Impact of human activities on wild ungulates in Nagarjunsagar Srisailam Tiger Reserve, Andhra Pradesh, India

, Ashraf, Deva, and Baiyana for their help in field data collection and the front line staffs of Nagarjunsagar Srisailam Tiger Reserve for accompanying us during field visits. We also thank Anant Pande, Ujjwal Kumar, Neha Awasthi, Muthu Veerapan, Shikha Bisht, and Vishnupriya Kollipakam, for their inputs in analysis and also Ninad Mungi, Swati Saini, Adarsh Kulkarni and Dhruv Jain for GIS inputs. We also thank State Director WWF-India, Hyderabad office for their support during the manuscript preparation and Akbar Sharif for assisting in Telugu translation.


INTRODUCTION
Global biodiversity declines are being driven by the direct and indirect effects of anthropogenic actions (Hooper et al. 2012). India supports an extremely high diversity of wildlife (inside and outside designated PAs); most of these species are found in higher densities here than elsewhere across their range (Srivathsa et al. 2023). Remarkable species richness can be found among herbivores, which are primary consumers at the base of many food chains (Putman 1989). Human activities including fuelwood extraction, fodder collection, cattle grazing, consumption of bush meat, and infrastructure development in natural areas can influence herbivore populations, habitat, behaviour, and relationships negatively (Meyer et al. 2013;Frey et al. 2017). In places where wild animals co-occur with humans and space is limiting, animals may minimize contact with humans by separating themselves in time and/or space (Kronfeld-Schor & Dayan 2003), often at a cost to their fitness. These activity shifts in wild animals have been studied using advanced monitoring tools such as GPS-satellite collars (Berger et al. 2003;Ungar et al. 2005) and camera traps (Edwards et al. 2020).
The time and activity budgets of species under different ecological conditions can provide insights into factors that influence predation, competition, metabolic requirements, and others (Aschoff 1989;Hayward & Hayward 2012;Kasiringua et al. 2017). Camera traps have been used as a tool for animal population estimation (Karanth & Nichols 1998;Rowcliffe & Carbone 2008), inventorying rare and elusive species (O'Brien et al. 2003), monitoring illegal activities (Jenks 2012;Hossain et al. 2016), and studying animal behaviour (Wegge et al. 2004). For species where direct observation is difficult, camera trap data has been used to study animal activity patterns (Rowcliffe & Carbone 2008;Frey et al. 2017). For species that cannot be individually recognized from coat patterns, camera trap-based encounter rates are used to compute a relative abundance index (RAI) that is often correlated with independent density estimates (Carbone et al. 2001;Rovero & Marshall 2009).
Nagarjunsagar Srisailam Tiger Reserve (NSTR) forms part of the Nallamala Hills of the Eastern Ghats in Andhra Pradesh. Despite being the largest tiger reserve in the country (area 3,728 km 2 ; , there is little ecological data available from the reserve (Srinivasulu 2001). Two forest-dwelling communities, the Lambadas and Chenchus, inhabit the core area of the Tiger Reserve. Impacts of humans and their animals on wild ungulates can be due to: 1) direct hunting, 2) hunting by free-ranging dogs, 3) competition with livestock, and 4) disturbance/competition caused by extraction of forest produce. These impacts may influence the demography of ungulates (decreased abundance and slow growth rates) changes in habitat use, and behavioural changes in time-activity patterns to avoid human activity periods (Madhusudan & Karanth 2002;Karanth et al. 2009;Dave & Jhala 2011;Ohashi et al. 2013;Ritchie at al. 2013).
