Under pressure: How human-wild-captive elephant social-ecological system in Laos is teetering due to global forces and sociocultural changes

1. Few empirical studies have described social- ecological systems (SESs) in transition. Some studies focused on external drivers that impact the SES and communities' responses to adapt to changes, including economic, land and conservation policies. Others have considered the effect of social and cultural changes on communities' capacity to sustain their activities. While sociocultural changes are increasingly common through globalization and world- wide economic development, there is an urgent need to better understand and document how these changes affect individual and community agency to adapt or transform a system that is facing a combination of powerful internal and external forces. 2. The human– Asian elephant relationship appears particularly illustrative of a complex SES because of the dual status of the elephant being wild or under human care, and the entanglement of ecological, cultural, social and economic dimensions. The ongoing and rapid political, socio- economic and environmental changes occurring in


| INTRODUC TI ON
Various conceptual approaches have been developed to understand and analyse the processes of transition or transformation in society (Patterson et al., 2017). Among them, social-ecological systems (SESs) are complex and integrated systems in which humans are part of nature (Berkes & Folke, 1998). SESs are dynamic, and changes occur across multiple scales in time, space and organizational levels (Gunderson & Holling, 2002;. In addition, people and their environment interact reciprocally and form complex feedback loops (Liu et al., 2007). In this context, there is a need to study how periods of gradual change interact with abrupt changes, and how human groups are able to adapt or even transform into new development pathways in the face of dynamic change (Folke, 2016). While adaptation refers to the capacity of social actors to learn, combine experience and knowledge and adjust social behaviour to pertain the SES fundamental structure and functions, transformation in contrast refers to a larger intentional change that creates a fundamentally different system when ecological, economic or social structures make the existing system unsustainable (Folke, 2016). Adaptation and transformation are intimately linked within SES; adaptation at one level may lead to a transformation at another level (Pelling & Manuel-Navarrete, 2011).
A growing number of empirical studies have described SES in transition. Early work on SES regime shifts focused on their ecological components (Armitage et al., 2012). Further studies identified the external drivers that impact the SES and the response of communities, including climate change, economic and political regime shifts Ramprasad et al., 2020;Zanotti et al., 2020).
However, few empirical studies have considered the effect of social and cultural changes, which are increasingly common through globalization and world-wide economic development, on the agency of individuals and communities to pertain their traditions and activities. For example, co-management initiatives, women and local communities' empowerment have been advocated to foster the internal response of communities to maintain sustainable and healthy SES (Eriksson et al., 2021;Fedele et al., 2017;Reid et al., 2014). Local social and cultural practices also support adaptive and transformative change in the face of changing internal and external forces (Apgar et al., 2015;Senos et al., 2006). However, at the individual level, actors at the centre of the system may be pulled away by the dynamics of change towards other livelihoods that are either less difficult, more culturally appropriate in a changing society or more remunerative (Ramprasad et al., 2020;White, 2015). For example, the social status of emerging livelihoods and the intergenerational erosion of interest in artisanal fishing led to an ageing population of fishers and difficulties to recruit young ones (Tam et al., 2018).
In a rapidly changing world on both the natural and social sides, there is an urgent need to better understand and document how social and cultural changes affect individual and community agency to adapt or transform a system that is facing a combination of powerful internal and external forces. In this context, we use the case of the human-captive and wild elephant relationship in Laos to shed light on the ongoing changes occurring in this particular SES under both external and internal drivers that are illustrative of a wider situation prevailing world-wide.
The human-Asian elephant relationship is akin to a SES because of the dual status of the elephant being wild or under human care, and the entanglement of ecological, cultural, symbolic, social, economic, emotional and inter-subjective dimensions. Elephants have been trained and used by humans for at least 4,000 years as war animals, mounts of kings or draught animals for transport and forestry (Trautmann, 2015). The capture of wild elephants to supplement the captive population has maintained demographic and genetic links between these two populations until the 1970s (Sukumar, 2011).
These links persist to date, particularly through the reproduction of captive females with wild males (Maurer et al., 2017). The plummeting number of Asian elephants is of global concern. From an ecological point of view, the viability of wild and captive elephant populations is threatened by habitat reduction, poaching and capture, increasing conflicts with local human communities and low birth rates (Choudhury et al., 2008). From the humanities point of view, the long-lasting relationship between humans and Asian elephants has been affected by many changes, among others the major social and economic changes occurring recently in range of countries, loss of traditional knowledge, new elephant working practice, and the raising of new actors like welfare and conservation organizations (Locke & Buckingham, 2016). 5. We discuss how the combination of external drivers, such as economic liberalization, land and conservation policies, and internal drivers linked to sociocultural changes could affect a SES in transition, leading to a fading interest of the new generation in their family heritage.

