Density and habitat selection of Endangered red panda in Imawbum National Park, Kachin State, Northern Myanmar

doi:10.21203/rs.3.rs-2722094/v1

Download PDF

Research Article

Density and habitat selection of Endangered red panda in Imawbum National Park, Kachin State, Northern Myanmar

https://doi.org/10.21203/rs.3.rs-2722094/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 25 Aug, 2023

Read the published version in Mammalian Biology →

You are reading this latest preprint version

Imawbum National Park is a biodiversity hotspot, supporting habitat of the Endangered red panda. The National Park is part of the Northern Forest Complex in Kachin State, Myanmar, in the Eastern Himalayan Region. Forest degradation and habitat loss resulting from illegal logging is the primary threat to red pandas in the area. Assessments of the species’ density and habitat status in Imawbum National Park have been hampered by inadequate data and the difficult and rugged terrain that is present. In this study, we estimate red panda density using a spatial presence-absence model and predicted the area and type of suitable habitat, using an infinite weight logistic regression model. Red panda density was estimated as 58 animals/100 km² in our study area, with the predicted area of suitable habitat being1373 km² inside and 633 km² outside the park. Elevation and snow cover both positively influenced the probabilities of occurrence and detection, respectively. These results highlight the significance of Imawbum National Park as an important area of habitat red panda and the need to protect and conserve both the national park and the surrounding areas effectively to ensure the species’ long-term survival.

Hindu-Kush Himalaya (HKH)

Suitable habitat

Temperate rainforest

Illegal wildlife trade

Effective conservation

The red panda (Ailurus fulgens) is endemic to the Eastern Himalayas, ranging from western Nepal through Bhutan, India, Myanmar, and the western Yunnan province of China (Dorji et al., 2012), where it is found at high elevations (2,200 to 5,000 m), inhabiting Himalayan temperate forests (Nijboer et al., 2010). The habitat is characterized by a temperature ranging from 10 to 25°C and an average annual rainfall of 3500 mm (Roberts and Gittleman,1984). The species is Endangered on the IUCN Red List (IUCN, 2015)) due to habitat loss, degradation and fragmentation, as well as poaching for illegal trade (Glaston et al., 2015), with a declining global population estimated at less than 10,000 individuals (Glatston et al., 2015). The species’ status in the wild has been defined predominately in China and Nepal, with a few studies conducted in other parts of its range, leaving major gaps in its conservation and management planning (Thapa et al., 2018b). To improve population estimates and inform conservation efforts, accurate assessments of the species’ status, habitat requirements and ongoing threats are crucial (Glatston et al., 2015; Wei et al.,1999 and Choudhury, 2001).

Myanmar, the largest country in mainland Southeast Asia, has a rich biodiversity, with a total country area of 676,577 km² and extensive natural areas of approximately 290,410 km² (Forest Department, 2020), more than 40% of which are mountain ecosystems (Beffasti, 2011). The mountainous area in the northern part of the country are connected to the Hindu-Kush Himalaya (HKH). It covers an area of 317,640 km² and comprises approximately 31,000 km² of continuous forest, of which over 12,623 km² is recognized as potential red panda habitat (Thapa et al., 2018a) out of the global estimate of 134,975 km² (Thapa et al., 2018b; Kandel et al., 2015).

The Northern Mountain Forest Complex is a significant stronghold for Himalayan species including vulnerable species Takin (Budorcas taxicolor) and the newly discovered and identified critically endangered species Myanmar snub-nosed monkey (Rhinopithecus Strykeri). This area also presents a confluence of three significant biodiversity hotspots, namely the Indo-Burma, Eastern Himalayan, and southwestern China mountain regions, creating a unique and diverse ecosystem (Myers et al.,2000). The intersection zone serves as a crucial habitat for many threatened species, making it an important stronghold for biodiversity conservation. As with the rest of the country, the biodiversity of this area is under high ongoing threats from commercial over-exploitation of natural resources (Rao et al., 2010; MOECF, 2014), and wildlife trade to China via the Sino-Myanmar border (Nijman, 2015; Sapai, 2012). To protect biodiversity, three protected areas: Hkakaborazi National Park (3,810 km²), Hponkanrazi Wildlife Sanctuary (2,700 km²), and Imawbum National Park (1,562 km²), have been established, creating a corridor with the Northern Mountain Forest Complex covering approximately 8,072 km² (Renner et al., 2007; Yang et al.,2019). However, has been a loss of 270 km² of habitat between 2001 and 2012 due to deforestation (Shreya, 2014). As such, the preservation of the Northern Mountain Forest Complex is of paramount importance for maintaining biodiversity in Northern Myanmar (Marchese,2015; Ye et al.,2021).

Imawbum National Park is an important component of the Northern Mountain Forest complex and is a recognized biodiversity hotspot that shares its borders with the Three Parallel Rivers, a World Heritage Site located in Yunnan, China, underscoring the area’s significance biodiversity conservation (Nijma, 2015; Lin et al.,2022). Forty-nine mammals and 176 bird species have been recorded within the boundaries of Imawbum National Park (Nijma, 2015). Of particular note is the presence of endangered red panda which was first recorded in the Imawbum National Park in 2010 (Lwin, 2010; Lin et al.,2022). Moreover, the Imawbum National Park is a part of the Hindu-Kush Himalayan region (Wester et al.,2019), which is of great significance for conserving the red panda in Myanmar. However, over the past decade, illegal logging, mainly by Chinese firms, has severely damaged forests inhabited by the red panda in Imawbum National Park and illegal logging continues in recent years (Myint, 2015; Myint, 2018). This commercial illegal logging and associated road construction have contributed significantly to the red panda’s habitat destruction in the region (Meyer et al., 2017; Ren et al., 2017).

