What drives local communities' attitudes toward the protected area? Insights from Bardia National Park, Nepal

Understanding the local communities' attitudes toward protected areas (PAs) is critical to PA management planning. However, what socio‐economic and demographic factors drive their attitude remains poorly studied. We interviewed 135 residents in three villages in the buffer zone of Bardia National Park, Nepal, to assess their perceived benefits and costs of the PA. We found that perceived benefits are skewed toward wealthy and influential people, whereas costs associated with wildlife‐caused damage (i.e., crop raiding) are borne disproportionately by poor people. The Tobit regression model showed that wealthy and educated residents who obtained direct tourism benefits from the park were significantly more likely to hold positive attitudes toward the PA. On the flip side, residents who had experienced conflict with park administration (park staff and military guards) and incurred a higher amount of crop loss in the recent past were likely to have negative attitudes. Our results showed that negative attitudes largely stemmed from park staff and military guards' behavior and inefficient service delivery from the national park office. In light of these findings, we suggest the park administration develop mechanisms to ensure efficient service delivery and pragmatic handling of community grievances concerning law enforcement and wildlife damage compensation. We also recommend directing park benefits to less educated and poor households and those suffering property losses from wildlife to improve park‐people relationships.


| INTRODUCTION
Establishing protected areas (PAs) is the main instrument for conserving biodiversity (Geldmann et al., 2013;Saura et al., 2017). However, the creation of PAs and their management have a wide-ranging positive and negative impact on local communities. For example, PAs can alter natural resource use rights (restrict or completely deprive of resource use), evict local communities (Allendorf et al., 2006;West et al., 2006), and wildlife can cause damage to crops and kill livestock (Nepal & Weber, 1995;Studsrød & Wegge, 1995;Woodroffe et al., 2005). Furthermore, given the high densities of people living in poverty around PAs, particularly in developing countries, PAs have more serious negative impacts (Cernea & Schmidt-Soltau, 2006;Karanth & Nepal, 2012). All these significant costs associated with the management of the PAs bear by the residents might result in negative attitudes and undermine the conservation goals (Dewu & Røskaft, 2018;Wang et al., 2006). Conversely, PAs can also generate employment and income for locals and provide ecosystem goods and services (Estifanos et al., 2020;Spiteri & Nepal, 2006).
In Nepal, PAs covers 23.39% of the total land (DNPWC, 2019), where a high density of natural resource-dependent local communities live alongside PAs (Heinen et al., 2019), particularly in Nepal's plains (Terai). Nepal's PAs initiated incentive measures, such as buffer zone program, revenue sharing, and provisions that allow the collection of thatch grass inside the PAs for a few days a year to meet the subsistence needs of local communities (Baral & Heinen, 2007;Heinen et al., 2019). The Bardia National Park (BNP) is the largest national park in the plains (Terai), characterized by high human population density (GoN, 2022). The population of mega-species, including tigers and one-horned rhinoceros, has increased (DNPWC & DFSC, 2018;Heinen et al., 2019), resulting in increased humanwildlife conflict incidences (Heinen et al., 2019;Prins et al., 2022;Shahi et al., 2021). Although there are few studies undertaken in BNP focusing on attitudes (Allendorf et al., 2007;Baral & Heinen, 2007), conservation benefits (Thapa Karki, 2013), and resource dependency on the park (Thapa & Hubacek, 2011), little is known what socio-demographic factors explain their attitudes toward the PA. It is also poorly understood whether it is due to a lack of conservation benefits or a combination of conflict with PA staff and military guards, unfair and inefficient PA service delivery that stem negative attitudes.
In this paper, we examine the residents' perceived benefits and costs from the national park and assess their attitudes toward the PA. We asked the following specific research questions: What are the main benefits and costs of the PA, and how do residents perceive these costs and benefits? Furthermore, what factors (e.g., age, gender, education level, ethnicity, wealth category, dependency on natural resources, and the distance from the park boundary) influence their attitudes toward the PA? Identifying and understanding factors affecting peoples' attitudes is critical to designing well-targeted conservation strategies and ensuring PA's effectiveness.
2 | METHODS 2.1 | Study area BNP (28 23 0 0 00 N 81 30 0 0 00 E), situated in western lowland Nepal, encompasses an area of 968 km 2 along with an additional 507 km 2 of the buffer zone ( Figure 1). BNP has undergone several changes in conservation approachesfrom first declared as a Royal Hunting Reserve in 1969 to present-day BNP in 1989. The altitudinal elevation of the park ranges from 152 to 1564 m from the mean sea level (BNP, 2016). Sal (Shorea robusta) makes up about 70% of the forest cover of the park with a mixture of grassland and riverine forests. BNP inhabit many threatened wildlife species, including tigers (Panthera tigris), one-horned rhinoceros (Rhinoceros unicornis), and elephants (Elephas maximus) (BNP, 2016). The southern end of the park extends into Khata forest corridor, an interspersion of the restored forest with agricultural settlements, between the border of the park and the Katarniaghat Wildlife Sanctuary.

