Operationalizing restoration on the roof of the world: an integrated framework for Qinghai-Tibetan Plateau grasslands

1 Introduction

1.1 The plateau as a global ecosystem provider

The Qinghai-Tibetan Plateau (QTP) is a cornerstone of regional and global ecological security, most evident in its role as ‘Asia’s Water Tower.’ Its alpine grasslands are the headwaters for ten of Asia’s largest river systems, including the Yangtze, Yellow, and Mekong Rivers, which sustain nearly 1.4 billion people downstream. These ecosystems perform critical hydrological regulation through soil water retention and baseflow maintenance, with water conservation services estimated at 345 billion m³ annually (Cao et al., 2019). The alpine grasslands of the QTP are zoned along a precipitation gradient into three primary types: the humid, vegetation-dense alpine meadow; the xeric, sparsely-vegetated alpine steppe; and the hyper-arid, largely barren alpine desert (see Figure 1). It is important to note that these ecosystem types differ substantially in their ecology, degradation drivers, and restoration responses, and generalizations across the entire plateau should be interpreted with caution (Wang et al., 2022; Duan et al., 2021).

Restoration efforts, such as those under the Sanjiangyuan project, have enhanced water conservation capacity by 15.6% (Cao et al., 2019), yet these gains are undermined by escalating pressure from climate warming projected at 0.3–0.4 °C per decade, threatening the long-term stability of this vital service. Concurrently, the QTP’s alpine grasslands constitute a massive carbon reservoir, storing an estimated 23.4 Pg of soil organic carbon (Zhou et al., 2023). Restoration practices such as grazing exclusion have proven effective, increasing aboveground carbon stocks by over 84% according to meta-analyses, positioning these landscapes as indispensable for climate mitigation. The plateau also harbors unique and iconic biodiversity, from the Tibetan antelope to the snow leopard, whose persistence is intrinsically linked to the health of the grassland matrix (Lu et al., 2023).

1.2 The critical disconnect in policy and practice

Despite its ecological significance, the governance of the QTP’s grasslands is marked by a persistent policy–implementation gap. China has deployed substantial state-led conservation programs, most notably the ‘Returning Grazing Land to Grassland’ (Tuimu Huancao) initiative, which channelled billions in subsidies. However, sectoral fragmentation endures, with agricultural, environmental, and water resource agencies often pursuing isolated objectives without integrating these interdependent goals (Chen et al., 2022). This fragmented approach has failed to reverse widespread degradation, with 90% of QTP grasslands showing signs of deterioration by the late 1990s. A key systemic tension lies in the misalignment between political and ecological timescales. Administrative systems prioritize short-term, visible outcomes, while genuine ecosystem recovery requires patient, decadal-scale investment (Zhang et al., 2017). This is epitomized by the widespread use of fencing: while optimal ecological benefits occur within a 4–8 year window, this cycle is poorly synchronized with typical program evaluation timelines, risking long-term habitat fragmentation and reduced landscape connectivity for wildlife (Zhao et al., 2018).

1.3 A pathway forward: integrating services into resilience

To secure the QTP’s vital ecosystem services, restoration paradigms must evolve from single-target interventions to integrated resilience building. This requires transcending sectoral silos to manage grasslands explicitly for their coupled hydrological, climatic, and biodiversity functions. Future policy must bridge the temporal divide by institutionalizing long-term monitoring and adaptive management frameworks that align incentives with ecological recovery cycles. The challenge is to move beyond treating symptoms of degradation and instead foster the inherent capacity of these alpine grasslands to sustain their provisioning, regulating, and supporting services for Asia and the world (Fayiah et al., 2020; Dong, 2023) (see Figure 2).

2 Review methods

This narrative review followed a systematic search strategy to identify peer-reviewed literature published between January 2020 and March 2025. The following electronic databases were searched: Web of Science, Scopus, Google Scholar, and the China National Knowledge Infrastructure (CNKI). Search terms were applied in both English and Chinese and included the following combinations: (“Qinghai-Tibetan Plateau” OR “QTP” OR “Tibetan Plateau”) AND (“grassland restoration” OR “grassland degradation” OR “alpine meadow” OR “alpine steppe”) AND (“fencing” OR “grazing exclusion” OR “reseeding” OR “adaptive management” OR “co-management” OR “ecosystem services” OR “herder livelihoods” OR “permafrost” OR “soil organic carbon” OR “NDVI” OR “remote sensing”).

