Niche Segregation and Habitat Suitability of the Red Fox (Vulpes vulpes) and Asiatic Jackal (Canis aureus), Two Sympatric Canids in Northern Punjab (Pothwar Plateau), Pakistan

© 2025 Academia Sinica, Taiwan Open Access Niche Segregation and Habitat Suitability of the Red Fox (Vulpes vulpes) and Asiatic Jackal (Canis aureus), Two Sympatric Canids in Northern Punjab (Pothwar Plateau), Pakistan Muhammad Farooq1, Tariq Mahmood1,* , Muhammad Sajid Nadeem1, Hira Fatima2, Faraz Akrim3, and Nadeem Munawar1 1Department of Zoology, Wildlife & Fisheries, PMAS Arid Agriculture University Rawalpindi 46300, Pakistan. *Correspondence: E-mail: [email protected] (Mahmood) E-mail: [email protected] (Farooq); [email protected] (Sajid Nadeem); [email protected] (Munawar) 2Department of Zoology, Division of Science and Technology, University of Education, Lahore 54770, Pakistan. E-mail: [email protected] (Fatima) 3Zoology Department, University of Kotli, Kotli, Azad Jammu & Kashmir, Pakistan. E-mail: [email protected] (Akrim) Received 6 November 2024 / Accepted 13 March 2025 / Published 30 July 2025 Communicated by Jian-Nan Liu Red fox (Vulpes vulpes) and Asiatic jackal (Canis aureus) are sympatric in many areas of their distribution range. Knowledge of the spatio-temporal niche segregation and habitat status of the species is important for effective conservation planning and management. In the current study, we investigated comparative spatio-temporal patterns of distribution and modeling habitat suitability of red fox and Asiatic jackal, in the Pothwar Plateau. Camera trapping, direct field sighting, recovering dead bodies, den sightings, scats and bioacoustics surveys were conducted from November 2018 to October 2020 to record data from four districts of the Plateau. The time of photos captured from camera traps was used to calculate the coefficient of temporal niche overlap using the overlap package in R software. To model the suitability of habitat, a total of twenty-six types of variables including 19 bioclimatic and seven other environmental variables were used. Results revealed a coefficient of temporal niche overlap between the two canid species as 4 = 0.47, 95% CI = 0.34–0.60. The red fox was found active during late night hours while the Asiatic jackal was found active during dawn and dusk, segregating their temporal niche. Habitat suitability modeling performed well in terms of AUC (0.840 and 0.824) and TSS (0.675 and 0.615) and identified land use and distance to poultry farms as major drivers of habitat suitability for both red fox and Asiatic jackal, respectively. Major factors determining habitat suitability of red fox and Asiatic jackal were land use cover and distance from poultry farms, respectively. Highly suitable habitats of red fox are present in the southern, central, and western parts of the study area while suitable habitat for Asiatic jackal is spread over the entire study area with few pockets of least suitable habitat. Furthermore, habitat suitability modeling revealed that 40.0% and 50.0% area of the Pothwar Plateau is highly suitable for red fox and Asiatic jackal, respectively. The study concludes that both canid species show temporal adjustments for their co-existence and suitable habitat for red fox is predominantly located on the southern side of the study area, whereas Asiatic jackal's suitable habitat is dispersed throughout the entire study area. Key words: Occurrence, Camera Trapping, Habitat Suitability Modeling, Niche segregation, Temporal niche segregation, MaxEnt Citation: Farooq M, Mahmood T, Nadeem MS, Fatima H, Akrim F, Munawar N. 2025. Niche segregation and habitat suitability of the Red Fox (Vulpes vulpes) and Asiatic Jackal (Canis aureus), two sympatric canids in northern Punjab (Pothwar Plateau), Pakistan. Zool Stud 64:17. doi:10.6620/ZS. 2025.64-17. Zoological Studies 64:17 (2025) doi:10.6620/ZS.2025.64-17 1 © 2025 Academia Sinica, Taiwan BACKGROUND Red fox (Vulpes vulpes) and Asiatic jackal (Canis aureus) (hereafter, golden jackal) belong to the family Canidae. Globally, both species are categorized as “Least Concern” (Hoffmann et al. 2018; Hoffmann and Sillero-Zubiri 2021) but listed as “Near Threatened” in Pakistan, where they are facing various threats, like increased human settlements, cultivation, overgrazing