| Buffer zone and study villages
Thakurdwara, Shivapur, and Suryapatuwa Village Development Committees (VDCs) were selected as study sites. After the promulgation of the new constitution in 2015, VDCs have been subsumed into rural municipalities and municipalities effective from 2017. But we continue to refer VDCs here to make a direct comparison to the previous studies. Currently, there are nine rural municipalities or municipalities that surround the park as the buffer zone. Under the new local government structure, Thakurdwara and Shivapur belong to Thakurbaba Municipality, and Suryapatuwa belongs to Madhuban Municipality. Three VDC were selected based on their similarities (i.e., proximity to the park headquarters and high wildlife densities) and differences (i.e., access to different forest resources-buffer zone community forest at Thakurdwara, park forest at Shivapur while government forest in addition to buffer zone community forest at Suryapatuwa and disparity of fund allocated from buffer zone management council-higher to Thakurdwara and Shivapur while lower to Suryapatuwa). Over 50% of the total population belongs to indigenous Tharu communities, followed by "high" caste groups (Brahmins and Chhetries), with a small representation of "low" caste groups (Dalits), and Tibeto-Burman origin ethnic groups (e.g., Tamang) (BNP, 2016;Studsrød & Wegge, 1995). The local livelihood depends on subsistence farming and forest resources such as timber, fuelwood, and grass from the park and nearby buffer zone forests. Paddy (Oryza sativa), Wheat (Triticum aestivum), Maize (Zea mays), Mustard (Brassica campestris), and Lentils (Lens culinaris) are the main crops cultivated. Most residents are farmers who rear livestock also, such as cows, buffalo, sheep, goats, chickens, and pigs.

| Data collection
We collected household lists from the ward offices of each municipality. We randomly selected households from each ward household list (i.e., serial numbers of households matching numbers generated by the random sample in R were the households selected for the F I G U R E 1 The map of the Bardia National Park, its buffer zone, and the villages where the study was undertaken questionnaire survey) (Table 3). A total of 135 households were selected, comprising 11.5% of 1177 households in six wards (villages). Two communities (villages/settlements) were selected from each Thakurdwara, Shivapur, and Suryapatuwa VDC based on distance from the park boundary. One sampling community was close to the park boundary (<2.5 km) and the other was relatively far distance from the park boundary (>2.5 km). We undertook face-to-face semi-structured interviews with the head of the household in six settlements. In their absence, another household member over 18 years of age was interviewed. The questionnaire consist of three sections pertaining: (1) household demographic and socioeconomic characteristics, (2) perceived benefits and costs of PA, and (3) attitudes toward the PA. After a pilot survey with 10 respondents, two group discussions, and consultation with park staff, we obtained a comprehensive list of benefits and costs from the park. These benefits and costs were later grouped into six T A B L E 1 The questions asked to respondents and the responses received to quantitatively assess the attitudes of residents toward protected area broad categories of benefits and costs. The study was conducted from January to March 2018.

