Ecological and anthropogenic factors influencing the Summer habitat use of Bos gaurus and its conservation threats in Chitwan National Park, Nepal

Study area

The study was conducted in CNP; the first national park of Nepal and one of the UNESCO World Heritage sites (Fig. 1). CNP and its buffer zone is situated in the southern part of Central Nepal which spreads over Chitwan, Nawalparasi, Parsa, and Makwanpur districts. The park (N27°20′19″ to 27°43′16″ longitude and E83°44′50″ to 84°45′03″ latitude) occupies an area of 952.6 km² in the Rapti Valley of the Siwalik physiographic region, while the buffer zone (N27°28′23″ to 27° 70′38″ longitude and E 83° 83′98″ to 84° 77′38′latitude) extends 729.37 km² area. Its altitude ranges from about 100 m in river valleys to 815 m in Siwalik hills (Bhuju et al., 2007).

The park is recognized as a global biodiversity hotspot, comprising healthy habitats for several globally threatened species, including the Bengal tigers (Panthers tigris), greater one-horned rhino (Rhinoceros unicornis), Asian elephants (Elephas maximus), gharials (Gavialis gangeticus), and Gaur (CNP, 2015). The contiguous habitat of Chitwan and Parsa National Parks and adjoining areas harbor the largest population of vulnerable fauna Gaur (DNPWC, 2020).

Data collection

Detailed field survey and recording of the habitat characteristics

To access the species-habitat relationship, we adopted ‘use vs available’ resource selection functions at the population level as our sampling strategy (Manly et al., 2002; Johnson et al., 2006; Northrup et al., 2013). Our sampling strategy involved overlaying a square of 4km ×4km geographical grid cells (hereafter sampling units) (Burnham, Anderson & Laake, 1980) over the entire CNP using ArcGIS version 10.8.1 (ESRI, 2020). Among the 59 sampling units, we purposively selected 11 sampling units namely Kasara, Dhoba, Bankatta, Dhruba, Sukhibar, Bhimle, Dumariya, Jarneli, Bhimpur, nearby Ghatgai and nearby Dhoba. The selection of sampling units was informed by both the historical distribution patterns of Gaur in CNP, and incorporating valuable insights gleaned from stakeholder consultations during preliminary field visits. To expedite the data collection process, the selected sampling grids were partitioned into two sectors: eastern (Dumariya, Jarneli, Bhimpur, Nearby Ghatgai and Nearby Dhoba) and western (Kasara, Dhoba, Bankatta, Dhruba, Sukhibar and Bhimle). To ensure standardized data collection across sampling units and minimize temporal bias, we employed a double-team approach. Two survey teams, each consisting of one co-author and three experienced field technicians (minimum three years of experience), conducted simultaneous surveys in the eastern and western sectors.

Within each selected sampling unit, we implemented two line transects, each spanning 4 km in length (Bayani & Watve, 2016). These transects were positioned 200 m apart to ensure non-overlapping sampling areas. To enhance efficiency, the team of four individuals was further divided into pairs, enabling simultaneous surveying of each transect.

Each pair recorded all live sightings of the animals within a 50 m distance and indirect signs (dung, and footprints) within a 5 m distance on either side of the transect (Pokharel & Chalise, 2010; Khulal et al., 2021; Neupane, Chhetri & Dhami, 2021; Neupane et al., 2022). This paired approach minimized the potential for double-counting individuals, as both transects progressed in the same direction. GPS coordinates of direct sightings’ locations and indirect signs were recorded to prepare the distribution map using Arc GIS 10.8.1 version (ESRI, 2020). The survey was conducted for 15 days, following the peak activity period of Gaur (Gad, 2012).