and drought, local trade for tail, hunting, retaliatory killing and accidental mortality due to vehicle collision, etc. (Sheikh and Molur 2005). They have an important role in the biological control of pests, seed dispersal, predator and prey, vector of rabies, works as cleaner by providing ecosystem services and have top-down effects on the ecosystem by regulating the population of herbivores (Ćirović et al. 2016; Elmhagen and Rushton 2007; Juan et al. 2006; MacInnes et al. 2001). Niche partitioning between two coexisting species with similar ecological requirements limits interspecific overlap by changing the extent of their use of a given resource to avoid or reduce competition. Numerous coexistence mechanisms have been proposed, including spatial segregation, variations in habitat use, behavioral adaptations and altered activity periods or movements, trophic segregation and specialization (Torretta et al. 2021). Habitat suitability modeling is an important tool in ecology and biogeography and a typical method for combining specific species with niche factors (Elith and Leathwick 2009; Wang et al. 2020; Zhang et al. 2018). Although there are many habitat suitability models (Maxent, BIOCLIM, GARP, and Climex) that have been developed to predict the distribution of species (Zhu et al. 2013), the Maxent model is one of the most popular tools, utilizing the principle of maximum entropy on presence-only data to estimate a set of functions that relate environmental variables and habitat suitability to approximate the species’ niche and potential geographic distribution (Phillips et al. 2006). Red fox and golden jackal are sympatric with each other throughout their distribution range (SilleroZubiri and Switzer 2004). Both species utilize different mechanism to avoid competition such as spatial, temporal or trophic segregation of resource use (Carricondo‐Sanchez et al. 2019; Shamoon et al. 2017 2018a b; Torretta et al. 2021; Tsunoda et al. 2018 2020 2024). Numerous studies have been conducted on the spatio-temporal segregation of both species in other parts of the world. Such interactions between both species may also exist in Pothwar Plateau of Pakistan. However, knowledge is scarce about the spatio-temporal segregation of both species in Pothwar Plateau which is important for effective conservation and management of species. Therefore, we hypothesized that (1) red fox and golden jackal have different distribution patterns in the Pothwar Plateau. We also expect that (2) red fox and golden jackal have segregated their spatio-temporal niche for co-existence. We similarly expect that (3) red fox and golden jackal have different suitable habitat in Pothwar Plateau. Therefore, the current study was conducted to fill afore mentioned gaps in the Pothwar region. MATERIALS AND METHODS Study area The current study was conducted in the Pothwar Plateau (32°10–34°9 N and 71°10–73°55 E) of the province of Punjab, Pakistan (Fig. 1). The eastern and western sides of the Plateau are flanked by the rivers Jhelum and Indus, respectively. To the north are the Margalla Hills and Kala Chita Ranges, and to the south are the Salt Ranges. The Pothwar Plateau comprises four administrative districts (Attock, Chakwal, Jhelum, and Rawalpindi) and encompasses an area of ~23,161 km2. The sub-humid climate of the northern part of Pothwar gradually becomes drier as the distance from the subHimalayan region increases. The average rainfall varies from 1500 mm in the northeast to about 380 mm in the southwest (Rashid and Rasul 2011). Out of the total, about 80% of rain falls from July to October (Sarwar et al. 2016). The summer temperature ranges between 15°C and 40°C while the range of winter temperature is generally between 4°C and 25°C but it can occasionally drop (Amir et al. 2019). The middle of the Pothwar Plateau is occupied by the structurally down-warped basin of the Soan River. The general terrain of the basin consists of interlaced ravines, which are locally known as khadar’s and are set deep in the soft Siwalik beds. The landscape is dissected and eroded by streams, which during the rains, cut into the land and wash away the soil. The streams are generally deep set and are of little or no use for irrigation (Britannica 2018). According to