Patterns of coexistence between two mesocarnivores in presence of anthropogenic disturbances in Western Himalaya

Species’ coexistence depends on species-specific resource utilization in a given habitat. Human disturbances in this context can constrain the realized niche by altering their community dynamics. In this study, we considered Western Himalaya as a case study to test the hypothesis that human disturbances influence mesocarnivore coexistence patterns. We regarded red fox and leopard cat as the focal species and assessed the coexistence patterns in low and high human disturbance areas in three dimensions: spatial, temporal, and dietary habit. We used camera trap detections and mitochondrial DNA-based species identification of fecal samples. We used generalized linear mixed-effect modelling (GLMM), activity overlap, Levin’s niche breadth, and Pianka’s overlap index to capture the spatial, temporal, and dietary interactions respectively. We found that red fox and leopard cat coexisted by spatial segregation in low human disturbance area, whereas dietary segregation was the means of coexistence in high human disturbance area. We observed a broader dietary breadth for red fox and a narrower for leopard cat in high human disturbance area. The altered coexistence pattern due to differential human disturbances indicates intensive anthropogenic activities adjacent to natural forests. It can link to increased opportunities for shared spaces between mesocarnivores and humans, leading to future disease spread and conflicts. Our study contributes to scant ecological knowledge of these mesocarnivores and adds to our understanding of community dynamics in human-altered ecosystems. The study elucidates the need for long-term monitoring of wildlife inhabiting interface areas to ensure human and wildlife coexistence.