Inclusion criteria required that studies: (1) were conducted within the QTP boundaries or directly addressed QTP grassland systems; (2) reported empirical findings from field experiments, remote sensing analyses, or socio-economic surveys; and (3) were published in peer-reviewed journals or as book chapters with identifiable methodology. Studies were excluded if they were purely modeling exercises without field validation, predated 2020 (except for foundational theoretical frameworks and policy context), or were not available in English or Chinese. Supplementary pre-2020 literature was included selectively to provide the necessary ecological and policy context.

Evidence synthesis followed a weight-of-evidence approach. Remote sensing studies were evaluated for sensor type, spatial resolution, time, and validation against ground-truth data. Field experiments were assessed for study duration, replication, grassland type, and degradation stage. Socio-economic studies were evaluated for sample size, data collection method, and community representativeness. For each major intervention (fencing, reseeding, soil amendment, grazing optimization, and adaptive co-management), the strength of evidence was rated as well-supported (multiple independent studies with consistent findings), emerging (limited or mixed evidence), or interpretive (authors’ synthesis and inference), as indicated in the text.

The proposed Adaptive Co-Management (ACM) framework presented in Section 8 is a new conceptual synthesis by the authors, integrating principles from Ostrom’s common-pool resource governance (Ostrom, 2009, 2017), resilience theory (Folke, 2006), and polycentric governance literature (Armitage et al., 2008; Wyborn, 2015; Hasselman, 2017). It is not a direct adaptation of a single existing framework but draws on and extends these established theoretical foundations to the specific institutional and ecological context of the QTP.

3 Ecological baselines

3.1 The QTP grassland system

The QTP’s grassland biome is defined by the complex interplay of distinct alpine vegetation zones, a permafrost-dependent hydrological foundation, and a long history of human adaptation. This system is primarily divided into two major zones: the relatively humid alpine meadows, dominated by Kobresia and Carex sedges, and the drier alpine steppes, characterized by Stipa and Festuca grasses. Each zone faces unique degradation pathways: meadows tend toward the severe ‘black soil beach’ (Heihumai) syndrome, and steppes toward shrub encroachment (Dong, 2023). The entire system rests on a vast permafrost layer that underlies roughly 60% of the plateau and regulates soil hydrology, temperature, and nutrient cycling. For millennia, this environment co-evolved with traditional nomadic pastoralism defined by seasonal mobility and collective land management, a system finely attuned to grassland phenology and resilience (Wang et al., 2022).

3.2 Pressures and syndromes of degradation

This finely balanced system is now undergoing a profound transformation driven by intersecting climatic and anthropogenic pressures. The most severe degradation manifests in three primary syndromes. First, the formation of ‘black soil beach’ (Heihumai) represents a catastrophic, self-reinforcing feedback loop in alpine meadows (Ganjurjav et al., 2016): vegetation loss exposes dark soil, leading to increased erosion, extreme soil temperatures, and severely impaired hydrology, preventing natural recovery. Second, in the northeastern steppes, altered disturbance regimes facilitate widespread shrub encroachment by species such as Caragana, thereby diminishing forage quality and altering ecosystem structure (Zhu et al., 2023). The most fundamental pressure, however, is climate-driven permafrost thaw, which deepens the active layer and triggers thermokarst development. This process irreversibly alters the plateau’s hydrological base, leading to changes in drainage and habitat loss that can outpace restoration interventions. Together, these syndromes signify a systemic shift that challenges both ecological function and the sustainability of human livelihoods on the plateau (Dong et al., 2012).

4 Degradation trends and restoration interventions

4.1 Recent vegetation dynamics and drivers

Remote sensing analyses using the (NDVI) and (EVI) reveal a complex, heterogeneous landscape from 2020–2024, with pronounced greening in areas under grazing exclusion and localized browning where recovery has plateaued or reversed. Disentangling the drivers shows that both climate forcing and anthropogenic activities exert significant, spatially variable pressures, with their relative importance varying substantially between alpine meadow, alpine steppe, and alpine desert zones. Effective intervention requires moving beyond uniform policies to location-specific diagnostics that account for these interacting drivers (Piao et al., 2014; Chen et al., 2014) (see Figure 3).