the 2017 census report by Pakistan Bureau of Statistics (PBS), 10,006,624 people are residing in the area (PBS 2017). Inhabitants of the area are mainly agrarian but many people are moving into industry and mining (Ali 2004). Distribution patterns of red fox and golden jackal Field surveys were conducted in the study area from November 2018 to October 2020 to record the presence of the red fox and golden jackal. Methods such page 2 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan as camera trapping, direct field sighting, recovering dead bodies, den sightings, and bioacoustics surveys were used. A detailed description of the camera trapping methodology is provided in the subsequent 'Temporal Niche Overlap’ section. Bioacoustics surveys were conducted for golden jackal by following the Harrington and Mech (1982) standard procedure. Stimulus howling was played by an mp3 player and was broadcast across the open environment from an elevated position with the help of a megaphone. Response calls of the jackals were recorded on a digital sound recorder (Sony ICDPX470 Stereo Digital Voice) (Passilongo et al. 2015). The geographical coordinates of each observation were recorded using a handheld GPS device (Garmin 12-Channel GPS). Collected data was subjected to ArcGIS (Geographical Information System) software version 10.5 for generating distribution maps of each species. Kernel density was calculated to show the hotspots of distribution signs on map. Temporal Niche Overlap Camera trapping Camera trapping was conducted at selected sampling sites in each district of the study area, based on direct and indirect signs of species presence. A total of seven infrared motion-triggered camera traps were used, although the number of cameras installed per night varied due to weather conditions and other disturbances. We employed three camera types: 1) Bushnell HD Trophy Camera Model-119537 (n = 2); 2) Bushnell HD Trophy Camera Model-119836 (n = 2); and 3) Browning Trail Camera Model-BTC-IXV (n = 3). Cameras were installed 30–40 cm above ground level (according to the breast height of animals) on tree trunks or bushes, approximately 1–2 meters apart from the expected passing route. The cameras were set to motion detect, rapid-fire, and photo mode, with a trigger speed of 0.2 seconds. We positioned the cameras to maximize the view of animals, enabling identification of individuals. Poultry offal, collected from butcher shops, was used as Fig. 1. Map of Study area: Map of four administrative districts of the Pothwar Plateau (Attock, Rawalpindi, Chakwal and Jhelum), the study area. N page 3 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan bait. A lure station was created by leveling the ground in front of the camera, approximately 1 meter away, and erecting a boulder with bait to maximize capture probability. Tall grass was removed from the trails in front of the camera to reduce false image captures. The trapping rate of the fox was calculated by dividing the total number of independent photo captures by the total number of trap nights. Cameras were installed for 5–7 consecutive days at each study site (Katuwal and Dahal 2013). Following Meek et al. (2014), pictures taken within two minutes were considered a single event. Time stamp was enabled on the photos to record the time and day of capture. Analysis of Camera Trapping Data The time of photos captured from camera traps was used to calculate the temporal niche overlap between the red fox and golden jackal in the study area. To estimate this overlap, we employed the kernel density function (Ridout and Linkie 2009). They defined the overlap coefficient (Δ) as the area under the overlapping curves of kernel density estimates from two samples. The coefficient values range from 0 (no overlap) to 1 (complete overlap). Δ(f,g) = ∫min{f (x),g(x)}dx Where f(x) and g(x) are density curves species f and g respectively. In the present study, since both samples exceeded 75 observations, we employed the overlap coefficient 4 (Dhat4) to estimate the overlap of activity patterns between the two species. We calculated the 95% confidence interval and mean of the overlap coefficients using 1000 bootstrap samples (Meredith and Ridout 2021). The extent of overlap can be classified into the following