Introduction
The position of a species in a community lies along a set of dimensions of environmental variables (Schoener, 1974). Each species exhibits an area of occupancy along the environmental dimensions called an ecological niche (Gray & Lowery, 1996). When the ecological niche is similar to other species, they are known as sympatric species. The coexistence of sympatric species is only possible till a threshold of niche similarity as predicted by the limiting similarity theory (Macarthur & Levins, 1967). The theory describes that the sympatric species should differ along one or more dimensions in their respective ecological niches to coexist. Else, the species is excluded from the community according to the competitive exclusion principle (Gause, 1934;Hardin, 1960) through divergent natural resource selection resulting in niche partitioning (Davies et al., 2007). On the contrary, competing or sympatric species might coevolve by achieving spatial, temporal differences in activity (Kronfeld-Schor & Dayan, 2003) or by dietary segregation (Walker et al., 2007). In this context, co-occurrence explains spatial coexistence where the occurrence of similar species in an ecosystem may be the cause of multiple factors, such as natural evolution resorting to differences or similarities among the animals in character through interspecific interactions influencing niche partitioning by means of competition and avoidance or niche sharing with association (Monterroso et al., 2015;Tammeleht & Kuuspu, 2018;Torretta et al., 2016). Understanding species coexistence is essential as it shapes the ecological and demographic dynamics of sympatric species through the effects of resource exploitation and interference interactions (Grassel et al., 2015;Holt & Polis, 1997).
Terrestrial carnivores exhibit complex community structures and interaction between species can be either positive (Allen et al., 2015) or negative (Pasanen-Mortensen et al., 2013). Mesocarnivores are available in more significant numbers and guilds of multiple species (Torretta et al., 2017). They are more sensitive and quick to respond to environmental changes (Randa & Yunger, 2006). They play a crucial role within an ecosystem as engineers, regulating the lower trophic levels through the cascading effect to sustainably structure the natural functions within an ecosystem (Monterroso et al., 2015). Being similar in size and function, they exhibit the phenomenon of coexistence (Torretta et al., 2016). Also, mesocarnivores are becoming increasingly important in ecosystems due to the ongoing loss of large carnivores. However, interactions within mesocarnivore communities remain poorly understood (Tammeleht & Kuuspu, 2018).
The Himalaya, the most biodiversity-rich landscape (Pandit et al., 2014), harbors a wide variety of mesocarnivore species (Shao et al., 2021a, b). Several studies have focused on single-species ecology (Bashir et al., 2014;Ghoshal et al., 2016;Roy et al., 2018), while fewer have emphasized interactions between species (Noor et al., 2017;Vernes et al., 2021). Moreover, human habitation is expanding toward the natural forests across the Himalaya (Angelici & Rossi, 2020;Cronin, 1979;Schaller, 1980). In this context, there is evidence that the presence of human habitations adjacent to natural forests have altered native mesocarnivore inter-specific interactions (Farris et al., 2015; for coexistence. For instance, human-induced activities altered resource partitioning among mesocarnivores in the wildland-urban interface in the Santa Cruz Mountains (Smith et al., 2018). Also, human disturbances like housing density, residential yard, and dog presence affected mesocarnivore interactions (Parsons et al., 2019). In particular to disturbed habitat, influence of anthropogenic disturbances on mesocarnivore interactions was reported in Gantchoff and Belant (2016) where coexistence of Geoffroy's cats and culpeo foxes was facilitated by dietary segregation while showing high spatial and temporal overlap. Similarly, mesocarnivore interactions were influenced in sites with fewer human settlements .
We deployed camera traps and conducted carnivore fecal sample collection in a contiguous habitat gradient of the Great Himalayan National Park Conservation Area (GHNPCA) in Western Himalaya. The administrative boundary of GHNPCA divides it into an ecozone and a national park. The ecozone is high in human disturbance due to the presence of approximately 50 villages (henceforth, high human disturbance area). The national park is low in human disturbance due to the absence of villages and roads and only seasonal usage by locals and tourists (henceforth, low human disturbance area) zones. We found red fox (Vulpes vulpes) and leopard cat (Prionailurus bengalensis) to be the most frequently detected terrestrial mesocarnivores in the study area and hence considered for analysis. Other mesocarnivores found in the study area are yellow-throated marten (Martes flavigula) and golden jackal (Canis aureus). In general, red fox and leopard cat can coexist with a high degree of spatial co-occurrence near human habitations Vitekere et al., 2020). Being nocturnal, they also overlap in time . In diet, rodents contribute majorly to both the mesocarnivore's food habits, while red fox feeds on a wide range of food items compared to leopard cat (Bashir et al., 2014;Rajaratnam et al., 2007;Reshamwala et al., 2018). We assumed that the coexistence of red fox and leopard cat would be facilitated by niche segregation considering similar body size and diet (Shao et al., Page 3 of 16 397 Vol.: (0123456789) 2021a, b) of species if competitive exclusion holds. In addition, we also assumed that mesocarnivore coexistence would be influenced by differential human disturbances, as shown in Vitekere et al. (2021), where red fox and leopard cat positively co-occurred with human presence. Therefore, we hypothesized red fox and leopard cat to show positive relation to each other's detection (spatial association, otherwise spatial avoidance) and positive relation to human (spatial association, otherwise spatial avoidance) in high human disturbance area. Secondly, we hypothesized red fox and leopard cat to overlap highly in their temporal activity. And finally, we hypothesized niche segregation among the two mesocarnivores through a dietary pattern as a mechanism of coexistence in high human disturbance area. We then compared the coexistence pattern of red fox and leopard cat to low human disturbance area and expected to perceive the same pattern. We assessed the coexistence of red fox and leopard cat in high and low human disturbance areas in three dimensions by using human presence as a disturbance factor: spatial (presence of human), temporal (activity of mesocarnivores and human), and diet (proportion of food item). This brings to our following objectives for this study: (1) to assess codetection pattern of red fox and leopard cat in high and low human disturbance areas, (2) to determine the temporal overlap between red fox and leopard cat in high and low human disturbance areas, and (3) to determine dietary pattern overlap between red fox and leopard cat in high and low human disturbance areas.

Study area
The study was conducted in the Great Himalayan National Park Conservation Area (GHNPCA) to understand the spatial, temporal, and dietary interactions of the sympatric mesocarnivores along a disturbance gradient. GHNPCA is a UNESCO world heritage site (UNESCO, 2011) and located in the Kullu district of Himachal Pradesh, Western Himalaya, India (Fig. 1).
The area of GHNPCA covers four catchments (river), viz., Parvati, Jiwa, Sainj, and Tirthan. We selected Tirthan for the intensive study due to the similarity in habitat characteristics with the entire GHNPCA (Singh & Rawat, 1999). Tirthan catchment (300 km 2 ) represents a highly variegated landscape with lower temperate Chir pine (Pinus roxburghii), Banj oak (Quercus leucotrichophora), and open scrubs at lower elevation (<2000 m) to upper temperate Fir (Abies pindrow), Kharsu oak (Quercus semecarpifolia) forests, and alpine meadows at high elevation (2500 to 4000 m). The national park (low human disturbance area) is majorly comprised of temperate to alpine habitats and is devoid of any villages or paved roads (2500 to >4000 m). In contrast, the ecozone (high human disturbance area) comprises of lower temperate habitat that contains all the villages (<2500 m) (Singh & Rawat, 1999;Vinod & Sathyakumar, 1999).

