Long‐distance dispersal by a male sub‐adult tiger in a human‐dominated landscape

Abstract Conservation of wide‐ranging species and their movement is a major challenge in an increasingly fragmented world. Long‐distance movement, such as dispersal, is a key factor for the persistence of population, enabling the movement of animals within and between populations. Here, we describe one of the longest dispersal journeys by a sub‐adult male tiger (Panthera tigris) through GPS telemetry in Central India. We analyzed movement metrics, directionality, and space use during three behavioral stages of dispersal. We also used the clustering method to identify resting and kill sites (n = 89). T1‐C1 dispersed a straight‐line distance of 315 km over 225 days, moving an average of 8.38 km/day and covering a cumulative displacement of ~3000 km. Movement rate during post‐dispersal was faster (mean = 0.47 km/h) than during dispersal (mean = 0.38 km/h) and pre‐dispersal (mean = 0.13 km/h), respectively. The overall movement rate during the night (0.44 km/h) was significantly faster than during the day (0.21 km/h). Likewise, during dispersal, the movement was faster (mean = 0.52 km/h) at night than day (0.24 km/h). The average size of clusters, signifying resting and kill sites, was 1.68 ha and primarily away from human habitation (mean = 1.89 km). The individual crossed roads faster (mean = 2.00 km/h) than it traveled during other times. During the post‐dispersal phase, T1‐C1 had a space use of 319.48 km2 (95% dBBMM) in the Dnyanganga Wildlife Sanctuary. The dispersal event highlights the long‐distance and multiscale movement behavior in a heterogeneous landscape. Moreover, small forest patches play a key role in maintaining large carnivore connectivity while dispersing through a human‐dominated landscape. Our study underlines how documenting the long‐distance movement and integrating it with modern technology can improve conservation management decisions.


| INTRODUC TI ON
Habitat loss and increased human land use are primary threats to many wildlife species, altering their movement and behavior (Tucker et al., 2018). Animal movement is further reduced by habitat fragmentation, degrading habitat quality and linear features such as roads (Andersen et al., 2017;Crooks et al., 2017;Fahrig, 2003;Jønsson et al., 2016;Shepard et al., 2008). Mammalian carnivores tend to have large home ranges and long dispersal distances, making them vulnerable to landscapes that have been fragmented by anthropogenic changes (Crooks et al., 2011;Dobson et al., 2006;Ripple et al., 2014). Long-distance dispersal is central to several ecological processes (Levin et al., 2003;Nathan, 2003), including gene flow (Trakhtenbrot et al., 2005), colonization of new areas, range shift, and functional connectivity between populations (Clobert et al., 2012;Noss et al., 1996;Weaver et al., 1996). The long-term population persistence in fragmented landscapes may depend on individuals traversing through a human-dominated landscape to reach suitable habitats (Watts et al., 2015).
In India, protected areas (PA) designated as tiger reserves cover 2.21% of the geographical region, and 35% of the total tiger population resides outside these PAs .
The tiger populations are mostly confined to small protected areas because the habitat outside those areas is highly fragmented, which affects their movement within and outside protected areas Thatte et al., 2018). Consequently, it is challenging for tigers to move through a fragmented and human-dominated landscape. Moreover, the long-term survival of the tiger population in a fragmented landscape depends on successful dispersal from one area to another, thus maintaining the connectivity between subpopulations and isolated populations.
Dispersal involves three successive behavioral stages: departure, transience, and settlement (Clobert et al., 2009). Accordingly, the dispersal behavior of sub-adult tigers can be classified into three distinct behavioral phases, that is, pre-dispersal, dispersal, and postdispersal. The pre-dispersal phase is identified by the movement of individuals within their natal area. The dispersal phase involves movement from its area of birth to another habitat, where it reproduces and establishes a new territory (Howard, 2015;Waser & Jones, 1983). Post-dispersal phase is identified by movement in an area having a stable and defined home range over time. The movement behavior during dispersal in a highly interspersed mosaic of forested areas, agriculture fields, and human settlements is rarely documented. Studies on tiger dispersal are limited, and few have described long-distance dispersal primarily based on camera trap data and VHF/GPS telemetry (Sarkar et al., 2021;Singh et al., 2021;Smith, 1993;Wang et al., 2015). Moreover, genetic studies have shown gene flow over long distances (Anuradha Reddy et al., 2016;Gour et al., 2013;Joshi et al., 2013).
The advancement of GPS technology made it possible to gain new insights into tiger dispersal and how animals perceive and navigate through a landscape. Additionally, long-distance dispersal routes might help identify land use and landscape features that provide connectivity between non-contiguous populations (Graves et al., 2007). This study reports the longest dispersal distance recorded by a sub-adult male tiger from the Vidarbha region of Maharashtra, India. We documented movement and space use from the pre-dispersal to the post-dispersal phase and described the chronological event during dispersal. We analyzed the movement data from the GPS collar with environmental covariates (land use and vegetation cover) depicting the landscape through which the individual moved. Our study also quantified the characteristics of clusters (resting/kill) and the effect of linear features, that is, roads, on movement. We conclude by discussing the long-distance dispersal in a human-dominated landscape and its implications for conservation and management.