categories: low (Δ ≤ 0.5), moderately high (0.5 < Δ ≤ 0.75), and very high (Δ > 0.75) (Monterroso et al. 2014). All statistical analyses were performed using the ‘overlap’ package (v. 0.3.4) in R statistical software (v. 4.0.0) (Meredith and Ridout 2021). Habitat Suitability Modeling Initial Examination of the Observed Environmental Values All statistical analyses and modeling were performed using R version 4.0.0, a widely used statistical software platform (R Core Team 2020). An initial examination of the observed environmental values for each species was carried out using the extract function from the ‘raster’ package (Hijmans et al. 2015). Environmental ranges of species were checked before running models in order to identify possible errors in the output. Acquisition and Processing of Data Bioclimatic data comprising 19 variables (Table 1) were downloaded from the WorldClim database by using the ‘raster’ package (Hijmans et al. 2005), and collinearity among the variables was assessed using Variation Inflation Factors (VIF). After collinearity assessment, three bioclimatic variables were selected for further analysis: bio_01 (annual mean temperature), bio_07 (temperature annual range) and bio_12 (annual precipitation). Normalized Difference Vegetation Index (NDVI) was calculated from the Landsat 8 images which were downloaded from the GloVis-USGS website (https://glovis.usgs.gov/) by applying the filter of the 0 to 20% cloud cover. Digital Elevation Model (DEM) was retrieved from the USGS earth explorer database (https://earthexplorer.usgs.gov/) and was used to generate Vector Ruggedness Measure (VRM) via the terrain function in the ‘raster’ package (Hijmans et al. 2017). The land use cover was downloaded from the Esri website (https://livingatlas.arcgis.com/ landcover/). To assess the collinearity with the other variables, the categorical land cover data was converted into proportional data (continuous data) by decreasing resolution to 3 × 3 km and calculating the proportion of each high-resolution cell (10 m resolution) within the coarser 3 × 3 km resolution cells. Local shape files for human settlements and roads were downloaded from the Humanitarian Data Exchange website (https:// data.humdata.org/). A shape file for water bodies was obtained from the DIVA GIS website (https://www. diva-gis.org/), and data on poultry farm locations were collected directly in the field. These shape files were used to calculate Euclidean distances using the fasterVectToRastDistance function from the ‘fasterRaster’ package (Smith 2018). Subsequently, all environmental variables were resampled to a 0.5 arcminute resolution using the bilinear method and then masked to the shape file of the Pothwar Plateau. The explanatory variables were then stacked and tested for multicollinearity using the vifstep and vifcor functions from the ‘usdm’ package, with a threshold VIF of 10 and an absolute correlation coefficient threshold of 0.7 (Naimi 2015); we excluded predictors that were strongly correlated but less relevant for modeling both canid species compared to other variables. Additionally, variables were also selected depending on the ecology of red fox and golden jackal (Charaspet et al. 2019; Honghai et al. 1999; Khattak et al. 2022; Krim et al. 1990; Mukherjee et al. 2018). page 4 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan Model Fitting and Tuning Maxent is a piece of software (Phillips et al. 2006) running a specific SDM algorithm on the supplied data and user-specified (or default) settings. It can also be seen as an algorithm itself. A range of environmental variables, which are crucial for understanding species distribution and abundance, are integrated into the model, ultimately contributing in conservation efforts (Debinski et al. 1999; Kaky et al. 2020). Habitat suitability model was developed for this study using following guidelines on Maxent parameterization (Feng et al. 2017; VanDerWal et al. 2009). Species occurrence records were spatially thinned to a single point per raster cell, minimizing spatial biases and ensuring data independence. Subsequently, Maxent models were generated utilizing the ‘dismo’ package, which provides a software wrapper for the Maxent algorithm (Hijmans et al. 2017). Bias correction ‘bias_con’ file was