Camera trapping
We conducted camera trapping in low and high human disturbance areas to understand the spatial and temporal interactions among sympatric mesocarnivores. Due to the absence of natural marking on red fox and difficulty in identifying rosette marks in leopard cat from single-sided camera traps, we considered the number of camera trap photo captures as individual detections for both the species. We deployed camera traps in 5 sessions from 2017 to 2019. The number of sampling locations (n) and days of effort (t) in the respective five sessions was April-July 2017; n = 59, t = 2986, October-December 2017; n = 78, t = 2589, April-July 2018; n = 40, t = 1791, October-December 2018; n = 81, t = 2737 and April-June 2019; n = 81, t = 1763. We deployed 340 camera traps from 2017 to 2019 (Figs. 1, S1, S2, S3, S4, and S5). Total camera trap effort was 11,866 trap nights (no. of camera traps × operational days). Total camera traps (n) and efforts (t) in the low and high human disturbance areas were as follows: n = 220, t = 9284 and n = 120, t = 2582 respectively. The minimum distance between the consecutive camera traps was 0.5 km and a maximum of 1 km. Camera trap details of the 5 sessions are in Table S1.

Carnivore fecal sample collection
To understand the dietary interactions, we opportunistically collected fecal samples from the low and high human disturbance areas along the humanmade trails from 2017 to 2019 with a walk effort of 685.38 km. We collected 683 fecal samples following the dry sampling protocol (Biswas et al., 2019) and transferred them to the laboratory within a maximum of 2 weeks of collection and stored them at −20°C till further process. Details of fecal sample collection and respective walk efforts in the 5 sessions are in Table S2. To avoid misidentification of mesocarnivore feces through morphological identification in the field (Morin et al., 2016), we subjected the fecal samples to species confirmation using molecular markers (Cytochrome b, 146 bp) (Farrell et al., 2000).

Mesocarnivore species confirmation
We chose only fresh samples for species identification using a molecular marker due to the degradation of the outer layer of the faces in old samples. Out of 683 fecal samples, we extracted DNA from 382 fresh samples and considered the remaining old samples (n = 301) for morphological species identification of mesocarnivore. We performed the DNA extraction by swabbing the outer surface of the feces and following the protocol described by Ball et al. (2007) and Biswas et al. (2019). We performed PCR reactions in 10 µl reaction volumes containing 5 µl of Qiagen master mix (Qiagen Inc., Hilden, Germany), 1 µl of bovine serum albumin (BSA), 0.8 µl each of forward and reverse primers (Cytochrome b, 146 bp), 0.4 µl of RNase free water, and 2 µl of template DNA. We followed Yadav et al. (2021) to sequence the amplified PCR amplicons. Finally, we identified the sequences by comparing them in the NCBI database using the BLAST tool (http:// blast. ncbi. nlm. nih. gov/ Blast. cgi).
We identified the remaining old samples using the genetically confirmed samples as a reference, as combining both methods is helpful for carnivore identification (Descalzo et al., 2021;Lonsinger et al., 2021; Oja et al., 2017). We then matched the genetically confirmed mesocarnivore fecal samples with the old samples (n = 301), using weight and structure to identify the species.

Prey identification
We used both the genetically and morphologically confirmed mesocarnivore fecal samples for prey estimation through medullary hair identification. We thoroughly washed the hairs in the fecal samples and selected 20 hair strands randomly. The hair strands were cut into small pieces, dipped in xylene for 15-20 min, and observed under a microscope (40 ×) for medullary pattern identification (Bahuguna et al., 2010).

Data preparation and data visualization
We found red fox and leopard cat to be the most frequently detected terrestrial mesocarnivores; hence, they were considered for further analyses. To understand the effect of human presence on the co-detection of red fox and leopard cat, we performed data visualization using red fox and leopard cat and human photo captures (detections) in low and high human disturbance areas. We plotted detections of red fox against leopard cat and human for each location in low and high human disturbance areas, respectively (Figs. S6 and S7). Similarly, we visualized leopard cat detections by plotting them against red fox and human detections for each location in low and high human disturbance areas respectively (Figs. S8 and S9).