| Study area
The study was carried out within and outside of protected areas of Eastern Vidarbha Landscape (EVL), Maharashtra, India. This region is a part of the Central Indian Tiger Landscape, dominated by teak (Tectona grandis) and bamboo (Dendrocalamus strictus). The region encompasses an area of approximately 97,320 km 2 , and forest cover accounts for 27.5% of the total area . The landscape is a mosaic of agricultural lands, human settlements and wildlife areas (Habib et al., 2017). Large carnivores in the region include tiger (Panthera tigris), which co-occurs with other species like leopard (Panthera pardus), sloth bear (Melursus ursinus), gaur (Bos gaurus), and several other ungulate species.

| Field methods
We captured and collared sub-adult male tiger T1-C1 on February 25, 2019. The individual was immobilized using a combination of Medetomidine hydrochloride, Ketamine hydrochloride, and Xylazine. The dosages were based on visual observation of body weight (150 kg) and age. The drug was remotely injected using an air-pressurized Dan-Inject projector (Model IM). The radiocollaring was part of a more extensive study in which sub-adult tigers were collared to understand the dispersal patterns, space use, and movement in a human-dominated landscape. Tigers were classified as sub-adult following (Sadhu et al., 2017). Animals were fitted with Iridium GPS radio-collars (Vectronics, Germany) and had a high spatial accuracy of location (±5 m). The collar was programmed to record locations every 1-3 h depending on the dispersal behavior of the individual. During the pre-dispersal phase, when the tiger was within its natal area, we received location every 3-h interval. During dispersal, when it moved out of its natal area, we programmed to receive intensive location every 1-h interval.
During the post-dispersal phase, locations were received every 3-h time interval.

| Analysis
We calculated movement metrics (movement speed and turning angle) within each of three phases: pre-dispersal, dispersal, and post-dispersal. We identified these three phases by calculating the net squared displacement (NSD) in ArcGIS 10.6.1 with ArcMET tool (Wall, 2014). NSD is movement metrics that helps to understand movement behaviors over time (Bunnefeld et al., 2011). Inflection points and increase in the NSD over time help identify movement modes such as migration, dispersal, nomadism, and range residency (Börger & Fryxell, 2012). We identified dispersal movement and age from the peak in the NSD graph, which indicates the commencement of dispersal from its natal area.
We calculated daily distance traveled (sum of displacement in 1 day) and daily displacement (linear distance between the start and end locations for each 24-h period) during all three phases. We considered a 24-h period from the first location of dawn and the first location of the following dawn. We also estimated the mean movement speed (km/h) across the three dispersal phases. To calculate the movement speed, we scaled the step length (Euclidean distance between successive locations) divided by the time the individual took to complete the distance due to the varying interfix intervals (Leblond et al., 2016). The movement parameters were calculated using adehabitatLT (Calenge, 2015) and the animal movement tool (Signer et al., 2019) in the R programming software (R Core Team, 2020). We used the Kruskal-Wallis test to compare the mean movement speed during the pre-dispersal, dispersal, and postdispersal phases as data were not normally distributed. Next, we used the Mann-Whitney U test to compare day and night movement speeds. To estimate the space use in each of the three phases, we applied the dynamic Brownian Bridge Movement Model (dBBMM) to estimate the utilization distribution in the package move (Kranstauber et al., 2012). The dBBMM requires a time-stamped series of animal locations and the estimated telemetry error associated with each location. The dBBMM allows the variance of the Brownian motion (σ2m) to vary along the movement path for user-defined subsets of n locations. We then included these values in a Brownian bridge movement model to estimate 95% utilization distributions (UDs). We considered the 95% utilization distribution as space use in three dispersal phases.
To understand the directionality or orientation of the animal, we calculated the turning angles across the three phases. We estimated the circular mean, from the von Mises distribution of tuning angles (Mardia & Jupp, 1999). The mean turning angle is between −π and π and defines the degree of linearity in an animal movement. A mean turning angle of zero implies a strong persistence in direction, whereas a mean turning angle of −π or π suggests that movement is undirected. We used the package circular (Lund et al., 2017) to calculate the circular mean and test for directionality. All calculations and statistical analysis were carried out in program R 4.0.4 (R Core Team, 2020).
We identified resting and kill clusters formed during dispersal using the GPSeqClus package (Clapp et al., 2021). The algorithm uses time-series location data to sequentially aggregate locations to build clusters based on three user-defined criteria: search radius, temporal window, and minimum number of locations. We used the average step length per hour of T1-C1 as the search diameter for cluster identification. The maximum temporal window for cluster identification was 3 days, and we considered eight locations as minimum cluster locations. Resting and kill sites were validated from field observation. We calculated average cluster size, time spent, number of visits, and distance to the nearest forest edge and human settle- and January 2020 at a spatial and temporal resolution of 30 m and 32 days, respectively (Gorelick et al., 2017). From the start date of each cluster, the nearest available mean NDVI was generated. NDVI value of one indicates a high density with green leaves, and a negative value indicates no vegetation.