Due to human-related activities, the animal density in NSTR seems to be low (Srinivasulu 2001). Yet, earlier studies from this site indicates that ungulate sightings were common in the early morning hours close to waterbodies (Bhargav et al. 2009). But due to livestock grazing and hunting pressure the detection of prey was very low and hence proper density estimates were not obtained (Bhargav et al. 2009;Jhala et al. 2011Jhala et al. , 2020. Due to the presence of armed militant groups in NSTR until recently, few studies could be conducted and therefore information on ungulate densities in this area were lacking. The objective of tiger reserves in India is to use the charismatic tiger as an umbrella species to protect ecosystems. A demographically viable tiger population requires space for a minimum of 20 breeding female tigers (Chapron et al. 2008;Bisht et al. 2019) which translates to an area of about 1,000 km 2 with an average of 50 km 2 as a female breeding territory in Indian forests. This area should support ~450 medium sized ungulates per tiger, and the minimum requirement for a breeding population of tigers is around 34,000 (Jhala et al. 2021). The All India Tiger Estimation Report (Jhala et al. 2020) reports that there were 38 unique tigers captured in the study area resulting in a density estimate of 0.91 tigers per 100 km 2 (SE ± 0.14) and due to low prey sighting on transects the prey density was not estimated (Jhala et al. 2020). NSTR is the only tiger reserve in the state of Andhra Pradesh that has a reasonable number of tigers, and when combined with the tiger reserve of Amrabad in the state of Telangana can potentially accommodate more tigers in the future.
High-density tiger populations and humans do not mix well. To create space for a source population of tigers while providing better livelihoods for forestdwelling people, a scheme of incentivized voluntary relocation of human settlements from the core areas of tiger reserves is implemented by the National Tiger Conservation Authority (Jhala et al. 2021). The relocation incentive scheme (currently INR 15,00,000 or ~ 20,000 US$ per adult) was not applicable to the tribal communities of Lambadas and Chenchus since their J TT presence in NSTR was not considered to be detrimental for tiger conservation due to the perception that tribal communities lived in harmony with nature and for the utilitarian reason that they were useful as labour for reserve maintenance and management (E.g., patrolling & protection, habitat management activities, and forest fire management activities) since bringing labour from outside is expensive. Also, owing to presence of armed militant groups, the implementation of human resettlement scheme was difficult as militants depended on local forest dwellers for resources and did not permit them to relocate. Now that militancy in the area has been subdued, the administration can initiate incentivized voluntary relocation of all interior settlements to outside of the tiger reserve for better livelihood options and for creating space for wildlife (Pandey et al. 2013;Jhala et al. 2021).
The present study is a first of its kind in the Eastern Ghats landscape that evaluates relative abundance of wild herbivores, their activity patterns, and their behavioural responses to human-related activities. Our study was constrained by the large size of the protected area and the low abundance of ungulates (Kothari et al. 1995;Karamsi 2010;, making traditional robust approaches like distance sampling impractical due to the large amount of effort required, compounded by low detections of skittish ungulates. Under conditions where ungulates are traditionally hunted, the use of line transect-based distance sampling can be biased, since wild ungulates are extremely vigilant and would likely detect the observer before they can be detected and flee, thus potentially be unavailable for sampling. To understand the ecology of a wild ungulate species, the factors that influence the dynamics of its population or the ecosystem it represents are crucial. Our a priori hypotheses were that ungulate abundances would be lower in areas of high human use, and that ungulates would adjust their activity to avoid periods of high human activity. With this ecological understanding in mind, our study aims to: a) estimate the relative abundance of wild ungulates in the park using camera traps and b) quantify the impact of human activities on the abundance & behaviour of wild ungulates. This study would help us to better understand the low densities and slow recovery of ungulate populations in NSTR and provide recommendations for management interventions.