Co-detection modelling
We modelled the detections of red fox and leopard cat at a given camera trap site on a given occasion as a function of interspecific effects using a generalized linear mixed-effects model (GLMM) (Bolker et al., 2009;Cusack et al., 2017;Tattersall et al., 2020). We considered photo captures with a minimum interval of 15 min as independent detections. We tested the models based on occasion lengths of 7, 14, 21, and 28 days to see if occasion length influenced the codetection pattern between red fox and leopard cat (Cusack et al., 2017). We collapsed the detections of red fox and leopard cat of each camera trap location into 7 days, 14 days, 21 days, and 28 days, respectively, and prepared in the framework required for GLMM. We performed separate models for red fox and leopard cat in low and high disturbance areas. We modelled the red fox detections (n = 344) as a function of inter-specific effects of leopard cat detections and the disturbance effect of human detections. Similarly, we modelled the leopard cat detections (n = 524) as a function of red fox detections and the disturbance effect of human detections (Table S3). We included a random intercept for the camera trap station in all models to account for the non-independence of detections at the same site and standardized all the covariates. We excluded occasions from the analysis when a camera was inactive. We used 16 models: 2 mesocarnivores × 2 areas × 4 occasion lengths and implemented in GLMM framework using package "lme4" (Bolker et al., 2009) in R (v.4.0.5). We performed the significance test using p-values for estimated coefficients using an approximation of the Wald statistic (coefficient estimate divided by its standard error).

Model validation
We performed a homogeneity test using residuals vs fitted values to check for any pattern in the residuals due to model misspecification (Zuur, 2012;Zuur et al., 2007Zuur et al., , 2013. We did not find any clear pattern in the residual vs fitted values plots (indicating homogeneity) for red fox and leopard cat in either of the low (Fig. S10) and high (Fig. S11) human disturbance areas. We also checked for spatial dependency using semi-variogram plots (residual vs distance) using package "gstat" (Pebesma, 2004) in R (v.4.0.5). The semi-variogram plots indicated no spatial dependency in the photo captures of red fox and leopard cat to the distance between sampling locations in low and high human disturbance areas (Figs. S12 and S13).

Temporal interaction
We assessed the temporal overlap between red fox and leopard cat using the time data in camera trap detections for low and high human disturbance areas. In addition to comparing the activity patterns of both mesocarnivores, we analyzed each mesocarnivore with human activity. To minimize pseudo-replication biases, we removed any subsequent photos of the same species that occurred within 15 min. The number of red fox and leopard cat detections was >75 in both areas; hence, we chose Δ 4 to estimate the overlap coefficient (Meredith & Ridout, 2021). The overlap coefficient ranges from 0, meaning the absence of overlap, to 1, meaning complete overlap. We generated 95% CI by bootstrapping 10,000 samples to check for the precision of the estimates (Dias et al., 2019;Mori et al., 2020). The species' overlap coefficient was considered low if Δ 4 < 0.50, intermediate if 0.50 < Δ 4 < 0.70, and high if Δ 4 > 0.70 (Monterroso et al., 2014). We evaluated the coefficient of overlap (Δ 4 ) using package "overlap" in R (v.4.0.5).

Food items estimation and dietary interaction
We calculated the total number of food items in red fox and leopard cat feces in low and high human disturbance areas. We performed an accumulation curve for food items found in red fox and leopard cat feces to ensure the fecal sample size represented most of the prey items for red fox and leopard cat using the package "vegan" in R (v.4.0.5).
We assessed the dietary breadth and overlap for red fox and leopard cat in low and high human disturbance areas. We first calculated the dietary breadth index of red fox and leopard cat using the frequency of occurrence (FOO) of food items in the genetically confirmed fecal samples to understand the breadth of prey consumed across the disturbance gradient. We calculated the niche breadth using Hurlbert's standardized niche breadth (B A ), a measure of Levins' formula that is on a [0,1] scale (Hurlbert, 1978): B A = (B − 1)/(n − 1). Here, B = 1/(P i ) 2 , where "P i " is the proportion of each food item (number of records of each food (FOO)/total number of records of all food items) in the feces of red fox and leopard cat and "n" is the number of prey items (Smith et al., 2018). We then calculated the relative frequency of occurrence (RFO) of each food item in the genetically confirmed red fox and leopard cat fecal samples in low and high human disturbance areas, respectively. We calculated RFO using the formula i/j*100, where "i" is the sum of the frequency of specific prey in all the fecal samples and "j" is the total sum of the frequency of all prey (Mukherjee et al., 1994). We used the RFO of each prey to estimate the Pianka's overlap index of the diet of red fox and leopard cat in low and high human disturbance areas, respectively. We calculated Pianka's overlap index using package "pigrmess" in R (v.4.0.5).

