INTRODUCTION

The increased prevalence of human–wildlife conflict around the globe (Dickman 2010) is highly relevant to both wildlife conservation and social justice. Human–elephant conflict (HEC) is a major threat to the conservation of elephants as well as to the people that share the landscape with them (Wilson et al. 2015, Munyao et al. 2020). Across their ranges in both Africa and Asia, elephants require substantial amounts of resources to survive, including from areas outside protected areas (Douglas-Hamilton et al. 2005, Kshettry et al. 2020). Thus, the survival of these species is contingent on safe places outside protected areas where the economic and socio-cultural conditions are conducive to co-existence with people (Rangarajan et al. 2010, Okello et al. 2015).

Unfortunately, human–elephant interactions in many parts of Africa and Asia appear to generally be growing more negative, with factors ranging from elephant habitat loss to human (and, at a local level, sometimes elephant) population growth leading to more direct competition over resources (Shaffer et al. 2019). More conflict has increasingly meant loss of property, livelihood, and mental and physical well-being of local people (Barua et al. 2013). Human–elephant conflict also leads to the loss of both human and elephant lives (Pinter-Wollman 2012, Wilson et al. 2015). This is perhaps starkest in India: between 2018–2020, it is reported that 1,082 people have lost their lives to HEC nationwide (compared with 31 people over a period of 27 mo in Mozambique; Dunham et al. 2010, Ministry of Environment, Forest and Climate Change (MoEFCC) 2021). Retaliatory killings by poisoning and electrocution (Gureja et al. 2007, Kalam et al. 2018) pose a major threat to India’s elephants, killing 214 elephants in the same time period (MoEFCC 2021).

Several methods have been used across the world to mitigate HEC, such as acoustic deterrents, culling, translocation, physical barriers, and psychological barriers (Shaffer et al. 2019, Nath et al. 2009, Chelliah et al. 2010). A growing number of efforts involve the use of non-lethal solar-powered electric fences, which deter elephants from entering farms and villages and are seen to be more cost effective than sturdier barriers (Kioko et al. 2008, Sapkota et al. 2014). Non-lethal electric fences act as a deterrent for elephants by giving a high voltage (>5,500 volts), pulsed, non-lethal shock when touched. As a determined elephant can often charge through the fence without lasting pain, these fences are more a psychological barrier for elephants than a physical one (Desai and Riddle 2015), and their effectiveness relies on their regular maintenance so that the fence regularly delivers an effective shock (Mumby and Plotnik 2018). Fences are generally built using the same materials to a similar standard, varying in shape—either enclosing a village or running along a forest boundary. Local residents suggest that a well-maintained fence can be 80–95% effective in deterring elephants (K. Goala, personal communication, 2020). Yet despite the requirement that low-cost fences be well maintained to be effective, many if not most fences are known to be poorly maintained (Neupane et al. 2018; District Forest Officer (DFO) Konwar, Assam Forest Department, personal communication, 2019).

This leads to our central question: given their relative effectiveness, why are some of the non-lethal electric fences used to reduce HEC well maintained while others are not? We address this question by treating it as a public goods puzzle. Due to the high fixed costs involved in purchasing the technical components (ca. $1,500 USD), each fence is typically set up with the help of non-governmental organizations (NGOs) or government bodies and designed to protect a village rather than individual households (Assam Hathi Project 2008). Fence maintenance thus fits the description of a “public good” where the good is characterized by (a) non-subtractability, i.e., an individual deriving utility from the good does not impact the availability of the good for another individual, and (b) difficulty of excluding individuals that do not contribute to the provision of the good (McGinnis 2011). In principle, this public goods problem can be addressed through some combination of direct government provision (i.e., the government maintains the fence), civil society participation (e.g., an NGO ensures maintenance), and collective action by the community. In this study, we seek to understand empirically what leads to successful fence maintenance in the context of diverse, low-income, rural communities.

Other studies have explored factors that shape the successful provision of public goods, finding success is more likely in contexts where the provision of the public good is matched optimally with users (Sarker and Itoh 2001); users pay for the public good (Ponnusamy et al. 2016, Chai and Schoon 2016); and users are able to coordinate their action, either because of strong leadership or institutions (Jack and Recalde 2015) or characteristics such as small numbers or ethnic homogeneity (Miguel and Gugerty 2005). As such, we test the a priori hypothesis that successful fence maintenance requires the following three conditions be met:

1. the fence affects only the movement and activities of those who benefit from the fence, i.e., fences are geographically compatible with the user group (Feiock et al. 2009);
2. the community contributes cash toward the fence, causing the community to be invested in the success of the fence (Sawada et al. 2013); and
3. the community is more able to coordinate fence maintenance (including the enforcement of sanctions) successfully, either because:
   • the community is ethnically homogenous (Habyarimana et al. 2007); and/or
   • the community has proactive leadership (Agarwal 2001).

These hypotheses were tested using both (a) qualitative comparative analysis (Schneider and Wagemann 2012) and (b) a combination of the social-ecological systems framework (SES; Ostrom 2009a, Hinkel et al. 2015) and ethnographic methods (participatory observation and semi-structured interviews). In addition, the latter methods were used to identify other factors that might help explain fence maintenance. Through this, we endeavor to find conceptually robust and practical solutions that could broadly help ensure successful provision of public goods that reduce human–wildlife conflict.

