Introduction

The Asian elephant (Elephas maximus) is listed as an endangered species globally and is protected under Schedule I of the Indian Wildlife Protection Act of 1972, which provides the highest level of protection. It has been designated as the National Animal by the Government of India. As a habitat generalist, it holds symbolic significance in the religious and cultural spheres across Asia, and it continues to captivate audiences- from laypersons to researchers due to its magnificent appearance, intricate social structures, and its role as a flagship species in wildlife conservation. Asian elephants fulfil vital ecological functions as efficient seed dispersers across extensive ranges within their ecosystems (Campos-Arceiz & Blake, 2011). Nevertheless, habitat degradation, loss of natural habitats, agricultural expansion, and the diversion of forests for human development activities, including infrastructure projects, mining, hydroelectric projects, and plantations- continue to exert pressures on elephants both within and outside their natural habitats (Luo et al., 2022; de Silva et al., 2023). This situation poses a grave threat to elephant populations across their distribution, ultimately impairing their capacity to adapt to rapidly changing environments. Consequently, many elephant populations are fragmented and forced to inhabit human-dominated landscapes, resulting in conflicts between humans and elephants (Desai, 1991; Sukumar, 2006; Madhusudan et al., 2015; Fernando et al., 2019). Human development projects, such as hydropower stations and agricultural expansion, further fragmented elephant populations, disrupted population contiguity, and affected the safety and livelihoods of local communities (Leimgruber et al., 2003; van de Water & Matteson, 2018; de La Torre et al., 2019; Vasudev et al., 2023). Despite cultural reverence in Asia, their numbers are declining due to reduced tolerance linked to these conflicts (Desai, 2000). Furthermore, forest loss resulting from the expansion of agricultural and commercial plantations, such as coffee and tea, has significantly impacted elephant movement and heightened human-elephant conflicts (Bal et al., 2011; Kumar et al., 2010; Puyravaud et al., 2019). These fragmented environments often result in suboptimal habitats for large mammals, such as elephants, disrupting habitat connectivity (Xu et al., 2021; Vasudev et al., 2021).

The state of Karnataka, well known for its extensive biodiversity, harbours the largest population of elephants, estimated at around 6,000 (Qureshi et al., 2025), representing one-fifth of the total Indian population (Baskaran, 2013). This population is distributed across three distinct groups, including regions within and surrounding the Bhadra Tiger Reserve, Shimoga, and the Hassan-Kodagu-Mysore-Mandya-Bangalore areas (KETF, 2012). The landscapes of Kodagu, Hassan, and Chikmagalur are characterised by significant alterations to natural habitats, primarily driven by agricultural expansion, particularly coffee plantations (Puyravaud et al., 2019). This transformation has led to the creation of fragmented habitats, often described as “forest islands” amidst production landscapes (Mudappa & Raman, 2007). These islands provide critical resources for elephants but also pose challenges for their movement and access to food and water. Earlier studies indicated that there are around 470 elephants in the coffee-rice paddies-dominated landscape of the region (Baskaran, 2013). The population forms a continuous link between elephant populations in Nagarahole and the Kodagu side, and in the Bhadra Tiger Reserve in Chikmagalur, with plantation-agriculture-forest mosaics in between. Research indicates that Asian elephants in this region exhibit remarkable behavioural adaptations to navigate the challenges posed by anthropogenic landscapes (Bal et al., 2011).

The Hassan Forest Division, adjoining the Kodagu Forest Division to the south and the Chikkamagalur Forest Division to the north, has been considered an important area for connecting elephant populations between the Nagarahole and Bhadra Tiger Reserves. The Hassan region, primarily characterised by a mosaic of coffee and rice paddies , alongside forest fragments and monoculture habitats, supports approximately 55-65 elephants and nearly 2,50,000 people in 466 villages that experience intense human-elephant conflict in the form of loss or injury to human and elephant lives and crop damage caused by elephants (Baskaran, 2013; Appayya & Desai, 2007; Srinivasaiah & Sinha, 2012; Anonymous, 2023). Krishnan et al. (2019) highlighted the importance of remnant forest patches and monoculture plantations, such as Eucalyptus and Acacia, which serve as the only refugia and provide feeding and shelter for elephants. In this region, the conflict arises not only from crop damage and loss of human life but also from retaliatory actions taken by farmers such as elephant drives, resulting in injuries or deaths of elephants, besides exerting pressure on the state forest department to capture and remove elephants as a strategy to mitigate human-elephant conflict (Appayya & Desai, 2007; KETF, 2012). Furthermore, the presence of large-scale barriers such as solar-powered fences and railway barriers, and the expansion of the National Highway 75 (NH 75), influence elephants’ use of fragmented habitat mosaics and their ranging patterns in relation to landscape changes, which remain understudied.