Spatial interaction
The GLMM framework described the co-detection pattern between red fox and leopard cat in low and high human disturbance areas (Fig. 2).
In the low human disturbance area, we expected red fox and leopard cat to show positive spatial interaction, but the results indicate that the red fox's presence negatively influenced leopard cats' detections ( Fig. 2a, b and Table S3) across all occasion lengths. Also, we expected the red fox and leopard cat to show positive spatial interaction with humans. The results indicate that the red fox was negatively influenced by human detections initially on 7-and 14-day and later showed a positive relationship on 21-and 28-day occasion lengths (Fig. 2a). Whereas, the leopard cat was positively influenced by human detections throughout (Fig. 2b). Similarly, in high human disturbance area, we expected both the mesocarnivores to show positive relation to each other's detections. Interestingly, in this case, red fox and leopard cat positively influenced each other's detections on all occasion lengths (Fig. 2c, d and Table S3). On the whole, red fox and leopard cat showed spatial avoidance to each other in low human disturbance area, whereas they showed spatial association in high human disturbance area. Overall, there was no clear pattern of spatial interaction of red fox and leopard cat to humans. In both the low and high human disturbance areas, the direction of the codetection pattern (positive or negative) between red fox and leopard cat was the same across all occasion lengths. However, the magnitude of the co-detection pattern (positive or negative) varied across occasion length and area (Fig. 2). Temporal interaction.
For the temporal interaction, we hypothesized high activity overlap between red fox and leopard cat in both areas. We found that the activity overlaps between red fox and leopard cat was high in low human disturbance area (∆ 4 : 0.83, CI: 0.76-0.89) (Fig. 3a). This indicated their crepuscular and nocturnal behavior with activity peak between 18:00 and Page 7 of 16 397 Vol.: (0123456789) 06:00. Human activity was primarily diurnal which was evident from low activity overlap of both the species with human (∆ 4 = 0.13 [0.09-0.17] for red fox and human and ∆ 4 = 0.14 [0.11-0.15] for leopard cat and human) (Fig. S14). The human activity started after 06:00 and continued till 18:00. However, in the high human disturbance area, the activity overlap between the two mesocarnivores was higher (∆ 4 : 0.91, CI: 0.85-0.96) due to the flatter kernel density curve of the red fox as compared to low human disturbance area (Fig. 3b). The compression of the activity density curve of the red fox was due to shifting in the peak active time of red fox at 00:00 h in the low human disturbance area to a range of time: 18:00 to 06:00 h in high human disturbance area. Similarly, the activity overlap between red fox, leopard cat, and human was also low in high human disturbance area (Fig. S14). The overlap coefficient (delta, ∆ 4 ) is the area lying under both the density curves in Fig. 3a, b.

Mesocarnivore species confirmation
Out of 683 fecal samples, we identified 586 mesocarnivore fecal samples using molecular (success rate = 93.9%) and morphological approaches (Table S5). The morphological identification was based on the structure and dry weight of genetically identified red fox and leopard cat fecal samples. The average weight of the genetically identified red fox and leopard cat feces was 4.49 (SE = 0.33) and 10.05 (SE = 0.45). The details of the structure of red fox feces are in Figs. S16 and S17, and that of the leopard cat in Fig. S18. The confirmed 586 mesocarnivore fecal samples included the following: 359 samples that amplified for red fox (n = 112) and leopard cat (n = 247), and 227 samples that were identified morphologically for red fox (n = 139) and leopard cat (n = 88) respectively using genetically confirmed samples as reference. Combining both methods, we identified 251 red fox and 335 leopard cat fecal samples (Tables S6 and S7). The morphological details of the red fox and leopard feces are in Tables S8 and S9.