created by following the guidelines of Young et al. (2011). We generated 2,000 geographically randomized background points. Models were tuned using the ENMeval function Table 1. Variables used for Maxent: List of Bioclimatic (N = 19) and other environmental variables (N = 7) that were analyzed for the habitat suitability modeling of the two sympatric red fox and golden jackal in the current study Sr. No. Variable Description 1 bio_1 (Annual Mean Temperature) Average temperature across an entire year at a specific location 2 bio_2 (Mean Diurnal Range) Average difference between the monthly maximum and minimum temperatures throughout a year 3 bio_3 (Isothermality) How large the day-to-night temperatures oscillate 4 bio_4 (Temperature Seasonality) Standard deviation of the monthly mean temperature multiplied by 100 5 bio_5 (Max Temperature of Warmest Month) Highest temperature recorded during the hottest month of the year at a specific location 6 bio_6 (Min Temperature of Coldest Month) Lowest average temperature recorded during the coldest month of the year at a specific location 7 bio_7 (Temperature Annual Range) Difference between the average temperature of the hottest month and the average temperature of the coldest month in a given location 8 bio_8 (Mean Temperature of Wettest Quarter) Average temperature during the quarter of the year with the highest precipitation 9 bio_9 (Mean Temperature of Driest Quarter) Average temperature during the quarter with the lowest precipitation throughout the year 10 bio_10 (Mean Temperature of Warmest Quarter) Average temperature during the three-month period with the highest average temperature in a year 11 bio_11 (Mean Temperature of Coldest Quarter) Average Temperature of the Coldest Quarter 12 bio_12 (Annual Precipitation) Annual precipitation 13 bio_13 (Precipitation of Wettest Month) Total amount of precipitation recorded during the single month with the highest rainfall in a given location throughout the year 14 bio_14 (Precipitation of Driest Month) Total amount of precipitation recorded during the single driest month of the year at a given location 15 bio_15 (Precipitation Seasonality) Calculated using the coefficient of variation to measure the variation in monthly precipitation totals over a year 16 bio_16 (Precipitation of Wettest Quarter) Total amount of precipitation received during the three-month period with the highest rainfall in a given location each year 17 bio_17 (Precipitation of Driest Quarter) Total amount of precipitation received during the driest three-month period of the year at a given location 18 bio_18 (Precipitation of Warmest Quarter) Total amount of precipitation received during the three warmest months of the year in a given location 19 bio_19 (Precipitation of Coldest Quarter) Total amount of precipitation received during the coldest three months of the year at a given location 20 NDVI (Normalized Difference Vegetation Index) Measures of health and density of vegetation 21 Distance from settlements Euclidian distance from settlements 22 Distance from road Euclidian distance from road 23 Distance to water body Euclidian distance from water body 24 Distance from poultry farms Euclidian distance from poultry farms 25 Land use cover Physical characteristics of what is on the Earth’s surface, including vegetation, water bodies, bare soil, and human-made structures 26 VRM (Vector Ruggedness Measure) Quantification of ruggedness or complexity of terrain page 5 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan from the ‘ENMeval’ package (Muscarella et al. 2014). This involved identifying the most informative feature selection variables and regularization multiplier (betamultiplier) values, which were selected based on the lowest delta Akaike Information Criterion (AICc) using five random k-folds. Evaluation of Model Performance The Maxent datasets were split into training (75%) and testing (25%) subsets using a 4-way partitioned k-fold approach. The final models included linear and quadratic (lq) features, with the regularization multiplier (beta-multiplier) initially varied between 1 and 2 for sensitivity analysis and ultimately set to 1 for the finalized model. We selected these features for the final Maxent models for each species based on the statistics we