Illegal trade in wildlife products can have a devastating impact on endangered species. The red panda is targeted by local hunters generally for its skin and is sometimes caught in traps meant for other species (Lin et al., 2022). The absence of comprehensive and systematic surveys has limited our knowledge of the abundance, status, distribution and availability of suitable habitats of the red panda within the Northern Mountain Forest Complex, including Imawbum National Park (Wei et al.,2014; Thapa et al.,2018a). Although the red panda is protected by law in Myanmar (Beffasti, 2011), there is currently a lack of specific plans for its long-term conservation.

Therefore, this study aimed to estimate the red panda’s population status in Imawbum National Park, using data collected through camera trapping, investigate the species’ habitat selection, and study the effects of human disturbance in order to predict the extent of suitable habitat within and surrounding the park.

Study Area

The study was conducted in Imawbum National Park, 1,562 km2, (26.35˚ N to 26.87˚ N and 98.29˚ E to 98.65˚ E) and adjacent areas of its national park (Lin et al., 2022).

The area lies between the N’Mai Hka River and the Myanmar-China border, adjacent to the Gaoligonshan Nature Reserve in Yunnan, China. The area has an elevation ranging between 1,672 and 4,004 m (the highest peak) with five distinct forest types: subtropical forest (930 km²), temperate rainforest (447 km²), rhododendron, pine and conifer forest (97 km²), shrubland, grassland and bamboo (45 km²) and alpine scrub (29 km²). In addition, it has rocky areas, 12 km², where trees cannot grow. The subtropical forest and rhododendron, pine, and conifer forest are found at elevations between 1,800 and 2,700 m, while the temperate rainforest occurs at 2,700 m. During the winter season, higher parts (> 2754 m) of the area are covered by snow from December to March (Nijman, 2015; Myint, 2018).

Camera Trapping

Red panda locations were recorded using digital camera traps (Bushnell Passive Infra-Red). The survey was conducted in November-December 2018 (42 stations), March 2019 (27 stations) and November 2019 (20 stations). A total of 89 camera traps were located in the study area at elevations between 2,540 and 3,467 m. The camera traps were set 500 m apart using a systematic random design, away from wildlife trails and human forest tracks. Only two camera traps were located less than 500 m from each other, due to the steep and rugged landscape (Fig. 1). To prevent snow from covering the camera traps, they were mounted on tree trunks at over one meter above ground level. Elevation and forest types were recorded at each camera trap location. The camera traps were set 24 hours, by using 3-second intervals to capture between each photograph. Photographs of the target species taken at over 30-minute intervals were considered as independent detections (O’Brien et al.,2003). In addition, we investigated the activity patterns of the red pandas during daytime (06:01 AM to 06:00 PM) and nighttime (06:01 PM to 06:00 AM) using timestamps from the photographs.

Red panda density estimate

Red panda density was estimated with the spatial presence-absence model, SPA (Ramsey et al., 2015), using data from the camera trap survey. The spatial presence-absence model was an extension of a spatial count model (Chandler and Royle, 2013), based on a spatially explicit capture-recapture model. We analyzed the spatial presence-absence in a Bayesian framework following Ramsey et al. (2015). The model assumed spatially correlated detections could infer an individual’s range center from multiple camera stations. Ramsey et al. (2015) showed the model could make unbiased density estimation when the trap spacing is less than the radial length of the animal’s home range. Based on Zhang et al. (2009) and Bista et al. (2022), we adopted an average home range size of 1.4–2.2 km² with a radius of approximately 670–840 m for the red panda. Thus, our trap spacing (500 m) was less than the home range radius as suggested by Ramsey et al (2015). However, spatial count models can conservatively misidentify neighboring individuals as recaptures of a single individual, leading to density underestimation (Sun et al 2022). Aware of this issue, we estimated our baseline information to benefit this little-studied endangered species (Glatston et al., 2015).

We ran models with an informative prior distribution on the spatial-scale parameter (σ), [gamma (1.75, 2.2)] based on telemetry data, to accommodate the average home range size of 1.4–2.2 km² (Zhang et al 2009; Bista et al 2022). We specified a vague prior [dunif(0,1)] for the baseline detection probability (g0). We used a prior density of 0.3 red panda per one square kilometer based on Glatston et al. (2015) and a weak informative prior [dbeta(2.2, 5.8)] for ψ (psi). We assumed a maximum (M) plausible abundance value of 300 individuals and the probable abundance (N) of 60 pandas (see more details at https://mmeredith.net/blog/2021/Priors_for_DA.htm). We defined the state-space of the study area as the extent of the trapping polygon with a 5 km buffer, equivalent to adding ~ three home ranges from the trapping polygon, resulting in a 241 km² state-space for the study area. We used data augmentation with M = 300 and ran six chains of Markov chain Monte Carlo (MCMC) for 170,000 iterations each, with a thinning rate of 1, an adaptation period of 2,500 iterations, and a burn-in period of 70,000 iterations. The SPA analysis was implemented in R software version 4.1.2 (R Core Team, 2022) using JAGS 4.3 (Plummer, 2003) and the ‘jagsUI’ package (Kellner 2021). Model convergence was evaluated by examining the trace plots of the chains and the R-hat value, where a value close to 1 indicates convergence (Brooks and Gelman, 1998). To ensure that the posterior distribution was drawn from a large enough effective sample size to get a reliable inference of the summary statistics for the mean and its 95% credible interval, we checked the MCMC iteration of each parameter until all Monte Carlo standard errors (as a percentage of the posterior standard deviation, MCEpc) were under 1.5% (Lunn et al., 2013), implemented with the ‘mcmcOutput’ package (Meredith, 2020). We reported the precision using the coefficient of variation (CV = standard deviation/mean), with CV < 20% generally considered acceptable for informing management (Williams et al., 2002).

