Abstract
As climate change impacts are intensifying, flood adaptation increasingly relies on the interaction between planned and autonomous adaptation, yet empirical evidence on such interaction remains limited. This study draws on fieldwork and policy document analysis to examine how planned and autonomous adaptation interact across different flood adaptation stages in multi-ethnic rural communities along the Tea-Horse Road in southwest China. The findings reveal mixed adaptation, in which planned and autonomous measures coordinate in resource allocation, labor mobilization, and institutional arrangements, thereby supporting community resilience. However, this interaction remains loosely integrated, with limited collaboration between community and governmental actors. Governance is dominated by planned adaptation, which often overlooks local ecological norms, cultural preferences, and knowledge systems, undermining communities’ self-reliant resilience and causing maladaptation. Such maladaptation is rooted in structural imbalances reinforced by historical inertia and path dependence. The study conceptualizes adaptation as a cross-scale process and argues that strengthening mixed adaptation requires redefining criteria for successful adaptation, recognizing the value of autonomous adaptation, and establishing regular negotiation mechanisms. Drawing on empirical evidence from rural China, the study provides theoretical insights into early forms of mixed adaptation with policy-relevant implications for community resilience in the Global South.
Keywords
Introduction
Climate change and intensified human activity have led to growing frequency and severity of extreme weather events, with floods being among the most damaging (Alifu et al., 2022; Rogers et al., 2025). Over the past 40 years, the concept of resilience has gradually become an important topic in research on flood events (Adger, 2000; Folke, 2006). Some researchers advocate that through proactive strategies, such as infrastructure improvement, technological advancement, and the strengthening of social organization, humans can enhance their adaptive capacity to increasing climate risks over long period (Chen et al., 2021; Yang et al., 2024).
The majority of disastrous flood impacts occur at the community level, making communities the fundamental unit of flood adaptation (Kwok et al., 2018; Laurien et al., 2020). Community resilience refers to the ability of a community to absorb shocks, accelerate recovery, and adapt to future challenges in the face of flood events. The formation of community resilience is a dynamic and multi-dimensional process, involving the interplay of individual and collective capacities, as well as external factors (Djalante and Thomalla, 2010). To improve community resilience, community-based flood adaptation has increasingly demonstrated cross-scale characteristics in recent years (Dietz et al., 2020; Tian et al., 2019). This trend is reflected in the collaborative participation of diverse actors, the combination of multiple measures, and the emergence of governance structures that span all phases of flood events (Adger et al., 2005; Orlove, 2022). These adaptive processes are shaped through ongoing actions and interactions that span geographic space, operate within social networks, and are embedded in multi-level governance systems (Doshi and Garschagen, 2024; Di Gregorio et al., 2019). By establishing and expanding interactions that link individuals and households, community-level formal and informal organizations, and broader national and institutional levels, community resilience can be enhanced (Bidwell et al., 2013).
Given the aforementioned diversity and complexity, enhancing community resilience first requires clarifying a fundamental question: “Who takes what action, when, and how?”. Existing studies have shown that government agencies, residents, businesses, and informal groups each play specific roles at different stages of the governance process (Petzold et al., 2023; Wannewitz and Garschagen, 2023). Among them, governments often take the lead throughout the entire process of planning, implementation, coordination, and evaluation, and tend to prioritize short-term interventions such as early warning, infrastructure development, and emergency response, reflecting the governance characteristics associated with the Disaster Risk Management (DRM) paradigm (Biagini et al., 2014). At the same time, individuals, households, and locally based informal groups—particularly in rural and multi-ethnic communities—have developed a wide range of everyday practices grounded in local knowledge and kinship networks (Juhola et al., 2022; Lindegaard and Sen, 2022). These forms of response tend to unfold gradually and across generations, aligning more closely with the long-term adaptation pathways emphasized in the Climate Change Adaptation (CCA) literature (Chen et al., 2021).
In the DRM and CCA literature, community flood adaptation is often framed through two pathways: planned adaptation and autonomous adaptation (also referred to in some studies as top-down and bottom-up approaches, respectively) (Fankhauser et al., 1999). Planned adaptation generally refers to deliberate and purposive interventions undertaken in response to projections of future climate change and associated risk assessments. It is usually led by governments or formal institutions and relies on policy instruments, scientific knowledge, and institutionalized resources (Innes and Booher, 2010; IPCC, 2007; Koontz and Newig, 2014; Smit et al., 2000). Planned adaptation is typically manifested in large-scale, long-term, and institutionalized policy actions (Engle, 2011). Autonomous adaptation refers to self-initiated adjustments by households or communities in response to ecological and socio-economic changes, primarily aimed at sustaining livelihoods. Such practices originate from direct perceptions of environmental change and are deeply embedded in local knowledge, social relations, and cultural practices (Adger et al., 2005; Atanga, 2023; Bawakyillenuo et al., 2016; Forsyth and Evans, 2013; IPCC, 2001; Malik et al., 2010). Although CCA and DRM address both short-term risk reduction and long-term adaptation at the community level, existing research has often treated them separately due to their distinct theoretical and policy traditions.