We investigated elephants’ habitat preferences and intensive use of the landscape in relation to changes in the Hassan landscape over the years. The study aimed to address key objectives, including a. understand the elephants’ use of fragmented mosaics of natural and anthropogenic habitats, b. analyse the spatial distribution of barriers in relation to elephant refugia, and c. spatial variation in the intensity of use by elephants across villages.

Materials and methods

Study Area
The study area comprised 466 villages (Figure 1), covering 1,190.5 km² across the Kodlipet-Alur-Sakleshpur-Alur- Arakalgud-Yeslur-Belur taluks, dominated by coffee plantations on the upslopes and rice paddies grown in the fallow regions, which together account for 76 % of the study area (approximately 64,400 ha). Additionally, there are 42 monoculture plantations, such as Eucalyptus and Acacia, varying in size from 0.7 ha to 122 ha, 29 isolated forest fragments, and abandoned or non-functional coffee plantations (more than 20 years old) which play an important role as refugia for 55 – 65 elephants when they move through the production landscape (Krishnan et al., 2019). In addition to elephants, the study area is frequently used by various species, including leopards, gaur, sambar deer, and mouse deer, which depend on these habitat refugia for movement. These refuge areas are distributed from north to south and northwest of the study region. The natural vegetation, in the form of forest fragments and Reserved Forest Areas, varies in size and collectively covers 5,240 ha. The Hemavati River, which flows through the middle of the study region and its backwaters, covering 3,224 ha, is a lifeline for people and elephants. While coffee is an annual crop, rice paddies are grown as a seasonal crop between July and December, during the southwest and northeast monsoons. Most of the coffee is owned by small and medium-sized growers, as well as a few national companies such as Tata Coffee Limited and Indian Builders Corporation Estate (IBC). The region receives about 300 cm of rainfall annually. The region supports around 2,00,000 people living in widely scattered villages and individual localities in plantations, mainly comprising coffee planters, paddy farmers, and maize growers, who also cultivate banana, areca, and pepper as intercrops. The inevitable dependence of people and elephants on land and resources, frequent negative interactions, and changes in landscape features posed significant challenges to coexistence between the protagonists.

Methods

Data collection
a) Elephant monitoring
Elephants were tracked daily through direct observations or indirect signs, such as dung, tracks, and feeding signs, for over three years, from 2023 to 2025. We have also relied on the state forest department field staff and an informant network across 466 villages to locate elephant presence and conflict occurrence in the study region. On each tracking day, information on age-sex composition, herd size, habitat type, name of the place or village and GPS coordinates of elephant locations at regular intervals along movement paths was systematically recorded. Once we located elephants during the day, we backtracked along their movement paths, using fresh dung, footprints, and feeding signs, to the location where they had been seen the previous day to record night locations (Kumar et al., 2010; Krishnan et al., 2019). All field data recordings were entered into a datasheet created using Epicollect, an open-source software (https://five.epicollect.net/), to improve data collection accuracy (Aanensen et al., 2009; Center for Genomic Surveillance, 2026). Identification of elephants was carried out using photographs and physical markings such as ear shape, persistent lumps, cuts on the tail or ears and degree of ear folding. (Moss, 2006; Goswami et al., 2007).

b) Digitising land use elements in the study region
The extent of natural and anthropogenic habitats, settlements, and physical barriers, such as solar-powered fences, railway barricades, and National Highway 75, was digitised using a combination of Survey of India topographic sheets, Google Earth imagery, and ground-truthing. Although the local village name was documented for each entry, we geographically linked the elephant locations to the official shapefile for Hassan district, supplied by the Karnataka State Government (https://kgis.ksrsac.in/kgis/), to obtain the corresponding administrative village name. Since some village names were similar, we modified them by adding numeric suffixes (1, 2, …) for data analysis. This designation has been used throughout the results presented in the paper. Shapefiles were created and verified using GPS locations collected during fieldwork.

