三江源区地处青藏高原腹地，是长江、黄河和澜沧江的发源地和我国重要的生态安全屏障。草地（主要包括高寒草甸、高寒草原和温性草原）作为三江源区分布最为广泛的生态系统，不仅直接支撑着当地数百万牧民的生计，同时其提供的诸多生态系统功能和服务也深远的影响着我国的中下游地区。近几十年来，在气候变化和人类活动的双重影响下，三江源区草地生态系统发生了严重的退化。为了遏制草地的持续退化，2005年以来国家实施了三江源生态保护与建设工程。然而，目前针对三江源退化草地恢复成效评估的研究仍存在着诸多不足之处。

本研究采用空间代替时间的方法，在三江源生态恢复工程典型区域进行了长时间序列的野外调查，以期分析不同恢复方式（人工恢复和自然恢复）和恢复时间对退化高寒草甸、高寒草原和温性草原生态系统结构、生态系统自组织过程、生态系统功能及其权衡关系的影响，揭示生态系统功能恢复的调控机制，评估三江源生态恢复工程的效益。主要研究结果如下：

（1）恢复措施显著影响了各类退化草地的群落结构，有效的改善了退化草地的大部分生物因子（除了高寒草原的物种丰富度、温性草原的鼠洞密度）、目标物种和非生物因子（除了高寒草原的土壤pH），但是和健康草地相比仍然有明显的差异。同时，不同恢复方式对各类草地生态系统结构因子的影响存在差异。自然恢复能够更有效的恢复高寒草甸目标物种的丰富度、多度和生物量，而人工恢复对高寒草原的目标物种丰富度、温性草原的目标物种多度和生物量提升效果更加明显。

（2）常见物种的空间格局在不同处理间和不同恢复年限间有所差异。总体来看，常见种以聚集分布为主，其次为随机分布，没有均匀分布。除了自然恢复高寒草原的植物群落没有显著关联外，各类草地的植物群落均表现为显著正关联。对于常见物种之间的种间关联，人工恢复高寒草原的显著相关物种对（包括极显著相关物种对）最多，其次为自然恢复高寒草原和自然恢复温性草原，人工恢复和自然恢复高寒草甸、人工恢复温性草原的显著相关的物种对最少。

（3）恢复措施显著提升了退化草地的部分生态系统功能和生态系统多功能性，但仍显著落后于健康草地。人工恢复和自然恢复对高寒草甸和温性草原的大部分生态系统功能和多功能性的提升效果没有显著差异。对于高寒草原，人工恢复对生态系统功能的提升效果优于自然恢复。总体来看，恢复措施对高寒草甸生态系统功能的提升效果优于高寒草原和温性草原。

（4）恢复措施、生物因子和非生物因子有效的解释了各类草地的大部分生态系统功能和多功能性。恢复措施对草地生态系统功能恢复的调控机制分为两种：一种是恢复措施（包括恢复方式和恢复时间）对生态系统功能没有直接显著的影响，而是通过影响生物和非生物因子对生态系统功能产生影响；另一种是恢复措施既可以直接对生态系统功能产生显著的影响，也可以通过生物和非生物因子调控生态系统功能的恢复。高寒草甸生态系统多功能性（35.6%）主要受到莎草科和禾本科多度占比、土壤含水率和土壤pH的调控；高寒草原生态系统多功能性（54.3%）主要受到恢复方式和裸斑面积占比的调控；温性草原生态系统多功能性（42.8%）主要受到莎草科和禾本科多度占比的调控。

（5）恢复措施一定程度上影响了各类草地不同处理间地上生态系统功能之间的权衡关系。对于牧草生物量和植物养分含量（叶片氮含量、叶片磷含量和叶片钾含量）之间的权衡关系，在各类退化草地均表现为植物养分含量更加受益。相较于地上生态系统功能之间的权衡关系，各类草地地下生态系统功能之间的权衡关系对不同处理的响应不明显。尤其是在高寒草原，不同处理间所有地下生态系统功能之间的权衡关系都没有显著差异。不同类型草地地上-地下生态系统多功能性之间的同步性存在差异：自然恢复高寒草原、人工恢复和自然恢复温性草原的地上和地下生态系统多功能性表现出较高的同步性，而人工恢复和自然恢复高寒草甸、人工恢复高寒草原地上和地下生态系统多功能性的同步性较差。