K E Y W O R D S
Asian elephant, commodification of nature, human-animal studies, Laos, shifting regimes, social and cultural values, social-ecological system, wild-captive interactions The SES approach provides a novel and relevant framework for analysing the dynamics of the human-wild-captive elephant system in Laos. Indeed, wild and captive elephants are closely interlinked within the landscape. The disciplinary roots of SES are in ecology, but the approach has been broadened by inclusion of the social sciences (Patterson et al., 2017). Therefore, the SES approach provides a framework suited to the abrupt social, ecological and economic changes that have occurred in Laos over the past few decades.
Through this study, we aim to provide an empirical description of a SES in transition under the influence of powerful external drivers, from economic liberalization to land and conservation policies. We also strive to shed light on the impact cultural and social changes have had on this relationship, leading to a fading interest of the new generation in their family heritage. We believe our case study is illustrative of a wider prevalent situation faced by SES in transition due to globalized economic, social and cultural changes.
We focus specifically on describing the changes in perceptions and practices among Lao elephant owners and mahouts who are at the heart of the human-captive-wild elephant relation using a combination of SES and anthropological approaches based on semi-structured interviews. Acknowledging the prominence of symbolic, historical and intimate dimensions of the relationships, we aim to provide a new insight, using a more systemic approach linking ecological, social, cultural and economic components of the system.
To do so, we first describe the SES that prevailed 20-30 years ago, which was based on the principles of common access to natural resources and social control over nature and spirits. Some anthropological materials collected during the study allow us to highlight the importance of values and symbolism of the species. Secondly, we enquire how, since the opening of the country to the market economy, the rise to dominance of nature commodification and the partitioning of natural resources led to major changes in the SES.
Thirdly, we analyse the impacts these changes have had on the elephant handling system using a four-dimensions model based on (a) specific modes of production and (b) taskscape in the sense of Ingold (1974Ingold ( , 2000, (c) the modes of relation between humans and animals following Stépanoff's (2017) categorization and (d) the landscape. Finally, we discuss how the combination of external drivers, such as economic liberalization, as well as land and conservation policies, and internal drivers linked to cultural and social changes has impacted this relationship, leading to a fading interest of the new generation in their family heritage.

| SUPP ORTING CON CEP TS AND THEORIE S
For more clarity, we define in this section several concepts and theories that are mobilized throughout our analysis and discussion. We first introduce some key external drivers of the SES: neoliberalism, commodification of nature and land management policies. Then we present supporting concepts, inspired from studies on pastoralism, that offer comparative categorizations to build a conceptual model of the human-elephant handling system. Harvey (2005) defined neoliberalism as the theory of politicaleconomic practices that proposes that human well-being can best be advanced by liberating individual entrepreneurial skills with an institutional framework characterized by strong property rights, free markets and free trade. Neoliberalization includes politicaleconomic practices such as expansion of market valuation to spheres that were formerly unaffected by commerce such as protected forests and ecosystem services (Gómez-Baggethun & Ruiz-Pérez, 2011;Robertson, 2006). Assigning prices and values to natural commodities and allowing their trading within a free market has been called the commodification of natural resources (Liverman, 2004;Robertson, 2006). Concomitantly, government policies promote the partitioning, regulation and even privatization of access to natural resources that had previously been managed as commons (Ingold, 1980). These landscape-dividing policies tend to control and often exclude local people, generating misunderstandings, oppositions or even conflicts with cultural values and the needs of local communities (Brockington & Igoe, 2006;Chan et al., 2007;Infield et al., 2017). Therefore, institutions, policies and information brokers (NGOs, media, researchers) can ultimately affect the relationship between humans and nature (Campbell, 2005;Cumming, 2016;Linnell et al., 2015).