Study Area
NSTR is the largest tiger reserve in the country (3,728 km 2 ), demarcated as core and buffer administrative units of 2,444 km 2 and 1,284 km 2 , respectively. It is located in the southern Eastern Ghats (15.88333-16.71666 N, 78.50000-79.46666 E) in the state of Andhra Pradesh. Our study area covered 2,500 km 2 within two administrative units, namely, Markapur and Atmakur divisions, including the extended Tiger Reserve core area constituted by Gundala Brahmeswaram Wildlife Sanctuary (GBM), Velgode, and Bairlutty ranges (Image 1).
The terrain of NSTR can broadly be classified as hills, plateaus, valleys, gorges, and escarpments. The vegetation type is southern tropical dry deciduous, tropical moist deciduous, and tropical thorn forests (Champion & Seth 1968). Forest contributed to (84%) of land cover in the study area which is mostly deciduous and scrub/degraded forest followed by agricultural land (1%), waste land (12%), water bodies (2%), and built up (1%). In total, forest covers 84% of the study area. These data were calculated using Arc GIS (v.10.1) (ESRI 2011).
The major portion of rainfall is received from the south-west monsoon that commences from the second half of June and continues up to the first week of October. Then there is a short dry spell for a month. The northeast monsoon is active from November to the first half of December, mainly on the eastern slopes of Nallamala Hills. The mean annual rainfall ranges from 590-760 mm (Jhala et al. 2020 (Pandey et al. 2013).
The study area encompasses 15 major villages that were home to two scheduled tribes (Subramanyachary 2013), the Chenchus and Lambadas, with few other scheduled castes and their livestock, mainly composed of cattle, buffalos, and goats & sheep. Location of human settlements is mostly determined by proximity to perennial water and productive flat lands, which are also prime habitat for wildlife (pers. obs.).
Andhra Pradesh is home to 12 primitive tribal groups (PTGs), with Chenchu being one of the PTGs recognized J TT by the Indian government. Later in 2006, the Indian government proposed renaming the primitive tribal group as primitive and vulnerable for 75 tribal groups in India based on their dependency on hunting, gathering food from the forest, growth of their population, and literacy level. The purpose of this classification was to provide assistance so as to uplift the tribal community in different sectors like education, health, livelihood, skilled labour, agriculture, housing, while retaining their culture (Ministry of Tribal Affairs 2015). These communities are mostly confined to the foothills and low-lying areas of Nallamalla Hills covering Prakasam, Kurnool, Mahaboobnagar, Rangareddy, Guntur Nalgonda districts of both Andhra Pradesh and Telangana states (Raju et al. 2009).
Historically, Chenchus were nomadic hunters and food gatherers inhabiting forested areas, where they ate honey and tubers, and hunted wildlife for food (Murty 1981). Most Chenchus now live in permanent settlements called gudem or pentas, which are a cluster of huts made from bamboo and grass, however, they continue to engage in collecting honey, grass, fruits, nuts, and leaves as supplements to their livelihood (Suryakumari et al. 2008). Chenchus still carry traditional bows and arrows when they move inside the forest that can be used for hunting. Lambada tribes are called by different names, such as Sugalis and Banjaras in other parts (Lal 2015). These tribes spread across Andhra Pradesh and Telangana states in southern India (Vaditya 2019). They live in exclusive settlements of their own called 'Thandas' (Shankar 2016). Present day occupation of majority of Lambadas in general is cultivation and pastoralism (Karamsi 2010).
Inside the core area of NSTR, there are around 5,650 households, with a total population of more than 25,000 people and 2,977 cattle, while another 69 villages with 1,26,000 cattle are present in the buffer zone of the tiger

J TT
reserve (Bhargav et al. 2009;Mathur et al. 2018). The entire tribal population within the tiger reserve depends on forest resources for survival, which are shared with wildlife (Srinivasulu 2001;Sudeesh & Sudhakar 2012).

Data Collection Estimation of wild ungulate density
The smallest administrative unit, i.e., forest beats, were used to systematically distribute line transects (n = 142) to survey the study area. The length of each line transect was between 1.5 to 3 km. Each transect was walked once during the early morning (0600-0800 h) between December to February of 2014 and 2016. All sightings of animals, the group size, radial distance to the centre of the group and bearing were manually recorded on a datasheet. Radial distances to animals were measured using a hand-held range finder (Bushnell RX1000). Bearings were recorded using a hand-held compass (Suunto KB 20).