used to validate our model: areaunder-the-curve (AUC) values within receiveroperating characteristic (ROC) curves and the true skill statistic, TSS = TP+TN-1, where TP = proportion of true positive predictions and TN = proportion of true negative predictions. AUCDiff values were also computed to check either model is over fitting or under fitting. The Jackknife procedure was implemented for the percentage contribution of each variable in determining the habitat suitably. Models were replicated with replacement using the bootstrap method. True skill statistics is defined based on the components of the standard confusion matrix representing matches and mismatches between observations and predictions (Fielding and Bell 1997). Model outputs were defined as habitat suitability on a 0–1 probability scale (1 = highest suitability; 0 = lowest suitability) (Fig. 2). Finally, the habitat suitability index (HSI) map was reclassified into three distinct categories: highly suitable, moderately suitable, and least suitable by using ‘raster’ package. RESULTS Distribution patterns of red fox and golden jackal Data revealed that red fox and golden jackal occur in all four districts of the Pothwar Plateau. A total of N = 341 field signs of the occurrence of both canid species were recorded at 45 different locations including N = 131 field signs of red fox and N = 210 for golden jackal (Table 2). Red fox was recorded at 36 of the 45 sampling sites surveyed in the study area, with an altitudinal range of 235–923 m above sea level (asl). High abundances of red fox signs were recorded at Ara Basharat (n = 10), Jalal Pur Sharif (n = 9), Sarkal (n = 8), Moorat Mor (n = 6), Lawa (n = 6), and Ahmadabad (n = 5) (Fig. 3a and b; Table 2). In contrast, direct and indirect signs of golden jackal were found at 42 of the 45 surveyed sites, with an altitudinal range of 194-904 m asl. Notably, high density of golden jackal signs was recorded at Jatli (n = 15), Dhurnal (n = 15), Khairi Murat (n = 13), Sayed Kasran (n = 12), Lawa (n = 12), and Ara Basharat (n = 10) (Fig. 3c and d; Table 2). Temporal niche overlap Temporal niche overlap was calculated by using the overlap package in R by using camera trapping data. A total of 79 capture events of red fox and 81 capture events of golden jackal were used in the analysis (Table 3; Fig. 4). Both of these samples were more than 75 observations, so the 4 estimate, Dhat4 was used. Our results showed that the red fox was found active during the late night while golden jackal was found active during dawn and dusk (Fig. 5). The coefficient of overlap ( 4 = 0.47, 95% CI = 0.34–0.60) showed that the degree of diel activity overlap between the red fox and golden jackal was low, indicating a 47% temporal overlap between the two species. Habitat Suitability Modeling Initially, 26 variables, comprising 19 bioclimatic and 7 other environmental variables were considered Fig. 2. Flow Chart of Habitat Suitability Modeling: Flow Chart of Habitat Suitability Modeling of red fox and golden jackal in study area. page 6 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan Table 2. Field signs red fox and golden jackal: Direct and indirect signs of red fox and golden jackal recorded in four districts of the Pothwar Plateau Sr. No. Site Name Latitude Longitude Elevation (m) Scats Camera Trapping Howling Den Direct Field Sighting Foot Prints Road Kill/ Dead Body Red fox Golden Jackal Red fox Golden Jackal Howling Red fox Golden Jackal Red fox Golden Jackal Red fox Golden Jackal Red fox Golden Jackal Attock District 1 Dhoke Pathan 33.12421 72.34756 441 2 5 + -+- - + + ---- 2 Gali Jahangir 33.42368 72.63858 523 -7 + + - -------- 3 Dhoke Afghan 33.93554 72.41761 477 3 8 + - - -------- 4Makhad Shareef 33.12115 71.74447 517 -3 + + - -------- 5 Moorat Village 33.47991 72.88713 477 1 2 + -+-------- 6Ganda Kas 33.64954 72.14657 405 ---+ + -------- 7Ratwal 33.51771 72.69780 423 -7 + + + -------- 8 Ikhlas Pindi Gheb 33.25423 72.30924 420 - - +-+---+---- 9Ahmadal 33.28663 72.47887 533 ---+ + -------- 10 Fate Ullah 33.77718 72.57200 655 ---+ + -------- 11 Moorat Mor 33.49688 72.90563 441 4 6 -+- - - +-+ + - - 12 Basal 33.55228 72.23705 638 2 2 + -+- - + + ---- 13 Khairi Murat 33.46367 72.78375 904 1 12 -+-- - +----- Chakwal District 14 Dhermond 32.93083 