Dietary interaction
We found 19 and 11 food items in red fox feces and 12 and 7 in leopard cat feces from low and high human disturbance areas, respectively. From the accumulation curves of food items against the number of fecal samples, we observed that most of the curves reached the asymptote indicating the capture of most of the food items in red fox and leopard cat feces (Figs. S19 and S20). We observed plastic (RFO = 0.16%) and dog hair (RFO = 2.96%) in red fox feces in low and high human disturbance areas. We found a broader niche breadth of consumed food items by red fox (B a = 0.26) in high human disturbance area compared to low human disturbance area (B a = 0.25) ( Table 1). In comparison, the niche breadth of the leopard cat was narrower in the high human disturbance area (B a = 0.08) than in the low human disturbance area (B a = 0.13) ( Table 1). We also found rodents to be the most frequently consumed prey for the red fox and leopard cat in both areas (Figs. 4 and 5) (Table S10). For instance, the RFO of rodents in the red fox diet was 27.67% and 30.77%, and of leopard cat, it was 48.67% and 53.68% in low and high human disturbance areas, respectively. We observed a higher overlap in the dietary pattern using Pianka's overlap index between red fox and leopard cat in low human disturbance area compared to high human disturbance area (Table 1).

Discussion
It is often difficult to determine whether the observed species' coexistence pattern results from interspecific interactions or alternate processes (Steen et al., 2014). In the current scenario, human disturbances alter species' coexistence patterns (Carricondo-Sanchez et al., 2019;Parsons et al., 2019;Penjor et al., 2022;Sévêque et al., 2020;Smith et al., 2018). In this context, understanding the species interaction patterns at a fine scale might be central to delineating the processes responsible for species coexistence. We assessed the fine-scale coexistence pattern between red fox and leopard cat in high human disturbance (ecozone) and low human disturbance (national park) areas considering Western Himalaya as a case study and compared to test the influence of human disturbances on species coexistence. We assessed the mesocarnivore coexistence in three dimensions: spatial, temporal, and dietary habits in high and low human disturbance areas.

Spatial interaction
The co-detection modelling indicates that red fox and leopard cat positively influence each other's detections and co-occur with human in the ecozone, which aligned with our first hypothesis (Fig. 2c, d). Explanation of these results is possible such as an increasing human population and concomitant demand for land and natural resource extraction make fragmented landscapes, reduce forage availability and prey items for carnivores, and increase the interaction between humans and carnivores (Woodroffe, 2000;Woodroffe & Ginsberg, 1998), as a result, there is insufficient space available for native wildlife use. The explanation for responses to human presence will be that the spaces available for wildlife use around villages are also heavily used by humans. Therefore, these become shared spaces for the native wildlife and humans. In this context, mesocarnivores like red foxes and leopard cats behave as urban adapters in disturbed habitats (Duduś et al., 2014) and spatially coexist with species-specific human tolerances. Leopard cats showed a stronger spatial association with humans than red foxes, based on positive interaction on occasion length of 7 days for leopard cat and 14 days for red fox (Fig. 2c, d) (Cremonesi et al., 2021). On the contrary, inside the park, red fox and leopard cat did not co-occur and spatially avoided each other (Fig. 2a, b). This might be because, the absence of villages inside the park offers more available spaces for species usage . Therefore, red fox and leopard cat are not restricted to using limited spaces and coexist independently with spatial segregation. Also, leopard cat co-occurred with humans across all the occasion lengths, but the magnitude of relation was low as compared to ecozone, whereas red fox co-occurred only at 21-and 28-day occasion lengths. The low human disturbance inside the park explained the low magnitude of the relationship. Both the mesocarnivores reflected speciesspecific tolerance to humans, based on the choice of occasion lengths, and were consistent in low and high disturbance areas. The result was analogous to Vitekere et al. (2021) and Hua et al. (2020), where red fox and leopard cat coexisted and were tolerant of human presence. Our results were in accordance with our first hypothesis of mesocarnivore spatial association in the ecozone but contradicted when compared to the national park. We found that human presence did not influence the coexistence pattern of the mesocarnivores as they co-occurred with humans in both the low and high human disturbance areas (Vitekere et al., 2021). Wherein, the habitat type played a crucial role in the spatial association of red fox and leopard cat in the ecozone and spatial segregation in the national park (Katna et al., 2022). Our results postulate the importance of habitat quality in shaping species co-occurrence patterns.