Relative Abundance of Wild ungulates
Sampling using camera traps was done across the study area between January to July 2014 in an area covering 713 km 2 . A total of 345 camera locations were sampled, with a double camera unit (Cuddeback attack 1149, Cuddeback ambush 1194) deployed at each location for about 40 days. Since this exercise's primary objective was to obtain a population estimate of tigers, camera placement was mostly on game trails, dry stream beds, and dirt roads to maximize photo captures of carnivores. However, we believe that the photo capture data on ungulates to address our study's objectives and comparisons with other sites would remain unbiased as placement locations were similarly selected across the study area and in other tiger reserves across India (Jhala et al. 2021). We checked cameras every 3-7 days to download data and check battery status. All photographs were segregated to species, and information on time, date, and coordinates, recorded for each image.
Livestock were not free-ranging in NSTR, but taken out to graze by herders from corrals in each settlement every morning and brought back by dusk. Herders were often accompanied by dogs. Since livestock movement was constrained by the distance they could move from their corrals and from water sources, human, dog, and livestock activity was mostly concentrated within a certain radius from settlements. Cattle, buffalo, goat, and sheep escorted by herders were accompanied by the first author from early morning when they left the corrals to late evening when they returned to their corrals. A hand-held GPS unit was used by the first author to record the daily grazing circuit from villages in the winter and summer of 2014 and 2015. The grazing circuit was mapped using ArcGIS (v. 9.3), the average displacement distance of livestock herds from settlements/villages was computed, and each settlement was buffered by the 95% upper bound of this distance to delineate a zone of high human use. A total count of all livestock in each season was done for each village and cattle shed across NSTR at a time when livestock were corralled to determine the total livestock population.

Data analysis Wild ungulate density estimation
Analysis was done using the conventional distance sampling approach in Program DISTANCE (v. 6.0) (Buckland et al. 2004). Due to low detection of ungulate species in NSTR on line transects we pooled observations from three sampling periods (Jhala et al. 2011 and sampled by first author in 2016) from NSTR and used line transect observation data from seven other sites in the country (Table S1, S2) which have the similar habitat type to NSTR for fitting species detection functions in program DISTANCE to estimate effective strip width. Shape criteria were examined for heaping, and any outliers were right-hand truncated where necessary (Buckland et al. 2004). Three key functions (Half normal and hazard rate all with cosine and Hermite polynomial series adjustment) were considered for analysis. Model selection was evaluated using Akaike's information criteria (AIC), while Kolmogorov-Smirnov statistics were used to assess the goodness of fit of each model (Buckland et al. 2004). Subsequently, this pooled effective strip width was used to obtain year wise density estimates of ungulate species in NSTR.

Relative Abundance of Wild ungulates
Relative abundances of the wild ungulates in the study area were estimated from 2014 camera trap data, as photo capture rates which were computed by summing independent photo-captures of each species and dividing this sum by the camera trap operational days. We defined an independent photo-capture event as follows 1) consecutive photographs of different species or different individuals of the same species; 2) Consecutive photographs of individuals of the same species taken more than 30 minutes apart (O'Brien et al. 2003); and, 3) non-consecutive photos of individuals of the same species.
We used independent photographs of species to calculate relative abundance index (RAI) from camera trap images. RAI was computed as the number of 23152 J TT independent photo captures of a species in 100 trap nights (Carbone et al. 2001). The total effort invested was determined by multiplying total camera operation day. Camera traps were segregated into two strata based on their location as i) within the high human -use areas and ii) those outside this zone as low human impact areas. RAI of ungulates was also computed separately for these two zones. We hypothesised that RAI values of ungulates would be lower in high human impact areas and RAI values of human disturbances (photocaptures of humans, domestic dogs, and livestock) would be higher in high human impact areas.
The RAI was computed for each camera trap location for each species in both high and low human use zones, for testing if RAI differed between high and low human use zones we used non-parametric Mann-Whitney U-test (Zar 2022).
Camera trap-based data collection overcomes biases induced by the skittish nature of wild ungulates which can result in non-availability for sampling on line transects, but unfortunately RAI does not allow for rigorous inference on absolute abundance. To test the hypothesis that RAI is a reliable index of absolute density we regressed the RAI values of Chital and Sambar (species with a reasonable sample size of observations) with absolute density estimates of these species obtained from line transect distance sampling from other similar forest types where absolute density estimates from distance sampling were also available (Jhala et al. 2020). A significant positive relationship between RAI and absolute density would lend support to the hypothesis.