| External drivers: Neoliberalism, commodification of nature and land management policies
Beginning in the late 1980s, governments in south-east Asia and international institutions undertook major political-economic changes and shifts in line with neoliberalism (Nevins & Peluso, 2018).
Changes in land-use patterns, deforestation and loss of biodiversity are among the main drivers of changes in tropical ecosystems (Fisher et al., 2011;Lambin & Meyfroidt, 2011). Since the 1980s, land conversion for agriculture and pasture has replaced or degraded tropical forests and grasslands (Gibbs et al., 2010). Recent studies have demonstrated the importance for large mammals of retaining native habitat within agricultural land (Ferreira et al., 2018;Galán-Acedo et al., 2019). Depending on the structure and configuration of the matrix (agroforestry, polyculture or monoculture), human-modified landscapes provide functional diversity for wildlife foraging, dispersal and reproduction (Ferreira et al., 2018).

| Linking social and ecological relations in a conceptual model: Production, Taskscape, TEK and landscape
Using the example of arctic reindeer herders, Ingold (1974) emphasized the relationships between the system of ecological relations linking the human population with herds and pastures, and the system of human social relations governing access to the land and to animals. The landscape arises alongside what Ingold (2000) called the taskscape, or the entire ensemble of tasks done by humans in the environment, where tasks are the constitutive acts of dwelling. The tasks include technical and social activities and their mutual interlocking. The concept of taskscape encompasses traditional ecological knowledge (TEK). TEK refers to the place-based knowledge that is rooted in local cultures and generally associated with long-settled communities that have strong ties to their natural environments (Antweiler, 1998;Orlove et al., 2010;Sillitoe, 1998).
According to Berkes (2008), TEK is the understanding of ecological relationships and systems of managing resources by communities. A resource management system includes an appropriate set of practices, tools and techniques. TEK also includes social attitudes, beliefs and conventions of behaviour.
Based on the human-reindeer system, Ingold (1974) described the conditions driving the changes from a hunting economy to pastoralism and ranching. Ingold's description of hunting, pastoralism and ranching economies refers to different modes of production.
The pastoralism mode of production is based on a common access to pasture with an exclusive right of disposal of the owner over his animals and progeny. While in pastoralism owners aim to increase their animal stock through natural reproduction, in a ranching economy owners intend to accumulate monetary capital. Recent research on pastoralism shows that the conventional relation of protective domination in which humans drive, protect and feed their livestock should indeed be characterized according to different levels of animal autonomy, leading to five different modes of relations ranging from seasonal freedom to captivity . Because local practices that constitute the taskscape impact the landscape in return, the landscape is perpetually under construction (Ingold, 2000).