Habitat selection

We investigated red panda micro-habitat selection using an occupancy method (MacKenzie et al., 2018). Detection histories of red panda was recorded using camera traps over a period of 167 days, with each day representing a separate occasion. A binary system was used to represent the recording, ‘1’ for detection and ‘0’ for non-detection events. A Bayesian framework was used to estimate the species’ probability of detection and probability of occupancy for single-season occupancy models (Kellner et al., 2022). Models containing one sampling covariate (snow) and two site covariates (elevation and habitat types) were implemented using the ‘ubms’ package (Kellner et al., 2022) in the R software. We considered snow as a sampling covariate to determine whether it has any effects on red panda detection. We constructed a categorical covariate with binary values (1 or 0) representing the presence or absence of snow on each day, respectively. For site covariates, the elevation was standardized before running the models by subtracting the mean and dividing by two standard deviations (Gelman and Hill, 2007). Elevation and habitat types were highly correlated. As a result, we did not include both in the same model.

We compared the detection model that included snow as a predictor variable (ψ ~ 1, p ~ snow) and the null model (ψ ~ 1, p ~ 1). In subsequent models, we included both the sampling covariate and site covariate to assess the latter’s effect. We compared all models with a null model using leave-one-out cross-validation (LOO) (Vehtari et al., 2017). We used 8,000 MCMC iterations of three chains to estimate the parameters. Model diagnostic was performed by assessing the trace plot and R-hat values. To validate the model goodness-of-fit, we used residual plots of the posterior distribution, following Wright et al. (2019), with and without covariates (Gelman and Hill, 2007).

Mapping the probability of red panda occurrence

Species recorded and available locations

Red panda recorded locations were derived from camera trap data collected during two different periods: randomly set between 2010 and 2018 and systematically set in 2018–2019 (see above). Out of 374 locations, red panda were detected in 129 locations. Beforehand, we tested for spatial autocorrelation to determine if these locations were independent and the appropriate distance between them. The spatial autocorrelation testing showed 1,250 m as the appropriate distance between each location. Next, we created a 625 m circular plot around those recorded locations and selected the circular plots that did not overlap each other. In addition, we also selected the circular plots with an overlap of less than 5%, following Sukumal et al. (2019) and Lwin et al. (2021). For circular plots showing overlaps of more than 5%, we selected only those with the latest records of red panda. Finally, we derived 40 independent recorded locations used in modeling the occurrence probability of the species, using an infinite weight logistic regression model (Hefley & Hooten, 2015) to investigate the relationship between the habitat covariates and the probability of occurrence based on used and available locations. A total of 40 independent recorded locations (presence-only data) were input as used location data. We created a buffering polygon around these 40 independent recorded locations with a radius of 10 km; then, the available locations (background points) were systematically created within this buffer area with a distance 1,250 m between them. In total, 1,236 available locations were obtained and compared with the 40 used locations by running the probability of occurrence model. To identify which of the predictor variables (elevation or forest type) has the strongest influence on the species’ probability of occurrence, we extracted information on the habitat characteristics of elevation and forest types within the 625 m radius circular plots around the used and available locations. The forest type was derived from data provided by FFI Myanmar (2015), with a total of seven forest types. We combined temperate and subtropical forests into one type as these two were mixed and did not show distinct characteristics in our study area. Agricultural landscapes were not included as a forest type for the species occurrence model, but instead set as a factor disturbance variable for the species occurrence. In the end, five forest types were extracted and used as continuous variables in the analysis. Digital Elevation Map downloaded from WorldClim (2005) was used to extract the elevation of the used and available locations. Disturbance variables such as distance to village and distance to agricultural areas were also used as continuous predictor variables. The locations of villages throughout Imawbum were derived from the village profile survey and local hunter interview survey (Lwin, 2011; Lin et al., 2022). We used the nearest tool in ArcGIS software to measure the distance from the center point of the used and available locations to the disturbance variables.

Before running the analysis, we checked the correlation among the continuous predictor variables. When predictor variables were highly correlated (r >|0.5|), we ran the analysis by including those correlated variables in separate models. To compare the effect sizes of the continuous predictor variables on the same scale, we standardized the variables by subtracting the mean from each value and dividing by the standard deviation (Gelman, 2008). Outliers were removed before running the analysis.

We compared the constant models (intercept-only model without predictor variables) to models containing variables of interest using the difference in the Akaike Information Criterion (AIC) and AIC weights. The best-performing model with the lowest AIC and highest AIC weights was selected. We also used the area under the curve (AUC) to evaluate the best-performing model. AUC values range from 0.5 (model no better than random) to 1.0 (model with high accuracy), with AUC 0.7–0.8 indicating an acceptable model performance, 0.8–0.9 indicating an excellent performance and > 0.9 indicating an outstanding performance. All statistical analysis was conducted using R software (R Core Team 2022).