METHODS

Study Area

India is home to a majority of the world’s population of wild Asian elephants, Elephas maximus, with a population of close to 27,000 (MoEFCC 2017). In the northeast Indian state of Assam, large-scale deforestation and loss of habitat appear to be forcing many of the state’s approximately 5,000 elephants to enter human-dominated spaces looking for food and shelter (Choudhury 2004, Kushwaha and Hazarika 2004). Sonitpur district, Assam, lost 403 km² of forests (or >35% of its forest cover) between 1994 and 2019 (Srivastava et al. 2002, World Resources Institute 2020) and is considered an epicenter of HEC. Most villagers in this area engage in sali rice paddy cultivation, which takes place between the months of June and December. Rice is generally the only grain crop that is cultivated in the year, is crucial to the food security of households, and has a strong cultural significance. The presence of elephants increases several-fold during the sali harvest months of October to December, with large herds moving through the landscape (Talukdar and Barman 2003). As a result, a spike can be seen in the levels of HEC in this time period (Zimmermann et al. 2009). A majority of the villages use conventional local methods to mitigate HEC, including the use of fire, firecrackers, shouting, catapults, and occasionally spears, bow and arrows, and guns.

Faced with high levels of HEC, several villages in this landscape, with external support from NGOs and government bodies, have established low-cost, non-lethal electric fences as an HEC mitigation strategy over the last 14 yrs. To date, several hundred kilometers of fences have been constructed in the region (DFO Konwar, Assam Forest Department, personal communication). These fences have been established with the help of WWF-India and other NGOs to differing degrees, with these organizations providing some combination of funds to cover part of the capital costs; technical knowledge; and, in a few cases, a stipend to a member of the local community to cover maintenance and operational costs. The arrangements for each fence were made in an ad hoc manner, based on local conditions and the intuition of the implementers. Fence lengths range from 0.6 km to 20 km; some enclose the area of interest completely, whereas others run along a boundary between the village and forests or protected areas. About two-thirds of the fences that were at least 2 yrs old were noted to have failed (DFO Konwar, Assam Forest Department, personal communication, 2019)—that is, they did not carry a voltage high enough (>5,500 V) to deter elephants. The other one-third still had >5,500 V, primarily because they were well maintained.

WWF-India participated in the establishment of fences in 42 villages across Nagaon, Biswanath, and Sonitpur districts of Assam. Due to time constraints, we couldn’t conduct our study across this entire population. To home in on a sufficiently diverse sample, we conducted exploratory visits to each village to collect preliminary data from November–December 2019. These comprised (i) preliminary information on the status of the villages with respect to the hypothesized determinants of maintenance and (ii) other drivers that we came to believe might contribute to fence maintenance outcomes. We inductively coded this preliminary data, where we identified potential variables to guide our selection of sample villages (Galvin et al. 2018; see Append. 1). Ultimately, we chose 19 villages based on logistic feasibility while ensuring variation in the following characteristics:

From our a priori hypotheses:

1. Fence shape;
2. Model of contribution toward fence by users;
3. Ethnic diversity of community;
4. Apparent presence/absence of proactive leader.

From the exploratory phase of data collection:

1. Pattern of raiding by elephants;
2. Technical capacity for fence maintenance within community;
3. Presence of political elite;
4. Presence of active committee for fence maintenance;
5. Presence of Forest Department for fence maintenance.

To understand long-term maintenance, villages were only included if fences had been established at least 2 yrs prior to fieldwork, as fences are often well maintained for the first few months post-establishment before falling into disarray (Desai and Riddle 2015; Assam Forest Department, personal communication, 2019).

The selected villages were studied over a period of 7 mos from November 2019 to June 2020 (Fig. 1). They comprised fences in tea estates, agricultural fields, industrial areas, forest-fringe interfaces, protected areas, and degraded forests. A majority of the residents were farmers, tea plantation workers, daily-wage workers, or workers in nearby industries, with most residing in stand-alone houses often with homestead gardens.

Data Collection

The study involved (i) measurements of the dependent variable (fence maintenance) and (ii) identification and assessment of the independent variables (factors that may determine fence maintenance). The latter involved both an examination of the a priori hypothesis and a grounded theory-based approach to understanding other variables of interest.

Measuring fence maintenance

To assess whether each village’s fence was well maintained, we evaluated fence maintenance through both direct assessments of the functionality of fences as well as assessments of the quality of fence maintenance. The latter “indirect metrics” were used to complement the former “direct metrics” as only a limited number of visits to each fence (mean = 3.3 times, range 1–7 times) were feasible over the sampling period; we did not want unrepresentative direct measurements to disproportionately drive our fence assessments. The direct metrics were summarized as a “tech score” for each fence: fences received one point if they had an average voltage over 5,500 volts, and one-third of a point each for a well-functioning solar panel, battery, and energizer. Indirect metrics were measured using a “human maintenance score,” which comprised three factors thought to influence fence functionality: (i) trimmed undergrowth to prevent the leakage of electric current, (ii) position and sturdiness of posts, and (iii) position of the insulators. We calculated the proportion of sampled units that were well maintained for each of these three factors and averaged the three proportions to yield a score. Other relevant contextual factors were also noted for each fence. The tech scores and human maintenance scores were used to classify each fence as “well maintained” or “poorly maintained” using independent blind and non-blind assessment by authors HK and DS, who worked with the communities to implement the fences. In the blind test, HK and DS were asked to assess whether a village had a well-maintained fence based on the empirical data from that fence (without disclosing the village name); in the non-blind test, they stated whether they believed a named village maintained its fence properly based on their experience with that village (cf. Append. 2 for the detailed methods).

