An Emerging Issue of Human-Leopard Conflict in the Human-Dominated Landscape of Mid-Hills: A Case Study from Tanahun District of Nepal

Information on the spatial and temporal patterns of losses caused by leopard ( Panthera pardus ) in terms of human attacks and livestock depredation in the human-dominated landscape of the mid-hills of Nepal is essential in formulating and implementing efective mitigation measures. Tis study aimed to assess the spatial and temporal patterns of leopard attacks on humans and livestock and the economic losses incurred by livestock depredation between 2015 and 2019 in the Bhanu municipality of Tanahun District. We adopted a household survey ( N =110), key informant ( N =10), and focus group discussion ( N =4) for this study. We purposively chose two confict wards: 2 and 4, based on the severity of the attacks by the leopard. Within each ward, we selected the households randomly and conducted a semistructured questionnaire survey in September 2020. A total of 8 incidents of human attacks and 142 incidents of livestock depredation were recorded, with six human casualties in ward 2 and 1.45 incidents of livestock depredation per household in ward 4. Te maximum attack was observed during 2019 both on humans ( n =6) and livestock ( n =67). Leopards mostly attacked children below 9years, living within 200m of the nearest forest edge, with the highest attack during the autumn months (62.5%). During the fve years, leopard killed goats that represent 83.1% of total livestock loss categories. A signifcant diference was found in the frequency of attacks on livestock over the years ( χ 2 =87.60, df=4, and P ≤ 0 . 01), months ( χ 2 =16.53, df=11, and P � 0 . 12), and time of day ( χ 2 =48.47, df=3, and P ≤ 0 . 001) with the highest attack during the year 2019 (47.18%), July (14.08%), and daytime (42.96%), respectively. Households living nearer to the forest edge ( < 200m) lost more livestock (72.54%). Te monetary value of a total of 8142 USD (74 USD per household) was lost due to livestock depredation, with major monetary loss at a distance > 400m from the forest edge. We suggest adopting mitigation measures like predator-proof livestock corals while stall feeding and strengthening conscientious livestock herding practices during grazing, encouraging livestock insurance schemes, educating local communities about leopard behavior, caring for and protecting children intensively in the leopard attack sites, improving the prey base in the wild, and timely management of man-eater leopard to reduce the confict in the study area and the country.


Introduction
Human-wildlife confict (HWC) refers to interactions between humans and wildlife that cause negative impacts on human livelihoods, wildlife, and their environment [1,2]. HWC often results from human encroachment or loss of natural habitats, leading to increased human-wildlife interactions from competition over shared resources [3][4][5].
Tese conficts include human injury, death, livestock depredation, crop-raiding, property damage, and retaliatory killing of wildlife [6,7]. Although such conficts may involve any wild animal, there are more common with large carnivores due to their diet and overlapping home ranges with human settlements [8]. Among diferent HWC incidents, human casualties, and injuries, fear of wildlife attacks and livestock depredation is considered one of the most serious problems across the world [9][10][11][12]; resulting in reduced support of local communities for wildlife conservation [13]. HWC threatens the existence of wildlife as well as the livelihoods of people [14,15]; hence, it is considered one of the major conservation challenges worldwide [16][17][18][19].
Leopard is one of the most prevalent carnivores [23], distributed throughout Southern Asia, India, and sub-Saharan Africa [24]. It is categorized as vulnerable by the IUCN (2016) due to >30% of its global population decline over three generations caused by habitat loss, hunting, prey depletion, and confict with humans [25,26]. Tis species preys on mediumsized ungulates (1-45 kg) [27,28] and often lives in habitat edges near human settlements [25]. It prefers to mate in January and February, although it has no fxed breeding season [29]. Among the 14 recognized subspecies of leopard [30], Panthera pardus pernigra is found in Nepal [31]. Teir habitat is greatly reduced from their original distribution [32]; however, their population is now increasing after the implementation of community forestry in the 1990s, which provided additional habitat for leopards [33,34].
Human-leopard confict (HLC) is one of Nepal's most frequently recorded serious human-carnivore conficts [13,32,35]. It is directly afecting very poor local communities [10,[36][37][38][39][40]. An average of 25 people per year were killed by leopard from 1994 to 2004 [41], and 18 were killed and three injured between 2011 and 2014 in Baitadi District [42]. According to DNPWC [43], leopard was the principal predator for depredation on 78% of the livestock in Nepal. HLC results in the killing of leopards [10,44], and also they are facing increasing risks of mortality [45]. Among the 51 dead leopards from 2006 to 2013 in Nepal, about 65% of them were killed by humans [46]. Tere is an increasing incidence of HLC, even in areas with no previously reported incidents [3,33].
Leopard attacks livestock while grazing in the forest as well as inside human settlements, posing risks to human lives [47]. Six children were killed and 57 livestock were depredated by leopard in one and half years in Bhanu Municipality, Tanahun [48]. Due to increased instances of attacks, the District Administration Ofce (DAO), Tanahun, as per Section 9 (4) of the Local Administration Act (1971), had directed the locals to kill leopards if they enter human settlements [49]. All the damages stated above are evident in the study area, but no research related to such confict issues has been conducted to date. Te major objectives of our study were: (a) identify the spatial and temporal patterns of livestock depredation and human casualties between 2015 and 2019 in the Bhanu municipality of Tanahun District and (b) quantify the economic loss faced by households due to livestock depredation. Further, this study tried to answer research questions: (a) When and where the livestock and human attack incidents occurred in recent years? (b) How does the economic loss vary accordingly to spatial location, time, and livestock categories? Tese research outputs will support concerned authorities in designing and implementing efective mitigation measures and safeguarding the livestock of inhabitants, particularly residing nearby the forest.