| Study area and species
Laos is a mountainous tropical country located in the Indochinese peninsula. Of its total population, 83% live in rural areas at low densities (Ducourtieux et al., 2005).
Since the revolution in 1975, Lao People's Democratic Republic, or Laos, is ruled under the Lao People's Revolutionary Party. With the end of Soviet aid after 1990, the political framework changed radically with the opening of the country to the market economy and economic liberalization. Within a few years, the regime turned to international assistance, undertook structural reforms of its economy and institutions, particularly financial institutions, and gradually opened up to foreign capital and tourists (Evrard & Pholséma, 1997).
Land reform, and more precisely the individualization of land rights and the systematic zoning and mapping of forests, became a precondition for agricultural intensification (Ducourtieux et al., 2005).
Concurrently, forest cover declined from 70% of land cover in the 1940s to 41.5% in 2005, with an acceleration since the 1990s (Robichaud et al., 2009). The main direct drivers of deforestation are the expansion of agriculture-both by individual farmers and by large agribusinesses-and the development of industrial tree plantations and large hydropower, mining and infrastructure projects (Lestrelin, 2010;Ramcilovic-Suominen & Kotilainen, 2020). Shifting cultivation and selective logging are blamed by the government and International NGO for degrading forests, but many scholars dispute this claim (Castella et al., 2013;Lestrelin, 2010;Ramcilovic-Suominen & Kotilainen, 2020). Access to land and natural resources by foreign investors, referred to as 'land grabbing', and neoliberal and extraterritorial modes of governance have driven agrarian transformation and impacted community livelihoods (Baird & Barney, 2017;Schoenberger et al., 2017). The development of agro-plantations, the reduction of shifting cultivation and the privatization of land have affected the dwellings of both wild and captive elephants.
Laos, historically known as Lan Xang, the 'Land of a million elephants', shelters both wild and captive elephants with high levels of interactions between the two populations (Maurer et al., 2017). In Lao language, captive elephants (Xang ban) are literally called 'village elephants' where 'ban' means village or home. In this paper, we choose to adopt emic interpretations that reflect the cognitive and linguistic categories of the natives, as opposed to etic interpretations that are developed by the researcher for purposes of analysis (Harris, 1976). Therefore, we adopt the Lao terminology of 'village elephants' instead of 'domestic elephants' or 'captive elephants' that is used in biology. Wild elephants are called 'Xang pa' where 'pa' means forest. Wild and forest being the same word in Lao language, we use the terminology of wild elephants in this article. In Laos, the interpretation of wild and forest is historically illustrated by the opposition between the muang (district or city) and the pa (forest), a dialectical contrast rather than a strict dichotomy between domestic, civilized settlements and the wild, untamed forests (Singh, 2009).
The pa is a place of backwardness that lacks social order, but is also a source of desired natural resources. The muang-pa dialectic interlinks natural and social worlds emphasizing the social process of the domestication of the wild pa into the civilized muang. The perception of the forest, and of wild and village elephants, has shaped the relationships between the species, their owners and their caretakers, who are called 'mahouts'.
Up until the end of the 20th century, Laos sheltered large populations of wild and village elephants sharing the same habitat, with 4,000 and 850 individuals respectively (Santiapillai, 1993). In 2010, the wild population numbers from 600 to 800 individuals, living mostly in isolated and fragmented forest areas (Khounboline, 2011).
Village elephants are privately owned in Laos. Most village elephants belong to owners from the Sayaboury province (Maurer, 2018).
Ownership is shared by 3.6 people on average, usually from the same family (Suter et al., 2013). Time-sharing is the most common arrangement among co-owners. Both working duties and profits are shared according to their agreement (Maurer, 2018). During the 1990s, at the time of the country's opening, families often handled a herd of several (up to 10) elephants (Maurer, 2018). Owners were in charge of caring and working with their elephant. In such cases, the owner was also a mahout. A minority of owners employed mahouts and did not perform husbandry tasks. For clarity, we use the word 'mahout' for employees who do not own the elephant.

| Data collection
The study is based on long-term work (15 years) of the first author with elephant owners and mahouts' communities completed in 2015 by 5 weeks of ethnographic survey focused on semi-structured interviews to explore (a) owners' and mahouts' knowledge and perception of the species, (b) handling practices, (c) interactions between wild and captive elephants and (4) the future of mahoutship and elephants in the country (see interview guide in Table S1). Additional questions and topics were addressed in the course of the discussion of specific points of interests or because of the respondent's willingness to talk about a particular subject. Questions were adapted, and some topics were not addressed for respondents who did not own an elephant, such as officials, monks, retired owners or training masters. Some questions or subtopics were not addressed with some respondents because of lack of time or misunderstanding, or reluctance of the respondent to answer. This method was chosen to allow respondents to provide a broad range of answers on complex issues such as their perceptions and practices, with as little influence as possible from the investigator (Huntington, 2000).
Interviews typically lasted for 30 min or more and were usually conducted in the field, next to the elephant's resting place. Sampling was done using a stratified method based firstly on the district of residence, as it appears from the first author's experience that handling practices may show spatial variation. The first author conducted a total of 52 interviews. In the Sayaboury province, he interviewed 13 respondents from Thongmixay district, 11 from Paklay and Phieng, 11 from Hongsa and nine from Ngeun district ( Figure 1).
He also conducted six interviews in Pathoumphone district in the southern province of Champassak. This district hosts 20 elephants, but sheltered more than a hundred village elephants at the end of the 20th century. Secondly, sampling was stratified according to the respondent's status to capture potential variation in their answers.
We prioritized elephant owners over mahouts, as they are predominant in making strategic decisions regarding the management system. The first author interviewed elephant owners (40), mahouts (eight), officials (two) from the Ministry of Agriculture and Forestry at national level and from Sayaboury province and one monk living in a village known for its elephant-keeping tradition. Because elephant handling is mostly a male activity, we had only one female respondent who co-owns an elephant with her brothers. Respondents were of various ethnic origins (Taï Lao, Taï Dam, Taï Leu, Khamou and Hmong). Interviews were conducted in Lao language by the first author, recorded on a Zoom H1 handy recorder, and fully translated into French and English.
In the absence of an ethics committee incharge of anthropological studies at Montpellier University at the time of the study, all methods were carried out in accordance with the legal requirements and guidelines for anthropological surveys from the institutional protocols for research of the National Centre for Scientific Research (CNRS) (Baude et al., 2006). All informants provided informed verbal consent before participating in the interview. Only respondent first name was asked. Answers were anonymized during analyses.