（6）随着恢复时间的增加，各类草地的生态系统结构因子、生态系统功能和生态系统权衡关系呈现出复杂的变化趋势，而非简单的线性变化。

（7）恢复措施显著提升了三种退化草地的牧草生物量、生物多样性、土壤固氮和生态系统总体的价值量，提升了自然恢复高寒草甸与人工恢复和自然恢复高寒草原水源涵养的价值量。自然恢复对高寒草甸生物多样性和水源涵养价值量的提升效果更加明显；人工恢复对高寒草原牧草生物量、水源涵养和生态系统服务总体价值量与温性草原牧草生物量和生物多样性价值量的提升效果更加显著。人工恢复和自然恢复均显著提高了三种草地的生态系统服务净收益，且两种恢复方式之间没有显著差异。三江源生态恢复工程实施以来效益显著，为189.43亿美元，其中高寒草甸为172.53亿美元，高寒草原7.89亿美元，温性草原9.02亿美元。

综上所述，三江源生态恢复工程成效显著，不同恢复措施显著的改善了退化草地的生态系统结构、自组织过程、生态系统功能。同时，草地的各项因子随着恢复时间表现出复杂的变化趋势，且与健康草地相比仍然差异明显。因此，需要认识到生态恢复的复杂性和艰巨性，长期的生态监测和合理的适应性管理措施必不可少。

Three-river Headwater Region (TRHR), located in the hinterland of Qinghai-Tibetan Plateau, is the source of Yangtze River, Yellow River and Lancang River and important ecological security of China. As the most widespread ecosystem in TRHR, grasslands (mainly including alpine meadows, alpine steppes and temperate steppes) not only supports livelihood of millions of herdsmen, but also provides many ecosystem functions and services which profoundly affected the middle and down areas of China. In recent decades, grasslands in TRHR has experienced severe degradations due to the joint effects of human activities and climate changes. To combat this situation, Chinese government has implemented ecological conservation and construction projects in TRHR since 2005. Unfortunately, there are still a lot of deficiencies in the researches about restoration effectiveness evaluation of degraded grasslands in TRHR.

 Based on space-for-time approach, long-term field monitoring in typical implementation areas of TRHR ecological restoration projects was carried out to analyze the effects of restoration methods (artificial restoration and natural restoration) and restoration time on ecosystem structures, self-organized processes, dynamics and trade-offs of ecosystem functions in alpine meadows, alpine steppes and temperate steppes, reveal the restoration mechanisms of ecosystem functions and evaluate the profit of TRHR ecological restoration projects in this study. The main results were as follows:

(1) Restoration measures significantly affected the community structures of all kinds of degraded grasslands, and efficiently improved many biotic factors (except for species richness of alpine steppes and pika burrow density of temperate steppe), target species and abiotic factors (except for pH of alpine steppes), but there were still significant differences in those factors between restored grasslands and healthy grasslands. Meanwhile, effect of restoration methods on the ecosystem structure factors of all the grasslands were different. Natural restoration could more effectively restore species richness, abundance and biomass of target species in alpine meadows, while artificial restoration had more clear effect on the target species richness in alpine steppes, abundance and biomass of target species in temperate steppes.

(2) Spatial pattern of common species varied with different treatments and restoration years. Generally speaking, common species were mainly clustered, followed by random clustered, without uniform distribution. Except for the insignificant association of plant community in naturally restored alpine steppes, plant community of all the other grasslands showed significantly positive association. For the interspecific associations among common species, the most significant species pairs (including extremely significant species pairs) were found in artificially restored alpine steppes, followed by naturally restored alpine steppes and naturally restored temperate steppes, and the least significant species pairs were found in artificially restored and naturally restored alpine meadows and artificially restored temperate steppes.