Temporal peak activity pattern
We used camera trap images and their associated information from the metadata of the images like date, time of the photograph to understand the temporal activity of six wild ungulate species in NSTR. The time of the photo capture was used to create a 24-hour activity pattern graph as well as analysis using Oriana software (v. 4.0). Oriana uses circular statistics to enumerate the dispersions such as mean vector length (r) along with confidence intervals. The mean vector has two properties: direction and length of the mean or angle, and the mean vector length (r) denotes the clusters of observation around the mean, which ranges from 0 to 1, where 1 is the frequency of observations very close to the mean and 0 is when observations are scattered across the study. In the rose plot the arc on the outer edge extending to either side of the mean represents the 95% confidence limits Oriana software (v.4). The output provided activity clustering along with mean peak activity time for wild ungulates and human related activities factors within a 24-hour cycle, facilitating a quantitative statistical comparison of their temporal activity.

Activity pattern and temporal overlap
We estimated the proportion of time active and activity pattern of ungulates across the day from camera trap data using the Activity package (v1.3.1) (Rowcliffe 2022) in Program R (v. 1.4). This provided information on how much time an ungulate species remains active in a day while the activity pattern describes the distribution of activity across the 24-hour period. Analysis of data was done separately for the two human impact strata. We hypothesised that ungulates in high human impact zones would alter their active behaviour and activity to avoid peak human associated activity periods (human, dog, and livestock activity peaks). Temporal overlap of ungulate activity with anthropogenic disturbances using different packages like Overlap (v. 0.3.3) (Ridout & Linkie 2009) and ggplot2 (v 3.3.3) in Program R (v 1.4.) software was estimated. We used the overlap coefficient (Δ), ranging from 0 -no overlap to 1 -complete overlap (Ridout & Linkie 2009) to estimate the overlap for each wild ungulate species in both high and low human-use areas with human related activities. Since samples used for overlap analysis were more than 75 independent photo-captures for most of the wild ungulate species in both high and low human impact areas we used D-hat 4 estimator for all species (Ridout & Linkie 2009).

Livestock Population and Grazing radius
The total livestock population in NSTR was 4,403 in summer and 3,934 in winter. The livestock population comprised of 44.5% goats, 31.4% cattle, & 24.0% buffalo during summer and 35.8% goats, 35.4% cattle, & 28.8% buffalo during winter. Average livestock grazing circuit was 4.0 (SE ± 0.12) km. Livestock ranged more in summer 4.6 (SE ± 0.22) km than in winter 3.5 (SE ± 0.23) km. The average foraging radius combined for both seasons was 1.8 (SE ± 0.07) km. The 99% upper bound on the foraging radius was 2.01 km. Camera traps within a buffer of this maximum foraging radius (2.01 km) around each human settlement / cattle shed were considered to be within high human activity zone (Image 1).

Relative Abundance Indices
We obtained 35,306 usable photographs with an effort of 10,681 trap nights. Humans were photocaptured the most (Table 2). Wild ungulates constituted 37% of this data. The highest number of captures were of Sambar (38%) followed by Chital (26%), Wild Boar (18%), Chousingha (9%), Mouse Deer (5%), and Nilgai (4%). RAI was highest for Wild Boar (10.0), while it was lowest was for Nilgai (1.5) ( Table 2). Human impact was recorded throughout NSTR (in the form of human, livestock, and domestic dog photo-captures), and was similar across the reserve for humans and domestic dogs since RAIs of humans and domestic dogs were not significantly different near settlements and away from settlements (Table 2, Figure S1). Livestock RAI was significantly higher in the proximity of settlements ( Figure S1). Amongst wild ungulates only Chousingha and Nilgai had significantly higher RAI in low human use areas while Wild Boars had significantly higher RAI in high human use areas (Table 2, Figure S1).