Figure 1. Location of the villages in Hassan district where elephant presence has been frequent

We used the Locus Map app (Asamm Software. Locus Map: Outdoor Navigation (Version 4.x) (Mobile app). https://www.locusmap.app/) on smartphones to record the tracks of solar fences around coffee estates. Before recording the tracks for each fence, we obtained permission from the respective estate owners or managers to collect information on the date or month, and year of construction. Since installing fences is both costly and time-consuming, often taking several months, sometimes, the owners of the coffee estate might not have recalled the precise month, but were aware of the year of installation, particularly in the case of old fences that were constructed several years ago. In such instances, we have considered January of that year to be the month of fence installation. Then, we walked along the fences to generate vector layers. Subsequently, we used QGIS 3.34.3 (Q GIS Development Team, 2023) to create polygons for each fence and calculate their lengths, perimeters, and areas.

We have stratified the study area into six major habitat categories (Figure 2) including coffee (643.7 km², 54.07%), monoculture refuges such as acacia, teak, Eucalyptus, and abandoned coffee plantations which are more than 20 years old (7 km², 0.59%), agriculture (262.49 km², 22.05%) primarily dominated by rice paddies with other crops of maize, ginger, banana, coconut, areca, backwater area of Hemavati reservoir (32.24 km², 2.71%), natural vegetation areas of forest fragments and Reserved Forests (52.44 km², 4.4%) and others (townships, roads, residential clusters in villages, and livestock open grazing areas (192.65 km², 16.18%).

Data organisation
We analysed data spanning three years, from January 2023 to December 2025. Since the spatial locations of animals are inherently autocorrelated, we randomly thinned the data prior to any analysis. For each date, we randomly selected two locations (one daytime and one nighttime) for each elephant ID (herds or males). We excluded locations of solitary male elephants that could not be identified in the field from subsequent analysis. We have examined one daytime and one nighttime location to analyse habitat use by elephants over a 24-hour cycle for each individual or herd ID, thereby minimising spatial autocorrelation.

The final data subset used for the analysis comprises 24,400 randomly selected locations that fell within the village boundaries of 466 villages currently being monitored for elephants. Of these locations, 53% were from the dry season, and the remaining were from the wet season (Ndry=12,981, Nwet=11,419). We had a near-equal split between day and nighttime locations (Nday=12,776, Nnight=11,624). Nearly 60% of locations were adult males or male groups (Nmales=14,647), and the remaining were herds (Nherds=9,753).

a) Seasonal Classification
We classified the months from January to June as the dry season and from July to December as the wet season. The seasonal classification included not only precipitation patterns but also the cropping patterns of the two major food crops, maize and rice, which are grown seasonally in the study area. To analyse rainfall patterns in the study area from 2023 to 2025, we used the Multi- Source Weighted-Ensemble Precipitation (MSWEP) dataset to assess precipitation and identify seasonal periods. Paul et al., (2024) reported that MSWEP shows the highest agreement with Indian Meteorological Department (IMD) rainfall data in the Western Ghats region of India. We therefore used MSWEP over the IMD gridded rainfall dataset because MSWEP has a higher spatial resolution (0.1°) than the IMD dataset’s 0.25° (Paul et al., 2024). The study area receives most of its annual rainfall from May to October (Figure S1), which influences the cropping pattern of food crops such as rice paddies and maize. Most villages towards the west, adjoining the Western Ghats, receive higher precipitation than those towards the east (Figure S2). Farmers start sowing only at the end of June or early July, when sufficient rainfall has accumulated. Crop harvest begins in late November and continues until the end of December for rice paddies and maize, while coffee picking and pepper harvest last until March.