Upon direct sighting of a Gaur, we marked the sighting location using a prominent topographical feature, and subsequently established a 5-meter radius circle with the sighting point as the center. Likewise, for indirect signs such as dung and footprints, we employed the same methodology, creating a 5-meter radius circle with the sign at its center (Yahnke, 2006; Bernard et al., 2014). For each of those circles, one additional circle plot with the same size was set with the center localized 100 m apart from the former circle in a randomly chosen direction (Neupane, Chhetri & Dhami, 2021). These plots represented average habitat samples, independent of the presence/absence of the Gaur. In each of these circular plots, the presence of Gaur was set to 1, if Gaur or any indirect sign was observed (Used plots), or to 0, if not (Habitat availability plot). In circles of both types (Used plots and Habitat availability plots), nine habitat parameters were measured namely canopy cover, ground cover, presence/absence of trees, presence/absence of predator sign, nearest distance from water body, road/path/firelines and settlement, habitat type and depth of leaf litter (see Table 1 for detail) to determine their effect on the probability of Gaur occurrence in a particular habitat. The canopy cover was measured with the help of a spherical crown densiometer. Similarly, ground cover, presence/absence of trees (any tree > 10 cm in diameter), presence/absence of predator signs (fresh scat, and pugmarks) (1–5 days old), and habitat type were determined based on visual observations by the team members possessing similar field experiences, similar to method adopted by Dhami et al. (2023a); Dhami et al. (2023b). The nearest distance to the water body, road/path/firelines, and settlement was calculated using the Nearest tool in Arc GIS 10.8.1 version about the presence location of the species. The shapefile of the water body was extracted using DEM (Digital Elevation Model, 12.5 m resolution) and Landsat Image 8 (USGS, 2021). Likewise, the shapefiles of roads/paths/firelines and settlements were extracted from an Open street map (OSM, 2021). The depth of leaf litter was measured following the method suggested by Marimon-Junior & Hay (2008).

Identification of conservation threats

After conducting thorough preliminary consultations with stakeholders and comprehensive field visits, supplemented by a review of pertinent documents including the Gaur Conservation Action Plan for Nepal (2020–2024), a total of eight significant conservation threats to the Gaur were identified. Subsequently, a workshop convened, gathering officials from CNP (n = 2), staff from the National Trust for Nature Conservation (n = 1), member of the Buffer Zone Management Committee (n = 1), field biologists (n = 3), and nature guides (n = 2). Participants were equipped with a printed list detailing the eight identified threats and were tasked with assigning each threat a rating on a scale of 1 to 10, indicating its impact on both the Gaur population and its habitat. With 10 representing the most severe threat and 1 indicating the least impact, participants were empowered to rank each threat accordingly. Following this evaluation, the average score for each threat was calculated, selecting top five threats for the subsequent relative threat ranking procedure (Annex).

Data analysis

Analysis of conservation threats

All the conservation threats that were identified from the detailed field survey and the first phase of the preliminary survey (consultations with local stakeholders) were listed and ranked through the relative whole-site threat ranking method suggested by WWF (2007) by considering criteria: scope, severity and urgency (Neupane, Chhetri & Dhami, 2021) (see Table 2 for detail). This method was also adopted by Chhetri et al. (2020) and Khulal et al. (2021) for threat identification and ranking. Together we identified six major threats which were then assigned a relative ranking from high (5) to low (1) referring to 5 as the most severe threat and 1 as least for all three criteria. Later, the ranking scores of each criterion were summed up and were reclassified into four classes very high, high, medium, and low.

Factors influencing the occurrence of Gaur in Chitwan National Park

Initially, three component models were generated through model averaging (models with ΔAIC ≤ 2). Among these three component models, the model with an additive effect of canopy cover, nearest distance to road/path/firelines, habitat type, and presence or absence of predator established itself as best fitting model with the lowest AICc = 312.56 and highest AIC weight 0.53. The competing model included canopy cover, depth of leaf litter, nearest distance to road/path/firelines, habitat type, and presence or absence of predator and had an AICc value of 314.15 and AIC weight of 0.24 (Table 3). Further, the area under the Receiver Operating Curve (ROC) for the best-fit model was calculated to be 0.75 with an accuracy value of 0.79 (79%) indicating an acceptable discriminatory ability (Fig. 2).

Among the predetermined nine variables, the model-averaged coefficient revealed that canopy cover, nearest distance to road/path/firelines, habitat type, and presence or absence of predator only influenced the occurrence of Gaur significantly (Table 4). Results showed that the probability of Gaur occurrence increases with an increase in moderate to high canopy cover, riverine and Shorea robusta dominated forest, and nearest distance to road/path/firelines, whereas decreases with an increase in the presence of the predator.

Major threats to the survival of Gaur

Among the five major threats identified during the field survey and preliminary survey (consultations with local stakeholders), uncontrolled burning was identified as a critical threat to the survival of Gaur. Similarly, invasive species and human disturbance were identified as major threats. Likewise, climate change and infrastructure development were ranked as medium and low threats to the species, respectively (Table 5).

