Assessing spatio-temporal patterns of human-leopard interactions based on media reports in northwestern India

: Large carnivores in human-use areas make for sensational print media content. We used media reports to examine human-leopard interactions in Rajasthan, India. We extracted news reports on leopard-related incidents from January 2016 to December 2018. Incidents (n= 338) were categorized, mapped


INTRODUCTION
Conserving wildlife while simultaneously balancing human welfare presents several challenges (Peterson et al. 2010), especially when large, potentially dangerous wildlife share spaces with humans (Treves & Karanth 2003). Besides ecological considerations, conserving such wildlife species requires an understanding of social contexts such as local politics, culture, religion, and ethnic attributes (Dickman 2010;Chapron & López-Bao 2014). Large carnivores are generally wide-ranging, and many species live in human-dominated landscapes (Ripple et al. 2014). Predators like Coyotes Canis latrans in Canada, Tibetan Black Bears Ursus thibetanus in Japan, Grey Wolves Canis lupus in parts of North America & Europe, Jaguars Panthera onca & Pumas Puma concolor in South America, and several carnivore species in Africa have been documented to share spaces with humans (Vynne et al. 2011;Alexander & Quinn 2012;Sakurai et al. 2013).
'Shared spaces' by their very nature entail some amount of negative interactions between humans and carnivores (Peterson et al. 2010). When these negative interactions intensify, they transform and perpetuate as negative attitudes, especially when people perceive that state agencies or managers prioritise wildlife over human safety and wellbeing (Madden 2004). Historically, conservation policies across countries and regions have approached the issue by considering carnivores and their habitats, and humans to be disparate entities with little to no overlap (Athreya et al. 2013;Chapron et al. 2014). But over time, studies have quelled this notion and provided empirical evidence for an evolving concept of human-carnivore associations that demonstrate the fluid nature of the dimensions they share (Ghosal et al. 2013;Athreya et al. 2015;Dhee et al. 2019).
Three of the four large felid species in India-the Asiatic Lion Panthera leo persica, Tiger Panthera tigris, and the Common Leopard Panthera pardus-share space with high densities of humans and livestock (Banerjee et al. 2013;Joshi et al. 2013;Athreya et al. 2015). Among them, leopards show the highest overlap with humanuse areas in the country (Karanth et al. 2013;Athreya et al. 2015;Kshettry et al. 2017). Leopards are highly adaptable, wide-ranging predators occurring in a variety of landscapes, from forested protected areas to densely human-populated urban centres (Jacobson et al. 2016). Their body size, hunting strategies and highly eclectic dietary preferences allow them to thrive on a broad spectrum of prey species (wild and domestic). They can potentially live in human-dominated landscapes with relatively low levels of damage to human life and property under certain conditions (Athreya et al. 2013(Athreya et al. , 2016Kshettry et al. 2018;Puri et al. 2020). Their populations, however, continue to dwindle due to habitat loss, direct persecution or illegal trade of body parts across the distribution range; the IUCN Red List therefore categorizes them as Vulnerable (Stein et al. 2011). A combination of (a) limited knowledge about their interactions with humans, (b) absence of viable mitigation strategies to counter negative interactions in shared landscapes, and (c) consequences from damages caused to people and livestock (Stein et al. 2011) create conservation concerns, necessitating the exploration of approaches to foster human-leopard co-existence in human-dominated landscapes.
Print media holds immense power and clout in creating and sustaining narratives based on how it represents human-leopard interactions in humandominated areas (Crown & Doubleday 2017). Previously, studies have addressed this by assessing the type, nature and tone of content communicated through the media (Bhatia et al. 2013;Hathaway et al. 2017) and asserted that issues related to large carnivores are more 'worthy' of a news report if the incidence represents an overall negative view. This can potentially create a bias against wildlife and undermine conservation goals to an extent (Bornatowski et al. 2019). The public who consume such content get an incomplete understanding of the actual gravity of the incident(s) and this can either attenuate or amplify their perception of risk, leading to diminished human acceptance of wildlife (Knopff et al. 2016). Other studies have used media reports to map the distribution of the leopards and patterns of depredation using spatial models, while also assessing social and management factors associated with negative interactions (Athreya et al. 2015). Therefore, media is not only a source of information but can also provide a broader view of human-leopard interactions and can potentially inform conflict mitigation strategies.
In this study, we examined human-leopard interactions reported in the media from the state of Rajasthan in northwestern India. The State's local print media extensively covers leopard-related incidents, and has a considerably wide readership. We used media records to assess spatial and temporal patterns of leopard-related incidents in the State, and then analysed negative interactions in terms of losses or damages faced by leopards and people. Finally, we examined current management strategies that involve leopard captures and translocations, providing an analysis of the locations of physical captures, the reasons behind the captures,

Ethics Statement
The study was implemented using archived reports from print media and some forest department records. The study does not contain personal information of any human subjects. Approval of animal care and use committee and of human ethics committee was not required.