Temporal interaction
The nocturnal behavior of the mesocarnivores supports the high temporal overlap between red fox and leopard cat in the ecozone and the national park (Zhao et al., 2020). However, we observed a higher overlap and shifts in temporal patterns of both the mesocarnivores in the ecozone, indicating more frequent activities. The increased temporal overlap might be due to the abundant anthropogenic resource availability (carrions, garbage dumps, kitchen wastes) around villages (Verdade et al., 2011), and red foxes and leopard cats are not forced to find the same prey (Finnegan et al., 2021). This was supported by the studies by Lorica and Heaney (2013), Reshamwala et al. (2018), andPenjor et al. (2022) where red foxes and leopard cats (Lorica & Heaney, 2013;Reshamwala et al., 2018) were reported to consume multiple food items near human habitations, thereby showing lower temporal competition for resource utilization in the same place (Finnegan et al., 2021). Our results aligned with the second hypothesis of temporal overlap between red fox and leopard cat that varied with differential human disturbances. Combining the results from co-detection modelling and temporal overlap, we found that in the ecozone, red fox and leopard cat co-occurred in space and time. As the competitive exclusion principle states, species coexistence is only possible through niche segregation in at least one dimension (Gantchoff & Belant, 2016;Hardin, 1960). Therefore, we finally hypothesized the dietary segregation of red fox and leopard cat, especially in the ecozone, and compared to the national park.

Dietary pattern
We first explained the species-wise dietary pattern using niche breadth and RFO of prey items of respective mesocarnivores in the ecozone and the national park. After which, we elucidated the dietary overlap between the two mesocarnivores in both areas using Pianka's overlap index. We found rodents to be the most frequently consumed prey species by red fox and leopard cat, which was analogous to studies by Ghoshal et al. (2016) and Bashir et al. (2014). Interestingly, rodent consumption was comparable for red fox in the ecozone and the national park but the consumption was higher for leopard cat in the ecozone as compared to the national park. The result was further supported by the broader dietary breadth of red fox in the ecozone and the contracted dietary breadth of leopard cat in the ecozone than in the national park. The result is due to the fact that the prey species in the less disturbed montane habitats (alpine and temperate) of the Western Himalaya are different from that in highly disturbed habitats located in the lower reaches of the study area. Inside the national park, among prey species, pika, Siberian weasel, and stone marten are found along with rodents in the alpine and temperate habitats (Vinod & Sathyakumar, 1999), contributing substantially to their diet (Bashir et al., 2014;Bischof et al., 2014;Ghoshal et al., 2016;Hisano & Newman, 2020;Reshamwala et al., 2018;Roy et al., 2018). Also, blue sheep, musk deer, Himalayan tahr, Himalayan serow, and Himalayan grey goral are present inside the national park upon which the mesocarnivores scavenge (Aryal et al., 2010;Hisano & Newman, 2020;Shao et al., 2021a, b). Whereas in the ecozone, rodents were the only natural prey species, and anthropogenic food items included livestock carrion, garbage dumps, and kitchen waste. For red fox, we observed a shift in natural prey consumption from pika and goral in the national park to livestock and primate (Himalayan langur and rhesus macaque) in the ecozone. The shift in prey species and broader dietary breadth of red fox revealed its behavior as an opportunistic feeder and adapter (Aryal et al., 2010) (Tables S10 and 1). On the other hand, leopard cat consumed rodents and birds more frequently and did not replace their natural prey with other anthropogenic food items ( Fig. 5 and Table S10). This might be attributed to the fact that rodent and birds are available abundantly around agricultural plots and human habitation (Lorica & Heaney, 2013). Therefore, leopard cat being a habitat specialist (Bashir et al., 2014) consumed only a few prey items and displayed lower dietary breadth in the ecozone than the national park. This pattern was linked to the Pianka's overlap index where red fox's and leopard cat's diet overlapped highly in the absence of human habitation (national park) as compared to a humandominated area (ecozone). Additionally, the presence of plastic, although in small quantity, in red fox feces in the national park indicates the influence of human disturbances in the natural forested habitats. Our results explained the coexistence of red fox and leopard cat through dietary pattern segregation in high human disturbance area, thereby supporting our third hypothesis. In contrast, these mesocarnivores coexisted through spatial segregation in low human disturbance area.