Study Area.
Te study was conducted in the Bhanu municipality, where the attack of leopard on humans and livestock was frequently reported (Figure 1). It lies in the northern part of the Tanahun District of Gandaki Province, Nepal, with latitude and longitude of 28°01′12.3″N and 84°25′25.4″E, respectively; comprising a total population of 6298 within 1800 households [50]. Most of the people are involved in agriculture-crop cultivation and livestock farming for their major sources of income. Te livestock rearing practice mainly involves stall feeding within the livestock shed, while seasonally grazing in the private farmland during the winter months after harvesting crops. Sal (Shorea robusta), Chilaune (Schima wallichii), and Katus (Castanopsis indica) are the major plant species in this area. common leopard, Asiatic black bear, rhesus monkey (Macaca mulatta), barking deer (Muntiacus muntjak), porcupine (Hystrix brachyuran), golden jackal and so on are the major fauna species found in this area.

Data Collection.
A preliminary survey was carried out to know the general information and existing situation of leopard-related incidents in the study area before the actual data collection. First, we conducted a focus group discussion including staf of the Division Forest Ofce (DFO), the municipality, the Federation of Community Forestry Users Nepal, local elites, and women to identify the humanleopard confict hotspot in the Tanahun District. We recorded the registered lists of confict incidents in DFO from 2015 to 2019 and selected wards no. 2 and 4 for our study based on the higher number of leopard attack incidents ( Figure 1). Te "ward" refers to the smallest administrative unit of local government in Nepal. Te actual feld study was conducted in September 2020. Since the socioeconomic characteristics (economic well-being and livelihood strategies) of the households were not much diverse; therefore, we randomly selected only 110 households (n = 55 from each ward) out of 1800 households following the method suggested by Koirala et al. [35] and Baral et al. [22] for the questionnaire survey. We assume that the selected households are not spatially correlated and could represent the diferent extent of human-leopard confict in the study area. We used both structured and semistructured questionnaires that were divided into two major sections. Te frst section included the demographic and livestock status of respondents, including the value of livestock owned. While the second part contained incident details such as time and place of attack, frequency of attack, severity of human and livestock attack (casualty/injury), category of livestock depredation, and economic losses incurred based on the monetary value of diferent categories of livestock. Similarly, a separate checklist with open-ended questions was prepared to aid in providing additional information on the existing situation as well as the past fveyear trend (2015-19) of livestock depredation and human casualties by a leopard, and the status of livestock holdings, and the retaliatory killing of leopard. We selected 10 key local stakeholders who were familiar with the leopard attack incidents within the study area. Te representatives from local bodies include; Division Forest Ofcer (n = 1), rangers (n = 3), local political leaders (n = 2), elites (n = 2), and the secretary of each ward (n = 2). In addition, four focus group discussions (n = 2 in each ward) were carried out each consisting of leopard attacked victims (n = 10), executive members of Community Forest User Groups (n = 2), and ward members (n = 2) to supplement and validate the information obtained from key informant interviews and household surveys. Each focus group discussion was performed for about 3 hours. Information about the general trend of the leopard confict, the consequences of the attacks on the victims' livelihood those of and their family members, and the economic loss incurred from livestock depredation were discussed and recorded through the memory recall method. Tis method assumes that the respondents could reproduce precise information [35]. Further, incident sites were identifed by the participatory method including local people. GPS coordinates for each incident site were taken, and the distance to a water source, forest, settlements, and road was calculated from ArcGIS 10.8 for spatial analysis of confict data. Te condition of livestock sheds in the afected households was also thoroughly observed.