Figure 2. The distribution of six major habitat types, such as coffee, rice paddies, forest fragments, monoculture plantations, backwaters, and others, such as roads, residences, and grazing lands in the study region (left panel). An aerial image showing the predominant land use type of coffee on the upper slopes and rice paddy farming in the fallow regions in the study region (right panel).

 Data analysis
All data wrangling and statistical analyses were conducted using R statistical software version: 2026.01.1 (R Core Team, 2023) We used Manly’s habitat selection analysis (Manly et al., 1993) to examine how elephants use natural and anthropogenic habitats. This analysis measures habitat preference by comparing the proportions of locations in each habitat with the proportion of available habitat. A selection ratio of 1 indicates that a habitat class is used in proportion to its availability, whereas values greater than 1 indicate preference and less than 1 indicate avoidance. We conducted the analysis in three ways— using pooled data across all three years and splitting locations by day and night. We used the adehabitatHS package (Calenge, 2024) in R for the habitat analysis.

Elephant locations were spatially intersected with the village-level shapefile to identify the administrative villages associated with each location, using the sf package (Pebesma, 2018; Pebesma & Bivand, 2023) in R. We aggregated the total number of locations within each village for each season within each year to obtain the total count of elephant locations. This count was further scaled by the total area of the village to obtain village-level density or intensity of elephant use. We ran a Getis-Ord (Gi*) analysis (Getis & Ord, 1992) to identify spatial clusters of elephants’ use intensity in villages. This method uses both the feature value and the values of neighbouring features to identify spatial clusters of hot and cold spots. It generates a standardised z-score (Gi* z-score) centred on 0, with positive values indicating hotspots and negative values indicating cold spots. Since the z-score is centred on 0 (mean=0), it can be used to identify statistically significant clusters at z=±1.96 (p<0.05) and z=±2.576 (p<0.01). This analysis was implemented using the spdep package (Bivand & Wong, 2018) in R. We defined neighbours based on Queen’s contiguity method, i.e., neighbouring villages share either a common boundary or a common vertex (in contrast to Rook’s contiguity method, which assumes neighbours share only a common border).

To understand the spatial distribution of solar fences, we calculated the nearest (Euclidean) distance from each fence polygon to the nearest refuge polygon (monoculture or natural vegetation patches).

Results

Habitat selectivity by elephants
Manly selectivity ratios from pooled data indicated an overall preference for monoculture plantations (Manly selectivity index: wi=13.24, S.E.=0.29, p<0.001) and natural vegetation (forest fragments; Manly selectivity index: wi=4.24, S.E.=0.06, p<0.001). Coffee, being the dominant land use type in the study area, was used proportional to its availability in the landscape (wi=1.02, S.E.=0.02, p<0.001) whereas other land use types of agriculture (wi=0.63, S.E.=0.01, p<0.001), backwater (wi=0.38, S.E.=0.02, p<0.001) and other habitats (wi=0.22, S.E.=0.007, p<0.001) were avoided (Figure 3(a), Table S1a)

Selectivity of habitats by elephants during the day mirrored overall patterns, with strong positive selection for monoculture patches (wi = 19.89, SE = 0.48, p < 0.001) and natural vegetation fragments (wi = 6.68, SE = 0.09, p < 0.001), with near proportional use of coffee (wi = 1.01, SE = 0.008, p >0.05), and avoiding agriculture (wi = 0.01, SE = 0.002, p< 0.001), backwater (wi = 0.006, SE = 0.004, p< 0.001), and other habitats (wi = 0.25, SE = 0.01, p< 0.001; 9 (Figure 3(b), Table S1b).

At night, elephants exhibited a high positive preference for monoculture patches (wi = 5.94, SE = 0.29, p<0.001), natural vegetation (wi = 1.56, SE = 0.05, p<0.001) and agricultural fields (wi = 1.32, SE = 0.02, p< 0.001) while coffee was used almost in proportion to its availability (wi = 1.03, SE = 0.009, p<0.001). In contrast, elephants showed a significant avoidance of backwater areas (wi = 0.788, SE = 0.05, p < 0.001) and other habitats such as residential localities, roads, and open grazing lands (wi = 0.19, SE = 0.01, p< 0.001; Figure 3(c); Table S1c).