Predicting species distribution

In order to predict the probability of red panda occurrence over the entire Imawbun mountain range (4,031 km²), we created a 1,107.5 m square grid (to be standardized with the area of 625 m radius circular plots) over our study area, which also covered the Imawbum National Park. The probability of occurrence, based on the best-performing model, was extracted from each of these square grids. We defined a total of four potential suitable habitats over the entire study area: (1) highly suitable areas, probability of occurrence > 0.75, (2) medium suitable areas, probability of occurrence between 0.50 and 0.75, (3) low suitable areas, probability of occurrence between 0.25 and 0.50, and (4) very low suitable areas, probability of occurrence < 0.25.

Red panda density estimation

A total of 151 independent red panda detections were recorded at 39 camera trap locations (43.8%) out of the 89 locations surveyed. The elevation range of detections was between 2,704 m and 3,349 m. From all detections, only one camera trap captured two red pandas at the same time, while the other camera traps captured only one red panda. The red panda activity patterns were recorded during the daytime (76%) and nighttime (24%).

The mean density of red panda within the 241 km² study area was estimated at 58 ± 17 (95% CI 28–91) individuals per 100 km², with a coefficient of variance (CV) of 29%. Our posterior distributions were sufficiently drawn as the Monte Carlo standard error percentage was less than 1.1% for all parameters.

Habitat selection

The model containing elevation and snow [ψ(ele) p(snow)] as predictor variables proved the best-fit model with the lowest AIC value and highest AIC weight of 58% (Table. 1). Elevation and snow had a positive influence on the habitat use probability (probability of occupancy; β = 1.51, SD = 0.70, 95%CrI = 0.25–3.00) and detection probability (β = 3.14, SD = 0.20, 95%CrI = 2.74–3.54), respectively (Figs. 2 & 3). The model's goodness of fit was assessed by examining the residuals plot against fitted values.

Probability of occurrence and suitable areas

The best model for predicting the probability red panda occurrence over the entire Imawbun mountain range included elevation, temperate and subtropical forests, and distance to the village as predictor variables (Table. 2). The probability of occurrence was higher at high elevations (β = 2.83, SE = 0.55) and increased temperate and subtropical forest areas (β = 1.22, SE = 0.58) but lower closer to the village (β = 1.15, SE = 0.32).

The total area of the Imawbun mountain range is 4,031 km². Of these, the highly suitable area was 875 km² (22%), the medium suitable area was 1,130 km² (28%), the low suitable area was 903 km² (22%), and the very low suitable area was 1122 km² (28%) (Fig. 4, Table. 3). Thus, high – medium suitable habitat for red panda account for 50% of the entire Imawbun mountain range. Within Imawbun National Park (1,561 km²), we identified highly suitable areas, 721 km² (46%), medium suitable areas, 652 km² (41%), low suitable areas, 180 km² (12%), and very low suitable areas, 9 km² (1%) (Table. 3). The area outside Imawbum National Park comprised highly suitable areas (155 km², 7%), medium suitable areas (478 km², 19%), low suitable areas (724 km², 29%), and very low suitable areas (1113 km², 45%) (Table. 3).

Although this study does not cover the entire Northern Mountain Forest Complex, it provides important information on the status of red panda in the Imawbum Mountain range, a predicted stronghold of the species inside and outside the national park within the forest complex. Moreover, ours is the first reliable density estimation of red pandas in the Imawbum Mountain range and is an improvement over previous estimates based on telemetry data. Our baseline density estimates might not be precise enough to inform management because the CV was larger than the recommended value. However, we are confident that our approach could improve the precision in the future with a larger sample size.

Our estimated red panda population density of 0.58 animals/km² is relatively high when compared to estimates for other parts of its range: between 0.33 and 0.50 animals/km² for the Wolong Nature Reserve in China (Reid et al., 1991), between 0.23 and 0.36 animals/km² for Langtang National Park in Nepal (Yonzon & Hunter, 1991) and between 0.26 and 0.60 animals/km² for Singalila National Park in India (Bahuguna et al., 1988; Pradhan et al., 2001). Even though there are no density estimates for red panda in Bhutan, the country has taken significant steps towards conserving this species with the development of relevant action plans (Dorji et al.,2012; NCD, 2019). Several factors possibly influenced our higher density estimates, including the method used and the effects of human disturbance on the red panda population. For example, the abundance of bamboo species, the main food of red panda and a ban on cattle grazing have benefited the red panda population in Singhalila National Park (Pradhan et al., 2001). The low red panda density estimates in China and Nepal can be attributed to the presence of grazing areas, roughly 34% of Langtang National Park in Nepal, the scarcity of bamboo, and habitat destruction (Reid et al. 1991; Yonzon & Hunter, 1991). Poaching, hunting, habitat loss and fragmentation in China, higher demand for farming and firewood in India, and tourism development, livestock grazing, road construction and over-exploitation of forest resources in Nepal have caused red panda density in remaining habitats to be low (Wei et al.,1999; Choudhury, 2001: Bista et al.,2017a). Finally, the higher estimated density we recorded in the Imawbum region could be a direct consequence of its situation adjacent to the Gaoligongshan Nature Reserve in China, which has the largest portion of red panda suitable habitat globally (Kandel et al.,2015; Thapa et al., 2018a). In addition, the area is relatively undisturbed owing to the low human population density in surrounding villages, the presence of suitable forests, and the absence of intensive hunting directed toward the species (Myint, 2015; Lin et al., 2022). However, Imawbum National Park lies between the Irrawaddy River to the west and the Salwin River to the east, isolating the red panda population. In the long term, however, this isolation could end as increasing human activities such as deforestation, road construction and other forms of land development will likely result in habitat fragmentation and a decline in the red panda population (Acharya et al.,2018; Bista et al.,2022).