| Principles guiding the SES in the 1990s: Common access to natural resources and social control over spirits
Human-elephant relations in the 1990s describe a SES based on two main principles-a common access to natural resources and the social control over nature and spirits. The period of the 1990s, at the time of the country's opening, was expressed by the interviewer with phrases such as: 'in the nineties', 'some 20-30 years ago' or 'when your father and grandfather carried out this activity', depending on the age of the respondent.
Elephants were formerly part of the family heritage (Xang moladok khopkwa). They were released most of the time and lived in small herds from owners of the same family or from the village. Village elephants were not feed with cultivated fodder but foraged in the forest or bush surrounding the village. Mahouts visited them once a month. Elephants were either completely free (poy), or partially restricted with hobbles (catyae) on the front legs or a chain (sow) that the elephant dragged. The choice of the mode of restraint depended on the animal's behaviour and on the distance to crops. Males and 'naughty' elephants were usually restrained with hobbles to avoid their escape, attacks on humans or crop raiding. Such an extensive mode of handling was possible because large forest areas and fallow lands were available where elephants could browse freely and sustain themselves. Elephants were used for occasional domestic work and raised at little or no cost thanks to their free-ranging abilities and free reproduction with their village conspecifics, or through crossmating with wild males.
The close relationship between elephants and humans is very much alive in some areas of Laos, such as Sayaboury province, and the perception of the species by mahouts and elephant owners is rooted in a long history and tradition of coexistence between humans and elephants.
Lao people distinguish wild (forest) elephants (Xang pa) from their captive (village) conspecifics (Xang ban), but 71% of respondents considered that village and wild elephants were the same species (sanît: kind, sort or species when related to animals). Interestingly, the further away respondents live from areas where wild elephants currently occur, the more they considered village and wild elephants as different species. When we asked what these differences were, we noticed 24 answers referring to physical attributes with no clear consensus, such as the head and/or the ear being either bigger or smaller. Otherwise, 20 answers were related to coexistence with hu- This popular cosmogony illustrates the dualist view between the civilized world, where humans were sent with their domestic companions, and the forest or wilderness that shelters afterwards some fugitives. More specifically, these stories illustrate the paths between the village (ban) and the wild or the forest (pa). Elephant owners do not conceive of either 'village' or 'wild' as an immutable status. Among 24 answers, 13 indicated that wild elephants can become village elephants after capture and training, with the exception of big or aggressive elephants that cannot be easily captured and tamed. Seven respondents also added that a wild elephant needed to be released from its wild spirits to become a village one.
Actually, an individual elephant's status is not inherited from its progenitors. Indeed, even if the owner of a female elephant can claim ownership of its progeny, the calf is considered a village elephant only from the moment it is trained and lives with humans, whether its father is a wild or a village elephant. Finally, the taming of a village-born elephant appears to be a mix of breaking-in techniques and rituals aimed at clearing the elephant from its wild spirits: A calf that hasn't been taught yet doesn't belong to After its taming, the village calf seeks the protection of the spirits of the house (phi heuan) such as the owner's family (Lainé, 2018a).
The domestic status is transient and the path to domestication is reversible. Indeed, 11 respondents confirmed known cases of village elephants returning back to the wild. Decades ago, village elephants were released in the forest over long periods of time and could meet with wild ones. It was not uncommon for them to escape and follow their wild congeners, especially gestating females. The latest known case of a village elephant never retrieved was 5 years ago. For eight respondents, such escaped village elephants become fully wild and forget the language of humans, whereas six respondents considered that they still remember the language. Two respondents also noticed that their wild spirits came back. The ownership that is transmitted from the mother to its calf is therefore only prerequisite for an elephant to be considered as a village one. The status of forest or village elephant is eventually a question of perpetuating the control over spirits. Elephant masters are engaged in a perpetual attempt to harness and control the wild nature of their elephants, which illustrates the dialectic between the wild and the domestic, between untamed natural resources and village prosperity described by Singh (2012).
According to our analysis, by the end of the 20th century, the interactions between elephant owners, wild and village elephants were based on the principles of cohabitation, habitat and resources being shared, with an attempt by the community to control the resources ( Figure 2).