Figure 3. (a) Overall selectivity of major habitats by elephants in the Alur-Sakaleshpur-Yeslur-Belur region of Hassan as indexed by Manly Selectivity Ratio (wi) with 95% confidence intervals. (b & c) Day and nighttime habitat selectivity, indexed by Manly selectivity ratios (wi), with 95% CI in the Hassan region. Data analysed for the period 2023-25 from pooled random locations.

 Solar fences: proliferation and spatial patterns
The number of solar-powered fences around coffee estates in the study region has gradually increased over the years (Figure 4). We have mapped 277 solar-powered fences, with a cumulative length of 382 km, covering 32 km² (~5% of the total study area), installed around plantations between 2010 and 2025 to prevent elephants from entering coffee estates (Table 1). Most of the fences have been installed in the north-south direction, aligned with the distribution of monoculture refugia and forest fragments (Figure 4). Over the past decade, we have witnessed an 886% increase in the areas protected by solar fencing.

The Euclidean distance to the nearest solar fence from each refuge fragment underscores the increased proximity of solar fences to these fragments, with a range of 0-13 km and a mean of 1.31 ± 2.68 km (SD). Nearly 20% of all refuge fragments share a border with a solar fence, while an additional 30% have fences located within 300 meters (median = 0.31 km).

Hotspot Cluster analysis of the village-level use intensity of elephants
Village-level spatial hotspots of elephant use intensity were identified using the Getis-Ord Gi* analysis (Figure 5). There were no “cold spots” in the landscape, indicating that most villages experienced elephant use proportional to the village area. We identified the hotspot villages, which exhibit seasonal fluctuations between the dry and wet seasons. In 2023, the highest-use villages (hotspot villages) were distributed on either side of National Highway 75 in the Alur-Sakhleshpur range. In 2024, hotspot villages were shifted to villages north of the highway in the Sakhleshpur-Belur range. This pattern of intensive use of villages in the northern side of the study area, continues into 2025, with a noted reduction in the total number of identified hotspot villages in the recent period.

During the 2023 dry season, elephant locations were recorded across 224 villages, with 40 villages showing significantly higher elephant densities. These villages were located on either side of the national highway. Similarly, in the following wet season, 53 villages emerged as hotspots of elephant presence among the 239 villages used by elephants. These hotspot villages were spread across both sides of NH-75 and extended into the eastern region.

In 2024, during the dry season, elephant sightings were recorded in 209 villages, of which 50 were identified as spatial hotspots. Most of these hotspots are situated to the north and northwest of NH-75, with only four hotspot villages located south of the highway. During the wet season of 2024, all 34 hotspot villages out of a total of 215 elephant-occupied villages were found north of the highway.

In 2025, a pattern similar to 2024 was observed, with hotspot villages primarily located north of the highway in both the dry and wet seasons. During the dry season, 34 of the 174 villages had significantly high elephant activity. In the wet season, 31 of 149 villages were intensively used by elephants.

Discussion

The Asian elephant (Elephas maximus), a habitat generalist species, demonstrates a remarkable ability to adapt to varying ecological conditions and modify its behaviour accordingly. However, the persistent challenges posed by fragmented habitat mosaics, primarily dominated by human activity, raise significant concerns about elephant survival in altered landscapes. In the Hassan region, the predominance of coffee and rice paddies , juxtaposed with forest remnants, monoculture habitats, and human habitations, creates a complex landscape that complicates elephants’ use of fragmented mosaics and movement.

Despite the limited availability of natural forest patches and monoculture plantations such as Acacia and Eucalyptus refugia, which comprise less than 7% of the total landscape in the study area, these refugia played a critical role in elephant movement. Overall, elephants preferred forest fragments and monoculture

Table 1. Annual patterns statistics of total solar fence coverage in the study areas, along with year-to-year percentage increase between 2010 and 2025. 