Factors affecting maintenance: testing the a priori hypothesis

We defined the main variables in our a priori hypotheses as follows:

1. Geographically effective design (GEO): the fence was situated such that it included only communities that desired the fence and did not lead to a major hindrance to the livelihood of those communities (e.g., fence did not block access to firewood);
2. Cash buy-in (CASH): whether the community had contributed cash toward the upkeep of the fence, suggesting investment in the success or failure of the fence;
3. Ease of collective action: communities were able to act collectively to maintain the fence based on one or both of the following two factors:
   • Ethnic homogeneity (HOMO): homogenous communities (>90% of one ethnicity) might find it easier to cooperate and hence undertake collective action due to lower transaction costs, greater trust, and shared social norms;
   • Proactive leadership (LEAD): communities with a leader who proactively championed fence maintenance facilitated maintenance. (Heuristic: when community members unanimously or nearly unanimously named the same person as associated with and actively involved in fence maintenance.)

From November 2019–January 2020, in each of the 19 study villages, we conducted participant observations, unstructured interviews, and informal conversations, and we examined documentary materials where available (see Append. 3 for the detailed methods) to assess each of the study villages on all the variables of interest.

Using the assessments of whether each fence had a well-maintained fence (above) and whether they possessed or lacked each of the a priori hypothesis factors, we then ran a qualitative comparative analysis, or QCA (Schneider and Wagemann 2012), to examine our a priori hypothesis, namely:

GEO * CASH * (HOMO + LEAD) → Fence Maintenance

In the above Boolean construct, “*” refers to “and” and “+” refers to “or”; i.e., that geographic compatibility, cash buy-in, and collective action (enabled by either ethnic homogeneity, proactive leadership, or both) were necessary and sufficient conditions for fence maintenance. Qualitative comparative analysis is a set theoretic method that uses Boolean algebra to identify the simplest combination of provided factors that explains the results for the outcome of interest (Schneider and Wagemann 2012). This approach is particularly useful in such complex systems in which multiple pathways might lead to the same outcome (Berg-Schlosser et al. 2009).

Factors affecting maintenance: identifying independent variables via grounded theory

From November 2019–January 2020, we also used a grounded theory (Glaser and Strauss 1967) approach to develop a model describing what explains variation in fence maintenance. Our methods were primarily ethnographic and included qualitative interviews and participant observations of behaviors such as fence guarding and maintenance (320 h). Given the social and ecological nature of the study system, Ostrom’s SES framework was used to guide our identification of potential variables (Ostrom 2009, Hinkel et al. 2015). Authors involved in fence implementation did not participate in this field work so as not to bias the responses of those interviewed. Additionally, footage from camera traps (1,055 trap-nights) placed along fences to gauge elephants’ interactions with fences incidentally provided insights on people’s interactions with the fence (Bernard 2006, Newing et al. 2010; cf. Append. 3 for the sample sizes).

From January–June 2020, we tested and refined our provisional grounded model by using unstructured and informal conversations, documentary materials (40 documents collected over 140 d), semi-structured interviews (n = 266; cf. Append. 3 for detailed methods and Append. 4 for the survey instrument). In each village, we conducted interviews with every fifth household, starting from the perimeter of the village (with greatest exposure to HEC) and working toward the center. On being denied consent (n = 6) or the house being empty (n = 18), an adjacent house was sampled. Each village was visited repeatedly across the fieldwork period and also at different times of the day so as to avoid a time-induced bias. Sampling in a particular village was stopped when we reached saturation—that is, each additional effort yielded little new information relevant to the research question, and we were able to “make sense” of the data (Glaser and Strauss 1967, Newing et al. 2010). Once this was done across all sample villages, our grounded model was finalized.

RESULTS

Fence Maintenance

Of the 19 fences assessed using the tech and human maintenance scores, 7 were classified as well maintained (37%), and 12 were poorly maintained (63%).

Seventeen of the 19 fences were classified the same way (as being either well maintained or poorly maintained) in blind and non-blind tests. Fences were ultimately classified as well or poorly maintained based on the blind tests: fences classified as well maintained had very high human maintenance scores and perfect tech scores (see Fig. 2).

A priori Hypothesis

Of our 19 sample communities with fences, 16 were found to have fences with geographically effective design, 7 provided some sort of cash buy-in for the fence, 13 were ethnically homogenous, and 12 were deemed to have proactive leaders.

Our QCA resulted in the following result:

LEAD (~HOMO*~CASH + CASH*~GEO) → Fence Maintenance

The solution term above proposes two pathways for maintenance: (i) maintenance will occur even if there is no cash buy-in and a non-homogeneous community so long as there is proactive leadership, or (ii) maintenance will occur if there is cash buy-in and no geographically effective design if there is proactive leadership. This result is thus incompatible with our a priori hypothesis. Furthermore, the combination of factors identified by the QCA as leading to fence maintenance does not seem to make logical sense. For instance, there is no reason to believe that cash buy-in would lead to maintenance when the fence design impedes non-users but then prevent maintenance in heterogenous communities. The result of the QCA is thus likely an artifact of chance (Ragin 2000), suggesting that our a priori variables do not meaningfully predict maintenance (refer to Append. 5 for detailed analyses).