In recent years, a growing body of research has emphasized that planned and autonomous adaptation are not mutually exclusive in community flood adaptation. Instead, they frequently interact in practice, manifesting in forms of collaboration, and even conflict (Malik and Smith, 2012; Mersha and van Laerhoven, 2018). Differences in actors’ roles, resource mobilization, and knowledge systems significantly influence the integration and effectiveness of adaptation efforts. Planned adaptation can go wrong because of poor design and a lack of understanding of local context. Autonomous adaptation can be maladaptive when implemented with limited information, lack of support from social networks, or cross-interference between actors who fail to communicate with each other (Fenton et al., 2017; Schipper, 2020). These dynamics not only challenge the conventional binary classification but also prompt renewed scholarly attention to the interplay between adaptation pathways (Juhola et al., 2022). A recent global review of 1682 adaptation cases by Maskell et al. (2025) shows that more than one-third exhibit characteristics of mixed adaptation, defined as adaptation responses that combine elements of both autonomous and planned adaptation. Such actions often involve the coexistence or interaction of self-driven, everyday practices with externally supported and institutionalized measures. The study develops a qualitative typology that identifies nine modes of interaction between the two forms, ranging from coexistence and mutual support to conflict and constraint. This reconceptualizes adaptation as a continuum rather than a dichotomy, providing a more flexible and practice-oriented framework for understanding adaptation processes at the local level. At the same time, it highlights potential risks of inequity and maladaptation (Maskell et al., 2025).
Scholars increasingly call for multi-dimensional analyses of these interactions, particularly focusing on institutional coordination, cultural tensions, and governance outcomes, while also highlighting the associated risks of injustice and maladaptation (Eriksen et al., 2015; Magnan et al., 2016). Although increasing theoretical attention has been given to mixed adaptation, empirical examinations of how such modes operate in specific contexts remain limited. This gap is particularly evident in Global South communities, especially in multi-ethnic and rural areas, where governance actors often operate with highly heterogeneous logics and value systems. Such heterogeneity complicates the coordination of mixed adaptation and may further exacerbate existing social vulnerabilities (Araos et al., 2017; El-Masri and Tipple, 2002; Kundzewicz et al., 2014).
To address this gap, this study takes rural communities along the Tea-Horse Road (Chama gudao, hereafter THR) in Southwest China as a case to investigate how planned and autonomous adaptation interact in flood adaptation processes (Huang et al., 2025). The study area is characterized by a relatively continuous civilizational history, while its distinctive geographical and institutional conditions have produced a governance configuration in which state interventions coexist with diverse local practices—constituting a typical setting for the emergence of mixed adaptation. In this study, we argue that the key essence of mixed adaptation lies in its cross-scale interactions. Based on fieldwork and policy document analysis, this paper addresses two core questions: (1) At the community level, who takes what kind of action, at what time, to construct mixed adaptation? (2) In highly heterogeneous communities, how does mixed adaptation operate in practice, and what are its key contributions and constraints in shaping community resilience? By empirically identifying and analyzing patterns of mixed adaptation, this study aims to contribute to theoretical understandings of adaptive pathway interactions and provide policy-relevant insights for enhancing community resilience in the Global South.
Research area and materials
The Tea-Horse Road area and its flood environment
The THR area in Southwest China is a historically significant trade corridor linking tea-producing areas in Yunnan, Sichuan, and parts of Xizang with markets in Southeast and South Asia (Jiang, 2014; Yang et al., 2019). It covers the southeastern Qinghai-Tibet Plateau, the Sichuan Basin, and the Yunnan-Guizhou Plateau, featuring diverse landscapes, climates, and cultural traditions (Sigley, 2021). The elevation in the region leads to climatic variation, ranging from subtropical lowlands to temperate and cold highlands, shaped by both southeast and southwest monsoons. Most rainfall occurs between May and October, often in the form of intense night time storms (Xu et al., 2018; Yang et al., 2021). Combined with steep terrain, these conditions make the area highly susceptible to floods, landslides, and other hazards triggered by heavy rains (Mao et al., 2014; Yin et al., 2021).
The region is home to more than twenty ethnic groups, including Han (Hanzu), Tibetan (Zangzu), Yi (Yizu), and Bai (Baizu) (Zhang, 2024). These communities are typically organized around clan-based settlements, primarily reliant on agriculture, with tightly knit social structures and strong cultural continuity. Floods have continuously exacerbated the vulnerability of these often economically disadvantaged, multi-ethnic communities (Zhuang et al., 2010). In response to increasing flood risks, adaptation in the THR region has increasingly exhibited characteristics of multi-actor collaborative governance. On one hand, in accordance with the Flood Control Law of China, the government, as the primary actor responsible for flood management, has institutionalized governance systems at the township level, established dedicated agencies, and continuously implemented various interventions (Wu, 2023). On the other hand, local communities have developed culturally embedded and intergenerationally transmitted autonomous adaptation strategies (Li, 2025; Zhou, 2021), rendering flood adaptation highly localized and experience driven. The coexistence of planned and autonomous adaptation has jointly shaped the foundational structure of flood adaptation in the region.
Research materials and methods
This study adopts a qualitative research design that combines fieldwork and policy document analysis (Table 1). Fieldwork for this study was conducted from July to August 2023 across several flood-prone communities along the THR in Southwest China, covering areas in Dali and Lijiang in Yunnan Province, as well as Ya’an and Xichang in Sichuan Province. The research employed a combination of participatory observation, semi-structured interviews, and group discussions. Respondents included three main groups: long-term residents living in high-risk flood zones, township-level officials involved in grassroots flood adaptation, and influential figures within local society such as clan leaders, religious authorities, and community elders. Through observation, the research team gained insights into the natural environment and everyday social interactions within the communities. Interviews with township officials focused on institutional flood management, including infrastructure development, early warning systems, and emergency response mechanisms. Interviews with residents and local opinion leaders explored their flood experiences, risk perceptions, daily coping strategies, and systems of local knowledge. These qualitative data were essential for identifying the behavioral logics of various actors involved in adaptation, particularly in relation to how local experience, cultural norms, and institutional arrangements interact in the formation of flood responses. Among these, an in-depth ethnographic investigation was conducted in eleven communities of Husa (户撒) Township, Dehong Prefecture, Yunnan Province, which constitutes the primary empirical basis of this paper. Husa is a multi-generational settlement of the Achang (Achangzu) ethnic minority in China and has also been subject to government interventions. The demographic characteristics of the respondents and the sampling criteria are presented in Table 2. In total, 66 interviews were conducted over a 28-day fieldwork period, each lasting between one and three hours. The data were documented in the form of photographs, audio and video recordings, and detailed field notes. Field interviews are cited in-text with anonymized codes (e.g. HS19, Clan Chief, Dehong, July 2023).