Figure 4. Locations of a railway barrier (pink line), the National Highway 75 (black line), and solar-powered fences that have been installed gradually around coffee plantations between 2010 and 2025 (different colours for different years) in relation to the distribution of monoculture plantations such as Eucalyptus, Acacia, and abandoned coffee plantations (light green) and forest fragments (dark green)

Figure 5. Spatial clusters of hotspot villages with the highest elephant activity across seasons, identified among the 466 monitored villages in the Hassan landscape from 2023 to 2025.

refugia that served as both shelter and feeding grounds, while coffee use was proportional to its availability. In contrast, agricultural habitats, backwater habitats, and village premises were largely avoided, highlighting the species’ avoidance of human-dominated areas.

Elephants’ use of different habitat classes showed moderate diurnal variation and mirrored the overall pattern. Elephants exhibited a strong preference for monoculture plantations and natural forest fragments, with a disproportionately higher number of elephant locations in these habitat classes, despite their relatively small land area and patchy distribution. Even though selection strength decreases at night, these patches remain highly preferred compared to other land-use classes. Agricultural fields are avoided during the day (Wi=0.011) but remain moderately preferred at night (Wi=1.32). Coffee plantations that dominate the landscape were used in proportion to their availability during both day and night, owing to the presence of grass and water resources that attract elephants (Bal et al., 2011). However, elephants avoided the backwater area of the Hemavati Reservoir and other land-use classes, such as residential areas, roads, and open grazing lands, probably as a strategy to avoid risks associated with habitations and sand mining activity at night in the backwater areas, as elephants tend to alter their behaviour by swiftly moving through human-dominated landscapes between areas of tree cover (Graham et al., 2009; Fernando et al., 2023).

The Bi (standardised Wi ratio) values estimate the relative contribution of each habitat category, which allows us to rank habitat classes in order of preference (see Table S1). With six habitat classes in our analysis, a Bi = 1/6 (0.167) indicates a neutral preference, meaning elephants were distributed among these land-use classes in proportion to their availability (Manly et al., 2002). Across all three analyses — overall, daytime, and nighttime — our findings underscore the importance of habitat refugia, such as monoculture plantations and forest habitats, which help elephants navigate anthropogenic habitats. Even during the night, we observed an increase in Bi values for coffee and agricultural habitats (Bi=0.10 and 0.12, respectively), alongside a corresponding decrease in the intensity of use of natural vegetation patches (Bi=0.14); monoculture remained the most preferred habitat class (Bi=0.55). The software package used in the analysis does not generate confidence intervals (CI) for the Bi values, limiting further inferences about elephants’ use of the coffee, agriculture, and natural vegetation classes at night.

The findings from this study corroborate our earlier research on elephant habitat preferences from the Alur-Sakhleshpur- Kodlipet region of the Hassan landscape (Krishnan et al., 2019). Similar patterns have been observed in the tea-plantation-dominated Valparai region, where natural vegetation and riverine habitats are essential to sustaining elephant populations, while anthropogenic habitats are predominantly used at night for movement (Kumar et al., 2010). This behaviour enhances their survival in fragmented landscapes, reflecting adaptive strategies in response to environmental and habitat changes (Fernando et al., 2008; Neupane et al., 2019; Pokharel & Sharma, 2026). In the study area, although the proportion of monoculture plantations is lower than that of forest habitats, their north-south distribution and the availability of secondary vegetation and shade influenced elephants’ use of these habitats and their movements through anthropogenic habitats. Monoculture refugia scattered across the landscape are forest department lands; hence, they are relatively undisturbed with minimal human presence and vehicular disturbance. Furthermore, these plantations have abundant secondary vegetation such as bamboo, which is planted by the Forest Department and provides foraging opportunities for the elephants.

The predominant orientation of forest fragments and monoculture refugia—primarily north-south and westward— has significant implications for elephant movement across the study area. Therefore, it is imperative to establish connectivity between these fragmented habitats wherever feasible to facilitate unhindered movement of elephants. In scenarios where physical connectivity proves unattainable, collaborative efforts with local coffee planters and paddy farmers should be initiated to promote functional connectivity between refugia. Such measures will not only enhance elephant movement but also help mitigate human-elephant conflicts in the region (Vasudev et al., 2023).