Finally, one element of the a priori hypothesis was directly contradicted by the qualitative data collected. We found evidence that homogenous communities do not necessarily find it easier to cooperate, as sanctions against individuals that fail to pay for maintenance of the fence are rarely enforced. For instance, Rekha^† from Jalokhiabasti^† elucidates, “...since all of us belong to the same community, and the same namghar [temple], we all know each other well. Therefore, even though the rule is to pay a full day’s wage as fine for shirking their duty to monitor the fence, it is rarely enforced.”

Factors Contributing to Fence Maintenance Identified Via Grounded Theory

Our grounded approach suggested that successful fence maintenance could happen due to any of three mechanisms: (1) a small number of community fence maintainers ensure functionality of the fence, (2) community maintenance is enabled by self-organizing, and (3) maintenance is done by the Forest Department (cf. Fig. 3). Of the seven successfully maintained fences, two were maintained by community maintainers funded by individuals/organizations, two by community maintainers funded by the community, one by the community themselves, and two by the Forest Department. Of the 12 unsuccessfully maintained fences, none were found to be maintained by community maintainers funded by individuals/organizations, four by community maintainers funded by the community, eight by the community themselves, and none by the Forest Department (although other communities with Forest Department-maintained fences examined during exploratory study had failed fences; cf. Table 1).

The results of our grounded approach suggest that successful fence maintenance is not fully predicted by the pathway it follows but rather by whether the key actors (political elites, community maintainers, the Forest Department, or the community as a whole) perceived the benefits of fence maintenance to be greater than the costs (cf. Append. 6 and 7 for more on perceived costs and benefits). Notably, in contrast with our a priori hypothesis, the perceived benefits and costs were diverse (see Table 2)—no three or four variables seemed to adequately explain whether fences would be adequately maintained. Instead, the most important factors leading to maintenance or non-maintenance could be highly idiosyncratic. In one community, the community maintainer noted that he was motivated to ensure the fence was effective because he had lost a family member to an elephant attack a few years ago. He said he did not want “anybody’s family to suffer like me... people don’t understand the consequences until a disaster happens... now, where the authorities have given us this fence for our safety, we must take care of it as the benefit is for us only.”

Despite such idiosyncrasies, the weight of the evidence suggests some patterns in whether fences succeeded or failed in the case of each pathway. Community maintainers seemed to be most motivated by the direct payments they received and the goodwill they earned from the community for their work. The latter is not necessarily just a complement to the former. One community maintainer named Sing^† from Gorumara^† notes, “Dipen^† and I anyway don’t do this for the money, it is for the safety of the people of the village—and they know this and respect us for doing this. So we don’t mind the fact that they have stopped contributing toward our stipend, Rs 5–10 is nothing anyway.” In contrast, unaccountable community maintainers were able to extract payment from the community even when they failed to maintain the fence. In one village, the maintainer was also the village head and hence held power over the community members. In the other, the community did not possess the technical knowledge required to discern between well-maintained and poorly maintained fences, making them unable to hold the maintainer accountable.

When the community as a whole was responsible for fence maintenance, three social-ecological factors were particularly relevant to whether communities perceived the benefits of fence maintenance to outweigh the costs. First was the perceived pattern of elephant raiding: in order for benefits of the fence to be salient, elephant raiding instances needed to be either very frequent or somewhat frequent but unpredictable: if elephants visit only rarely, the costs of fence maintenance may be perceived to be higher than the benefits. If elephants visit infrequently but very predictably, fence maintenance and other mitigation measures need only be concentrated at the time of high risk, meaning fence maintenance efforts might flag at other times.

Second, communities had to perceive that fence maintenance significantly reduced HEC—i.e., that the fence could be effective at reducing HEC, giving communities a sense of safety, security, and peace of mind. Fences were seen as an especially useful tool in small villages that found it difficult to effectively chase elephants away. Noted a resident of Baghmara^†, which comprised six households and a hospital, “We are a small cluster [of houses] and don’t have young men to drive the elephants away like the neighboring villages. Therefore, the fence is even more important for us—prior to the fence, the Sister’s house was broken four times in one month!” Trust in the effectiveness of the fence could become an issue if elephants breached the fence—even if the breach was due to poor maintenance. Even occasional breaches could lead to reduced perceived benefits from the fence, which could lead to reduced maintenance, causing a vicious cycle. Additionally, some members believed that even poorly maintained fences could be deterring elephants due to prior negative experiences with live fences, suggesting that, “the elephants now know not to come here at all.” This belief could also potentially lead to a decline in the perceived benefits of maintenance.

Finally, communities were more likely to perceive the benefits of the fence to outweigh the costs if they had an opportunity to make and enforce the rules governing the fence. This provided community members with a sense of participation and ownership and allowed for consensus rules that fairly distributed costs and benefits across members, reducing instances of non-participation and sabotage. Community-constructed rules included the opportunity to take part in deciding where the fence should be established and how it should be maintained. Such a process could also help community members develop a positive relationship with the people establishing or maintaining the fence, as well as cause villagers to anticipate social sanctions by the community if they failed to comply with agreed rules and processes. The most important rules had to do with financial contributions to fence maintenance. In communities where the process to construct rules was not sufficiently participatory, a few individuals found they were the only community members delivering on their commitments to maintain the fence. These individuals felt taken advantage of, so they stopped maintenance altogether, resulting in the fence failing. This is termed as the “sucker effect” and appears to be a more salient problem in smaller villages, contributing to the failure of three small-village fences in our study. Relatedly, if the community chose to pay a specific person to act as community maintainer, the maintainer had to be seen as trustworthy and having a good track record, even in non-fence related activities. In two instances where the maintainer was caught shirking duties or siphoning money, the trust was broken, payments ceased, and fence maintenance did not take place. Once this trust was broken, despite the benefits outweighing the costs for a maintainer, the cost of transitioning to another maintainer were too high for the community, leading to the fence failing.