Factors influencing the probability of occurrence of Gaur

Food, water and cover are the essential habitat factors that mostly influence the occurrence and distribution of Gaur (Ebil Yusof, 2009; Prayoon et al., 2024). However, the preferable habitats of Gaur are site-specific that are influenced by type of vegetation, forest cover and proximity to water bodies. Among the pre-determined nine predict variables that were chosen in our study, we found that the probability of occurrence of Gaur was significantly influenced by four of them, where it increased with increase in moderate to high canopy cover, Riverine and S. robusta species dominated forest, nearest distance to road/path/firelines, while decreased with increase in presence of predator.

Our study showed that Gaur preferred moderate to high canopy cover. The finding is similar to the study of Paliwal & Mathur (2012), who found the presence of Gaur showing a positive relation with canopy density classes less than 30% and 40–60% in Tadoba-Andhari Tiger Reserve (TATR), India. Unlike other smaller herbivores preferencing heterogeneous habitats, Gaurs are selective in their habitats and prefer mixed hardwood forests with the dominant species Shorea robusta, which has also supported the findings of (Bhattarai & Kindlmann, 2012). Likewise, a study on Gaur conducted in the summer season also found that it preferred grassland and S. robusta forest habitats in the Bandhavgarh Tiger Reserve of Central India (Sankar et al., 2013). Moreover, Ariffin et al. (2021) found Gaur more preferring riverine areas in Peninsular Malaysia. Steinmetz (2004) reported Gaur more commonly from mixed deciduous, dry dipterocarp, and semi-evergreen forest, in the Xe Pian National Protected Area in Lao PDR. Zangmo et al. (2017) reported more Gaur from subtropical warm broadleaved forests in Bhutan. Overall, deciduous forest provides a more favorable habitat for Gaur and other herbivores in Asia (McShea, Davies & Bhumpakphan, 2011). Our study has also identified similar results of Gaur preferring the Riverine and S. robusta species habitats. However, the studies conducted in different areas of India mentioned that it did not prefer dense forests (Imam & Kushwaha, 2013; Sankar et al., 2013; Haleem & Ilyas, 2018). Similarly, in contrast to our study, it was mentioned that Gaur preferred habitats with grasses, herbs, and shrubs and avoided woody species habitats (Gad & Shyama, 2011). Other studies have found that this species preferred habitat areas with low shrub and canopy cover and high grass cover (Goswami, 2007; Prayoon et al., 2024). Paansri et al. (2022) also noted that Gaur prefers grassland with lower canopy cover in Thap Lan National Park, Thailand. Likewise, Zangmo et al. (2017) found that Gaur prefers open areas in the Royal Manas National Park, Bhutan. Furthermore, Pla-ard et al. (2022) observed Gaur using dry evergreen forests more than other types, in Thailand. Prayoon et al. (2021) predicted Gaur presence more in moist evergreen forests (past), bamboo forests (present), and secondary forests (future), in Thailand. The contradictory findings of our study might be due to fewer grass species available during our survey period in the study area, which could attract the species in the forest area for feeding on barks, fallen leaves, and fruits (Gad & Shyama, 2009). There might be other possibilities that the species signs might be observed in the S. robusta forest areas with moderate to high canopy cover because they might have come to these areas for rest after feeding the nearby grasses and forbs species.

Gaurs are sensitive to human disturbances, so they usually prefer habitats located far from human disturbances including roads, paths, and settlements (Imam & Kushwaha, 2013). Our study further depicted that, distance to road/path/firelines has affected most significantly in the detection of Gaur. Further nearest distance to the road in between (0–400) m was seen to have a higher presence of Gaur. It was found that road distance is an important factor contributing about 15%, where habitat suitability increases away from roads for European Bison (Kuemmerle et al., 2018). Pla-ard et al. (2022) also reported Gaur appearing more near the road than far from the road, and the distance from the road impacted the chance of occurrence at 6.6%, in Khao Yai National Park, Thailand. Another study conducted in Yellowstone National Park also reported that bison (similar species to Gaur) mostly traveled on those land areas where there was availability of streams (Bruggeman et al., 2007). Gaur though prefers living in the core area, as reported by Trisurat et al. (2010), and travels through the road seasonally. Thus, this might be the reason that the indirect presences were observed nearest to road areas. Additionally, the roads inside the Gaur habitats are usually the fire lines and paths that are used by Gaur for travelling across different habitat patches inside the park.