72.21575 426 2 --- +---+---- 15 Sarkal 33.14803 72.90467 510 6 -+-+- - + + ---- 16 Ara Baharat 32.78206 73.08862 510 8 8 + -+----+- - + 17 Chumbi Surla 32.78719 72.79134 420 - - + + - - - +----- 18 Dhoke Fateshah 32.92398 72.15728 391 4 8 + + - -------- 19 Dhurnal 32.74083 72.13212 199 -13 -+ + -------- 20 Mutan Khurd 32.99256 72.01190 494 2 8 + + + ------+- 21 Sarkal Kasar 33.19077 72.72844 442 3 5 -+- -------- 22 Saghar 32.92575 72.25561 428 -1- - - - +------ 23 Thoa Mehram Khan 32.83335 72.23482 640 - - +-+-------- 24 Lawa 32.67763 71.96703 310 5 10 -+ + - - +----- Jhelum District 25 Ahmadabad, Graveyard 32.49867 72.83253 194 ---- +-------- 26 Andhri 32.79037 73.32580 424 -1 + - - -------- 27 Goorha Atum Singh 33.02134 73.30945 525 2 -+- - - - +----- 28 Jalal Pur Sharif 32.66531 73.42029 565 6 -+- - - - +----- 29 Mal Maira 32.83738 73.29863 411 2 1 + - - - - - - +--- 30 Potha 32.80794 73.25960 430 2 - - +-+---+ + - - 31 Rakh Parial 32.84432 73.15923 425 4 -+-+---+---- 32 Rehana Munda 32.98432 73.59257 257 ---+- -------- 33 Ahmadabad 32.52200 72.86322 235 4 9 + + - -------- 34 Kotila 32.76208 73.28798 292 3 1 + -+-------- 35 Mal Maira 32.83738 73.29863 411 4- - +- -------- 36 Sohawa 33.12883 73.42798 529 ---+ + -------+ 37 Dina 33.03630 73.55748 554 ---+- - +------ Rawalpindi District 38 Bhangali Shareef 33.24203 73.10698 529 -1 + -+-------+ 39 Dera Abbas 33.19248 73.31375 638 ---+ + -------- 40 Dhala 33.43101 72.91279 565 - - +-+-------- 41 Koont farm 33.12038 73.01442 517 2 5 - - + + ---+- - + 42 Jatli 33.18445 73.09298 542 -14 + -+-------- 43 Keral 33.65702 73.44326 923 - - +- - -------- 44 Mari Janjir 33.24677 73.14604 550 2 - - + + -------- 45 Sayed Kasran 33.11560 73.02741 485 4 10 -+ + ----+--- Total signs (N = 341; red fox = 131, golden jackal = 210) 78 147 34 24 25 2 2 10 6 6 2 1 4 page 7 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan for habitat suitability modeling. After assessing collinearity by using the Variance Inflation Factor (VIF) and considering the ecological requirements of both species, we removed 16 variables and retained only 10 variables for our model. Habitat suitability modeling performed well in terms of AUC (0.840 and 0.824) and TSS (0.675 and 0.615) (Fig. 6). In terms of AUCDiff value, the model was better fitted for the red fox (0.0213) compared to the golden jackal (0.0973). Major factors determining habitat suitability Jackknife tests of variable importance for the red fox showed that land use cover made the greatest contribution (24.6%), followed by annual precipitation (bio_12, 23.2%), Vector Ruggedness Measure (14.2%), distance from human settlements (7.8%), distance from water bodies (7.3%), temperature annual range (bio_07, 7.1%), distance from roads (5.0%), distance from poultry farms (4.2%), annual mean temperature Fig. 3. Distribution and Kernel density maps: GIS-based map showing (a and c), comparative distribution of red fox and golden jackal in the Pothwar Plateau and (b and d), Kernel density analysis of their field signs. Table 3. Camera trapping effort of red fox and golden jackal recorded in four districts of the Pothwar Plateau District Nights Camera station Photo capture Photo Captured of Non-target species False triggered Trap event (Red fox) Trap event (Golden jackal) Trap nights Attock 25 14 150 25 92 25 31 350 Chakwal 22 10 132 27 92 13 19 220 Jhelum 16 10 198 21 70 18 15 160 Rawalpindi 12 5 114 12 55 23 16 60 Total 75 39 594 85 309 79 81 790 N N N N page 8 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan (bio_01, 4.2%), and vegetation cover (NDVI, 2.5%) (Fig. 7a; Table 4). Land use cover showed the highest contribution in determining the habitat suitability of the red fox. Level-11 (rangeland or scrub forest) of land use cover had a peak value of 0.60. The habitat suitability for the red fox increased with increasing annual precipitation (bio_12) and reached a peak value of 0.65 at values ranging from 400 to 800. Habitat suitability Fig. 4. Camera trapped photos: Photos of (a), red fox and (b), golden jackal captured by camera trap in the study area. Fig. 5. Temporal Niche overlap: Temporal activity patterns of red fox and golden jackal in Pothwar Plateau. Solid line shows the activity of red fox while the dotted line shows the activity of golden jackal in 