Data Analysis. MS Excel 2016 and Statistical Package for
Social Sciences [51] were used to analyze data. Descriptive statistics such as mean, percentage, frequency, and crosstabulation were used to interpret the result. Further, we classifed livestock depredation by leopard into 5 categories: International Journal of Zoology goat, bufalo, cattle, chicken, and pig, since these livestock were the major fnancial source for their livelihoods at the household level. Although other pets like dogs and cats were also attacked, they did not contribute to the household economy in rural settings: we did not consider them for this study. Te chi-square tests of independence were applied to compare the frequency of livestock attacks by a leopard with time (year, month, and time of day), location (house, farmland, or forest), and distance from the forest at a 5% level of signifcance. Further, we categorized the months; Dec-Feb, Mar-May, June-Aug, and Sept-Nov as winter, spring, summer, and autumn seasons, respectively. Similarly, we divided the 24 hours of the day into 4 equal classes (time of day): morning (4 am-10 am), day (10 am-4 pm), evening (4 pm-10 pm), and night (10 pm-4 am) for a relative comparison of attack risks. In addition, the relative risk of livestock depredation was calculated for diferent categories of livestock concerning the numbers owned by the household in the study area. We did not account for retaliatory killings of leopard and further analysis as there were no ofcial records of such event in the Divisional Forest Ofce.

Te Monetary Value of Livestock.
Te existing market value of all types of livestock was estimated by interviewing villagers and local livestock traders. It was found that the average value for each category of livestock is similar in both wards of Bhanu municipality through consultation with respondents. Te local selling prices were considered for an individual animal. Te rate of each category of animal was fxed as a goat (weight and sex), cattle and bufalo (age or milking or not), and pig and chicken (weight) ( Table 1). During the survey, each respondent was asked to provide the sex, weight, age, and condition (milking or not) of particular depredated livestock, which was verifed through the records kept at DFO and the respective ward ofce. Finally, we converted all the monetary value into USD (1 NRS � 0.0086 USD).

Extent and Nature of Human Casualties.
A total of eight incidents occurred during 2015-2019, out of which, half of the cases resulted in death and half were of minor injury ( Table 3). Five of the eight victims were male and three female, with six of the victims being children <9 years old. In half of the eight cases, attacks occurred when the person was by themselves. Tis appeared to be a strong indicator of whether the person was killed or not (3/4 attacks on lone people resulted in death, while 1/4 attacks on people accompanied by friends resulted in death (Table 3).   Table 4). Livestock losses to leopards varied signifcantly among species (χ 2 � 357.78, df � 5, and P value <0.05).

Spatial Pattern.
A relatively high proportion of livestock predation was reported from ward no. 4, with an average of 1.45 livestock heads per household than ward no. 2 (1.13 livestock/household) ( Table 4). According to respondents (victims), more than half of the livestock depredation by leopard (55.63%) was caused inside the livestock shed in the last fve years period (Table 5). Tere were marked distance efects in depredation (χ 2 � 14.11, df � 3, and P � 0.002) with maximum killings of about 72.54% of the livestock occurring at a distance of 0-200 m from the nearest forest, followed by 20-400 m (23.24%), and above 400 m (4.23%) (Figure 2).      (Table 6). Similarly, the highest depredation on livestock was found in July (n � 20), followed by June (n � 16), and the lowest in January (n � 6) ( Figure 3). Although some fuctuations were observed, we could not fnd any signifcant diferences between the frequency of livestock depredation over the months (χ 2 � 16.53, df � 11, and P � 0.12). Te highest livestock depredation events were found to have occurred during the daytime (42.96%) followed by nighttime (32.39%) ( Figure 4). Likewise, there was a signifcant efect of time of day on the frequency of attacks (χ 2 � 48.47, df � 3, and P ≤ 0.001).  (Table 7). Te economic loss was higher near the forest area than in the area far from the forest. Te mean economic loss of households located within 200 m of the forest (USD 91) was higher in comparison to those far away from the forest (Table 8). About half a percentage of economic losses occurred in corral, followed by forest (23.03%) and farmland (23.03%) ( Table 9).