Over the years, land-use changes were observed between 2010 and 2025, during which a large number of solar-powered fences, a 35 km stretch of rail barricades along the Hemavati Reservoir backwaters, and the expansion of National Highway 75 from two-lane to four-lane impacted elephant movements. Solar-powered fences were installed around coffee plantations to deter elephants from entering the estates. Studies have highlighted the adaptive responses of elephants to the specific spatial arrangement of natural and plantation forests within mosaic habitats, as the distribution of food and water resources, along with barriers and patterns of human activity over time and space, shape their ranging and foraging behaviours (Alfred et al., 2012; Kanagaraj et al., 2019; Ram et al., 2024; Mimeault & Weladji, 2025). The inability of elephants to exhibit wide-ranging movements could trigger ecological cascading effects, and their movement systems are highly dependent on specific bottlenecks that are increasingly threatened by human activities in fragmented elephant landscapes. Elephants tend to adapt their movements in response to resource availability in specific habitat patches, often choosing paths that maximise foraging efficiency while minimising conflict with human activities (Fernando et al., 2008). Thus, understanding elephants’ forage selection in human-dominated landscapes is critical, as it enables managers to prioritise habitats, secure forest resource areas, and reduce human-elephant conflicts.

In the Hassan region, most of these fences were spatially oriented north-south and towards the northwest, in accordance with the spatial distribution of forest remnants and monoculture refugia. Between 2010 and 2025, 277 fences of varying sizes around coffee plantations gradually appeared, covering approximately 32 km², representing an overall per cent increase in fence length of 21,122% over this period. The majority of these fencing structures are located near critical refuge areas, significantly limiting elephants’ access to these vital habitats and resources.

The spatial configuration of fences significantly affected their habitat use and movement throughout the study region. As a result, there was a shift in elephant movements and the intensity of village use over the years, with a higher concentration of elephant locations towards the north and northwest, rather than across the study area (Krishnan et al., 2019). Such land-use modifications not only affect elephants’ habitat preferences but may also drive them into neighbouring areas, heightening the potential for human-elephant conflict (Fernando et al., 2015; Osipova et al., 2018).

Moreover, the presence of solar fences around coffee plantations may compel elephants to use roads and residential or village premises, increasing the likelihood of encounters between people and elephants in the Hassan region. The presence of calves or young elephants may force herds generally to avoid areas with fences. Conversely, male elephants are known to navigate risks by breaking fences when moving between natural and human-altered habitats. These alterations in elephant movement patterns present significant challenges for effective management and conservation strategies, while also shifting conflict dynamics to areas where local communities may lack the experience to handle such interactions. Moreover, confinement to limited habitats can lead elephants to forage in closer proximity to human settlements, resulting in habituation to human presence (Vanak et al., 2010).

The conversion of National Highway 75 from a two-lane to a four-lane road (commenced in 2019 and completed in 2023) has further fragmented elephant populations and restricted their access to habitats and movement in villages located to the north of the highway. By 2023, elephants were found on both sides of the highway. However, in 2024 and 2025, elephant use intensity in villages shifted to the north of the highway. Though elephants were seen in most villages, their intensive use was observed in a few. Over the years, elephants showed distinct seasonal patterns of village use. During the dry season, hotspot villages were primarily observed on the western side, where large coffee estates provided adequate grass and water resources, as well as shelter for elephants (Bal et al., 2011). During the wet season, the intensity of elephant presence in villages shifted largely to the east, due to the presence of seasonal crops such as maize and rice paddies, resulting in increased human-elephant conflicts. Local cropping patterns are contingent on rainfall patterns, which in turn influence elephant movement and determine crop loss incidents within the region (Webber et al., 2011; Bohrer et al., 2014; Anoop et al., 2023; Matsika et al., 2024). The onset of early rains enables farmers to commence planting; however, the subsequent harvesting period after the peak rainy season tends to attract elephants to agricultural fields, thereby exacerbating human-elephant conflict (Webber et al., 2011; Matsika et al., 2024).