Research data sources (selected).
Demographic profile of key informants.
*Selection criteria key: 1 = Flood experience, 2 = Relocation, 3 = Floodplain residence, 4 = Special status, 5 = Willingness.
Secondly, government documents were primarily collected from national and provincial government sources, including the PKU (Peking University) Law database (Beida fabao) and the official websites of emergency management departments in Yunnan, Sichuan, and Xizang. The research focused on policies, regulations, governance plans, and technical guidelines related to flood prevention and control issued between 2010 and 2023. Keywords used for document retrieval included “flood”, “meteorological disaster”, “flood control”, and “hazard relief”. To ensure analytical relevance, only documents that explicitly addressed disaster governance at the community level were included. These materials were used to analyze the institutional arrangements underpinning government-led flood adaptation.
Given the exploratory nature of the questions posed in this study, a qualitative methodology was deemed most appropriate (Creswell, 2012). During the analysis phase, the study combined grounded theory and the analytical lens of critical political ecology. On the one hand, drawing on grounded theory (Glaser and Strauss, 1967), interview transcripts and observational notes were systematically coded to construct an analytical framework inductively for understanding the interaction between planned and autonomous adaptation. On the other hand, informed by critical political ecology (Valdivia and Perreault, 2015), the study examined how state-led techno-institutional adaptation strategies are embedded in local social contexts, highlighting the tensions between flood knowledge production, resource allocation, and governance power. This dual approach helps explain the emergence of mixed adaptation modes in complex socio-political settings.
Resilience measures and actors in coping with floods
With the accumulation and institutionalization of governance experience, flood adaptation in THR communities has gradually evolved into four key phases: prevention, preparedness, response, and recovery. The following section maps the key measures and the interactions among actors across these stages, highlighting the mixed and cross-scale nature of adaptation that involves both planned and autonomous processes.
Flood prevention
According to policy documents, flood governance along the THR adheres to a “prevention-first” principle, focusing on risk identification and small-watershed engineering to mitigate flood impacts and safeguard agriculture and local livelihoods with water conservancy measures (Xu et al., 2020). Within a tiered governance framework, provincial and municipal water authorities are responsible for planning flood control projects, allocating dedicated funds, and promoting implementation through performance evaluation mechanisms. Interviews with township officials further confirmed that township governments oversee local risk assessments and technical standardization. Using data on rainfall, terrain, population, and economic activities, flood management agencies produce risk maps to delineate safety zones, evacuation routes, and identify high-risk groups. 1 At the community level, participant observation showed that village committees are responsible for installing warning signs, preparing broadcast messages, and communicating early warning information as part of their routine responsibilities.
Beyond the scope of planned adaptation, the adaptation of individual households is often fragmented and remains invisible in official narratives. Detailed participatory observation reveals that flood prevention behaviors are embedded in local values, cultural rituals, and rhythms of production. Most families, guided by Theravada Buddhist teachings, uphold an ecological ethic that shapes their relationship with the natural environment. Eco-friendly communities are ultimately the result of autonomous adaptation and have contributed to reducing flood risks (Huang, 2014). For instance, according to an interview with a Yi clan chief, when a spring is discovered, measures are taken to protect it, and the surrounding forest is designated as a protected water source area. 2 The Miao ethnic group (Miaozu), on the other hand, uses “Water Books” (shui shu) to record flood-related information, helping guide the timing of planting and harvesting activities. 3
Flood preparation
Policy guidelines specifying flood preparedness in the THR area involves three core components: rainfall monitoring, early warning dissemination, and evacuation. At the community level, basic rainfall monitoring stations and information systems are in place. Interviews with township officials and villagers revealed that during the flood season, flash flood warnings are issued based on weather forecasts and river level changes and are disseminated to residents through multiple channels, including loudspeakers, mobile phones, and village networks. 4 Most communities have designated shelters capable of accommodating evacuees for more than ten days. Evacuation routes are specified for each household, and annual flood drills are conducted. Field interviews further indicated that communities operate under a “Ten-Household Joint Defense” system, in which every ten households form a team led by a designated coordinator responsible for receiving early warnings, relaying evacuation orders, and organizing mutual support during emergencies. Additionally, “Flood Information Cards” are distributed to households, detailing family members’ information, contact details for local team leaders, and designated evacuation routes.
In the preparedness phase, informal kinship organizations also participate in adaptive actions. An elderly Buddhist said that the custom of collective food storage has evolved into a form of customary law. 5 One flood survivor stated, “After the autumn harvest in August, each Achang household contributes fifty kilograms of rice to a communal flood reserve.” 6 Similarly, the Dulong (Dulongzu) implement a “communal eating system” (gongshi zhidu), gathering and storing wild edible resources each March under tribal leadership. 7 Participant observation revealed that in the humid environments inhabited by the Yao (Yaozu), grain is typically stored in sealed granaries known as “he cang”.
