (RP). Based on the analysis of the spatial and temporal distribution characteristics of ecosystem services, spatial autocorrelation analysis, Spearman correlation analysis, principal component analysis (PCA) and self-organizing mapping (SOM) were used to examine the dynamic changes of the distribution pattern of individual ecosystem services, the relationship between paired ecosystem services and ecosystem service clusters, respectively, in different time nodes (2000, 2010 and 2017) and on different spaces scales (pixel, grid and sub-watershed scale). The temporal and spatial scaling dependence of trade-offs and synergies of ecosystem services are discussed. The results show that: (1) From 2000 to 2017, grass yield, water yield, water retention and sand fixation in the study area showed a decreasing spatial distribution trend from southeast to northwest, while soil retention, carbon sequestration, habitat quality and recreational potential had no obvious spatial distribution trend. In terms of time variation, besides the gradual decline of habitat quality, grass yield, soil retention and sand fixation showed a trend of decreasing first and then increasing, while water yield, water retention, carbon sequestration and recreational potential showed a trend of increasing gradually. (2) On the three nodes of 2000, 2010 and 2017, taking ES data at the pixel scale as an example, the global Moran's I index shows that all eight services can maintain significant spatial agglomeration in three years, but the degree of spatial agglomeration of each service has changed. LISA shows that the pattern of local association mode of each service also varies in area and location with time. The correlation analysis shows that the relationship between pairwaise ecosystem services will also change through time, which is reflected in the the change of the value of the correlation coefficient. The composition of service items and the area proportion of various service bundles have changed over time, and the importance of the main factors that can reveal potential trade-offs and synergies among multiple ecosystem services has also changed. At the same time, the composition of multiple ecosystem services in different land use types has changed over time. Grass yield, water yield, water retention and carbon sequestration at the grassland type all showed a small increase trend. The time variation of trade-offs and synergies require managers to update the protection and restoration boundaries of specific ecosystem services to adapt to the dynamic changes of cold and hot spots; help to answer the questions of when and where to deliver a series of ecosystem services aimed at improving; and comprehensively consider the restoration of water conditions and vegetation types for ecological restoration activities in semi-arid regions in order to weak the trade-offs between key services. (3) At the three spatial scales of pixel, grid and sub-watershed, taking ES data on 2010 as an example, the spatial agglomeration of ecosystem services tends to weaken with the increase of scale. At the sub-watershed scale, the habitat quality and recreational potential show random distribution. Some ecosystem services may have different local correlation patterns in analysis units of different scales, especially in cold and hot areas. There are fewer paired ecosystem services that can maintain significant correlations, and there are differences in the positive and negative types of correlations and the numerical values of correlation coefficients. The correlation between paired ecosystem services with positive correlation is gradually increasing, while the correlation between paired ecosystem services with negative correlation is gradually weakening, and the scale-up process will cause the two services to change from negative correlation to positive correlation. In large scale, only a few factors are needed to explain the change of ES item composition in each ecosystem service bundles. The dominant factors driving the separation of bundles have changed and the area proportion and location of various service clusters have changed greatly with the scale. The spatial dependence of trade-offs and synergies requires managers to develop top-down decision-making frameworks to set priorities for cross-scale ecosystem protection; different combinations of ecosystem services need to be designed at different spatial unit boundaries; and multi-functional landscape recognition based on sub-basin scales can achieve more spatial continuity. The above results show that single-scale analysis is easy to produce one-sided results on the trade-off and synergistic effects of ecosystem services. Multi-scale analysis is helpful to understand the complex interaction between multiple ecosystem services in semi-arid grassland, thus providing a theoretical basis for grassland landscape spatial planning and promoting regional sustainable development.