| Data analysis
We used a 4-point Likert-type ordinal scale (coded as 0, 1, 2, and 3) to assess the perceived benefits and costs of the park (Likert, 1932;Spiteri & Nepal, 2006). On the scale, 0 represents no, 1 represents the low level, 2 represents the medium level, and 3 represents the high level of benefits or costs perceived by the respondents. Finally, we calculated the weighted mean for each benefit and cost to find the main benefits and costs perceived by the respondents. Based on the information collected on landholding size, livestock numbers, family income, assets owned, and proportion of educated (grade 10 or above) individuals in the family, we categorized household into rich, middle, and poor.
To examine attitudes, we formulated nine structured questions (Table 1). Adding those questions gave a composite attitude score ranging from À11 to +11. Since the composite attitude score had upper and lower limits, we could not use the ordinary least square regression model. Instead, we built a Tobit regression model since it accounts for limited dependent variables (Baral & Heinen, 2007). The explanatory variables included in the model, the type of variable, and how we coded variables in our analysis are presented in Table 2. All the statistical analyses were performed using R (R Core Team, 2013).

| Household characteristics
Majority of respondents were male (57%, Table 3). The average age of respondents was 41 years (range: 18-74 years). Most (79%) of respondents were literate. Nearly all respondents (95%) were involved in mixed agriculture (farming and livestock rearing), mostly done at the subsistence level. More than a third (39%) of respondents had jobs (e.g., teachers, civil servants, army) or small businesses (e.g., shops, tea stalls/restaurants, homestays), whereas 11% depended on remittance sent from abroad by a household member and 9% had homestays (i.e., an eco-tourism-based business) as a major source of income. Additional socio-demographic information of respondents is provided in Table 3.

| Perceived benefits and costs
Respondents rated the Increase in the awareness of conservation (2.19) and Improvement in basic infrastructure (1.76) (e.g., roads, drinking water, transportation to a nearby city, and schools) as the main benefits received due to the existence of the park (Table 4). In contrast, very few respondents viewed tourism and incomegenerating training (e.g., sewing & tailoring, house electrical wiring, vegetable farming) as benefits indicated by low weight mean value (Table 4). Similarly, the T A B L E 2 Selected socio-demographic explanatory variables for Tobit regression models to examine respondents' attitudes toward protected area improvement in social service (e.g., education, health, and sanitation) and participation in natural resources management were also not reflected as highly perceived benefits by the respondents (Table 4). While increase in human-wildlife conflict (2.27) and limited access to forest resources (1.87) were the major perceived costs by the respondents, few respondents perceived the eviction and loss of land tenure as the main cost (Table 4). Respondents also considered conflict with park staff and military guards (1.24) and an increase in the cost of living (0.88) as other perceived costs, but on a relatively low scale (Table 4). Additionally, we documented the respondents' response regarding if they received any tourism benefits or experienced wildlife damage in the last 5 years. Of total poor respondents, only few respondents (6.4%) reported getting tourism benefits, while a considerable proportion of middle (27.6%) and rich (33.3%) category respondents reported getting tourism benefits. In contrast, we found that the wildlife damage experienced by poor respondents was substantially higher ($50%) than the non-poor residents (i.e., middle and rich).

| Respondents' dependency on natural resources
Firewood still was the primary source of energy. Residents also collect thatch grass for roofing, timber for house construction, and grass and tree fodder to feed livestock. The majority of households (54%) were exclusively dependent on firewood, followed by a substantial proportion of households (41%) dependent on a combination of firewood, liquified petroleum gas (LPG), and biogas, and only a few (5%) used LPG only. The households that did not have access to buffer zone community forest were more dependent on the illegal extraction of forest resources from the national park (Table 3). We found that 74% of Shivapur households were involved in the illegal extraction of fuelwood (from the national park), followed by 68% of Thakurdwara (from the park) and 45% of Suryapatawa households (from the park and government forest).