In our two cases of Forest Department-maintained fences, we found maintenance to occur because (a) of political pressure from the community or an administrative body to reduce HEC, or (b) perverse incentives/rent-seeking opportunities associated with the fence. In the case of one community, a politically empowered Eco-Development Committee could mobilize (violent) protests if the Forest Department did not prevent HEC, motivating maintenance of the fence. In another case, officials were reported to have maintained the fence so that superiors and local residents would not protest when officials exploited natural resources and funds made available to reduce HEC.

Even when the perceived benefits of maintenance outweighed the perceived costs, other conditions had to be met to allow for fence maintenance. For all fences other than those maintained by the Forest Department, fences must be designed such that they do not unduly hinder movement (of people or livestock). For instance, we accidentally captured evidence on the camera traps of the fence obstructing villagers carrying head-loads of firewood. As a result, the fence was sabotaged repeatedly in order to aid access. Sabotage was found to generally not be as significant a concern for Forest Department-maintained fences as the Forest Department rapidly repairs any breaches, and local community members are afraid of antagonizing “the powerful people in uniform,” hinting at the repercussions of getting caught. Finally, fence maintenance required that those responsible have the technical know-how required to maintain the fences.

DISCUSSION

Contrary to our hypothesis, successful maintenance of low-cost, non-lethal fences intended to reduce HEC could not be explained completely by whether fences were geographically well matched with their users, users paid for the fences, and community coordination was enabled by leadership or ethnic homogeneity. Instead, our results reflect the complexity of other social-ecological systems (Fleischman et al. 2014, Nagendra and Ostrom 2014). We found that fence maintenance is driven by a combination of ecological factors (pattern of elephant crop-raiding), political factors (pressures upon the Forest Department), the availability of technical ability, and the perceived costs and benefits of fence maintenance for several key actors. Even in our modest sample size of 19 villages, we found a diversity of explanations for fence maintenance and non-maintenance that defied simple summary. Some fences were successfully maintained through collective action, such as the collaboration of the entire community; some were maintained by a few dedicated community maintainers; and some by the State via the Forest Department. We found that fences could be successfully maintained due to the dedication of varying numbers of people, ranging from two highly motivated individuals to a small subset of approximately 10% of the households to nearly all the households in the village. In our initial exploratory survey of all 42 villages, there were also examples of unsuccessful maintenance from each of the three management arrangements we found.

The diversity of relevant perceived costs and benefits motivating action or inaction is of particular policy relevance. Emphatically, decisions were not necessarily made based on economic calculations. There was a general consensus that well-maintained fences helped reduce HEC by as much as 90%—but even when the benefits in terms of crops saved and protection of homes and safety almost certainly outweighed the costs in terms of time, effort, and money, this did not always lead to fence maintenance. For instance, when some community members felt that others weren’t contributing fairly to the fence, they withdrew their own support (i.e., the “sucker effect”), even though it might have been in their economic interest to ensure fence maintenance. Conversely, we found community fence maintainers that ensured fence functionality (in one case, even at his own financial expense) even if they weren’t paid, motivated by either social capital or their personal experiences with HEC. Although some fences succeeded despite non-payment, others failed despite fair payment due to a lack of accountability.

What, then, do these findings mean for conservation practitioners hoping to promote fence maintenance to stem HEC? Although fences could be maintained successfully via any of the three pathways we describe (Fig. 3), we believe working through community maintainers is the most dependable path to success. The appointment of one or two community maintainers can make it clear who is responsible for the fence and ensure a clear set of deliverables in exchange for pre-determined remuneration; such market norms—accompanied by a practical plan for accountability—can be easier for an NGO to navigate than unwritten social norms; these are typically complex and require a deeper understanding of intra-community dynamics, relationships, and the individuals’ interactions with one another (Ariely 2010). In selecting community maintainers, conservation practitioners should be deliberate: a trusted maintainer with a strong record of honesty and diligence might be the difference between a successful and failed fence.

In some contexts, however, practitioners might want to avoid force-fitting the community maintainer model. Where communities are small (between six and 30 households), maintenance by the full community seems more likely to succeed than for larger communities, as noted in other contexts (Olson 1965 in Ostrom 2009b). If the political incentives happen to allow for a government department (in our case, the Forest Department) to maintain the fence, then that could also be a functional option—but such political conditions such as being areas of strategic importance, rent-seeking opportunities, and political ramifications for non-maintenance are generally beyond the control of conservation practitioners from the civil sector.