4.2 Evaluating primary interventions: advantages, limitations, and trade-offs

4.2.1 Fencing (grazing exclusion)

Fencing is the most widely documented intervention on the QTP. Within a 4–8-year optimal window, it consistently facilitates seed bank recovery, biomass increase, and soil carbon accumulation across alpine meadow and steppe ecosystems (Cao et al., 2019; Zhao et al., 2018) (see Figure 4). However, long-term fencing (>10 years) produces diminishing returns and can lead to plant community homogenization, reductions in plant species diversity through competitive exclusion, and increased vegetation flammability (Qin et al., 2025). Fencing creates physical barriers that fragment wildlife movement corridors and restrict access of migratory species such as the Tibetan antelope (Zhao et al., 2018). Socially, fencing often imposes livelihood costs on herders who lose access to seasonal pastures, with subsidies shown to cover only a fraction of the income lost in many households (Yeh and Area, 2011). On severely degraded ‘black soil beach’ sites, fencing alone is insufficient for recovery; active reseeding is required (Zhou et al., 2023). Evidence for fencing efficacy in alpine steppe and alpine desert zones is more limited, and effects are substantially weaker than in alpine meadow contexts.

4.2.2 Reseeding and soil amendment

For severe ‘black soil beach’ (Heihumai) degradation, active reseeding with native species (Kobresia, Carex, Elymus) combined with phosphorus addition has demonstrated accelerated recovery of soil quality and vegetation cover compared to fencing alone (Lu et al., 2023; Zheng et al., 2022). Reseeding is costly and logistically demanding in remote alpine areas. Success rates vary substantially depending on elevation, slope aspect, soil moisture, and seed source provenance. Use of non-native or commercially sourced seed can reduce long-term biodiversity value. Phosphorus amendments may be less effective in soils with high organic matter content or where nitrogen is the primary limiting nutrient. Long-term monitoring studies are lacking, and the permanence of recovery without continued management intervention remains uncertain (Zhou et al., 2023). Evidence is primarily from alpine meadow contexts; efficacy in alpine steppe and desert is less established.

4.2.3 Adaptive grazing management

Adaptive grazing management, particularly through stocking rate optimization and herder cooperatives, offers a pathway to integrate livelihoods with ecological objectives. Evidence suggests that overall livestock density is more critical than rotational system design per se (Piao et al., 2019; Qin et al., 2025).

Adaptive grazing management depends on functioning collective institutions, which are undermined by China’s individualized grassland contract system (Li and Huntsinger, 2011). Optimal stocking rates are difficult to determine and vary substantially by grassland type, season, and inter-annual climate variability. Monitoring compliance with adaptive management rules requires continuous effort and social capital. Under conditions of severe degradation, reducing grazing alone may be insufficient to trigger recovery, and evidence for recovery on severely degraded sites through grazing optimization alone is limited. The transaction costs of establishing and maintaining cooperative management institutions should not be underestimated.

4.3 Synthesis and decision framework

No single intervention is universally optimal. The choice among fencing, active reseeding, or adaptive grazing must be guided by the specific degradation syndrome, grassland type (alpine meadow, steppe, or desert), and the social and institutional context. Moderately degraded steppes may recover with time-bound fencing, while severely degraded Heihumai meadows require intensive ecological engineering. Functional grasslands are best sustained through adaptive grazing cooperatives that optimize stocking rates. This targeted, syndrome-based approach is crucial for building resilient grasslands that sustain both ecosystem services and pastoral communities. The governance structure for implementing these targeted interventions is summarized in the Adaptive Co-Management cycle (Table 1). See the Evidence Summary Table (Table 2) at the end of this manuscript for a structured synthesis by intervention type. A companion Decision Framework (Table 3) is also provided, mapping the optimal restoration approach to each combination of degradation stage and ecosystem type. The associated monitoring indicators and performance targets used to track intervention outcomes are detailed in the Nested Monitoring System table (Table 4).