Temporal Activity Patterns
All wild ungulates except Chousingha showed bimodal activity. Chousingha were diurnal, Chital and Nilgai were

Species
Observations Model

J TT
crepuscular and diurnal, Sambar and Mouse Deer were primarily nocturnal, while Wild Boars showed activity at night and in the forenoon ( Figure S3). All human related activity (humans, domestic dogs, and livestock) were diurnal, beginning late mornings and extending into late evening (Figure 1, Figure S3). In agreement with our a priori hypothesis, within the constraints of some ungulates being diurnal, wild ungulates avoided all forms of human activities (Figure 1; Figure S3). The 95% confidence intervals of wild ungulate activities (except Chital) did not overlap any of the human related activities (humans, livestock, and domestic dogs; Figure S3). Chital activity in low human use areas overlapped only with the 95% confidence intervals of livestock active periods (Table 3, Figure S3). Overlap of ungulate activity with anthropogenic activities within the high-human impact zone was found to be higher for Chousingha (63%) and the lowest for Sambar (15%) ( Figure S4). For species like Chital, Chousingha, and Nilgai in the low human impact zone, the overlap with various anthropogenic disturbance factors, such as humans, dogs, and livestock activities combined, was found to be more than (60%) for Chital and least for Mouse Deer (18%), respectively ( Figure 1).

DISCUSSION
In the Anthropocene, exclusive space for biodiversity is one of the most limiting factors for conservation (Kipkeu 2014). Many protected areas set aside for wildlife conservation have people residing within them (Kothari et al. 1995). NSTR has 15 villages with human population of 5,650 families with a population of ~18,000 (Lal 2015), and a livestock population of ~4,500 within the tiger reserve. In addition to the resident settlements, NSTR is

J TT
also used by humans and their livestock from peripheral villages (Image 1). Human and livestock photo-captures outnumber all other species in NSTR (Table 2), which should be reason for concern.
Human-related activities contributed 63% of total independent photo-captures. Photo-captures of humans were high, followed by livestock and domestic dogs which were all primarily restricted to daylight hours. However, except for livestock, the presence of humans and domestic dogs was recorded across the protected area, suggestive of high impacts of human activities within NSTR and not limited to near the settlements. Though we found that wild ungulates avoided active periods of humans we found no statistical differences in relative abundance or activity for most wild ungulates between areas closer to human settlements (high human use) and further from settlements (low human use) suggesting a pernicious impact of humans across NSTR.
In a comparative scenario, in Kuno Wildlife Sanctuary with similar dry deciduous forest like NSTR, the Chital population was found to recover from a density of 4.6 km -2 (Banerjee 2005) to ~40 km -2  after human habitations were relocated from Kuno and protection from poaching enhanced. Subsequent to the data collection for this study, we obtained photo-capture of local communities indulging in hunting activities (Image S1). Wild Boar abundance was not in agreement with our a priori predictions since they had higher abundance in high human use areas. This was likely since human habitations were located in flatter productive terrain (Image 1) which is also the only habitat for Chital and Wild Boars as these species tend to avoid hilly areas.
Earlier studies on temporal activity of sambar in India (Schaller 1967;Shea et al. 1990;Lahkar et al. 2020) and on greater mousedeer from Borneo (Ross et al. 2013) report these species to be nocturnal. Our data confirm these inherent biological patterns of Sambar and Mouse Deer becoming active well after darkness and continuing their activity into dawn hours (Lahkar et al. 2020). By being nocturnal, both species avoid periods of high human activities. However, in human-free areas, Sambar are recorded to show activity during daylight hours as well (Griffiths & Schaik 1993). Though there were no totally human free areas in NSTR, we did observe greater daytime activity for Sambar in low human impact zone (Figure 1).
More significantly, the overall active period of Sambar reduced in high human use areas (Table 2), thereby reducing the duration available for foraging and other vital activities. At NSTR, chital were reported to be widely dispersed and to form small herds (Srinivasulu  Table 3. Temporal activity pattern of wild ungulates, livestock, domestic dogs, and humans in Nagarjun Sagar Srisailam Tiger Reserve. Mean vector length (r) denotes the clusters of observation around the mean, which ranges from 0 and 1 where 1 is the frequency of observations very close to the mean and 0 is when observations are scattered. The 95% confidence limit of the mean, overlap between HDZ (high human use zone) and LDZ (low human use zone) 95% CI signifies no statistical shift in peak activity between the two zones.