| Factors influencing attitudes toward PA
The mean attitude score of all the respondents was 1 (range: À10 to +8). Respondents from the three villages did not differ significantly in their average attitude scores toward the PA (Kruskal-Wallis χ 2 = 1.9781, df = 2, p-value = .3719). The Tobit regression model revealed that experience of conflict with park management (park staff and military guards), education level, mean crop loss, total fuelwood collection, tourism benefits received, and wealthy category were the significant variables T A B L E 3 Respondent's socio-economic characteristics (N = 135) in Bardia National Park during a field survey conducted in January- affecting the attitudes of the respondents (Table 5). However, age, ethnicity, gender, and distance from the park boundary did not have a significant influence on the attitudes (Table 5).

| DISCUSSION
Our results showed that suite of socio-economic variables influences residents' attitudes toward the PA. While our results generally agree with previous studies, we documented several interesting new insights. Residents who had conflict with park staff and military guards in the recent past was important determinant of negative attitudes toward the PA (Table 5). The majority of the respondents (53%) expressed that they have had a bitter experience of getting scolded and hit by park staff (frontline park staff) or military guards for extracting park resources. Some respondents had the experience of paying monetary fines and getting imprisoned for collecting firewood and fodder illegally. These kinds of negative interactions have been documented in the same study site, BNP (Allendorf et al., 2007), and elsewhere in Africa (Tessema et al., 2010), India (Karanth & Nepal, 2012), and South America (Fiallo & Jacobson, 1995). It perhaps calls for training and support to PA management authority to facilitate a trustworthy and good relationship with local communities (Woodside et al., 2021). Although the contemporary narrative paints pessimistic picture of park staff behavior and their approach to deal with local communities, recent scholarship calls for more nuanced planning and discussion to foster positive park-people relationships. This is apparent because park staff usually are overloaded with occupational stress, political pressure, and poor field conditions (Moreto, 2016). They may also have little incentive and training for compassion toward local communities in the midst of work overload. It is thus likely that overlooking their perspective of their relationship with local communities may not resolve the park-people conflict over resource access. We, therefore, argue that fostering this relationship would require policymakers and conservation practitioners looking beyond a single perspective to the park-people relationship (Moreto & Charlton, 2021;Mutanga et al., 2015Mutanga et al., , 2016Mutanga et al., , 2017Woodside et al., 2021). We found that perceived benefits from the park were more skewed toward wealthy and educated people with high social status. It is surprising given that even after more than 20 years of initiating the buffer zone program aimed to benefit the local community through livelihood and development, poor residents still feel that the benefits have not been fair (Thapa Karki, 2013). Although some poor economic status respondents said that they sometimes get some tourism benefits in the form of jobs, T A B L E 4 The perceived benefits and costs revealed by residents (N = 135) in three villages of in Bardia National Park the benefits they receive are meagerly $20,000 NPR (US $192) per year-too low to appreciably influence their attitude (Karanth & DeFries, 2010;Karanth & Nepal, 2012). They get seasonal employment in lowerpaid positions, particularly as housekeepers and cooks in hotels. This apparent insignificant change over the last two decades of the buffer zone program suggests that those in influential positions (e.g., village leaders, buffer zone user group chairperson, or non-resident hotel owners) continue to reap a larger share of benefits from the park. And this is corroborated by the fact that residents receiving direct benefits from the PA (i.e., tourism benefits) elicited a positive attitude. Overall, this suggests that there has been a trivial change in improving the income of poor households through the buffer zone program. A few respondents noted that they are largely unaware of park activities and programs as they are rarely invited to buffer zone community meetings. Importantly, residents who incurred higher economic loss (i.e., crop damage by wildlife) had negative attitudes. Our previous study has shown that most of the residents in Bardia depend on agriculture for subsistence, and the amount of crop damage by wildlife is considerable (Shahi et al., 2021). Furthermore, poor people are the ones who bear disproportionally high economic loss due to wildlife damage in relation to their wealth (e.g., landholding size) because of small landholding and greater dependence on the natural resource (Shahi et al., 2021).