In addition, conservation practitioners can pay attention to several key elements of the process of establishing a fence (cf. Append. 8 for detailed recommendations). When establishing a fence, practitioners can attempt to involve all members of the community in designing the fence and orienting them with regard to how it functions. This allows for multiple benefits. First, it can help ensure that the fence does not hinder livelihood activities to the degree that it attracts sabotage. Second, skilled implementers can attempt to elicit an understanding of the perceived costs and benefits of the fence and ensure that the interests of various community members are addressed in the design and management of the fence. Third, by establishing a clear shared understanding of the fence as infrastructure meant to serve the whole community, and by making clear the roles and responsibilities for fence maintenance to all community members, conservation practitioners can increase the (i) social costs of non-compliance and (ii) perception that compliance will be recognized by the community. We hope this will help prevent the sucker effect and the resultant vicious cycle of non-maintenance.

Conservation practitioners should also conduct training sessions and perhaps refreshers to ensure that those responsible for the fence learn and retain the technical ability to maintain the fence. Our findings suggest that communities might not openly ask to be given the skills they need to ensure fence maintenance.

Finally, when monetary contributions must be organized from the community to fund fence maintenance, we recommend that payments be structured as single large (perhaps annual) contributions instead of repeated small ones (e.g., weekly). Respondents indicated these larger payments would generally be feasible, and by reducing the transaction costs and making it easy for community members to monitor those failing to contribute (i.e., by increasing the social costs of non-compliance), such a system can help prevent the disintegration of collective action.

CONCLUSION

Our findings suggest that there are a variety of ways public goods can be successfully provisioned in low-resource rural settings. Public goods are often seen as best-met by government (Desai 2003, Flavin 2019), but even when political conditions do not allow this, community action or the dedication of well-incentivized individuals can work at the small scales we examined. However, that doesn’t mean providing such public goods is straightforward: overall, although some best practices in fence establishment and maintenance can be scaled (cf. Append. 8 for recommendations), the diversity of reasons that appear to contribute to the maintenance or non-maintenance of non-lethal fences aimed at reducing HEC should push conservationists to better understand the particular socio-ecological, political, and economic contexts in which they aim to work (Akama et al. 1996, Dickman 2010, Bennett et al. 2017, Dhee et al. 2019). More generally, participatory approaches (Usongo and Nkanje 2004, Milich et al. 2020) can provide practitioners in conservation and other fields the tools they need to identify the particular costs and benefits salient to the communities they work with and optimize the probability that the public goods they try to establish take root and are well sustained, particularly in low-resource settings.

^† Names changed to protect the identity of the respondent.

LITERATURE CITED

Agrawal, A. 2001. Common property institutions and sustainable governance of resources. World Development 29(10):1649-1672. https://doi.org/10.1016/S0305-750X(01)00063-8

Akama, J. S., C. L. Lant, and G. W. Burnett. 1996. A political-ecology approach to wildlife conservation in Kenya. Environmental Values 5: 335-347. https://doi.org/10.3197/096327196776679276

Ariely, D. 2010. Predictably irrational: the hidden forces that shape our decisions. Harper Perennial, New York, New York, USA.

Assam Haathi Project. 2008. Living with elephants in Assam. EcoSystems-India and North of England Zoological Society, Guwahati, Assam, India.

Barua, M., S. A. Bhagwat, and S. Jadhav. 2013. The hidden dimensions of human–wildlife conflict: health impacts, opportunity and transaction costs. Biological Conservation 157:309-316. https://doi.org/10.1016/j.biocon.2012.07.014

Bennett, N.J., R. Roth, S. C. Klain, K. Chan., P. Christie, D. A. Clark, G. Cullman, D. Curran, T. J. Durbin, G. Epstein, and A. Greenberg. 2017. Conservation social science: understanding and integrating human dimensions to improve conservation. Biological Conservation 205:93-108. https://doi.org/10.1016/j.biocon.2016.10.006

Berg-Schlosser, D., G. De Meur, Rihoux, B., and C. C. Ragin. 2009. Qualitative comparative analysis (QCA) as an approach. Pages 1-18 in B. Rihoux and C. C. Ragin, editors. Configurational comparative methods: qualitative comparative analysis (QCA) and related techniques. Sage Publications, Thousand Oaks, California, USA. https://doi.org/10.4135/9781452226569.n1

Bernard, H. R. 2006. Research methods in anthropology: qualitative and quantitative approaches. Rowman Altamira, Lanham, Maryland, USA.

Chai, Y., and M. Schoon. 2016. Institutions and government efficiency: decentralized irrigation management in China. International Journal of the Commons 10(1):21-44. https://doi.org/10.18352/ijc.555

Chelliah, K., G. Kannan, S. Kundu, N. Abilash, A. Madhusudan, N. Baskaran, and R. Sukumar. 2010. Testing the efficacy of chilli-tobacco rope fence as a deterrent against crop raiding elephants. Current Science 99:1239-1243.

Choudhury, A. 2004. Human–elephant conflicts in northeast India. Human Dimensions of Wildlife 9(4):261-270. https://doi.org/10.1080/10871200490505693

Desai, A. A., and H. S. Riddle. 2015. Human–elephant conflict in Asia. U.S. Fish and Wildlife Service, Washington, D.C., USA and Asian Elephant Support, St. Louis, Missouri, USA.