Participatory observation shows that during the preparedness phase, adaptive actions at the individual and household levels are particularly relevant. Most ethnic minority communities along the THR area are located at elevations between 1500 and 2000 meters, a range that supports both mountain hunting and lowland farming. Strengthening housing structures is a key strategy for coping with floods. Common stilted buildings feature sloped roofs of approximately 30 degrees to facilitate drainage. These two-storey structures typically use the upper level for living and the raised ground floor for storing tools or keeping livestock, effectively preventing moisture and water intrusion. In Qiang (Qiangzu) communities, densely arranged dwellings connected by narrow pathways create tightly knit social spaces that support mutual assistance. According to flood survivors, in recent years, some communities have constructed brick and stone pedestal houses with elevated foundations to reduce water damage. Additional measures include waterproofing roofs, sealing doors and windows, and reinforcing walls to enhance flood safety. 8
Emergency flood responses
During the rainy season, government agencies closely monitor rainfall patterns. Each community has two hydrological monitors who are required to reside in the community long-term and monitor the rivers and streams within the area. A flood monitor from the Achang community in Longchuan County, Yunnan Province, reported that during the dry season he only needs to patrol the mountains once a day, but during the rainy season he must conduct measurements at least three to four times daily. 9 He records videos of the actual conditions at various flood-prone locations and sends them to a dedicated WeChat group. On days of heavy rainstorms, he is required to be on duty 24 h a day. If any abnormalities are detected, he immediately notifies the villagers to prepare for evacuation by striking a gong or blowing a whistle. Upon receiving the signal, people promptly evacuate to shelters. Policy documents and participant observation confirmed that after such evacuations, the government dispatches a task force to mobilize resources quickly to assess the disaster, repair damaged roads, power, and communication facilities.
Influenced by cultural backgrounds, communities along the THR have developed social organizational structures based on the family clan system, which is vital for flood response. The ritual specialist in charge of ceremonies explained that, the residents of a village typically belong to the same family clan, or are related through marriage or close kinship ties. 10 Most villagers agreed that each member has both the rights and obligations to protect and be protected, to assist and be assisted. 11 In Yi group customary law, there is a locally circulated proverb: “When relatives and family come to support, famine will not claim lives” (qinren lai bangmang, jihuang bu shangren). 12 This saying was frequently cited by villagers during field interviews. Based on their flood experiences, they recognize that government assistance often cannot arrive immediately, so they must rely on their own capabilities to evacuate. Interviews with flood monitors reveal that during flood events, residents prioritize rescuing family members, especially children and the elderly, and then relocate essential belongings such as livestock and stored agricultural products to safer locations. They then assist their neighbors, particularly households with elderly members, those without adults, and those headed by women. Sharing food with neighbors is a strategy and behavior that helps people survive the first and second days of a flood. The elderly and children are evacuated to shelters. 13
Post-flood recovery
The township government plays a central role in post-flood reconstruction, especially via economic compensation and spatial relocation strategies. Policy documents indicate that smallholders receive a range of subsidies related to production, livelihood, and resettlement. In terms of housing, two main approaches are adopted. According to the civil affairs department, in communities with limited flood damage, on-site rebuilding is prioritized to preserve traditional settlement patterns and minimize farmland loss. In contrast, communities in high-risk flood zones are often relocated. Original settlements are dismantled and the land is converted into farmland. Resettled households are provided with low-interest loans to support the transition. The village committee director of Village G noted that the village committee supports recovery by providing free agricultural training programs to help smallholders restore production. Meanwhile, the agricultural cooperative contributes by supplying shared equipment, organizing bulk purchases of seeds and fertilizers, delivering technical guidance, and facilitating access to markets for selling produce.
Charitable organizations and private enterprises are also actively involved in post-disaster recovery efforts. For instance, a village committee director pointed out that the Red Cross in Yunnan not only provided emergency relief supplies but also helped to rebuild schools and hospitals. COFCO Group (Zhongliang jituan), on the other hand, offered seeds and equipment to farmers in Sichuan, supporting the restoration of agricultural production. 14 Additionally, tourism companies in Yunnan have worked with local governments to promote ethnic cultural tourism projects, boosting the recovery of the local economy. Different indigenous groups maintain close ties. Indigenous groups often engage in resource exchanges, creating networks that span geographic boundaries. One respondent mentioned borrowing money from relatives in Myanmar to repair homes and start businesses after a flood. 15 Another clan leader noted that in the days following one flood, they spontaneously sought out help from Southeast Asian countries and carried back rice for emergency aid. 16
For individuals and households, the pressure of post-flood reconstruction is often greater than that exerted by the flood itself. Since most people primarily rely on agriculture for their livelihoods, they frequently make adjustments to their livelihood strategies. The largest farming household in the area reported that adaptation actions include increasing crop diversification, changing crop varieties, modifying cultivation calendars, adjusting the use of fertilizers and pesticides, and purchasing agricultural insurance. 17 The Bai people in the Dali region have adopted more flood-resistant rice varieties following flood events. Additionally, local residents tend to cultivate cash crops. Findings from participant observation indicate that in many communities across western Yunnan, crops such as tobacco, cardamom, and rapeseed are grown to supplement household income. Some smallholders reallocate part of their labor or resources to livestock farming, increasing the number of livestock as a strategy to cope better with floods. Interviews with local residents reveal that a decade ago, most households primarily engaged in farming. However, over the past ten years, an increasing number of individuals have shifted to migrant labor, which has become a major source of income. In communities rich in ethnic resources, many residents have focused on developing tourism projects and promoting ethnic handicrafts to improve their livelihoods. 18
Figure 1 provides a comprehensive overview of how community flood adaptation has begun to exhibit characteristics of mixed adaptation, as reflected in the participation of multiple actors, coverage across multiple stages, and the application of diverse resilience measures. At the micro scale, individuals and households respond through everyday practices and livelihood adjustments; at the meso scale, clan and community organizations mobilize through kinship networks and collective action; and at the macro scale, township and provincial governments intervene through policies, planning, and large-scale projects.
Interactions of responding actors, flood management stages, and resilience measures in the THR area.
Mixed adaptation with collaboration and conflicts
Classification of flood adaptation types
Building on mixed adaptation, this section proposes a typological classification of flood adaptation pathways. Rather than a traditional binary, this classification serves as an analytical entry point for understanding how and why different actors interact. Most adaptation actions can be broadly categorized into two types: planned adaptation and autonomous adaptation. This distinction is based on differences in their goals, knowledge sources, organizational forms, and spatial configurations. Other actors, such as local enterprises and charitable organizations, also play roles in the process but fall outside the scope of this classification and are therefore not examined in detail.