In this study, the presence of predators was found to be affecting the habitat use of Gaur as the population of Gaur declined with its increase. Gaurs can quickly hear and detect the dangers of predators (DNPWC, 2020). So, they preferred to stay in the sites that are at lower predator risk as shown by our study. Tiger pug marks were spotted at different places in the study area. In CNP, there are 128 Tigers according to a recent survey (DNPWC and DoFSC, 2022). Karanth et al. (2004) also mentioned Gaur being preyed on by the largest carnivores of tropical forests like tiger (Panthera tigris) and leopard (Panthera pardus) in India. Tiger’s main prey are ungulates, including Gaur (Paansri et al., 2022; Suksavate et al., 2022). During the diet analysis of Tiger in CNP, Bhandari, Chalise & Pokharel (2017) found 2% of Gaur in its diet. Similarly, Andheria, Karanth & Kumar (2007) found 23.87% and 9.16% Gaur in the diet of Tiger and Leopard, respectively, in Bandipur Tiger Reserve, India. Prey species such as Gaur usually avoid predator-rich areas (Bhattarai & Kindlmann, 2013).

Conservation threats to Gaur

Among the five ranked threats in our study, we identified that uncontrolled burning was the most severe threat for the Gaur. Similarly, invasive species and human disturbance were recognized as the major threats to the species in our study area. Gaurs are very sensitive to habitat destruction including uncontrolled burning. For example, slash and burn shifting cultivation of hill tribes was one of the severe anthropogenic activities for threatened the Gaur population in the north-east India (Choudhury, 2002). Uncontrolled burning not only severely affected the wildlife species but also increased the spread and growth of invasive species in our study area (Murphy et al., 2013). DNPWC (2020) stated that controlled burning can encourage the growth of grass in the Gaur habitat. However, sometimes, fire spreads accidentally or intentionally by local people, in an uncontrolled manner. There are historical records of disastrous forest fires in protected areas of Nepal, including CNP, causing loss of wildlife (Sharma, 1996; BBC, 2012; GMFC, 2014). It is argued by some scholars (e.g., Yadav et al., 2008; Aryal et al., 2012; Sadadev et al., 2021) that fire in unsuitable timing has negative impacts on the survival of wildlife, as well as reduces the regeneration of preferred grass species and deteriorates habitat. Recently burned areas/grassland contain forage with higher quality but in low quantity (Allred et al., 2011), which is more used by smaller body-sized mammals than by larger body-sized, such as Gaur (Eby et al., 2014; Donaldson et al., 2018).

We found that invasive species and human disturbances are the major threats to Gaur. Invasive alien plant species (IAPS) like Lantana camara, Mikania micrantha, Chromolaena odorata are depleting the forest ecosystem (DNPWC, 2020). It was reported that in CNP, M. micrantha has caused a serious threat to wildlife habitat (Murphy et al., 2013; Khadka, 2017). IAPS deteriorates the habitat quality by replacing native plants (Schirmel et al., 2016; Gorchov et al., 2021). IAPS influences the abundance of wildlife, including ungulates (Garcês et al., 2020; Gorchov et al., 2021). In a recent study in the Barandabhar Corridor Forest, it was found that the abundance of ungulates decreased with increased cover of invasive species such as C. odorata, M. micrantha, and P. hysterophorus (Adhikari et al., 2022).

Human disturbance is another major factor in restricting the Gaur population in the wild. The human disturbances in our study area include presence of human for illegal collection of forest resources, jungle safari vehicles, jungle walk visitors, infrastructures such as machan or view tower. The threats in our study area have negative influence in the distribution and abundance of prey species of tiger including Gaur (Bhattarai & Kindlmann, 2013). Gad & Shyama (2009) also mentioned human disturbances could obstruct the movement of Gaur. Prey species like Gaur strive to avoid anthropogenic activities (Ahrestani, 2018; Suksavate et al., 2022). Gaur was reported to avoid human presences, in the study by Paliwal & Mathur (2012) in RATR, India. Likewise, Bhattarai & Kindlmann (2013) reported an abundance of prey species like Gaur, negatively associated with human disturbances in CNP. A similar finding was mentioned by Imam & Kushwaha (2013) in Chandoli Tiger Reserve and by Zangmo et al. (2017) in Royal Manas National Park. Moreover, it is stated that habitat loss due to human interference is a significant threat to the Gaur population in Asia (Choudhury, 2002; Duckworth et al., 2016).