Desai, M. 2003. Public goods: a historical perspective. Pages 63-64 in I. Kaul, editor. Providing global public goods: managing globalization. Oxford University Press, Oxford, UK. https://doi.org/10.1093/0195157400.003.0003

Dhee, V. A., J. D. Linnell, S. Shivakumar, and S. P. Dhiman. 2019. The leopard that learnt from the cat and other narratives of carnivore–human coexistence in northern India. People and Nature 1(3):376-386. https://doi.org/10.1002/pan3.10039

Dickman, A. J. 2010. Complexities of conflict: the importance of considering social factors for effectively resolving human–wildlife conflict. Animal Conservation 13(5):458-466. https://doi.org/10.1111/j.1469-1795.2010.00368.x

Douglas-Hamilton, I., T. Krink, and F. Vollrath. 2005. Movements and corridors of African elephants in relation to protected areas. Naturwissenschaften 92(4):158-163. https://doi.org/10.1007/s00114-004-0606-9

Dunham, K. M., A. Ghiurghi, R. Cumbi, and F. Urbano. 2010. Human–wildlife conflict in Mozambique: a national perspective, with emphasis on wildlife attacks on humans. Oryx 44(2):185-193. https://doi.org/10.1017/S003060530999086X

Feiock, R. C., A. Steinacker, and H. J. Park. 2009. Institutional collective action and economic development joint ventures. Public Administration Review 69(2):256-270. https://doi.org/10.1111/j.1540-6210.2008.01972.x

Flavin, P. 2019. State government public goods spending and citizens’ quality of life. Social Science Research 78:28-40. https://doi.org/10.1016/j.ssresearch.2018.11.004

Fleischman, F., N. Ban, L. Evans, G. Epstein, G. Garcia-Lopez, and S. Villamayor-Tomas. 2014. Governing large-scale social-ecological systems: lessons from five cases. International Journal of the Commons 8(2):428-456. https://doi.org/10.18352/ijc.416

Galvin, K. A., T. A. Beeton, and M. W. Luizza. 2018. African community-based conservation. Ecology and Society 23(3): 39. https://doi.org/10.5751/ES-10217-230339

Glaser, B. G., and A. L. Strauss. 1967. Discovery of grounded theory: strategies for qualitative research. Routledge, Abington-on-Thames, UK. https://doi.org/10.4324/9780203793206

Gureja, N., V. Menon, P. Sarkar, and S. S. Kyarong. 2007. Ganesha to Bin Laden human–elephant conflict in Sonitpur District of Assam. Wildlife Trust of India, New Delhi, India.

Habyarimana, J., M. Humphreys, D. N. Posner, and J. M. Weinstein. 2007. Why does ethnic diversity undermine public goods provision? American Political Science Review 101(4):709-725.

Hinkel, J., M. E. Cox, M. Schlüter, C. R. Binder, and T. Falk. 2015. A diagnostic procedure for applying the social-ecological systems framework in diverse cases. Ecology and Society 20(1): 32. https://doi.org/10.5751/ES-07023-200132

Jack, B. K., and M. P. Recalde. 2015. Leadership and the voluntary provision of public goods: field evidence from Bolivia. Journal of Public Economics 122:80-93. https://doi.org/10.1016/j.jpubeco.2014.10.003

Kalam, T., H. K. Baishya, and D. Smith. 2018. Lethal fence electrocution: a major threat to Asian elephants in Assam, India. Tropical Conservation Science 11:1-8. https://doi.org/10.1177/1940082918817283

Kioko, J., P. Muruthi, P. Omondi, and P. I. Chiyo. 2008. The performance of electric fences as elephant barriers in Amboseli, Kenya. African Journal of Wildlife Research 38(1):52-58. https://doi.org/10.3957/0379-4369-38.1.52

Kshettry, A., S. Vaidyanathan, R. Sukumar, and V. Athreya. 2020. Looking beyond protected ;areas: identifying conservation compatible landscapes in agro-forest mosaics in north-eastern India. Global Ecology and Conservation 22: e00905. https://doi.org/10.1016/j.gecco.2020.e00905

Kushwaha, S. P. S., and R. Hazarika. 2004. Assessment of habitat loss in Kameng and Sonitpur elephant reserves. Current Science 87(10):1447-1453.

McGinnis, M. D. 2011. An introduction to IAD and the language of the Ostrom workshop: a simple guide to a complex framework. Policy Studies Journal 39(1):169-183. https://doi.org/10.1111/j.1541-0072.2010.00401.x

Miguel, E., and M. K. Gugerty. 2005. Ethnic diversity, social sanctions, and public goods in Kenya. Journal of Public Economics 89(11-12):2325-2368. https://doi.org/10.1016/j.jpubeco.2004.09.004

Milich, K. M., K. Sorbello, L. Kolinski, R. Busobozi, and M. Kugonza. 2020. Case study of participatory action research for wildlife conservation. Conservation Science and Practice 3(2): e347. https://doi.org/10.1111/csp2.347

Ministry of Environment, Forest and Climate Change (MoEFCC). 2017. Synchronized elephant population estimation. Project Elephant, MoEFCC, New Delhi, India.