J TT
2001), which contrasts with our observations, since we found all ungulates to be very skittish, Chital in particular, were at very low densities and mostly observed as solitary or in very small groups, a major deviation from observations in other protected areas where chital tend to be the most abundant wild ungulate, often occurring in large herds. Chital have been reported as being diurnal with a bimodal activity at dawn and dusk (Schaller 1967) our results conform to this pattern. A shift in the activity of Chital (though not statistically significant) was observed between high and low humanuse areas of NSTR (Figure 1), with the evening peak of activity being less pronounced and more spread out into the late evening and early night in high human-use areas. High and low human-use areas actually differed only in terms of livestock use, with human and domestic dog usage being recorded across NSTR with no statistical difference across zones.
Livestock is sympatric with wild ungulates in most forested areas of India (Kothari et al. 1989), where they potentially compete for essential resources like food and water. Even though livestock grazes Indian forests to varying extents, their impact on wild native ungulates is less understood (Madhusudan 2004). Understanding the interaction between wild ungulates and livestock is complex and varied under different ecological conditions (Sankar 1994;Dave & Jhala 2011). Though we segregate our camera traps into high and low human impact zones we caution that human activity was recorded across NSTR and therefore we find little differences between low and high human impact zones in terms of timing of activity as well as active duration, these differences would likely have been more pronounced if compared between total human impact free areas and human use areas.
More importantly, our data show that all ungulates across NSTR avoided time periods having high human activities. Often diseases like foot and mouth can get transmitted between livestock and sambar (Johnsingh & Manjrekar 2015). NSTR has a large resident cattle population and during the monsoon an additional large number of cattle migrate from nearby villages to graze (Bhargav et al. 2009).
Presence of domestic dogs in protected areas shifts wildlife temporally or permanently from the available space they have (Banks & Bryant 2007). Our results show that domestic dogs were very active (41%) in high human use areas and domestic dogs usually accompanied humans (Table 2). Domestic dogs have been traditionally used by forest dwelling communities to hunt bushmeat. Even the odour of dog urine or faeces can trigger wild animals to avoid an area (Hennings 2016). Since domestic dogs occur at densities higher than natural predators, the frequency of attacks on wild prey species is also likely high, especially in and around protected areas (Ritchie et al. 2013).
We found that free-ranging dogs often accompany tribesmen armed with bow and arrows who move around unhindered inside the protected area on the pretext of collecting non-timber forest products. While conducting fieldwork AK witnessed incidents where dogs accompanied by local tribal communities chased Chital. Temporal activity pattern revealed that activity of dog's overlap more than 60% of the activity of Chital, Nilgai, and Chousingha. These ungulates being diurnal are limited in their ability to change their activity to avoid dog activity periods (that are only diurnal). Thus human impacts and predation through dogs would affect these diurnal species the most. Domestic dogs were often used for hunting wildlife by local tribal communities and their impact were likely significant in depressing ungulate densities as also reported (Madhusudan & Karanth 2002).
Many wildlife species face extinction because of human impacts; therefore, a prevailing belief is that many species cannot co-exist with people (Carter et al. 2012). Any human-related activity can disturb wildlife; one such significant depressant is hunting. Carnivore assemblages may be affected by direct poaching or through poaching of their prey. Diverse methods, including domestic dogs, bow and arrows, traps, and smoking of fossorial mammals, were traditionally used for hunting (Datta & Naniwadekar 2019). It is recognized that continued overhunting lowers animal densities and subsequently leads to local, regional, and overall species extinction (Diamond 1989;Rabinowitz 1995). A study from Nagarahole Tiger Reserve mentions that 78% of local communities interviewed preferred to hunt Mouse Deer by using domestic dogs (Madhusudan & Karanth 2002).
In NSTR mousedeer has the least overlap with domestic dog activity ( Figure S5) possibly to avoid predation. Hunting also changes the behaviour of wildlife as seen in Sika Deer in Bialowieza National Park where they became more diurnal once the park management restricted tourism and hunting (Kamler et al. 2007). Hunting influenced Wild Boar activity patterns where it was more diurnal during the non-hunting season in central Japan (Ohashi et al. 2013). The NSTR management acknowledges that the resident Chenchu tribals, who always carry a bow and arrows and are accompanied by domestic dogs whenever they move inside the forest, do hunt birds and monitor lizards (Pandey et al. 2013). The Lambada tribe, are reported to occasionally hunt small mammals during festive season (Bhargav et al. 2009). Despite the fact that we were unable to quantify ungulate poaching as a cause of their low densities, based on observations and camera trap photographs of such actions, poaching, combined with high livestock densities, and domestic dog related stress was most likely to be responsible for NSTR's low wild ungulate densities.