| A new state of the SES based on commodification and divided access to natural resources
The two last decades have seen major changes in these ecological and socio-economic factors.
Since the 1990s and the opening of the country to the market economy, Laos has been engaged in a major social, economic and The process of commodification has also strongly impacted breeding practices. The long gestation of females and the long period before calves are weaned are not compatible with an intensive workload. Therefore, owners of female elephants have to make an economic choice between selling their calf after weaning or earning an alternative income from logging during the 3-year gestation and weaning period (Maurer et al., 2017). During the last two decades,

| Impact of SES changes on the elephant handling system
We describe the elephant handling system using four interconnected dimensions: the mode of production and the taskscape, in the sense of Ingold (1980), the mode of relations between the mahouts and their animals, in the sense of Stépanoff (2017), and the landscape (Table 2).

| Mode of production: From pastoralism to ranching-like economy
During the 1990s, the economy of elephant handling could be seen as a form of pastoralism, where pastoralism is defined as a mode of production. Ingold (1980, p. 237) listed prerequisites for a pastoralism economy, namely: 1. Animals are convertible into basic raw materials, 2. capable of self-reproduction, 3. identifiable with forbears and progeny, 4. of relatively low reproductive potential, 5. tolerant of crowding and 6. capable of supporting itself in nature. All of these criteria were satisfied in the Lao elephant handling system, with the first condition being adapted to the specificity of village elephants, where animals are convertible into basic domestic services as draught animals. According to Ingold, the pastoralism mode of production is based on a common access to pasture with an exclusive right of disposal of the animals by the owner. The system observed in Laos during the 1990s conformed to such a mode of production, being closer to agropastoralism as elephant owners also cultivated rice. Families were engaged in a spiral of accumulation (Ingold, 1980) through TA B L E 2 Conceptual model of recent changes in the elephant handling system. The elephant handling system is characterized using four interconnected dimensions: the mode of production, the taskscape (in the sense of Ingold, 1980), the mode of relations between the mahouts and their animals (in the sense of Stépanoff, 2017)

Functional specialization of landscape = Land sparing
The defining principles of each dimension are indicated in bold for the period before 1990 and nowadays. is no longer raised by the owner or one of his sons, but by an employee. Mahouts employed in the tourism sector are mostly young and inexperienced (Suter et al., 2013). They lack the traditional knowledge to care for, and control their animal. Vanitha et al. (2009) showed that elephants handled by non-traditional mahouts in South India cause higher rates of casualties and deaths among the public (or of their own mahout). Knowledge of traditional medicine and elephants' reproductive behaviour are also replaced by veterinary skills, especially in tourism camps (Dubost et al., 2019;Lainé, 2017b;Suter, 2014). Conflict management with wild elephants is delegated to forest department officials. Similar to other Asian countries, traditional knowledge and the art of mahoutry are dying out rapidly (Lair, 2002;Vanitha et al., 2009). The taskscape is thus undergoing a recent process of functional specialization of the socio-economic actors.