Government-led, centralized planning and hierarchical governance are the main characteristics of planned adaptation. Its objectives encompass life safety, agricultural production security, and social welfare, supported by legal frameworks and administrative responsibilities. Provincial governments, township governments, and village committees, as representatives of technocratic governance, achieve these goals through monitoring, planning, and structural modifications. Rooted in environmental science research, the government focuses on environmental management and planning, mobilizing “nature” as an intrinsic element of state power. By studying, planning, monitoring, and modifying natural elements such as climate, precipitation, forests, soil, and rivers, governance is effectively implemented. This pathway is primarily event driven and characterized by foresight and structured planning.
Unlike planned adaptation, autonomous adaptation often exhibits micro-level and covert characteristics. The objectives of the participants—primarily individuals, families, and informal groups such as clans—go beyond life protection and livelihood security to encompass culturally embedded needs. These adaptive behaviors are embedded in lifestyles, customary laws, and values, and they flexibly evolve with environmental changes. These actions are characterized by their long-term and ongoing nature, manifesting as daily adaptation, labor adaptation, and value adjustments. Adaptive behaviors are distributed based on geographical and kinship networks, grounded in experiential and localized knowledge, and perpetuated through intergenerational inheritance and accumulation. Table 3 provides a detailed overview of the characteristics of the two types of adaptation.
Comparison of the two types of flood adaptation.
Collaboration between planned and autonomous adaptation
Although planned and autonomous adaptation differs in terms of goals, knowledge systems, and implementation mechanisms, they frequently intersect in practice, producing both collaboration and conflicts. The case of P Village exemplifies mixed flood adaptation shaped by the interplay of multiple actors, where collaborative mechanisms coexist with underlying conflicts.
P Village is an area predominantly inhabited by the Achang ethnic group. Interviews with the head of the Land and Resources Office revealed that a landslide occurred there in the 1970s, resulting in 17 fatalities. 19 Currently, the village has been designated as a flood-prone area. The government constructed a small reservoir and flood control dikes above the community, with drainage channels distributed around the village. Over time, the integration of these infrastructures with agricultural life has led to a tradition of community-led maintenance. Rather than relying on formal administrative arrangements, the operation and upkeep of these government-built projects are sustained through informal rules embedded in religious and clan-based networks. Each household is responsible for dredging and maintaining its assigned drainage sections before the rainy season. As one Buddhist villager explained, “We must take responsibility for the cleaning and maintenance work, otherwise we will be prohibited from participating in the temple’s blessing rituals. That is something we simply cannot accept.” 20
Through participant observation, it was found that, influenced by local cultural traditions, government-designated shelters are often co-located with temples, which serve not only as religious centers but also as communal gathering spaces. As a result, villagers tend to use these temples as temporary shelters during flood events, reflecting the integration of formal planning with local cultural preferences. In addition, the government's “Ten-Household Joint Defense” (shihu lianfang) system largely overlaps with existing kinship and geographical ties. The survey reveals P Village has formed a dual-layer flood management framework: the government-appointed village–community–group (cun–she–zu) structure and the religion-based zhuang–pian–mu network. The government-appointed flood monitor is often a young leader named L, selected for his bravery as well as his influence within the clan-based society. People of different age groups take on specific responsibilities during emergencies based on their identity and roles within the local social network. 21
Conflicts between planned and autonomous adaptation
While collaboration reveals the potential for mixed adaptation, the convergence of differing governance logics often gives rise to conflicts and, in some cases, undermines existing adaptive capacities. Afforestation has been one of the most widely promoted flood adaptation strategies in the THR area. A retired teacher recalled that since the 1990s, alpine meadows in P Village that were once used for free-range grazing were reclassified as restricted conservation zones under national soil and water conservation policies. As a result, villagers were forced to abandon household-based livestock farming. 22 Native tree species were largely replaced by monoculture plantations of Chinese fir, a species valued for its rapid growth and economic returns. However, its shallow roots led to increased erosion during the rainy season and water depletion in dry periods. Combined with upstream reservoirs’ strict water retention during droughts, this strategy contributed to persistent downstream water scarcity. In 2013, villagers recalled that after days of heavy rain, a landslide above P Village sent entire rows of Chinese fir trees downhill, devastating homes and farmland. 23 This case highlights how well-intentioned, planned adaptation can generate unintended ecological and social consequences.
Due to the shortsightedness of planned adaptation, post-flood recovery interventions in P Village are often limited, and long-term reconstruction policies lack clear direction. The director of the township emergency office believed that relocation is considered the most ideal recovery solution, as it aligns with census requirements, land-use planning, cost assessments, and the restoration of order. 24 The government proposed relocating P Village to N Village, a Yi settlement built for a previous relocation project near the border. However, the villagers disagreed with this plan. As L's family pointed out, “We have different beliefs and languages.” 25 P Village, known as the “Buddha’s Garden” of the Achang people, is seen not as a risk zone but as a productive and livable environment. This perception reflects their confidence in sustaining livelihoods based on long-term interaction with the floodplain. However, the definition of a “normal community” is inherently political. In terms of community adaptation, the state positions itself as the sole legitimate authority for ecological assessment and management, often excluding local cultures and livelihoods shaped by long-term interactions with nature from its framework. Ultimately, P Village was merged with N Village.