CONCLUSIONS
Our findings suggest that RAI estimates can help index abundance and can be used to estimate trends in wild ungulate populations. Our data and inferences show that impacts of human activities alter wild ungulate abundance and behaviour, as also demonstrated previously (Gaynor et al. 2018) The tropical dry deciduous forests are among the most impacted habitats by anthropogenic activities and are vulnerable to degradation (Chundawat et al. 1999). The forests near human settlements were more disturbed than those away from settlements. In the short-term, we recommend active removal of free-ranging dogs, control of poaching, and minimizing livestock grazing, for wildlife population revival.
Most forest dwellers prefer to relocate when given a genuine opportunity, since living within protected areas is difficult due to limited access to basic amenities like electricity, roads, health care, education, and markets. While within protected areas, their crops are raided by wild ungulates, and large carnivores often kill their livestock and sometimes humans (Madhusudan & Mishra 2003;Chapron et al. 2008). However, people rights activists argue that human resettlement from protected areas is unethical and is not required since forestdwelling communities live in harmony with nature and forest resource use by them is sustainable (Rangarajan & Shahabuddin 2006;Dattatri 2010). In certain instances, relocation results in transformation of the 'way of living' since relocation usually results in changing nomadic hunter-gatherer or pastoral communities to a more settled livelihood based on agriculture or labour.
Several communities such as Gujjars in Uttarakhand, Sahariyas in Madhya Pradesh, and Maldharis in Gujarat face a challenging transition that is often difficult to make (Rangarajan & Shahabuddin 2006). In line with this argument, the forest-dwelling tribes of NSTR (Chenchus and Lambadas) have not been offered the NTCA incentive of voluntary relocation. Thus, without any genuine feasible option to move out of the core area of NSTR, human settlements continue to grow within the tiger reserve, and their impact on forest resources remains unabated and increasing with time. To achieve the conservation objectives of the tiger reserve, i.e., to establish a long-term viable population of tigers that act as a flagship and umbrella species for the conservation of the ecosystem, higher abundance of wild ungulates is required, for this it seems important to mitigate the current human impacts in NSTR.
We propose that the incentivized voluntary relocation package of INR 1.5 million per adult (~ USD 20,000) (NTCA 2021) be made available to the forestdwelling communities of NSTR. This would open an option for potentially better livelihoods and lifestyles to these people outside of the tiger reserve and benefit both people and wildlife simultaneously. Future studies should be carried out by camera trap based monitoring each year, keeping the present study as a baseline, to understand the status and trends of carnivore and herbivore abundance after human impacts are reduced/removed within NSTR. Such monitoring should conclusively prove the depressant effects of humans on wildlife and document the recovery of the wild ungulate populations (Anonymous 2009). Figure S6. Scatter plot and correlation between distance sampling based density estimates for Sambar with relative abundance index (RAI) from camera trap data. Image S1. Camera trap image of hunting by local communities in this landscape.