| Directionality
During the pre-dispersal phase, the movement of T1-C1 was undirected (mean turning angle = −2.88) suggesting uniform distribution of turning angle. During the dispersal, the mean turning angle was 0.11, indicating strong persistence in direction and linear movement. Moreover, the directional persistence or tendency to move in a straight line was more at night (mean turning angle = −0.01) than during the day (mean turning angle = 0.24; Figure 1c).
During post-dispersal, the wide range in the mean turning angle (mean day = −0.60; mean night = 0.24) indicates a mix of tortuous and forward movement in the landscape.

| Chronological dispersal event
The dispersal event of T1-C1 occurred between two protected areas, Tipeshwar and Dnyanganga Wildlife Sanctuary, separated by a linear distance of 240 km. The movement during dispersal oc- However, before establishing a stable area in the sanctuary, he moved further 90 km west and southwest of the sanctuary, including the forested landscape of the Ajanta caves. After more than a month of exploring the landscape, T1-C1 movement was confined to the Dnyanganga Sanctuary and adjoining fragments of forest patches outside the PA and eventually had a defined core area within the sanctuary. We monitored the individual through camera trapping in the sanctuary for 10 months after the collar was removed. There was no report of the presence of tigers other than T1-C1 when it entered the sanctuary. However, co-predator like leopards was present in the area and adjoining forest patches.
While dispersing through a human-dominated landscape, T1-C1 crossed linear features such as railway lines (n = 7), interstate highways (n = 34), and primary roads (n = 67) before reaching the Dnyanganga Sanctuary. The individual crossed primary roads more at night (n = 51) than day (n = 16). Similarly, it crossed highways at night (n = 29) more often than day (n = 5) while dispersing in a human-dominated landscape. The movement rate significantly increased when crossing roads (mean = 2.00 km/h) than traveling in non-road areas (mean = 0.30 km/h; p < .05). Subsequently, the trajectory path of T1-C1 showed territorial movement behavior confined to the sanctuary and adjoining forested landscape. Therefore, after tracking for more than a year, the GPS radio-collar was removed using a drop-off mechanism.