Table 1. The adaptive co-management cycle.

Source: Authors’ synthesis, building on Armitage et al. (2008); Wyborn (2015); Hasselman (2017), and Sendzimir et al. (2018).

Table 2. Evidence summary: restoration interventions by grassland type and degradation stage.

↑, increase; ↓, decrease; SOC, soil organic carbon; NDVI, Normalized Difference Vegetation Index. Evidence strength: Well-supported, multiple independent studies with consistent findings; Emerging, limited or mixed evidence; Interpretive, authors’ synthesis.

Table 3. Decision framework: restoration approach by degradation stage and ecosystem type.

Sources: Compiled from Cao et al. (2019); Zhao et al. (2018); Lu et al. (2023); Zhou et al. (2023); Qin et al. (2025), and Wang et al. (2022).

Table 4. Indicator domains, measurement methods, frequency, and performance targets for adaptive co-management monitoring.

Thresholds are indicative starting values and should be co-designed with local stakeholders during the Co-Design phase. They should be adjusted based on grassland type and degradation stage (see Table 3).

5 The human dimension: policy impacts and local knowledge

The success of grassland restoration on the QTP is inextricably linked to the complex human dimension, where well-intentioned top-down policies often conflict with local socio-ecological realities. Foundational programs like Tuimu Huancao, rooted in equilibrium ecology models, promoted widespread fencing and grazing exclusion but frequently offered subsidies that were insufficient to offset herders’ income losses, fostering economic uncertainty and compliance issues (Yeh and Area, 2011). This policy architecture has been reinforced by instruments like the Ecological Conservation Redline and national park designations, which can restrict traditional land use without providing viable alternative livelihoods (Li et al., 2020). Furthermore, the historical shift from communal management to individualized grassland contracts has created landscape fragmentation and tenure insecurity, effectively disincentivizing the long-term, place-based stewardship inherent to traditional nomadic knowledge systems (Li and Huntsinger, 2011; Roche et al., 2023).

6 Socio-economic repercussions and livelihood transitions

Grassland conservation policies have generated profound socio-economic impacts, often placing disproportionate burdens on pastoral communities. While direct subsidy payments constitute 10–30% of household income, they often fail to offset opportunity costs, with regressive effects hitting the least-resourced households hardest. This economic pressure, combined with land-use restrictions, drives significant labor displacement and out-migration, particularly among younger generations, leading to a loss of critical herding labor and the intergenerational erosion of traditional knowledge. Beyond material impacts, these policies threaten the very foundation of Tibetan pastoral identity, which is intrinsically tied to livestock husbandry, thereby contributing to cultural disruption and psychological distress (Yeh and Makley, 2019; Bum, 2025).

7 The undervalued resource of indigenous ecological knowledge

In parallel to these disruptions, a vast repository of adaptive local knowledge remains systematically underutilized in formal restoration planning. Herders possess a sophisticated, place-based understanding encoded in traditional grazing calendars that synchronize livestock movements with plant phenology and seasonal weather patterns, exemplified by the nuanced three-pasture rotational system. This knowledge extends to recognizing ecological indicator species, predicting vegetation dynamics, and managing complex plant–animal interactions. Such insights are critical for adaptive, context-specific management but are seldom integrated into top-down policy frameworks, representing a significant loss of opportunity for designing more effective and culturally coherent interventions (Wang et al., 2022; Dong et al., 2023).

8 Reimagining incentives: from rigid payments to adaptive contracts

The prevailing model of Payment for Ecosystem Services (PES) is hampered by fundamental design flaws that limit its efficacy and equity. Standardized payment rates, short contract durations (typically 3–5 years), and minimal herder participation in design lead to substantial ‘slippage,’ where payments are made without verifying corresponding ecological outcomes. Emerging alternatives point toward a more promising future. Community-based contracting and performance-based payment schemes, which are co-designed with and sometimes administered by herder collectives, demonstrate potential for greater local relevance, improved monitoring, and enhanced legitimacy. Shifting from rigid, top-down subsidies to flexible, accountable contracts is essential for aligning livelihood security with long-term ecological stewardship (Yu et al., 2021; Guo et al., 2022).