The relocation brought numerous challenges. Firstly, although the new village was scientifically planned with more modernized housing and infrastructure, it lacked traditional Buddhist temples and did not include Christian churches for Yi villagers. A mother who lost her child in the flood said that this diminished cultural and religious spaces and hindered post-flood psychological recovery. 26 Secondly, the director of the local chronicles office pointed out that the relocation disrupted local networks based on ethnicity and kinship, making it difficult to restore the original dual flood adaptation in the short term. 27 One villager noted that the breakdown of local social networks interrupted the continuity of autonomous adaptation, including the regular organization of community-based preparedness activities that were previously sustained through collective participation. 28 Interviews with the elderly showed that flood adaptation knowledge in the village was framed in a modern technical Mandarin translation, which was largely unintelligible to indigenous residents, especially elderly community members whose knowledge systems and daily communication relied on native languages. In addition, flood-related information, such as early warnings, was disseminated through digital communication technologies. However, many residents, particularly the elderly and those without access to mobile phones or the internet, were unable to receive these alerts in time. 29
Reconstruction was delayed by a cost-sharing mechanism that shifted financial burdens onto villagers, while limited government support further undermined the prospects for long-term recovery. Villager L explained that he farms during the rainy season and works elsewhere during the dry season, with his income barely covering household expenses and his children's education. 30 Repairing his house was not a priority, and he could not afford insurance for his home or property. After the flood, large areas of land were covered with sand and stones, and post-disaster land restoration received little attention. As a result, villager M said that the once-abundant rice paddies and amomum fields saw their yields reduced by two-thirds, forcing residents to adjust their livelihoods. 31 After assessing the soil and climate, the government identified the area as suitable for tobacco cultivation and partnered with the China National Tobacco Corporation to develop the land. However, tobacco farming requires large quantities of pesticides and fertilizers, as well as high initial investment costs, which compelled most families to borrow from the company. Additionally, tobacco cultivation is inherently risky and its quality inconsistent, further increasing income uncertainty for residents. More critically, the village committee director of Village J explained that the shift from food crops to cash crops exacerbated local food insecurity, and long-standing rice-farming traditions also came under threat. 32 This recovery model ultimately reinforced the community's dependence on planned adaptation in terms of funding and technology, further limiting its autonomy in resource allocation and livelihood choices.
The coexistence of collaboration and conflict demonstrates that mixed adaptation is not a naturally occurring institutional evolution, but rather the result of continuous negotiation and contestation among diverse actors. In areas such as resource allocation, labor mobilization, and institutional arrangements, planned and autonomous adaptation show certain forms of synergy. For example, there is a linkage between official flood control networks and clan-based systems, a functional overlap between official shelters and religious spaces, and a complementarity between short-term government engineering projects and long-term community maintenance efforts. These mechanisms have, to some extent, enhanced the local embeddedness and implementation contribution of formal governance. However, currently mixed adaptation remains at a preliminary stage. In critical areas such as flood planning, post-disaster relocation, and livelihood recovery, planned adaptation continues to be implemented in a top-down, engineering-driven manner. These approaches often overlook—or even suppress—local ecological norms, cultural preferences, and indigenous knowledge systems. As a result, they undermine community identity and participation, ultimately weakening the resilience that stems from autonomous adaptation.
Discussion
Rethinking beyond the traditional binary
As climate change intensifies, developing more resilient flood adaptation strategies has become a central task on the current policy agenda (IPCC, 2022;Singh et al., 2021). This study views community-based flood adaptation as a cross-scale process involving small-scale individual interactions, interactions among formal and informal groups, and large-scale processes between local and national governance (Berrang-Ford et al., 2021; Dietz et al., 2020; Torfing et al., 2026; Young, 2002). Our aim is to explore how cross-scale interactions give rise to mixed adaptation, and to contribute to scholarly debates on its possibilities and mechanisms (Eriksen et al., 2021; Juhola et al., 2022; Watson et al., 2020).
In rural communities along the Tea-Horse Road, we observed that flood adaptation is not confined to a single level but rather constitutes a cross-scale process. Planned and autonomous adaptation intersect across different phases of flood adaptation. This interplay has integrated resources, labor, and institutions, laying a critical foundation for the development of community resilience. This finding provides empirical support for the concept of mixed adaptation in Global South contexts (Doherty et al., 2017; Lindegaard and Sen, 2022; Mersha and van Laerhoven, 2018), where individuals and households are key actors—reinforcing previous observations that adaptation in developing countries is often highly localized, gradual, and embedded in everyday life (Castro and Sen, 2022; Pörtner et al., 2022). However, unlike some studies that portray autonomous adaptation as inefficient or disorganized (Chambwera, 2010), our research highlights its important strengths in including marginalized groups, fostering self-organization, and building trust and learning (Ford et al., 2020; Pisor et al., 2022). Moreover, in contrast to many Global South countries where state capacity is limited and adaptation is predominantly led by individuals and households (McNamara and Buggy, 2017), rural communities along the Tea-Horse Road—despite their remote locations—demonstrate sustained and substantial involvement of planned adaptation efforts.
A further contribution of this study lies in its nuanced examination of the dynamic relationship between planned and autonomous adaptation. This paper reveals a form of mixed adaptation arising from the interaction between strong planned adaptation and strong autonomous adaptation, thereby contributing to the typological study of adaptation pathways. This form of mixed adaptation is not yet fully integrated but already shows initial signs of coordination in practice, which often emerges as fragmented and temporary rather than enduring coordination. Under this pattern, adaptation measures, though framed as risk-reduction strategies, may in fact exacerbate local vulnerability, marginalization, and dependency in practice—ultimately constituting forms of maladaptation (Barnett and O’Neill, 2010; Glover and Granberg, 2021; Juhola et al., 2016). For instance, state-led measures such as relocation policies and post-flood reconstruction frequently fail to align with local socio-cultural logics, and instead generate tensions with existing livelihoods, indigenous knowledge systems, and social networks. This finding is consistent with empirical studies in other Global South contexts (Kumamoto and Mills, 2012; Mosberg and Eriksen, 2015; Naess et al., 2015; Young, 2006), which emphasize that adaptation failures often stem from poor integration into pre-existing social structures. Such disjuncture is partly rooted in fundamental differences between state and community frameworks for understanding vulnerability (Malik and Smith, 2012).
