Influence of Covariates:
Summed AIC weight from the most competitive (ΔAIC<2) eight models showed that termites are most important covariates followed by fruit, disturbance, tree cover, TRI and EVI (Figure 2) . The averaged strength of positive influence of covariates in habitat occupancy was higher for termites followed by fruit, disturbance, TRI and EVI while it was negative for tree cover. The 95% confidence interval (CI) of averaged beta coefficients occupancy covariates overlapped zero except for termites indicating high confidence on the magnitude and direction of influence by termites but less confidence on similar influence by other variables (Figure 3) .
Food resources of sloth bears particularly termites followed by fruits had relatively strong influence on sloth bear occupancy. This is expected because sloth bears are opportunistic omnivores specialized for myrmecophagous diet particularly termites. Studies on feeding ecology also shows that termites are most frequent in diet throughout year while fruits dominate according to seasonal availability (Yoganand et al. 2005; Palei et al. 2014, 2020; Bargali et al.2004; Ramesh et al. 2012; Rather et al. 2020; Philip et al. 2021). In Chitwan, fruits are available for short duration from April-August while termites increasingly dominate the sloth bear’s occurring in 52% of scats in 1970s (Laurie and Seidensticker, 1977), 81% during 1990s (Joshi et. al 1997) and 92% in 2010s (Khanal and Thapa 2014). Sloth bears had negative association with tree cover indicating its preference for open grassland habitat. During the dry season, the soil in upland Sal forest habitat becomes stiff (Mall & Karki, 2016) and termites excavate deeper into the ground to keep up with lowering water table (Mugendi D.G 2020; Ahmed et al. 2018; Kalumanga, 2016; Dangerfield et.al.1998). While water table is relatively higher in alluvium grasslands and the soil is relatively loose making it favourable for sloth bears to obtain termites. Correlation analysis with tree cover indicated that not termites mounds (r=0.02) but fruits were more abundant in grasslands (r= -0.28) during this period. Axelsson and Andersson E.P. 2012 and Chakraborty and Singh 2020 also observed that termite mounds were higher in forested areas compared to edge or open habitats. However, Garshelis et al (1999) also showed that mound digging by sloth bears was higher in grasslands despite high density of mounds in the forest. They also reported high ratio of diggings to obtain underground colonies of termites during the dry season. Thus, the preference for habitats with open tree cover during the dry season may be due to the better accessibility of food resources.
We expected that disturbance would have negative influence on habitat occupancy by sloth bears but in contrast the averaged beta coefficient for the covariate indicated otherwise. Understanding effects and impact of disturbance on species is challenging as it is driven by multiple factors such as individual behavior, evolutionary history as well as the frequency, duration and scale of disturbance events (Graham et al.2021; Iwasaki & Noda, 2018; Sousa W.P. 1984). In relatively intact landscape such as in western ghats, India, sloth bears have shown to avoid disturbance (Das et al. 2014; Puri et al. 2015; Babu et al. 2015) while in human dominated landscapes they have been reported to tolerate some degree of disturbance (Bargali et al. 2012) often consuming cultivated crops (Palei et al. 2019), human food waste (Prajapati et al. 2021) and causing conflicts with humans (Debata et al. 2017; Dhamorikar et al. 2017). Human-sloth bear conflict is common throughout the year in chitwan national park suggesting that sloth bears perceive humans as threats (Acharya et al.2016; Silwal et al.2016; Lamichhane et al. 2018). However, sloth bears might be using disturbed habitat in moderation for requirements such as food, water and shelter as suggested by weak but positive correlation of disturbance with terrain ruggedness, vegetation productivity and termites (Figure 2). Rugged terrain provides sloth bears with resting and denning refuge (Baskaran et al. 2015; Bargali et al. 2012; Akhtar et al. 2007) sites and cover to hide their cubs from potential predators like tigers. We used coefficient of variation in terrain ruggedness and enhanced vegetation productivity across our grids as it better represented the heterogeneity in the habitat. They were positively related but did not have enough variation to produce significant effect on the sloth bear occupancy. The weak association of covariate might be because our study was more localized while these variables are more adequately captured at a larger scale (Rather et al. 2021; Srivathsa et al. 2017).