Conclusions

The results of the study highlight the need to manage elephants in fragmented landscapes, grounded in an ecological understanding of their behaviour and navigation through anthropogenic habitats. Given that most Asian elephants occur outside protected areas (Madhusudan et al., 2015), identifying critical habitats is essential for elephant management in fragmented landscapes. In the Hassan region, despite low coverage and distribution of natural forests and monoculture habitat refugia, they played a significant role in elephants’ habitat preferences and facilitated their movement. Hence, protecting these habitats and preventing their conversion to some other land uses is extremely important for the survival of elephants and for minimising conflicts. Given the large-scale degradation and loss of habitats suitable for elephants across their distribution ranges, many elephant populations have been ranging outside PA boundaries in anthropogenic landscapes, with complex behavioural adaptations that modulate their group dynamics (Srinivasaiah et al., 2019), time-activity budgets, and vocal repertoire (Pokharel & Sharma, 2026).

The installation of solar-powered fences around coffee plantations in key locations adjacent to critical elephant habitats has hindered elephant movement, pushing them into neighbouring areas. This has led to an increase in crop damage incidents and poses a significant threat to human safety. Therefore, it is imperative to avoid fencing large coffee plantations near natural forests and monoculture plantations, as this impedes elephants’ access to refugia and essential resources. In areas with existing fences, collaborate with coffee planters to redesign them to allow easy passage for elephants. Additionally, experimenting with temporary/seasonal fences, similar to those used in Sri Lanka (Fernando et al., 2025), may be considered to protect seasonal crops such as maize and rice paddies, which are often on a smaller scale and managed by marginal farmers. This strategy would help prevent elephants from entering croplands while still allowing them to move through the landscape.

The conversion of National Highway 75 has restricted elephant movement and forced them to concentrate in a few villages, resulting in intense human-elephant conflicts. Thus, there is a pressing need to construct overpasses or underpasses at critical elephant crossing points along the highway. Our study indicates that effective, proactive management strategies, grounded in a thorough analysis of local conditions, can help safeguard elephants while minimising conflicts between humans and elephants.

TO DOWNLOAD SUPPLEMENTARY MATERIAL, CLICK HERE.

Acknowledgement

We express our sincere gratitude to the following organisations and individuals for their generous support of the project over the years: Cholamandalam Investment and Finance Company Limited, Anita Dongre Foundation, Rainmatter Foundation, Rohini Nilekani Philanthropies, Mr Arvind Dattar, and Mr Venky Muthaiah. We sincerely thank the Karnataka Forest Department, especially the officers of the Hassan Forest Division and the Elephant Taskforce Teams, for their permission, cooperation, and assistance. We thank Vinod Krishnan and Mr Abhijith KJ for their help in the digitisation of the landscape features in the study area. We acknowledge the Coffee Planters Associations and individual coffee planters of Hassan for granting us access to their property and for their assistance and cooperation to the research team. We are also grateful to our colleagues at the Nature Conservation Foundation, particularly Mrs. Smita V Prabhakar, Dr T. R. Shankar Raman, Dr Divya Mudappa, for their valuable contributions and discussions. We extend our sincere acknowledgement to team members Rajkumar, Jai Shankar, Velumurugan, Mahesh Hosakoppalu, Pradeep, and Gopi for their support in the field.

CONFLICT OF INTEREST
M. Ananda Kumar is a guest editor at the Journal of Wildlife Science for the special issue that includes this article. However, he did not participate in the peer review process of this article except as an author. The authors declare no other conflict of interest.

DATA AVAILABILITY
Data is available from the corresponding author on request.

AUTHORS’ CONTRIBUTION
Deepak Bhat contributed to data collection, project administration, data curation, mapping, and analysis.
Aditya Ghoshal has conceptualised, cleaned the data, wrote the manuscript, and per-formed data analysis.
Nisar Ahamad and Nandini NC have contributed to the field data collection and data compilation.
Ganesh Raghunathan has contributed to project administration and to the review and editing of the manuscript.
Ananda Kumar was involved in funding acquisition, conceptualisation of the idea, su-pervision, validation, methodology, project administration, writing the original draft, and review and editing.

DECLARATION OF THE USE OF GENERATIVE AI & AI-ASSISTED TECHNOLOGIES:
The author(s) used ChatGPT (ver: GPT 4.0 ) to help with R codes which are used in the data cleaning, GIS analysis, and mapping purposes. The author(s) thoroughly reviewed and edited the content generated and take(s) full responsibility for the content of the publication.