Study area
The State of Rajasthan in northwestern India covers an area of around 350,000 km 2 (10.4% of the country's total geographical area; Fig. 1), and supports a population of 68,900,000 people with a density of ~200 humans per km 2 . The literacy rate in the State stands at 66.11%, with 75% of the population-mostly agricultural and pastoral communities-residing in rural areas (Census of India 2011). The State has 33 districts (administrative units) and 241 sub-districts (locally called 'tehsils'). Major geographic features of the State include the Thar Desert in the west, and the Aravalli Hill range that spans more than 850 km from the south to the northeastern parts. The vegetation in the State is diverse, ranging from desert dunes, scrub, deciduous forests, forest plantations, saline or swampy grasslands with a mosaic of seasonal agriculture belts, urban settlements, and barren lands. The region is largely arid/semi-arid, with low rainfall throughout the year (100-1,000 mm).
The print media in the State frequently covers leopard-related incidents. The types of reports vary from leopard sightings outside protected areas to attacks on humans & livestock and leopard deaths. They also report on the presence of leopards in human settlements, which often leads to capture/rescue and translocation operations. Reports of these incidents are primarily in Hindi, which is the most widely-spoken language in the State. The wide readership of print media, and its role in reporting leopard-related incidents, lend themselves to two considerations from a management perspective: first, it is important to analyse the content that is represented in newspapers; and second, the spatial and temporal patterns or trends of these reported incidents, particularly outside protected areas.

Data collection and processing
We selected Rajasthan Patrika (published in Hindi) as an ideally suited media publication for our study,

Human-leopard interactions in India
Chauhan et al.

J TT
because it has high readership across the State with an average issue readership of 7,586,000 people (Indian Readership Survey 2017) as compared to average issue readership of other regional publications (Punjab Kesari with 3,252,000 and Navbharat Times with 2,327,000 people). All the reports pertaining to leopard-related incidents were collected from the publication's online portal (www.patrika.com/rajasthan-news/), from January 2016 to December 2018. We extracted detailed information on various aspects of human-leopard interactions (see Table 1), including precise geographic location (name of village/town/city, sub district, district), date of event, time of event and type of event, along with associated details about the leopard(s) involved in the incident, and livestock/human victims. Media reports were classified into major thematic categories, multiple reports pertaining to the same incident were combined, and each incident was assigned a unique identification code before being archived (see Supplementary Table 1).

Spatial patterns of leopard-related incidents
Our objective was to use media reports to identify spatial patterns of human-leopard interactions in the State. We spatially mapped the number of media reports by assigning incidents to the corresponding districts and determined geographic coordinates of the village/town/city named in each media report using Google Earth (version 7.3.2). To minimize errors arising from villages with same/similar names, we used a directory of village names (from government records), tracing their locations by sub-district and district names as additional qualifiers. Reports for which we could not reliably assign spatial locations were excluded from the spatial maps; other relevant information from such reports was nonetheless used for examining aspects related to leopards and/or the livestock/human victims. We used geospatial software QGIS (version 3.4.6) to generate spatial maps for each category of leopardrelated incidents as detailed in Table 1.

Characterizing human and livestock attacks
We undertook detailed analyses of specific cases where leopard attacks on livestock/humans were reported. First, we sought to identify temporal patterns (if any) in these attacks. Using date and month information as reported in the newspapers, we broadly classified unique incidents of leopard attacks based on seasons (Summer-March to June, Monsoon-July to October, and Winter-November to February). Next, we analysed information on the specific locations of human attacks, the activity of humans during the attack and the demographics of human victims (sex and age group, i.e., male/female, infant/young/adult; where, infant: <5 years, young: 5-18 years, and adult: >18 years). Similarly, we also created a profile of all the livestock attacked, as presented in individual reports (species/ breed, and young/adult).

Fate of leopards outside protected areas
We examined three aspects related to the fate of leopards outside protected areas in Rajasthan. Given that mortality can severely impact leopard populations, we examined the causes of leopard deaths based on reportage. Second, we examined the reasons for management interventions that involved physical captures of leopards. For these events, specifically, we used data from forest department records of leopard captures and releases in the State to corroborate information reported in the media. Finally, we also recorded the locations where leopard cubs were reported and measured their distance to the nearest protected area. While doing so, we included protected areas from Rajasthan and also from the neighbouring states of Madhya Pradesh and Gujarat, since some locations could have been closer to the protected areas within these states rather than in Rajasthan.