9 An integrated adaptive co-management framework

9.1 Principles and application

A sustainable future for the QTP necessitates a shift to an integrated adaptive co-management (ACM) framework, built on robust theoretical foundations. First, applying Ostrom’s principles for successful commons governance reveals critical institutional deficits on the plateau, including a lack of congruent rules and insufficient local participation in rule-making (Ostrom, 2009). Second, resilience thinking underscores the peril of pushing these complex socio-ecological systems toward irreversible thresholds and highlights the paramount importance of preserving both ecological and community-based adaptive capacity (Folke, 2006). Finally, the concept of polycentric governance provides a structural model that advocates coordinated decision-making across nested scales—from national policy to household management—to better align institutional responses with the specific scales of ecological and social problems, thereby enabling necessary local experimentation and learning (Ostrom, 2017). Together, these principles form the cornerstone of a management paradigm capable of navigating the inherent uncertainties of the plateau, balancing restoration goals with livelihood security, and fostering the long-term resilience of its vital grassland ecosystems (Armitage et al., 2008; Wyborn, 2015).

9.2 The adaptive co-management cycle: a five-step framework for iterative resilience

To move beyond the rigid, fencing-centric paradigm, we propose a five-step Adaptive Co-Management (ACM) cycle: Co-Design, Act, Monitor, Evaluate, and Adapt (see Figure 5). The process begins with Co-Design, which fosters genuine collaboration among herders, scientists, and agencies to jointly define problems, set goals, and co-produce intervention strategies. This leads to the Act phase, where flexible, phased implementation, supported by contingency planning and adaptive capacity buffers, enables responsive on-the-ground action (Le Pira et al., 2023). Systematic Monitoring follows, integrating ecological, hydrological, and socio-economic indicators through remote sensing and participatory field methods to track multi-dimensional outcomes. In the Evaluate stage, herder perspectives and technical assessments jointly examine causal attribution, distinguishing intervention impacts from external drivers. Finally, the Adapt phase uses these insights to refine strategies—scaling what works, modifying what does not—and feeds lessons learned back into a renewed cycle of Co-Design, creating a continuous loop of evidence-based learning and adaptive management suited to the complex socio-ecological dynamics of the plateau (Sendzimir et al., 2018; Li et al., 2021).

9.3 Operationalizing the framework: mechanisms for adaptive governance

Effective implementation of the ACM cycle depends on establishing three core operational mechanisms (Armitage et al., 2008). First, Multi-Stakeholder Platforms must be institutionalized as dedicated forums for ongoing dialogue and negotiation among herders, researchers, and government agencies at provincial, county, and community levels, ensuring inclusive representation of diverse community voices, including women, elders, and marginalized households. Second, a Nested Monitoring System is required, strategically combining broad-scale remote-sensing metrics with ground-based ecological plots and participatory observations by herders (Wyborn, 2015; Hasselman, 2017). Third, Flexible Financing Models are essential, moving beyond rigid annual budgets to funds that can be dynamically reallocated in response to monitoring results, complemented by insurance schemes to buffer climate and market shocks, and outcome-based payments that directly incentivize the achievement of verified ecological and livelihood goals (Frow et al. 2010; Sirkiä and Laanti 2015).

9.4 Transforming policy: recommendations for integrated and adaptive governance

To realize the ACM framework, a fundamental policy transformation is required, anchored in four key recommendations (Sharma-Wallace et al., 2018). First, Grazing System Reforms must shift from blunt bans to nuanced adaptive management, supported by technical extension services, strengthened collective herder institutions, and market mechanisms that incentivize sustainable practices. Second, implementing Herder-Run PES schemes is critical, piloting community-based contracts with longer durations (8–10 years), performance-based components, and direct herder involvement in program design to enhance legitimacy and ecological outcomes. Third, securing Tenure Security through the formal recognition of collective or community grazing rights with legal provisions for coordinated management across individual contracts is essential to provide the long-term stability needed for stewardship. Finally, establishing robust Cross-Sectoral Coordination through integrated planning processes and shared performance metrics is necessary to dismantle the current fragmentation among agricultural, environmental, and water resource agencies and align their objectives toward the common goal of socio-ecological resilience on the plateau (Brunner and Lynch, 2013; Cookey et al., 2016).