Discussion
Leopard is one of the most feared species that causes the highest number of human mortalities in Gandaki Province. Our study reveals that children were the most common victims of human casualties in the study area, comprising 75% of all cases. Similar results have been reported by Athreya et al. [9] in Maharashtra of India. While attacking human, leopard shows a waiting nature and whenever it gets the best opportunity, it attacks and kills human [52]. Tus, the cases of human attacks when alone were more than those accompanied by friends in our study area. In context to the wards, the proportion of livestock depredation reported from ward no. 4 was relatively high with an average of 1.45 livestock heads per household than ward no. 2 which has more livestock. Tis shows high susceptibility of ward no. 4 to leopard attacks than ward 2 and the reason could be the higher coverage of the nearby community forest which might have supported the larger population of leopard; therefore, the frequency of attacks might have been increased. However, a detailed feld survey using standard scientifc protocol should be performed to prove this assumption. After the successful drive of the community forestry program in the mid-hills of Nepal, it can be argued that leopards have increased in these habitats [11]. Most forest patches of the mid-hills are close to human settlements [53], and studies have shown that leopards can live in human-dominated landscapes [10,54], and humanleopard confict is mostly induced by the depletion of natural prey populations, the lack of water and poor livestock herding and guarding practices [10].
Although chickens were the major livestock category comprising nearly 75% of total livestock, leopard attacked majorly on goats which only shared one-fourth portion of the total livestock categories. Tis event alone comprises 89.42% of the total loss which is similar to the fndings of Rahalkar [55] who also accounted for 80% of depredation in goats as a result of the leopard attack. Tis may be due to the preference for smaller-sized hoofed animals such as goats, sheep, and calves by leopards [56]. Goat belongs to the preferred prey size range (25-50 kg) for leopard as a result they can be killed and dragged to a safe place easily [10,35,57]. Further, goats in our study were typically freeranging, and attacks on free-ranging livestock by large carnivores are more common [58,59]. Tis is similar to the fndings of Kabir et al. [10] who have reported goats as the most depredated livestock species even when it was the second most abundant livestock. Research carried out in the Pokhara Valley by Adhikari et al. [60] and the bufer zone of Chitwan National Park by Lamichhane et al. [13] and Dhungana et al. [34] also refected goats as the major depredated livestock. Over the 5 years (2015-2019) in this study, there was some evidence that human-leopard confict might be increasing with the highest level being recorded in 2019. However, without additional data in the time series, it is impossible to be sure if this refects a real trend or a singleyear efect restricted to 2019. Moreover, particular historical events such as huge hail storms and post-earthquake were also taken as references for capturing the information on real incidents of leopard attacks in a particular year through the memory recall method [35]. And also, this increment might be due to the greater resilience and adaptability of leopards in comparison to other carnivores which allow them to survive in human-dominated landscapes [10]. Koirala et al. [35] and Ghimirey [31] also reported the increasing damage trend by leopard in the ACA even after the establishment of the Annapurna Conservation Area Project (ACAP) ofce. Likewise, studies of human-carnivore confict conducted in Bhutan and Pakistan also reported leopards to be the main predator of livestock [57,61].
In addition, the highest depredation on livestock by leopard in terms of the month was found in July, followed by June and November. In another study by Lamichhane et al. [13] in the bufer zone of Chitwan National Park, the highest livestock depredation by leopard was reported in December, followed by June and November. During the summer months, people make their movement in the forest and pasture land for livestock grazing, where they lack wellstructured shelters for their livestock, and also, they graze their livestock there for a long period making them more vulnerable to leopards [10]. Tus, leopard kills on livestock become maximum during the daytime in our study area. Similar to our study, Dar et al. [57] in Machiara National Park also found a majority of livestock depredation due to leopards during the summer months. It is because, in the winter months, the leopard has enough prey species in its natural habitat due to less disturbance by local villagers, so, they might not enter into a human settlement in search of food. Similarly, Sijapati et al. [62]; reported the highest livestock depredations during summer and spring in the eastern and western areas of Bardiya National Park. It is because livestock was kept inside small-sized low-wall corals during winter to shelter them from cold weather. Tis constrained their movements and further accelerated the likelihood of livestock depredation when confronted by leopards. However, another research by Consolee et al. [63] in the natural reserve of China showed the highest livestock depredation by leopards during spring, followed by summer, autumn, and winter, respectively. Tis is likely because, in spring, the ice melts