24 hours. The shaded area shows the overlapping. Fig. 6. Average training AUC: ROC verification of distribution of suitable habitat for (a), red fox and (b), golden jackal in the Pothwar Plateau. page 9 of 16Zoological Studies 64:17 (2025) © 2025 Academia Sinica, Taiwan Ridout MS, Linkie M. 2009. Estimating overlap of daily activity patterns from camera trap data. JABES 14:322–337. doi:10.1198/jabes.2009.08038. Roberts TJ. 1997. The mammals of Pakistan (revised ed.). Oxford University Press, Karachi, Pakistan. Sarwar M, Hussain I, Anwar M, Mirza S. 2016. Baseline data on anthropogenic practices in the agro-ecosystem of Pothwar plateau, Pakistan. J Anim Pl Sci 26:850–857. Scheinin S, Yom-Tov Y, Motro U, Geffen E. 2006. Behavioural responses of red foxes to an increase in the presence of golden jackals: A field experiment. Anim Behav 71:577–584. doi:10.1016/j.anbehav.2005.05.022. Shamoon H, Cain S, Shanas U, Bar-Massada A, Malihi Y et al. 2018a. Spatio-temporal activity patterns of mammals in an agroecological mosaic with seasonal recreation activities. Eur J Wildlife Res 64:1–10. doi:10.1007/s10344-018-1196-8. Shamoon H, Maor R, Saltz D, Dayan T. 2018b. Increased mammal nocturnality in agricultural landscapes results in fragmentation due to cascading effects. Biol Conserv 226:32–41. doi:10.1016/ J.BIOCON.2018.07.028. Shamoon H, Saltz D, Dayan T. 2017. Fine-scale temporal and spatial population fluctuations of medium sized carnivores in a Mediterranean agricultural matrix. Landscape Ecol 32:1243– 1256. doi:10.1007/s10980-017-0517-8. Sheikh K, Molur S. 2004. Status and Red List of Pakistan’s Mammals, based on Conservation Assessment and Management Plan, 312 pp, IUCN Pakistan. Sillero-Zubiri C, Switzer D. 2004. Canids: Foxes, wolves, jackals and dogs: Status survey and conservation action plan. 2nd ed. IUCN Canid Specialist Group, Gland, Switzerland and Cambridge, UK. Smith A. 2018. Fasterraster: Faster raster processing in R using GRASS GIS. R package version 0.4 1. Torretta E, Riboldi L, Costa E, Delfoco C, Frignani E et al. 2021. Niche partitioning between sympatric wild canids: The case of the golden jackal (Canis aureus) and the red fox (Vulpes vulpes) in north-eastern Italy. BMC Ecol Evol 21:129. doi:10.1186/ s12862-021-01860-3. Tsunoda H, Ito K, Peeva S, Raichev E, Kaneko Y. 2018. Spatial and temporal separation between the golden jackal and three sympatric carnivores in a human-odified landscape in central Blgaria. Zool Ecol 28:172–179. doi:10.1080/21658005.2018.15 04406. Tsunoda H, Newman C, Peeva S, Raichev E, Buesching CD et al. 2020. Spatio-temporal partitioning facilitates mesocarnivore sympatry in the Stara Planina Mountains, Bulgaria. Zoology 141:125801. doi:10.1016/j.zool.2020.125801. Tsunoda H, Peeva S, Raichev E, Kirilov KB, Uzunowa K et al. 2024. Anthropogenic activities facilitate temporal overlaps and spatial partitions among sympatric canids in a human-modified landscape of Bulgaria. Food Webs 39:e00344. doi:10.1016/ j.fooweb.2024.e00344. VanDerWal J, Shoo LP, Graham C, Williams SE. 2009. Selecting pseudo-absence data for presence-only distribution modeling: How far should you stray from what you know? Ecol Model 220:589–594. doi:10.1016/J.ECOLMODEL.2008.11.010. Wang G, Wang C, Guo Z, Dai L, Wu Y et al. 2020. Integrating maxent model and landscape ecology theory for studying spatiotemporal dynamics of habitat: Suggestions for conservation of endangered red-crowned crane. Ecol Indic 116:106472. Young N, Carter L, Evangelista P. 2011. A maxent model v3. 3.3 e tutorial (arcgis v10). Natural Resource Ecology Laboratory, Colorado State University and the National Institute of Invasive Species Science. Zaman M, Tolhurst BA, Zhu M, Jiang G. 2020. Den-site selection at multiple scales by the red fox (Vulpes vulpes subsp. Montana) in a patchy human-dominated landscape. Global Ecology and Conservation 23:e01136. Zhang G, Zhu AX, Windels SK, Qin CZ. 2018. Modelling species habitat suitability from presence-only data using kernel density estimation. Ecol Indic 93:387–396. doi:10.1016/ J.ECOLIND.2018.04.002. Zhu G, Liu G, Bu W, Gao Y. 2013. Ecological niche modeling and its applications in biodiversity conservation. Biodiversity Science 21:90–98. doi:10.3724/SP.J.1003.2013.09106. page 16 of 16Zoological Studies 64:17 (2025)