10 Key messages and practical implications

• Match restoration interventions to specific degradation types. For lightly degraded alpine meadows (>70% cover), use adaptive grazing management through herder cooperatives. For moderately degraded meadows (30–70% cover), apply time-bound fencing (4–8 years) followed by controlled grazing, combined with native reseeding (Kobresia, Carex species) and phosphorus amendment where needed. For severe ‘black soil beach’ (Heihumai) degradation, intensive ecological engineering, including native species reseeding and soil restoration, is required. For shrub-encroached steppes, targeted grazing management and shrub control are more effective than fencing alone.
• Implement the Adaptive Co-Management (ACM) framework. Follow a five-step iterative cycle: Co-Design, Act, Monitor, Evaluate, and Adapt. Establish multi-stakeholder platforms at provincial, county, and community levels. Deploy a nested monitoring system combining remote sensing (Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), ground-based ecological plots, and participatory herder observations. Use flexible financing mechanisms, including dynamic reallocation of funds, insurance schemes, and outcome-based payments.
• Reform grazing policies. Shift from blanket bans and rigid fencing to nuanced adaptive management. Strengthen collective herder institutions and secure community grazing rights through formal tenure recognition. Redesign Payment for Ecosystem Services (PES) models from short-term subsidies (3–5 years) to community-based, performance-based contracts with longer durations (8–10 years) and direct participation by herders.
• Institutionalize cross-sectoral coordination. Agricultural, environmental, and water resource agencies must adopt integrated planning processes and shared performance metrics aligned toward socio-ecological resilience to dismantle current fragmentation that undermines restoration effectiveness.
• Allow sufficient time for recovery. Successful restoration requires at least 5–7 years to achieve measurable improvements in plant cover, species diversity, and soil stability. Under the proposed ACM framework, restored grasslands can recover critical ecosystem services including water regulation (345 billion m³ annually; Cao et al., 2019), carbon sequestration (23.4 Pg soil organic carbon; Zhou et al., 2023), and habitat for iconic QTP biodiversity.

11 Conclusion

11.1 Synthesis of evidence

This review affirms the need to move beyond the simplistic fencing paradigm, which, while providing short-term recovery, incurs unsustainable ecological and social costs over the long term. Effective restoration requires the integrated management of ecological, hydrological, and livelihood systems, recognizing traditional pastoral knowledge as a critical asset (see Figure 6). The paramount finding is that the success of these socio-ecological efforts is fundamentally constrained by top-down, fragmented governance, pointing to the urgent need for reforms that establish participatory, outcome-based, and polycentric institutions. It bears emphasis that effective interventions must be tailored to specific grassland types (alpine meadow, steppe, desert) and degradation stages—one-size-fits-all approaches are demonstrably insufficient across the heterogeneous QTP landscape.

11.2 A unified path forward

The Qinghai-Tibetan Plateau is a global climate and hydrological commons, and its crisis demands a unified response. We call for a transformative policy shift that embeds local rights and knowledge as central restoration assets, establishing genuine power-sharing in decision-making. Research must prioritize long-term adaptive management experiments and climate-resilient restoration species. By adopting the Adaptive Co-Management (ACM) framework—a pathway that is both scientifically rigorous and socially legitimate—we can secure the plateau’s vital ecosystem services, offering a model for high-altitude grassland conservation worldwide.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. During the preparation of this work, the authors used DeepSeek to improve the language, conciseness, and structural flow of the manuscript. After using this tool, the authors reviewed, fact-checked, and edited the content as needed and took full responsibility for the scientific content and conclusions of the publication. AI tools were not used to generate figures, tables, or data. All figures were produced by the authors using, e.g., ArcGIS Pro, R, Adobe Illustrator, and bioeroder. All quantitative values, citations, and factual claims have been manually verified by the authors.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Necmi Aksoy, Duzce University, Türkiye

Reviewed by: Jian Chen, University College London, United Kingdom

Xi Wei, China University of Geosciences (Beijing) Energy Institute, China