 (3) Restoration measures significantly promoted part of ecosystem functions and multifunctionality of degraded grasslands, but they were still far lower than those of healthy grasslands. There was no significant difference in most of ecosystem functions and multifunctionality of alpine meadows and temperate steppes between artificial restoration and natural restoration. For alpine steppes, artificial restoration could improve ecosystem functions more efficiently than natural restoration. Overall, effects of restoration measures on ecosystem functions in alpine meadows was better than that in alpine steppes and temperate steppes.

(4) Restoration measures, biotic and abiotic factors could effectively explain most of ecosystem functions and multifunctionality. There were two regulation mechanisms of restoration measures on the restoration of grassland ecosystem functions. One was that restoration measures (restoration methods and restoration time) had no significant direct impacts on ecosystem functions, but affected them by influencing biotic and abiotic factors. The other was that restoration measures can not only have a significant direct impact on ecosystem functions, but also regulate the ecosystem functions through biotic and abiotic factors. Ecosystem multifunctionality of alpine meadows (35.6%) was associated with abundance ratio of Cyperaceaes and Gramineaes, soil moisture and soil pH. Ecosystem multifunctionality of alpine steppes (54.3%) was regulated by restoration methods and bare land ratio. Ecosystem multifunctionality of temperate steppes (42.8%) was significantly impacted by abundance ratio of Cyperaceaes and Gramineaes.

(5) Restoration measures partly changed the trade-offs between aboveground ecosystem functions of all the grasslands. For the trade-offs between forage biomass and plant nutrient concentrations (leaf nitrogen concentrations, leaf phosphorus concentrations and leaf potassium concentrations), plant nutrient concentrations were more beneficial in all kinds of degraded grasslands. Compared with trade-offs between aboveground ecosystem functions, trade-offs between belowground ecosystem functions in all the grasslands didn’t vary greatly with different treatments. There was no significant difference in all the trade-offs between belowground ecosystem functions among different treatments in alpine steppes. Synchronism between above- and belowground ecosystem multifunctionality of grasslands varied with grassland types: there were relative high synchronism between them in artificial restored alpine steppes, artificial restored and naturally restored temperate steppes, and relative low synchronism in artificial restored and naturally restored alpine meadows and artificial restored alpine steppes.

(6) Ecosystem structure factors, dynamics and trade-offs of ecosystem functions of all the grasslands showed complicated trends along restoration times, rather than simple linear change.

(7) Restoration measures effectively improved the value of forage biomass, biodiversity, soil carbon storage and total ecosystem services of three degraded grasslands and water conservation of naturally restored alpine meadows, artificially restored and naturally restored alpine steppes. Natural restoration had a greater improvement in value of biodiversity and water conservation of alpine meadows than artificial restoration; while artificial restoration had a greater improvement in value of forage biomass, water conservation and total ecosystem services of alpine steppes and forage biomass and biodiversity of temperate steppes than natural restoration. Both artificial restoration and natural restoration significantly improved the net profit of ecosystem services of three grasslands, and there was no significant difference between two restoration methods. The benefits of TRHR ecological restoration projects was really remarkable, reaching 18.943 billion dollars, including 17.253 billion dollars for alpine meadows, 0.789 billion dollars for alpine steppes and 0.902 billion dollars for temperate steppes.

In a word, the ecological restoration projects of Three-river Headwater have achieved remarkable results, and different restoration measures significantly improved the ecosystem structures, self-organization processes and ecosystem functions of degraded grasslands. Meanwhile, the factors of restored grasslands showed a complex trend along restoration time, and were far lower than those of healthy grasslands. Consequently, it is necessary to recognize the complexity and difficulty of ecological restoration, and long-term ecological monitoring and reasonable adaptive managements are required in further ecological actions.