This case not only reveals the limitations and blind spots of the conventional binary distinction between planned and autonomous adaptation (Juhola et al., 2022; Rahman and Hickey, 2019), but also reinforces the analytical value of recognizing mixed adaptation as a potential third pathway. By conceptualizing community-based planned and autonomous adaptation as part of a cross-scale process rather than as static and opposing pathways, we can clearly observe how state-led flood management interacts with people’s everyday practices of climate adaptation. These interactions entail both tensions and misalignments while also fostering cooperation and collaboration, thereby presenting the complex landscape of cross-scale adaptation. It is within this process of interaction that the foundations of community resilience are gradually constructed, while at the same time exposing potential risks of maladaptation.
Structural barriers to mixed adaptation
Building on the identification of the mixed adaptation pattern, this study further explores why mixed adaptation remains difficult to achieve in an integrated and constructive manner. Recent research increasingly emphasizes that adaptation is not merely a matter of technical choices or behavioral adjustments, but rather a deeply political governance practice (Eriksen and Lind, 2009; Eriksen et al., 2011; Ribot, 2014; Sovacool and Linnér, 2016). Fieldwork in the THR region reveals that the barriers to integrating mixed adaptation lie not only in the divergent governance logics between formal and informal actors, but also in deeper structural constraints.
The THR region has long been perceived by Chinese imperial authorities as a remote, unruly, and inhospitable frontier (Zhou, 2007). Since the Han dynasty (202 BCE–9 CE), it has been progressively incorporated into the imperial governance system, achieving institutional integration during the Yuan dynasty (1271–1368) (Sun, 2015), while its natural resources were continuously extracted to support the Central Plains (Miller, 2022; Schlesinger, 2017; Yang and Kim, 2018). The rise of the modern nation-state has further positioned the “state” as the dominant actor in human–nature relations (Hall, 1986). The planning and transformation of “nature” became a distinct domain of state power, serving as a foundation for legitimizing its authority and sovereignty (Duara, 1991; Pietz, 2002; Pomeranz, 1993).
Today, under the Communist Party of China’s agenda of ecological civilization building, the governance of nature remains a central means through which the state extends its territorial reach in this region and produces landscapes aligned with its vision of socio-natural optimization (Rodenbiker, 2023). Scott (1999) argues that modern states often simplify socio-natural complexity through technical interventions. Some studies argue that the emergence of various forms of “knowledge lenses”, such as statistics, survey techniques, cartography, forestry, and scientific agronomy, has provided the intellectual foundations for simplifying nature (Xun, 2015). However, Rodenbiker (2023) observes in this region that although state scientists and planners deploy techniques aimed at mechanizing nature and optimizing society and space, these interventions do not necessarily produce mechanistic outcomes. Beneath the veneer of a centrally orchestrated environmental governance effort, the processes remain highly indeterminate. This case study likewise shows that planned adaptation, while relying on technical measures to rationalize floods, unfolds in ways marked by contradictions, resistance, and creative reworkings in the interaction between state power and autonomous adaptation. This condition endows mixed adaptation with an inherent complexity.
Although the Chinese state does not simply “simplify” nature, but rather seeks to harness its complexity in order to optimize socio-ecological relations, the current form of mixed adaptation has so far been less than satisfactory. Planned adaptation cannot simply override or eliminate autonomous practices, but instead interacts with them in fragmented ways—sometimes reinforcing resilience, while at other times generating new vulnerabilities. In this case, the difficulty can be attributed to two main factors. On the one hand, indigenous groups have developed their own adaptations rooted in lived experience, often aimed at minimizing deep interventions by external governance mechanisms (Lattimore, 1951; Scott, 2010). For example, compared with irrigated agriculture, shifting cultivation was more commonly adopted by indigenous groups because it facilitated mobility and evasion; while rice may have been more productive, they often chose root crops that were less visible to the state and thus facilitated the avoidance of taxation and extraction (Ma, 2017; Scott, 1985). Over time, the institutional rationality of the state and the adaptive logics of local communities have shaped a form of relative autonomy, which David A. Bello (2005) conceptualizes as the “intermediate zone”. This historical legacy constitutes one of the key structural barriers to the full integration of planned and autonomous adaptation.
On the other hand, the path dependence of current planned adaptation further reinforces structural barriers. At present, flood adaptation is closely intertwined with regional development agendas. State-led technical measures, such as engineering projects and risk mapping, are highly valued within governance logics because they can produce visible short-term results, such as reducing flood losses and safeguarding agricultural production. Performance-based administrative evaluation systems further consolidate reliance on these technical pathways. However, this approach not only narrows the space for autonomous adaptation but also weakens the flexibility of adaptive practices, thereby exacerbating the difficulty of achieving fully integrated forms of mixed adaptation.
Enhancing the contributions of mixed adaptation
Although this study has revealed multiple challenges facing mixed adaptation in practice, these difficulties do not imply that collaborative governance between the state and communities is destined to fail. On the contrary, they indicate that mixed adaptation can be actively facilitated through mechanisms such as participatory negotiation, delegated authority, and the co-production of knowledge. This calls for a fundamental reconsideration of flood adaptation: Who makes the decisions? Who benefits? And how should the process be implemented? Our case study shows that for state-led interventions to genuinely enhance community adaptive capacity, they must go beyond the technical aspects of flood control and engage with non-technical factors that shape adaptation outcomes. They must also accommodate multiple temporal scales and diverse stakeholder perspectives. We particularly emphasize that institutional flexibility and fairness, pluralism in knowledge sources, and the expansion of local deliberative spaces are essential conditions for fostering mixed adaptation.