| Mode of relations: From seasonal freedom to captivity
Pastoralism is usually associated with a relationship of domination of humans over animals. In view of the relative autonomy of village elephants, it is questionable that this condition applied here. Indeed, alternative forms of herd management based on animal autonomy and an alternance of separation and control periods have been described for reindeer herding in northern Asia , and these systems share similarities with elephant handling in Laos.
The mode of relations defined by Stépanoff that best corresponds to the case of village elephants in the 1990s was an alternance of seasonal freedom and watching modes.
The privatization of forests and agricultural lands has impacted the way of life of mahouts and village elephants alike. At the very least, this situation compels village elephant owners to adopt a more intensive form of husbandry, as observed for other animals in Mongolia (Marchina, 2017) and to tether their elephants. Mahouts keep their elephants tied to a tree to prevent them from destroying neighbouring crops. Every day, mahouts have to move their elephant to a new location and let them drink. In the absence of work and income, this task is carried out at a loss, motivating many owners to sell their elephants.
Before the region was developed, there were no plantations, no market gardens, and elephants could be where reproduction and feeding are under exclusive human control. But permanent guarding and fodder supplementation induce high costs that are hardly bearable during the low tourism season or in the event of crises that affect tourist frequentation (Duffy & Moore, 2010). Ultimately, reproduction is controlled through the payment of stud fees. Ingold (1980) also mentioned that a ranching economy entails a strict control over reproduction and the selection of specific traits while pastoralists favour the animal's wild qualities.
Indeed, males and aggressive elephants are highly disliked in tourism, contrary to their suitedness for logging.

| Landscape: From a multifunctional landscape to functional specialization
In the 1990s, the landscape was a combination of forest and fallow lands from rotational agriculture where village elephants were released. Secondary forests and fallow lands also constituted an important part of the habitat of wild elephants, especially since the ranges of individual wild elephants extend well beyond protected areas (Evans et al., 2018;Fernando & Leimgruber, 2011).
Landscape was characterized by its multifunctionality, being the habitat for both wild and village elephants but also a resource for humans' livelihood through the collection of forest products, hunting or capture.
Since the late 1980s, agricultural intensification and land allocation policies have gradually led to the disappearance of secondary forests and fallow lands of shifting cultivation (Ducourtieux et al., 2005). Following the process of division of natural resources described above, elephant owners have been dispossessed from accessing forest resources in common lands, reinforced by the ban to access protected areas imposed by the government and NGOs, who want to keep this habitat exclusively for wildlife conservation.
It is forbidden to tie up elephants from Nakhanyang to Muang Thongmixay. Because, if people were allowed to enter the park, then they would do something else, like hunting. (#28, official, Paklay) Many village elephants have also been moved to semi-urban areas accessible for tourists, far from their natural habitat and their wild congeners. This situation impedes the interactions between wild and village elephants, notably mating opportunities. Similar to taskscape, the landscape is undergoing a recent process of functional specialization.
Such land segregation is akin to the concept of land sparing, which consists in sparing land surface with conservation priorities through the intensification of high-yield crop farming (Perfecto & Vandermeer, 2008;Phalan et al., 2011).

| D ISCUSS I ON
The relationship between humans and elephants has lasted for millennia based on a tight interlink between wild and village elephants (Trautmann, 2015). The combination of ecological, cultural and socio-economic dimensions makes the system even more complex.
The rapid political, socio-economic and environmental changes oc- External factors such as market forces as well as land and conservation policies became main drivers of the system. But our study also pointed out internal drivers affecting the capacity of communities to adapt to changes, such as the eroding willingness of the new generation to carry on with the harsh living conditions of their fathers and the consecutive loss of ecological knowledge.

| Characterizing a SES in transition
Despite a large body of literature theorizing SES in transition, it has been challenging to apply these paradigms to empirical cases because of the intricacy of the system's components and the multiscalar nature of changes (Pelling & Manuel-Navarrete, 2011).
Adaptation at one level may lead to a transformation at another level of the system (Ramprasad et al., 2020). Defining the system boundaries and its normative dimensions allows to overcome the tensions whether a social-ecological change is a transformation or adaptation (Zanotti et al., 2020). Our conceptual model aims at defining the sys- have become major drivers of population dynamics through their impact on fecundity rates. While fecundity is expected to rise again gradually in response to the rapid increase in the price of elephants, long-term equilibrium remains threatened by demand fluctuations, financial crises or economic stochasticity in the short term (Maurer et al., 2017). Lair (2019) (Gunderson, 2010). The growing specialization of the elephant economy makes the community more vulnerable and more dependent on fluctuating markets and external shocks.