We found educated respondents to have a more favorable attitude, as demonstrated by several previous studies (Dewu & Røskaft, 2018;Karanth & Nepal, 2012;Kideghesho et al., 2007;Mutanga et al., 2015). This is likely because educated people have a better understanding of the role that PAs play in conserving biodiversity and provisioning ecosystem services (Allendorf et al., 2012). Interestingly, we did not find any significant influence of gender, age, and ethnicity on respondents' attitudes, which corroborates the findings of a study conducted in the same national park (Baral & Heinen, 2007). Studies have shown the contrasting influence of gender and age on attitudes toward conservation (positive; Karanth & Nepal, 2012, negative;Gifford & Sussman, 2012;Newmark et al., 1993;Rao et al., 2003;Stem et al., 2003). The dependency on park resources also influenced residents' attitudes. Respondents more dependent on the park's forest resources (e.g., fuelwood) perceived the PA more negatively. Previous studies also demonstrate this finding (Allendorf et al., 2007;Karanth & Nepal, 2012;Spiteri & Nepal, 2008;Thapa & Hubacek, 2011). While previous studies show that residents residing closer to the park boundary often have more negative attitudes than those living in distant areas (Sarker & Røskaft, 2011;Spiteri & Nepal, 2008), we did not find any significant association between distance from the park and respondent attitudes. This difference may be attributable to only two distance categories in our study, which were relatively close to the park boundary (<5 km). It is also likely because a similar number of incidents of wildlife damage were observed in far and near settlements from the park boundary (Shahi et al., 2021). Increase in the awareness of conservation and Improvement in basic infrastructure were the main perceived benefits, whereas the increase in human-wildlife conflict and limited access to forest resources were the main perceived costs by the respondents. While benefits were felt at the societal level, costs were incurred at the individual level (Baral & Heinen, 2007). The livelihood of the local people is still dependent on natural resources as documented by earlier studies (Allendorf et al., 2007;Baral & Heinen, 2007). The unavailability or inadequacy of alternative source (buffer zone community forest) to fulfill their needs compels them to illegal collect forest resources, such as firewood. We found average fuelwood consumption per household (kg/day) in all three VDCs was significantly lower than Thapa and Hubacek (2011). This might be due to increased use of alternative fuel sources (i.e., LPG and biogas) or underreporting of fuelwood amount collected or both. Furthermore, there was decrease in average number of livestock (Thakurdwara = 5.3, Shivapur = 4 and Suryapatuwa = 6.2) and livestock unit than previous studies (Baral & Heinen, 2007;Thapa & Hubacek, 2011). The apparent decrease in livestock number might be because of fear of livestock depredation, the shift in livelihood from animal husbandry to tourism, alternative income generating activities opportunities (e.g., sewing and tailoring, house electrical wiring, aromatic plant cultivation [Mentha and Chamomile], and out-migration of youths for employment to abroad).
One of the limitations of our study was the relatively small sample size. We tried to overcome this by randomly selecting respondents of diverse demographic and socioeconomic background to make the sample representative of the local context. Nevertheless, we submit that caution should be exercised while generalizing the findings of our study to other PAs with entirely different socio-economic contexts. In this study, we used a quantitative approach (i.e., regression analysis) to determine the relationship between attitudes and socio-demographic factors of respondents. Supplementing this with rigorous qualitative analysis with a well-defined theoretical framework would have provided additional insights into multiple causal links between attitudes and various social, economic, and institutional factors. We, therefore, suggest future studies to explore this aspect.

| CONCLUSION
Overall, this study provides two important insights into the park-people relationship and the prevailing management approach of the PA. First, despite calls from previous studies to direct interventions to poor communities to ensure fair conservation benefits, perceived benefits from the park continue to be skewed toward wealthy and educated people with high social status. Second, residents' negative attitudes toward PAs largely stemmed from park staff and military guards' behavior, and inefficient service delivery concerning compensation for damages caused by wildlife. Together, these insights highlight what is lacking in the contemporary approach to PA management. Based on these insights, we suggest that fostering positive park-people relationships will require fair and equitable sharing of conservation benefits to local communities surrounding the PA. This can be achieved by directing benefits toward less educated, poor landholders, who are heavily resource-dependent and have had economic loss due to crop damage by wildlife. This will also require capacitating and re-orienting park administration toward effective service delivery and a more facilitative and collaborative approach to PA management.