At our site, red panda were detected at elevations ranging between 2,704 and 3,349 m, and occupancy was positively affected by the elevation gradient (Fig. 2). Bhatta et al. (2014) also show that red panda in Nepal have a preferred elevation range of 2,800 to 3,300 m and avoid areas outside of this range. Additionally, the occupancy of red panda in China, India, Nepal, and Bhutan has been recorded between 2,200 and 4389 m (Reid et al.,1991; Pradhan et al.,2001; Saroj et al.,2021; Dorji et al., 2011). Most studies over the species range suggest it prefers subtropical, temperate, and broadleaf forests with undergrowth of bamboo (Bhatta et al., 2014; Choudhury, 2001; Dorji et al., 2011; Thapa et al., 2018b). Similarly, our results also show red panda preference for tropical and temperate rainforests (Table. 2), which comprised 89% of the total area of Imawbum National Park, when compared to rhododendron, pine and conifer forests, shrublands, grasslands with bamboo and alpine scrub. We also found a positive relationship between red panda detections and the presence of snow, confirming its preference for higher elevations (Fig. 3). The positive effects of elevation and snow on the probability of habitat use and detection indicate their important roles in determining the probability of occupancy and detection in our study area (Fig. 2&3). Human disturbance, defined as the distance to the villages and agricultural landscapes, also affected the probability of the red panda’s occurrence. Human activities such as hunting, logging, shifting cultivation, and agricultural expansion have been identified as the major threat to the species’ habitat (Panthi et al., 2017; Panthi et al., 2019). Therefore, red panda have increasingly occupied areas far from human settlements (Dendup, 2016; Bista et al., 2017b).

Out of the total 4031 km² area of the Imawbum Mountain range, an estimated 2,006 km² (50%) was considered a moderate to high suitable habitat (Table.3, Fig. 4), of which 1373 km² was within Imawbum National Park and 633 km² in its surrounding areas (Table. 3). This area represents up to 16% of the total suitable habitat defined for Myanmar, 12,623 km² following Thapa et al. (2018a). The results highlight the importance of Imawbum National Park for red panda long-term survival and the need to protect and conserve both the national park and the surrounding areas effectively. In addition, due to its connection with the Gaoligongshan National Nature Reserve in China, the Northern Mountain Forest Complex can be considered an important stronghold for red pandas in the Hindu-Kush Himalayan. However, the remoteness of the region, inadequate wildlife conservation efforts, weak collaboration on biodiversity conservation by local people, and lack of law enforcement have continued to pose challenges to the long-term red panda conservation efforts. Moreover, the current political situation in Myanmar has stifled conservation efforts in the National Park since its establishment in March 2020 due to underfunding and understaffing. Being a remote mountainous region, the locals are dependent primarily on forest resources for their livelihood. Illegal extraction of forest resources, such as timber and bamboo, and slash-and-burn agriculture are increasingly becoming the main cause of habitat destruction and fragmentation in the Imawbum region. Access to the forest is made easier by old illegal logging roads constructed over a decade ago and still used to extract the remaining logs and bamboo (Nijman, 2015; Lin et al., 2022). Due to the harsh terrain, paucity of information, and limited accessibility of areas where red panda live, gathering detailed information and protecting them remain difficult.

In order to ensure the long-term survival of red panda and mitigate threats to them in Imawbum National Park, Myanmar, specific conservation plans should be developed, including community-based conservation to protect the species’ habitat. Involving local communities in conservation activities can be an effective strategy to accelerate the red panda’s conservation in remote areas with weak law enforcement. Although red panda are not targeted primarily for food by local hunters in Imawbum National Park, their skin and body parts are sometimes hunted for use in traditional Chinese medicine. Therefore, hunting and trade also need to be prevented along the Myanmar-China border. To reduce habitat loss and habitat fragmentation for red pandas in Imawbum National Park, the Myanmar government should curb illegal logging, and the establishment of long-term sustainable agriculture is essential.

In conclusion, Myanmar does not yet have specific plans for red panda conservation, so it is important to develop a concrete plan as soon as possible to mitigate relevant threats and ensure the survival of this species in the Northern Mountain Forest Complex, including Imawbum National Park and adjacent areas. Conserving red panda requires a comprehensive and integrated approach involving multiple stakeholders, including government agencies, local communities, and conservation organizations. A comprehensive conservation plan should include habitat protection and restoration measures, biological and ecological research, and community engagement.

Acknowledgements

We would like to thank the members of the Local Conservation Group from the Imawbum region who supported us in conducting the fieldwork. We wish to thank Helmsley Charitable Trust for providing financial support for the field survey and Fauna & Flora International for sharing the forest classification GIS data for data analysis. We are grateful to Win Naing Oo, Khaung Haung and members of our survey team for conducting the camera trap survey.

Author contributions Project design: Aung Ko Lin, Dusit Ngoprasert, Niti Sukumal, Stephen Browne, Tommaso Savini; data analysis and writing: all authors.