Implications and management strategies
In some carnivore guilds, coexistence is facilitated more by spatial or dietary segregation than temporal activity patterns, such as with canids in central Brazil (De Almeida Jácomo et al., 2004), or among mediumsized Mediterranean carnivores (Fedriani et al., 1999).
Our results indicate fine-scale mesocarnivore coexistence patterns where spatial and dietary segregation was observed specific to differential human disturbances. The results were comparable to Penjor et al. (2022), which indicates the influence of human disturbances on species' coexistence patterns in GHNPCA. The outcome of the study points toward intensive anthropogenic activities like increasing human habitation, and agricultural practices adjacent to natural forests, thereby affecting the coexistence of native wildlife (Rodriguez et al., 2021;Sévêque et al., 2020). In the context of Western Himalaya, which is one of the remaining biodiversity-rich landscapes (Rashid et al., 2013), the occurrence of such disturbances poses a threat to other wild animals inhabiting the area like large carnivores, leopards and Himalayan black bears (Naha et al., 2020;Sathyakumar, 2000). Knowledge about the coexistence of wild animals in human-dominated areas helps to understand the behavioral modifications toward interspecific interactions due to anthropogenic disturbances (Carter & Linnell, 2016;Carter et al., 2012). Additionally, the comparison of mesocarnivore coexistence between the national park and the ecozone revealed the adaptability of mesocarnivores for resource utilization in human-modified areas (Ghoshal, 2011;Hua et al., 2020), implying intensive space usage in the ecozone. These shared spaces by red foxes, leopard cats, and human-induced entities like livestock, and dogs can lead to disease spread and conflict (Chhabra & Muraleedharan, 2016;Nadin-Davis et al., 2021;Namusisi et al., 2021;Plumer et al., 2014). Management strategies aiming to shape the human-modified areas into highly heterogenous ecosystems can better facilitate fine-scale spatial segregation among species (Duelli, 1997). Also, proper garbage disposal strategies can reduce the readily available anthropogenic food resources. Thereby providing less opportunities to the opportunistic feeders and lessening the future disease spread and conflict outbreaks.

Limitations
Our study had a few limitations due to inherent challenges. Due to harsh environmental conditions, no fieldwork was conducted in the monsoon (August-September) and snowing (January-March) seasons. Therefore, this study was carried out only in the accessible months between April to July and October to November. We excluded data derived from cameras not functioning correctly (due to camera failure, battery failure, and heavy snowfall), resulting in the non-detection of the target species. Furthermore, camera placements, orientation, temperature differences, fecal sample detection, and degradation due to logistic limitations for storage added to species non-detection.

Conclusion
Patterns and mechanisms of coexistence are of particular research interest when involving carnivores of the same guild with comparable morphologies and overlapping diets (De Satgé et al., 2017). Additionally, assessing the species' coexistence in a humandominated landscape is crucial to comprehending the extent of carnivore adaptation. Our work provides a pattern of the spatiotemporal and dietary interactions between two sympatric mesocarnivores, red fox and leopard cat, across a human disturbance gradient in a mountainous landscape of the Western Himalaya. Our results suggest that dietary segregation played a major role in shaping red fox and leopard cat interactions in high human disturbance area. On the contrary, red fox and leopard cat coexisted through spatial segregation in low human disturbance area. Such adaptations to human disturbances point toward intensive anthropogenic activities adjacent to natural forests and opportunities for shared spaces between mesocarnivores and humans, leading to future disease spread and conflict issues. Including spatial, temporal, and dietary behavior in conservation and management plans may help to promote the coexistence of native wildlife and humans, especially in the ecozone, through controlling garbage dumps, increasing households, and expanding agricultural practices adjacent to natural forests to dissuade the native wildlife to these sites. While studies reporting the influence of anthropogenic activities on species interactions in the Western Himalayan landscape are largely lacking, we posit the need for long-term monitoring of wildlife inhabiting the human-modified areas to ensure human and wildlife coexistence in the future. Our work was, however, undertaken over a relatively small area of the Western Himalaya. Nevertheless, this initial understanding of mesocarnivore coexistence in 3 dimensions: spatial, temporal, and dietary habit, especially in a biodiversityrich landscape, can enlighten future carnivore conservation primarily through its applicability to alleviate potential human-wildlife conflict.