Ministry of Environment, Forest and Climate Change (MoEFCC). 2021. Rajya Sabha starred question no.140, 2021

Mumby, H. S. and J. M. Plotnik. 2018. Taking the elephants’ perspective: remembering elephant behavior, cognition and ecology in human–elephant conflict mitigation. Frontiers in Ecology and Evolution 6: 122. https://doi.org/10.3389/fevo.2018.00122

Munyao, M., M. Siljander, T. Johansson, G. Makokha, and P. Pellikka,. 2020. Assessment of human–elephant conflicts in multifunctional landscapes of Taita Taveta County, Kenya. Global Ecology and Conservation 24: e01382. https://doi.org/10.1016/j.gecco.2020.e01382

Nagendra, H., and E. Ostrom. 2014. Applying the social-ecological system framework to the diagnosis of urban lake commons in Bangalore, India. Ecology and Society 19(2): 67. https://doi.org/10.5751/ES-06582-190267

Nath, N. K., B. P. Lahkar, N. Brahma, S. Dey, J. P. Das, P. K. Sarma, and B. K. Talukdar. 2009. An assessment of human–elephant conflict in Manas National Park, Assam, India. Journal of Threatened Taxa 1(6): 309-316. https://doi.org/10.11609/JoTT.o1821.309-16

Neupane, B., B. Khatiwoda, and S. Budhathoki. 2018. Effectiveness of solar-powered fence in reducing human–wild elephant conflict (HEC) in northeast Jhapa District, Nepal. Forestry Journal of Institute of Forestry Nepal 15:13-27. https://doi.org/10.3126/forestry.v15i0.24917

Newing, H., C. M. Eagle, R. K. Puri, and C. W. Watson. 2010. Conducting research in conservation social science. Routledge, Abington-on-Thames, UK.

Okello, M. M., S. J. Njumbi, J. W. Kiringe, and J. Isiiche. 2015. Habitat use and preference by the African elephant outside of the protected area, and management implications in the Amboseli Landscape, Kenya. International Journal of Biodiversity and Conservation 7(3):211-226. https://doi.org/10.5897/IJBC2014.0795

Ostrom, E. 2009a. A general framework for analyzing sustainability of social-ecological systems. Science 325(5939):419-422. https://doi.org/10.1126/science.1172133

Ostrom, E. 2009b. Analyzing collective action. Paper presented at the 2009 conference of the International Association of Agricultural Economists, 16-22 August 2009, Beijing, China.

Pinter-Wollman, N. 2012. Human–elephant conflict in Africa: the legal and political viability of translocations, wildlife corridors, and transfrontier parks for large mammal conservation. Journal of International Wildlife Law and Policy 15(2):152-166. https://doi.org/10.1080/13880292.2012.678793

Ponnusamy, V., P. Chackrapani, T. W. Lim, and S. Saaban. 2016. Farmers’ perceptions and attitudes towards government-constructed electric fences in peninsular Malaysia. Gajah 45:4-11.

Ragin, C. C. 2000. Fuzzy-set social science. University of Chicago Press, Chicago, Illinois, USA.

Rangarajan, M., A. Desai, R. Sukumar, P. S. Easa, V. Menon, S. Vincent, S. Ganguly, B. K. Talukdar, B. Singh, D. Mudappa, and S. Chowdhary. 2010. Gajah. Securing the future for elephants in India. Report of the Elephant Task Force, Ministry of Environment and Forests, New Delhi, India. URL: https://www.environmentandsociety.org/sites/default/files/key_docs/Gajah.pdf

Sapkota, S., A. Aryal, S. R. Baral, M. W. Hayward, and D. Raubenheimer. 2014. Economic analysis of electric fencing for mitigating human–wildlife conflict in Nepal. Journal of Resources and Ecology 5(3):237-243. https://doi.org/10.5814/j.issn.1674-764x.2014.03.006

Sarker, A., and T. Itoh. 2001. Design principles in long-enduring institutions of Japanese irrigation common-pool resources. Agricultural Water Management 48(2):89-102. https://doi.org/10.1016/S0378-3774(00)00125-6

Sawada, Y., R. Kasahara, K. Aoyagi, M. Shoji, and M. Ueyama. 2013. Modes of collective action in village economies: evidence from natural and artefactual field experiments in a developing country. Asian Development Review 30(1):31-51. https://doi.org/10.1162/ADEV_a_00002

Schneider, C. Q., and C. Wagemann. 2012. Set-theoretic methods for the social sciences: a guide to qualitative comparative analysis. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9781139004244

Shaffer, L. J., K. K. Khadka, J. van Den Hoek, and K. J. Naithani. 2019. Human–elephant conflict: a review of current management strategies and future directions. Frontiers in Ecology and Evolution 6: 235. https://doi.org/10.3389/fevo.2018.00235

Srivastava, S., T. P. Singh, H. Singh, S. P. S. Kushwaha, and P. S. Roy. 2002. Assessment of large-scale deforestation in Sonitpur district of Assam. Current Science 82(12):1479-1484.

Talukdar, B. K., and R. Barman. 2003. Man–elephant conflict in Assam, India—is there any solution? Gajah 22:50-56.

Usongo, L., and B. T. Nkanje. 2004. Participatory approaches towards forest conservation: the case of Lobéké National Park, south east Cameroon. The International Journal of Sustainable Development and World Ecology 11(2):119-127. https://doi.org/10.1080/13504500409469816

Wilson, S., T. E. Davies, N. Hazarika, and A. Zimmermann. 2015. Understanding spatial and temporal patterns of human–elephant conflict in Assam, India. Oryx 49(1):140-149. https://doi.org/10.1017/S0030605313000513

World Resources Institute. 2020. Global forest watch. World Resources Institute, Washington, D.C., USA. URL: https://www.globalforestwatch.org

Zimmermann, A., T. E. Davies, N. Hazarika, S. Wilson, J. Chakrabarty, B. Hazarika, and D. Das. 2009. Community-based human–elephant conflict management in Assam. Gajah 30:34-40.