| Land and conservation policies influence community response to changes
The process of landscape functional specialization has reduced the diversity of habitats and thereby the adaptability of the SES.
Elephants favour a habitat with an intermediate disturbance regime, these habitats being mainly outside protected areas (Evans et al., 2018;Fernando & Leimgruber, 2011 (Jax et al., 2013;Lainé, 2018b). But the ability of communities to organize themselves and retain their ecological knowledge is fundamental to their resilience and fosters their adaptive response to changes (Barnes et al., 2017).

| Mahout/owner' agency to pertain their activities under new conditions
The number of owners and mahouts decreases nationwide, as well as the number of elephants. The situation in which there were many owners/mahouts, usually from the same family, owning several elephants, is being increasingly replaced by one owner for one or several elephants, each elephant being cared for by one full-time mahout. The functional diversity-the role and functions that species or social actors play in the system-is reduced. Some owners would prefer to maintain their former practices but face new socioecological conditions that make this difficult. Many owners are selling or renting their elephants abroad, especially in China. If most owners declared they like their job and way of life, they also prefer their sons to avoid enduring such harsh living conditions in the forest under the rain. Their descendants also value working on a computer in an office rather than in the forest.
As a result, the decline in father-to-son transmission undermines traditional ecological knowledge. The loss of know-how and the absence of technical innovations (Ingold, 1980) threaten the long-term sustainability of this new intensive and specialized mode of husbandry. Traditional medicine is replaced by veterinary skills (Dubost et al., 2019;Lainé, 2015). Veterinarians are also supervising reproduction, especially in tourism camps, amid plummeting rates in fecundity. We illustrate how elephant husbandry in Laos has become more contingent on specialized actors such as tourism companies, veterinarians, government officials and NGOs. We can wonder if this functional specialization of socio-economic actors could reduce the system's resilience in Laos, as shown in other contexts Grêt-Regamey et al., 2019).
Overall, the decline in mahout and owner numbers can be explained by a combination of factors such as the decrease in elephant population, new handling practices and growing economic and regulations constraints pushing them to leave their activity. We also shed light on additional internal drivers more connected to change in social and cultural values that found echoes in a cultural and economic globalized world. For instance, similar processes have been described for the SES of artisanal fisheries in Norway (Johnsen & Vik, 2013) or Chile (Tam et al., 2018). Attitudes and behaviours among fishers appear to have shifted with economic development, favouring education, professions of higher status and greater economic security (Tam et al., 2018). The lack of recruitment and loss of interest from the young generation is threatening the sustainability of the system from the inside.
There is an urgent need to promote collaborative governance, community-based management, knowledge integration among multiple stakeholders and address the young generation desires and expectations to allow elephant owner and mahout communities to adapt to both gradual change and unforeseen disasters.

| CON CLUS ION
Simple dichotomies between nature and society, local and global, and economy and ecology prevent us from handling the complexity of interrelations existing 'in-between' and favours the processes of 'purification' (Latour, 1991). In Laos, the social control over nature is replaced by a clear demarcation and purification process between the village and the forest poles. Thus deforestation can be regarded by Lao people as a sign of social improvement, since civilized and prosperous (chaleun) areas are expanding (Singh, 2009). However, the long-term survival of wild elephants may be compromised by repeated poaching of wild elephants and encroachment of their habitat, genetic isolation, increasing hostility towards the species and the alarming rise of conflicts between humans and elephants.
The ontological dichotomy between the wild and village animals is getting stronger. Wild individuals are perceived more and more as different species, living in different habitats.
Meanwhile, a broad range of stakeholders-mahouts and elephant owners, the Lao society, tourism companies, welfare organizations, conservation NGOs, governmental and international institutions-are promoting their own agendas, values and knowledge without proper dialogue and coordination.
Forbes (2009)  While facing new challenges, the long-lasting relationship between humans and Asian elephants will either adapt or transform into new pathways. But the shape of this future coexistence will depend above all on the relationships that will be woven between people, places and elephants.

ACK N OWLED G EM ENTS
This work was made possible through the financial support of

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R S ' CO NTR I B UTI O N S
G.M., O.G. and N.L. conceived the ideas and designed methodology; G.M. collected and analysed the data; G.M. and N.L. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

DATA AVA I L A B I L I T Y S TAT E M E N T
Informed consent given by respondents does not allow for the interviews to be made public.