RESULTS
Systematic searches of Rajasthan Patrika e-newspapers, available from the publication's online portal, yielded 306 media reports pertaining to leopardrelated incidents that occurred between January 2016 and December 2018. We recorded a total of 338 leopard-related incidents, after thorough processing and separating out instances where multiple unique incidents were documented in the same report(s). Most incidents were reported within a day's time since the occurrence of the event. We found that 26 of 33 districts in the State of Rajasthan reported leopard-related incidents. Almost all incidents across categories were in the eastern half of the State (Fig. 2), and the highest frequencies of events were clustered in the southern districts (see Table 2). Of the reports related to human attacks (n= 78) during the period of our study, around 50% were in the southern districts of Udaipur, Rajsamand, Dungarpur, Banswara, and Pratapgarh. Udaipur had the highest percentage of reports on human attacks (18%) followed by Rajsamand (14%), Alwar (8%), Banswara (8%), and Jaipur (6%). Human deaths in the period from January 2016 to December 2018 (n= 15) occurred in Rajsamand J TT (6), Alwar (3), Udaipur (2), Pratapgarh (2), and Jaipur (2). For livestock attacks (n= 79), the eastern district of Jaipur had the highest number of reports (28%) followed by the southern districts of Udaipur (14%), Rajsamand (12%), Banswara (7%), and Dungarpur (6%).

Attacks on humans and livestock
The number of human attacks reported was similar in summer and winter months; monsoon months had much fewer attacks. Most attacks on livestock were reported in the summer months, and least in the monsoon months ( Figure 3). The 78 media reports on human attacks involved 120 people (85 male and 35 female). Fifteen of these 120 (5 male and 10 female) resulted in fatalities. Most of the victims attacked (76 of 120) were adults, while 20 were young and one was an infant; the age of the victim was not reported in 23 cases. Except for 12% of attacks on people in forest areas and 12% of attacks for which exact location was unknown/not reported, all other attacks took place in predominantly human-use areas. Attacks in farmlands were the highest, at 38% ( Figure 4). Most (53%) of the victims were attacked while they were engaged in outdoor activities. About 11% and 14% of the people were attacked inside their

J TT
house and during a mob commotion around a leopard, respectively. Around 21% of the attacks did not have any reported details regarding the activity of the victim ( Figure 5).  More than 50% of media reports on livestock attacks pertained to goat and sheep deaths, followed by reports on calves of cows (~23%). A relatively lower number of reported attacks were on adult cows and buffalo calves (12% of reports each).

Leopard mortalities and physical captures
Media reports analysed in this study included 56 cases of leopard deaths (outside protected areas). Considering the subset that included reports where the cause of death was known (63% of total), 52% of deaths were due to vehicular collisions on roads or railway lines. Ten road-collision incidents occurred on national or state highways. The second most common cause was accidental deaths after falling into open wells (20%). Retaliatory killing by people accounted for 11% of deaths, and the other 17% of cases included electrocution, trapping and death following rescue operations. Three cases have been reported of probable death by poaching, with indications like missing nails, teeth, and heads. Around 37% of cases did not report the cause of death, or, indicated that the cause could not be conclusively determined.
We recorded a total of 32 cases of physical capture/ removal of leopards through direct intervention of the forest department personnel. More than 50% of these captures were undertaken following a leopard sighting in human-use areas (indicating leopard presence, but no damage to human life or property). Captures/removals following attacks on humans and livestock occurred in 25% and 19% of the cases, respectively, with one case where the reason for capture was not reported.
The presence of breeding leopards was recorded from 15 sub-districts, where 36 cubs were rescued by the forest department and eight cubs were found dead during the study period. We found that the average distance between the location from where cubs' presence was reported and the nearest protected area was ~39 km (range= 0-104 km).