Conflicts of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Acharya KP, Shrestha S, Paudel PK, Sherpa AP, Jnawali SR, Acharya S, Bista D (2018) Pervasive human disturbance on habitats of endangered red panda Ailurus fulgens in central Himalaya. Global Ecol Conserv 15:1–9. https://doi.org/10.1016/j.gecco.2018.e00420
Bhatta M, Shah KB, Devkota B, Paudel R, Panthi S (2014) Distribution and Habitat Preference of Red Panda (Ailurus fulgens fulgens) in Jumla District, Nepal. Open J Ecol 4:989–1001. http://dx.doi.org/10.4236/oje.2014.415082
Brooks SP, Gelman A (1998) General methods for monitoring convergence of iterative simulations. J Comput Graphical Stat 7:434–455
Bista D, Shrestha S, Sherpa P, Thapa JG, Kokh M, Lama TS, Jnawali RS (2017a) Distribution and habitat use of red panda in the Chitwan-Annapurna Landscape of Nepal. PLOS, pp 1–16. https://doi.org/10.1371/journal.pone.0178797
Bista D, Shrestha S, Kunwar AJ, Acharya S, Jnawali SR, Acharya KP (2017b) Status of gastrointestinal parasites in red panda of Nepal. PeerJ 5:e3767. https://doi.org/10.7717/peerj.3767
Bista D, Baxter GS, Hudson NJ, Lama ST, Weerman LJ, Murray PJ (2022) Space use, interaction and recursion in a solitary specialized herbivore: a red panda case study: Endangered Species Research, Vol. 47: pp 131–143, 2022 https://doi.org/10.3354/esr01171
Beffasti L, Gallanti V (2011) Myanmar Protected Areas: Context, Current Status and Challenges (PDF). Istituto Oikos, Biodiversity and Nature Conservation Association, Milano, Yangon
Bahuguna C, Dhaundyal S, Vyas P, Singhal N (1998) The red panda at Singalila National Park and adjoining forest: a status report. Small Carnivore Conservation 19:11–12
Choudhury A (2001) An overview of the status and conservation of the red panda Ailurus fulgens in India, with reference to its global status. Oryx 35:250
Chandler RB, Royle JA (2013) Spatially explicit models for inference about density in unmarked or partially marked populations. Annals of Applied Statistics 7:936–954
Dorji S, Rajaratnam R, Vernes K (2012) The Vulnerable red panda Ailurus fulgens in Bhutan: Distribution, conservation status and management recommendations. Oryx 46(4):536–543. 10.1017/S0030605311000780
Dorji S, Vernes K, Rajaratnam R (2011) Habitat Correlates of the Red Panda in the Temperate Forests of Bhutan. PLoS ONE 6(10):e26483. 10.1371/journal.pone.0026483
Dendup P, Cheng E, Lham C, Tenzin U (2016) Response of red panda Ailurus fulgens fulgens (Cuvier, 1825) to anthropogenic disturbances and its distribution in Phrumsengla National Park, Bhutan. Oryx 51(4):1–8
Forest Department (2020) Forestry in Myanmar. Ministry of Natural Resources and Environmental Conservation (MONREC), Naypyidaw
Glatston A, Wei F, Zaw T, Sherpa A (2015) Ailurus fulgens. The IUCN Red. List of Threatened Species. e.T714A110023718. https://doi.org/10.2305/IUCN.UK.2015-4.RLTS.T714A45195924.en
Gelman A, Hill J (2007) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press
Gelman A (2008) Scaling regression inputs by dividing by two standard deviations. Stat Med 27:2865–2873
Hefley TJ, Hooten MB (2015) On the existence of maximum likelihood estimates for presence-only data. Methods Ecol Evol 6:648–655
IUCN (2015) The IUCN Red List of Threatened Species. Version 2015-4. Available at: www.iucnredlist.org
Kandel K, Huettmann F, Suwal MK, Ram Regmi, Ganga, Nijman V, Nekaris KAI, Lama ST, Thapa A, Sharma, Hari P, Subedi TR (2015) Rapid multi-nation distribution assessment of a charismatic conservation species using open access ensemble model GIS predictions: Red panda (Ailurus fulgens) in the Hindu- Kush Himalaya region. Biol Conserv 181:150–161. 10.1016/j.biocon.2014.10.007
Kellner KF, Fowler NL, Petroelje TR, Kautz TM, Beyer DE, Belant JL (2022) ubms: An R package for fitting hierarchical occupancy and N-mixture abundance models in a Bayesian framework. Methods Ecol Evol 13:577–584
Kellner K (2021) jagsUI: A Wrapper Around 'rjags' to Streamline 'JAGS' Analyses. R package version 1.5.2, <https://CRAN.R-project.org/package=jagsUI
Lwin N, Aung SS, Aung TN, Aung ZM (2010) Hoolock Gibbon and Biodiversity Survey in Wusut Area, Nmai Hka River Catchment, North-east Kachin State, unpublished report. Biodiversity and Nature Conservation Association (BANCA) and Fauna & Flora International (FFI)
Lin AK, Lwin N, Aung SS, Oo WN, Lum LZ, Grindley M (2022) The conservation status of Red Panda in north-east Myanmar. In: ANGELA R. GLATSTON (ed) RED PANDA, Biology and Conservation of the First Panda, 2nd edn. Academic Press is an imprint of Elsevier, pp 475–488. https://doi.org/10.1016/B978-0-12-823753-3.00023-5
Lwin N, Sukumal N, Savini T (2021) Modelling the conservation status of the threatened hoolock gibbon (genus Hoolock) over its range. Global Ecol Conserv 10.1016/j.gecco.2021.e 01726
Lwin N, Grindley M (2011) Village Profiles, Myanmar Snub-nosed Monkey Area, Maw River Catchment, Kachin State, unpublished report. Biodiversity and Nature Conservation Association (BANCA) and Fauna & Flora International (FFI)
Lunn D, Jackson C, Best N, Thomas A, Spiegelhalter D (2013) The BUGS Book: A Practical Introduction to Bayesian Analysis. Chapman and Hall/CRC
MOECF (2014) Myanmar Convention on Biological Diversity 5th National Report, Republic of the Union of Myanmar, Nay Pyi Taw, 79 pp
Myint W, Thura T, Phyo EE (2015) Mount Imawbum Proposed Protected Area, Flora Survey Report, unpublished report. Fauna & Flora International (FFI)
Myint W, Maung N, Thura T, Phyo EE (2018) Vegetation and Botanical Survey report of new propose protected national park Mount Imawbum, unpublished report. Fauna & Flora International (FFI)
Meyer D, Momberg F, Matauschek C, Oswald P, Lwin N, Aung SS, Roos C (2017) Conservation status of the Myanmar or black snub- nosed monkey Rhinopithecus strykeri. Fauna & Flora International, Yangon, Myanmar and German Primate Center
MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LL, Hines JE (2018) Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence. Acdemic Press
Meredith M (2020) mcmcOutput: Functions to Store, Manipulate and Display Markov Chain Monte Carlo (MCMC) Output_. R package version 0.1.1, <https://CRAN.R-project.org/package=mcmcOutput