| DISCUSS ION
We recorded one of the longest dispersals of a male tiger, traveling a maximum linear distance of 315 km in a human-dominated landscape. The dispersal journey was remarkable for its length, duration, and movement through a highly interspersed mosaic of forested areas and agricultural fields. Evidence of such long-distance dispersal in tigers has been recorded previously in a few studies through radiotelemetry (VHF and GPS), intensively monitored populations (camera traps) and genetic studies across various landscapes in the Indian subcontinent (Gour et al., 2013;Reddy et al., 2012;Sarkar et al., 2016Sarkar et al., , 2021Singh et al., 2021;Smith, 1993). Such movement occurs in both sexes, with males moving longer distances from few kilometers to hundreds of kilometers (Table 3) and can be influenced by various factors. In general, male dispersal is related to intrasexual competition for mates, inbreeding avoidance, and resource competition (Dobson, 1982;Greenwood, 1980;Perrin & Vladimir, 1999;Pusey & Wolf, 1996). Information from long-distance dispersal thus provides knowledge and helps identify functional corridors, which are crucial to designing conservation policies for large-ranging species like tigers.
The dispersal journey of T1-C1 involved three distinct movement phases and traveled a cumulative distance of ~3000 km from its natal area to the site where it localized after dispersal. The dispersal route of a large-ranging carnivore like tigers highlights the multiscale nature of the individual movement and the ability to navigate a heterogeneous landscape. The dispersal event started during the monsoon and lasted until winter. During the period of monsoon, vegetation cover, availability of water, and productivity in dry deciduous habitat increases, making dispersal more permeable through a human-dominated landscape. Moreover, the agricultural fields with growing crops act as a cover for dispersal. Additionally, the availability of livestock increases as they are left to graze in the forested areas and form easy prey for dispersing tigers.
During post-dispersal, we found that the daily distance traveled and displacement was higher than in the dispersal and pre-dispersal phases. We also found a wide range in turning angles, suggesting the movement to be more tortuous and forward movement. This is probably because of the environmental features and mosaic of habitats within the landscape that influences movement. T1-C1 space use included multiple core areas (17.29 km 2 ) within the Dnyanganga Sanctuary and encompassed a larger fragmented area, including forest and agricultural fields outside the PA. Thus, it adapted a high territorial movement to cover a more extensive territory while crisscrossing agriculture fields and forested areas in a human-dominated landscape. Similarly, studies on cougars and lions exhibited higher speeds while traversing through fragmented human-dominated areas to reduce time spent in multiple-use areas (Kertson et al., 2011;Valeix et al., 2012). Moreover, temporal changes in movement speed allow the individual to cover longer distances and more fragmented forest patches occurring both inside and outside the sanctuary. Thus, it minimizes the anthropogenic disturbances by moving more at night (0.65 km/h) than during the day in a human-dominated landscape.
Similar movement behavior was reported in tigers that exhibited higher speeds outside PA and at night .
The movement was more linear with strong directional persistence during dispersal with a maximum mean movement speed of 3.50 km/h during the night. Altered movement patterns due to habitat fragmentation and human pressure have been observed in many carnivore species worldwide (Poessel et al., 2014;Tigas et al., 2002;Tucker et al., 2018). His movement was more confined and restricted within small forest patches and sheltered in these  to rest in these refuge patches during the daytime when human activity is maximum. As human activity reduces at night, the animal can move between forest fragments situated far away from each other.

Number of visits
Whereas, in forested areas, the habitat is contiguous, making movement more conducive during day and night. Thus, the animal spent less time resting in larger clusters, perceiving a low risk attributed to human activity and anthropogenic disturbances.
In addition to the landscape heterogeneity, linear features such as roads affect the rate of movement while crossing roads compared to when traveling in non-road areas. Moreover, T1-C1 avoided high-traffic roads by crossing roads at night as animal perceives roads and human activity as risk (Frid & Dill, 2002;Northrup et al., 2012;Thurfjell et al., 2015). Thus, linear features such as roads can act as barriers to animal movement either through mortality, displacement, or behavioral avoidance, depending on the landscape (Anderson, 2002;Forman et al., 2003;Scrafford et al., 2018;Shepard et al., 2008). Consequently, habitat mosaic and landscape structure greatly influence dispersal patterns by facilitating or restricting movement (Holderegger & Wagner, 2008).

| Conservation implications
The study highlights the importance of documenting long-distance

AUTH O R CO NTR I B UTI O N S
BH and PN conceived ideas and designed methodology; ZH carried out fieldwork, analyzed the data, and led the writing of the manuscript. PG helped in field data collection. RG and PP helped in capture and local field support. All the authors contributed critically to the draft and gave final approval for publication.

ACK N OWLED G M ENTS
We are thankful to the Ministry of Environment, Forest and Climate Change (MOEFCC) and Maharashtra Forest Department (MFD) for the permission to collar animals and conduct research. MFD is duly acknowledged for the funding provided to carry out the research project. We are also thankful to the Field Directors, DFOs, RFOs, forest guards, and watchers for their constant support, advice, and input during the fieldwork in the Vidarbha landscape. We are also grateful to our field assistants (Irfan, Mangesh, Noor, Rama, and Roshan) for providing constant support and necessary help throughout the study.
I am thankful to Dr. Hussain Reshamwala and Mr. Shaheer Khan for their critical thought in the analysis. We thank the Director, Dean, and Research Co-ordinator of the Wildlife Institute of India and Chief Wildlife Warden, Govt. of Maharashtra, for supporting the study.

FU N D I N G I N FO R M ATI O N
Funding was provided by Maharashtra Forest Department, Government of Maharashtra.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dispersal data of the endangered species contain location data outside protected areas, which are prone to poaching and human prosecution. Moreover, the movement corridor and forested areas are used by other tigers and will make other dispersing individuals prone to many risks including poaching. Because of conservation reason, our request may be considered not to share the location data.