DISCUSSION
Information on species like leopards is hard to obtain from fully human-dominated landscapes, especially at broad spatial scales within a short duration and in a cost-effective manner (Athreya et al. 2015). We analysed media reports to better understand leopardhuman interactions across Rajasthan. We show that media reports can serve as an important resource to obtain quick yet valuable information on wildlife across J TT a large landscape, where implementing other common field-based survey methods would be economically and logistically infeasible. The reports we analysed over the three-year duration indicated that leopard-related incidents were largely restricted to the eastern zone of the State panning two major areas. The first was in the southern districts of Rajsamand, Udaipur, Dungarpur, Banswara, and borders of Pali. The second area was in the northeastern districts of Jaipur, Alwar, Sikar, and Dausa. Both these areas constituted 50% and 22% of all reported incidents, respectively. This spatial pattern likely reflects habitat types and vegetation cover across the eastern and the northeastern zones, which have higher prey (wild and domestic) availability. The western zone of the State consists of largely open, barren, desert habitats, making it unsuitable for leopards, although, very few records were present from some parts of the western zone.
Telemetry studies of leopards from across their global range suggest that their home ranges could vary from 2 to 600 km 2 (average= 160km 2 , median= 54km 2 ) (Simcharoen et al. 2008;Steyn & Funston 2009;Weilenmann et al. 2010;Grey 2011;Stein et al. 2011;Habib et al. 2014;Rozhnov et al. 2015;Fattebert et al. 2016;Kittle et al. 2017). Considered together with the fact that leopard cubs were found up to 104 km away from the nearest protected area, our assessment provides evidence in favour of resident, breeding leopard populations in human-dominated areas of the State. Studies in Karnataka and Maharashtra have reported similar cases of resident leopard populations in completely human-dominated areas (Athreya et al. 2013(Athreya et al. , 2015. Although most of the leopard-related incidents we report were in the rugged, hilly areas of the eastern region, some leopards were also reported from the relatively open, arid western region, indicating a rather widespread distribution of the species across the State. Analysis of the reportage indicated that 29% (n= 90) was related to only leopard sightings. A total of 70% (n= 38) of the reports on livestock depredation involved attacks on goats and calves, indicating that the leopards are generally pursuing small or medium-sized domestic prey. With respect to human victims, we found that adult males accounted for 75% of all attacks. This may be due to two reasons: first, men generally intervened to 'handle' situations involving leopards in humanuse areas before the arrival of the forest department personnel, and men were also more prone to the attacks in the open ruckus created amidst a mob during rescue operations; second, many attack incidents took place when the victims were grazing their livestock, which is almost always carried out by the men in the family. On the other hand, young children were typically attacked when they were asleep or were left unsupervised. Taken together, 15 of the 120 attacks resulted in the death of the victim, suggesting that loss of human life, although a very grave issue, is not the norm.
There were many reports regarding the death of leopards (n= 56), and among the known causes, 43% were because of road collisions on the state or national highways. Similar results were found in a study in Karnataka (Gubbi et al. 2014). Such linear infrastructures, besides causing direct mortalities, can also impact the gene flow of large carnivores like Tigers and Leopards (Thatte et al. 2019;Jayadevan et al. 2020). There is greater need to mitigate mortality due to linear infrastructure and information of high leopard mortality due to roads/railways from our analysis of media reports shows the same for the state of Rajasthan. Based on the district-wise maps of mortality we present, mitigation measures such as increasing the number of speed breakers, installing road signages of animal crossing/presence and vehicle patrols at the national or state highways may be implemented by management authorities.
Locations with frequent physical captures of leopards somewhat correlated with high numbers of human and livestock attacks. Studies have shown that removal of carnivores may increase conflict due to social

J TT
disruption, and is also ineffective in reducing population densities because new individuals immediately occupy these vacant territories (Linnell et al. 1997). These new individuals would further alleviate conflict situations since they are not familiar with the area (Lindzey et al. 1992;Cooley et al. 2009;Athreya et al. 2010). Around 53% of the captures reported were carried out following a leopard sighting (with no attacks on people or livestock). Such removals, generally executed by the forest department under media pressure, may affect the attitude of local residents and generate more intolerance towards leopards. We argue that removals which do not involve any negative interaction among human and leopards will not help mitigate conflict. Large carnivores typically require large spaces, invariably creating some overlap with human-use areas and it is highly important to accept these dynamics at the policy and management levels (MOEF Guidelines for human-leopard conflict management 2011).

CONCLUSION
Our study provides some key insights on leopard ecology and human-leopard interactions in the state of Rajasthan. A more rigorous, field-based approach in the areas of high conflict could have helped better understand the socio-political and economical aspects, while offering information on ecologically favourable, leopard-friendly landscapes. Despite the limitations in the overall scope because of the methods we chose, our findings still offer a broad understanding of the status of leopards at a state-wide scale. We show that media reports can be used for quick and cost-effective information to assess the spread of leopard incidences across wide spatial scales. Leopards are widely distributed across Rajasthan, and negative interactions between people and leopards, although widespread, are mainly in the eastern part of the State. The maps we present can be useful for prioritizing management efforts in managing leopard populations, conflict situations and channeling monetary resources. Our results could serve as a basis for wildlife managers and the government to initiate a detailed assessment of leopard populations in specific locations, focused on human-use landscapes.
Finally, the media acts as a frontier for disseminating information to the public. Since, print media is the main source of news across the State, it plays a key role in shaping public perception towards wildlife. In our study, it appeared to be portraying a level of conflict that was higher than what data-based analysis revealed. A more responsible and measured reporting of leopard-related incidents by the regional media outlets could go a long way in bettering people's perceptions towards and acceptance of leopards in shared landscapes. Considered together, these multiple, interlinked strategies could be helpful to adequately mitigate negative interactions between humans and leopards, and aid in leopard conservation in the long term.