We offer three key recommendations. First, the criteria for evaluating “successful adaptation” should be redefined. Existing assessments of community resilience need to be expanded to include indicators such as ecological resilience, self-organizing capacity, and cultural continuity. These dimensions are critical for capturing the long-term sustainability of adaptation strategies. Second, institutional support for autonomous adaptation should be significantly strengthened. This requires eliminating biases against informal and community-based practices, recognizing their practical value in flood adaptation, and proactively incorporating the perspectives of marginalized and indigenous groups into policy design. Concrete steps may include allocating resources to document and disseminate local knowledge, facilitating mutual translation and recognition between local and official knowledge systems, and establishing dedicated funding schemes to support the continuation and enhancement of autonomous adaptation mechanisms. Such mechanisms may include the maintenance of clan networks, retrofitting of traditional housing, and sustaining culturally embedded livelihood. Finally, inclusive and institutionalized participatory mechanisms should be established. At the early stages of adaptation projects, collaborative platforms should be created that bring together local communities, researchers, and technical experts to ensure cultural compatibility and contextual relevance. In addition, a quota-based system for community participation should be implemented to guarantee that marginalized groups are meaningfully and representatively involved throughout the planning, implementation, and evaluation stages of adaptation initiatives. These efforts are essential for achieving a mixed adaptation pattern characterized by both strong planned and strong autonomous components.
This case provides empirical support for exploring mixed adaptation pathways in Global South communities. The “strong planned—strong autonomous” pattern identified in this study demonstrates that maladaptation is not always the result of governance absence. It can also emerge from the incompatibility between multiple adaptation pathways. Rather than treating mixed adaptation as a fixed institutional model, we argue that it should be understood as a processual construction. What may appear as an unintegrated or fragmented pattern is better viewed as an evolving process within adaptation dynamics. This perspective invites a rethinking of the scope and limits of state-led adaptation (Smit et al., 1999; Teebken et al., 2023). For communities in the Global South, it is especially important to recognize that meaningful adaptation must be rooted in local agency and development aspirations, rather than being instrumentalized as a component of state control or as a demonstration of policy performance. Future flood adaptation must remain alert to the risks posed by tensions between different adaptation pathways, and must pay greater attention to the roles, visions, and practices of diverse actors within mixed adaptation patterns. We encourage future research to move beyond static typologies of adaptation and instead focus on how multiple governance logics interact within spatial scales. Such an approach offers valuable comparative perspectives and policy insights for enhancing community resilience in Global South settings.
Conclusion
This study takes the multi-ethnic rural communities along the THR in Southwest China as a case and, combining fieldwork with policy analysis, examines the interactions between planned and autonomous adaptation. We observe that mixed adaptation exists along the THR, with the coexistence of collaboration and conflict being a defining feature of community adaptation. Collaboration occurs in areas such as resource allocation, labor mobilization, and institutional arrangements; however, it is only loosely integrated in its current form and is characterized by limited cooperation between community and government actors. Furthermore, current governance is still dominated by planned adaptation, which often neglects local society by inadequately considering their ecological norms, cultural preferences, and local knowledge. This unequal interaction undermines the original resilience of communities, which is characterized by self-dependence.
This paper moves beyond the binary classification of adaptation pathways and advocates for understanding adaptation as a cross-scale process. It reveals a form of mixed adaptation arising from the interaction between strong planned adaptation and strong autonomous adaptation, thereby contributing to the typological study of adaptation pathways. This form of mixed adaptation is not yet fully integrated but already shows initial signs of coordination in practice, which is often manifested as fragmented and temporary rather than enduring coordination. In addition, the paper offers a new explanation for the causes of maladaptation. We argue that maladaptation is not merely the result of limited resources or institutional absence, but is more fundamentally rooted in the exclusion of local knowledge systems, unequal power structures, and uneven access to institutional opportunities. These structural imbalances, shaped by historical inertia and path dependence, not only hinder the meaningful integration of diverse adaptation mechanisms but also risk eroding the self-dependence resilience that has historically been embedded in local communities.
The study provides empirical evidence of cross-scale interactions and uses this perspective further to conceptualize and empirically ground the concept of mixed adaptation. We argue that cross-scale dynamics represent both opportunities for synergy and challenges for governance. We recommend that, in building future mixed adaptation models, policymakers should redefine the criteria for “successful adaptation”, broaden the indicators used to assess community resilience, acknowledge the value of autonomous adaptation, and establish regular mechanisms for deliberation. Future research should further explore the dynamics of mixed adaptation in different political-ecological and socio-cultural contexts as well as spatial scales, in order to promote the development of more equitable, flexible, and sustainable frameworks for community resilience.
Footnotes
Acknowledgments
The authors sincerely thank the reviewers for their rigorous, thoughtful, and constructive comments, which helped refine the core arguments and improve the theoretical framework of this article.
Contributorship
Mei Ai was responsible for study design, data collection, manuscript writing, and visualization. Wenhan Feng contributed to study design and manuscript revision. Christine Heinzel provided proofreading and manuscript revision. Siying Chen contributed to manuscript revision. Liang Emlyn Yang contributed to study design, manuscript revision, and funding acquisition. All authors approved the final manuscript.
Ethical approval
All interview contents in this paper have been used with the explicit consent of the interviewed individuals and institutions. We have adhered to ethical standards and obtained the necessary permissions to ensure compliance with privacy and consent requirements. For reasons of confidentiality, personal identifiers and full transcripts of the interviews cannot be disclosed; only anonymized codes are used for citation.
Funding
This research is funded by the European Research Council (ERC) Starting Grant STORIES (grant no. 101040939) and the LMU-CSC Scholarship.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.