Myers N, Mittermeler RA, Mittermeler CG, Da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403(6772):853–858. https://doi.org/10.1038/35002501
Marchese C (2015) Biodiversity hotspots: a shortcut for a more complicated concept. Glob Ecol Conserv 3:297–309
Nijboer J, Dierenfeld ES (2010) Red panda nutrition: how to feed a vegetariancarnivore. https://doi.org/10.1016/B978-1-4377-7813-7.00014-8
NCD (2019) Red Panda Conservation Action Plan (2018–2023): Ensuring the future of red panda landscapes through national and regional collaboration. Nature Conservation Division, Department of Forests and Park Services, Ministry of Agriculture and Forests, Thimphu, Bhutan
Nijman V (ed) (2015) The conservation status of mammals and birds in the Imawbum Mountains, Kachin State, Northern Myanmar. Fauna & Flora International, Cambridge, UK, p 76
O’Brien TG, Kinnaird MF, Wibisono HT (2003) Crouching tigers, hidden prey: Sumatran tiger and prey populations in a tropical forest landscape. Anim Conserv 6:131e139
Panthi S, Wang T, Sun Y, Thapa A (2019) An assessment of human impacts on endangered red pandas (Ailurus fulgens) living in the Himalaya. Ecol Evol. 10.1002/ece3.5797
Panthi S, Khanal G, Acharya KP, Aryal A, Srivathsa A (2017) Large anthropogenic impacts on a charismatic small carnivore: Insights from distribution surveys of red panda Ailurus fulgens in Nepal. PLoS ONE 12(7):e0180978. 10.1371/journal.pone.0180978
Pradhan S, Saha GK, Khan JA (2001) Ecology of the red panda Ailurus fulgens in the Singhalila National Park, Darjeeling, India. Biol Conserv 98:11–18
Plummer M (2003) JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling. Proceedings of the 3rd International Workshop on Distributed Statistical Computing (DSC 2003), Vienna, Austria
Ren GP, Yang Y, He XD, Li GS, Gao Y, Huang ZP, Ma C, Wang W, Xiao W (2017) Habitat evaluation and conservation framework of the newly discovered and Critically Endangered black snub-nosed monkey, vol 209. Biological Conservation, pp 273–279
Renner SC, Rappole JH, Leimgru P, Kelly DS, Shwe NM, Aung T, Aung M (2007) Land cover in the northern forest complex of Myanmar: new insights for conservation. Oryx 41(01):27. https://doi.org/10.1017/S0030605307001603
Roberts MS, Gittleman JL (1984) Ailurus fulgens. Mammalian Species 222:1–8
Rao M, Htun S, Zaw T, Myint T (2010) Hunting, livelihoods and declining wildlife in the Hponkanrazi wildlife sanctuary, North Myanmar. Environ Manag 46(2):143–153. https://doi.org/10.1007/s00267-010-9519-x
Ramsey DSL, Caley, Peter A, Robley A (2015) Estimating population density from presence-absence data using a spatially explicit model. The Journal of Wildlife Management, 79(3), pp 491–499. doi:10.1002/jwmg.851
R Core Team (2022) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.URL https://www.Rproject.org/
Reid DG, Jinchu H, Yan H (1991) "Ecology of the Red Panda Ailurus fulgens in the Wolong Reserve, China". Journal of Zoology. 225 (3): 347–364. doi:10.1111/j.1469-7998. 1991.tb03821.x
Sapai M (2012) Impact of wildlife Trade on Conservation in Kachin State, Myanmar. TRAFFIC Bulletin Vol. 24 No. 2 (2012), pp 57–64
Shreya D Saving Myanmar’s red pandas by protecting land, educating people https://news.mongabay.com/2014/11/saving-myanmars-red-pandas-by-protecting-land-educating-people/
Sun C, Burgar JM, Fisher JT, Burton AC (2022) A cautionary tale comparing spatial count and partial identity models for estimating densities of threatened and unmarked populations / https://doi.org/10.1016/j.gecco.2022.e02268
Sukumal N, Dowell SD, Savini T (2019) Modelling occurrence probability of the Endangered green peafowl Pavo muticus in mainland South-east Asia: applications for landscape conservation and management. Oryx, pp 1–10. 10.1017/S003060531900005X
Saroj S, Arjun T, Damber B, Natasha R, Ang PS, Krishna PA, Shant RJ, Sonam TL, Sony L (2021) Distribution and habitat attributes associated with the Himalayan red panda in the westernmost distribution range. Ecol Evol. 10.1002/ece3.7297
Thapa A, Hu Y, Wei F (2018a) The endangered red panda (Ailurus fulgens): Ecology and conservation approaches across the entire range https//. doi.org/10.1016/j.biocon.2018.02.014
Thapa A, Wu R, Hu Y, Nie Y, Singh PB, Khatiwada JR, Yan L, Gu X, Wei F (2018b) Predicting the potential distribution of the endangered red panda across its entire range using MaxEnt modeling. Ecology and Evolution, –. doi:10.1002/ece3.4526
Vehtari A, Gelman A, Gabry J (2017) Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Stat Comput 27:1413–1432
Wester P, Mishra A, Mukherji A, Shrestha AB (2019) The Hindu Kush Himalaya Assessment (Mountains, Climate Change, Sustainability and People)., 10.1007/978-3-319-92288-1(), –. doi:10.1007/978-3-319-922881
Wright Wj, Irvine KM, Higgs MD (2019) Identifying occupancy model inadequacies: can residuals separately assess detection and presence? 2019 Jun; 100(6): e02703. doi: 10.1002/ecy.2703.Epub 2019 Apr 18
Wei FW, Traylor-Holzer K, Leus K, Glatston A (2014) Red Pandas in China Population and Habitat Viability Assessment (PHVA) Workshop Final Report. IUCN SSC Conservation Breeding Specialist Group, Apple Valley, MN
Wei F, Feng Z, Wang Z, Hu J (1999) Current distribution, status and conservation of wild Red Pandas Ailurus fulgens in China. Biol Conserv 89(3):285291. 10.1016/S0006-3207(98)00156-6
Williams BK, Nichols JD, Conroy MJ (2002) Analysis and management of animal populations. Academic Press, San Diego, California, USA
Yang Y, Ren G, Li W, Huang Z, Lin AK, Garber PA, Ma C, Yi S, Momberg F, Gao Y, Wang X, Li G, Behie A, Xiao W (2019) Identifying transboundary conservaiton priorities in a biodiversity hotspot of China and Myanmar: implications for data poor mountainous regions. Glob Ecol Conserv 20:1–13. https://10.1016/j.gecco.2019.e00732
Ye HL, Lin W, Guogang L, Kyaw WM, Kyaw S, Rui CQ (2021) Diversity, distribution and conservation of large mammals in northern Myanmar, Global Ecology and Conservation, doi:https://doi.org/10.1016/j.gecco.2021.e01736
Yonzon PB, Hunter ML (1991) Conservation of the red panda, Ailurus fulgens. Biol Conserv 57:1–11
Zhang Z, Hu J, Yang J (2009) Food habits and space-use of red pandas Ailurus fulgens in the Fengtongzhai Nature Reserve, China: food effects and behavioural responses. Acta Theriol 54:225–234. https://doi.org/10.4098/j.at.0001-7051.017.2008