favoring the growth of grass; as a result, the livestock roams around the reserve looking for green pasture. In the same way, the resident ungulates also disperse to pastoral lands outside protected areas where most livestock are found due to actively growing grasses, and together with reduced wild prey base and large home ranges, predators are turned towards livestock. In contrast to our study, Koirala et al. [35] reported the highest depredation during the winter months followed by spring and summer in ACA of Nepal. Tis is because people along with their livestock migrate from high altitudes to the lower region to avoid the extreme cold and on the way; they get attacked by the  Total  2015  21  0  0  0  1  22  2016  10  1  0  0  0  11  2017  13  1  0  0  0  14  2018  22  1  0  5  0  28  2019  52  3  0  10  2  67  Total  118  6  0  15  3  International Journal of Zoology leopard, particularly in the pastures where livestock are often left unguarded. Te carnivores are assumed to kill ungulates more frequently in winter due to harsh environmental conditions which cause ungulates to congregate [64,65].
Most of the daytime attacks occurred in the grazing land when the livestock are left unattended to graze while their owners engaged with other work. However, Ahmed et al. [66] found nighttime to be the best for the leopard to attack livestock. Te study further showed depredation of more than half of the livestock inside the livestock shed which suggests the need for better husbandry practices with predator-proof livestock corrals. Te most killing of     [63] in Tieqiaoshan Provincial Nature Reserve of China has estimated the mean annual economic loss to range from USD 706.58-1413.17 per person. Tis loss is comparatively higher as compared to our study as the primary source of income of local people in and around this reserve was cattle keeping, accounting for about 57.1% of total income. Another study by Koirala et al. [35] in the ACAP area reported an economic loss of about USD 130 from the leopard. Further, in Trans-Himalaya of Ladakh, Namgail et al. [67] estimated an annual monetary loss of USD 190/household/year from livestock depredation, particularly of valuable livestock such as yak and horse, by Tibetan wolf, snow leopard, and lynx. Distance from forest or wildlife habitat is considered an important determinant of predator attacks on livestock [68] and the intensity of the economic loss [12]. Our results showed that the majority of the leopard attack incidents were within 0-200 m distance from the forest edge, with an average economic loss of 59 USD. Results from Bardiya National Park and Waza National Park, Cameroon also reported that the probability of an attack increases signifcantly from a closer distance to a core area [69,70], which supports our study. In addition, Adhikari et al. [12] reported a signifcant amount of economic loss near forest edge (0-500 m) due to livestock depredation. In contract, our study showed that the economic loss is signifcantly high when the distance to the forest edge is >400 m; the reason could be the predation of only goat species which has high monetary value in response to other distance categories where livestock with low monetary value was also killed. Our result also showed that the majority of the livestock attack incidents were in corals, with an estimate of monetary loss equivalent to 3796 USD, followed by farm land. Tis fnding is similar to that reported by Upadhyaya et al. [68], who also accounted for the most livestock attacks in corals.
Tis study refects the scenario of human-leopard confict that has been rising in recent years by putting humans and livestock at risk. Tis problem is common in other areas of the country, and no efective interventions have solved the situation yet. Although this is one of the case studies of the human-leopard confict from the mid-hill region of Nepal, the fndings are applicable for the concerned government authorities and managers to address the issues of leopard attacks on humans and livestock in other regions of the country. However, a detailed long-term study including several years and large spatial coverage is needed for predicting the confict hotspots (risk zones) across the hilly region of the country. Further, a detailed scientifc study on the prey abundance and other habitat parameters of the leopard, including its behavioral ecology is required to supplement lacking information and promote efective management of human-leopard confict.

Conclusion
Based on our results, we can conclude that leopards primarily attacked goats, although chickens were the major livestock. Te frequency of attacks on livestock was high in the daytime and summer months. Also, the attack on the human during being alone was higher than being with friends. Although the risk of wildlife attacks may not be eliminated, eforts must be directed toward minimizing them to a level that people can accept. Hence, we suggest making loud sounds, displaying fashing lights during the night, and constructing predatory-proof corals to reduce attacks on both humans and livestock. Furthermore, people should be made aware through sensitization programs about the opportunistic attacking behavior of the leopard; to prevent them from going to the forest alone as well as not leaving their children unattended outside the houses during the evening.

Data Availability
Te data are made available in the data repositories https:// www.re3data.org.

Conflicts of Interest
Te authors declare that there are no conficts of interest. International Journal of Zoology 9