Table 1

Occupancy models of red pandas from camera trap surveys between 2018 and 2019 at Imawbum National Park, Myanmar. K is the number of parameters, AIC is Akaike information criterion values, delta AIC is the absolute difference in the AIC values of each model from the top best model, and AIC_wt is a measure of relative support for each model.
Models	K	AIC	Delta AIC	AIC_wt
psi(elevation) p(snow)	4	930.06	0.00	0.58
psi(hab) p(snow)	6	931.12	1.06	0.34
psi(.) p(snow)	3	934.22	4.15	0.07
psi(.) p(.)	2	1140.79	210.73	0.00

Table 2

Parameters that influence the probability of red panda occurrence model based on the used locations (random camera trap detections during 2010–2018 and systematic detections in 2018–2019) and available locations (background points) at Imawbum Mountain range, Myanmar.
Parameters	Estimate	SE	Lower CI	Upper CI
(Intercept)	-11.16	0.30	-11.82	-10.62
Elevation	2.83	0.55	1.82	3.98
Temperate-Subtropical Forest	1.22	0.58	0.24	2.54
Near-village	1.15	0.32	0.52	1.80

Table 3

Predicted red panda suitable habitat based on the percentile of the probability of occurrence inside and outside of Imawbum National Park
Suitable levels	Percentile	Area (km²)	percentage
Inside Park
Very Low	0.00–0.25	9	1
Low	0.25–0.50	180	12
Medium	0.50–0.75	652	41
High	0.75–1.00	721	46
Total area		1561	100
Outside Park
Very Low	0.00–0.25	1113	45
Low	0.25–0.50	724	29
Medium	0.50–0.75	478	19
High	0.75–1.00	155	7
Total area		2470	100
Very low probability corresponds to percentiles between 0.00 and 0.25, low probability corresponds to percentiles between 0.25 and 0.50, medium probability corresponds to percentiles between 0.50 and 0.75, and high probability corresponds to percentiles between 0.75 and 1.00.

Download PDF

Journal Publication

published 25 Aug, 2023

Read the published version in Mammalian Biology →

Reviewers agreed at journal
24 Mar, 2023
Editor assigned by journal
23 Mar, 2023
First submitted to journal
22 Mar, 2023

You are reading this latest preprint version

Density and habitat selection of Endangered red panda in Imawbum National Park, Kachin State, Northern Myanmar

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Method

Camera Trapping

Red panda density estimate

Habitat selection

Mapping the probability of red panda occurrence

Species recorded and available locations

Predicting species distribution

Results

Red panda density estimation

Habitat selection

Probability of occurrence and suitable areas

Discussion

Declarations

References

Tables

Status:

Journal Publication

Version 1