The work was carried out on the western slope of Mt. Gongga (29°01ʹ‒30°05ʹN, 101°29ʹ‒102°12ʹE), located on the eastern edge of the Qinghai-Tibetan Plateau in Sichuan Province, China. The main peak, Mt. Gongga, is the highest point in the Hengduan Mountains (Fig. 1). The vertical climatic zonation on the western slope of Mt. Gongga includes a sub-alpine cool temperature zone, an alpine sub-frigid zone, an alpine frigid zone, and a permanent snow zone. The major vegetation types are deciduous forest, conifer–deciduous mixed forest, conifer forest, alpine scrub, alpine meadow, flat or gently sloping rocky areas, and glacier. Mt. Gongga is also one of the most eastern glacial areas in China with five valley glaciers of a length greater than 10 km, including Hailuogou glacier, Mozigou glacier, Yanzigou glacier and Nanmenguangou glacier on the eastern slope, and Gongba glacier on the western slope (Li et al. 2010b). Dozens of mountains with elevations over 6000 m surround the main peak of Mt. Gongga, and together with the glaciers form a magnificent landscape and the environment of the Tibetan Snowcock.
The survey was carried out from late June to early November 2016 with a survey area around of 650 km2. We divided the survey area into 5 km × 5 km blocks using geographic information systems (GIS), and two transect trails were established in each block with a distance of at least 1 km between them. When the transect survey began, trained staff deployed two camera traps (Ltl-6210MG, China) on each transect trail somewhere likely to be used by animals at their own discretion. In total, 103 camera traps were deployed between 3925 and 5084 m in elevation (Fig. 1). Each camera trap was carefully hidden in a rock pile to prevent animal disturbance. The camera traps were placed 30‒40 cm above ground and set to work 24 h/day with a two-second delay between consecutive exposures. If any activity triggered the camera, three consecutive photographs followed by a 9 s video were recorded. We recorded the GPS location, elevation, slope, aspect, and vegetation type at each camera sample site. The beginning date of each camera trap was the date when the camera was deployed. At the end of the survey, the camera traps were tested to confirm that they were still operational; if not, the date on the last photograph was recorded as the last operational date. The time and temperature were recorded automatically by the camera traps and shown on the photographs.
Camera traps that failed to collect data were first removed from the final dataset. Due to disturbances from curious animals and equipment failures, only 92 camera traps produced data included in our analysis. We defined a detection at a camera trap as one individual photograph of one species during a 30-min period. After importing and marking photographs captured by camera traps in DigiKam 5.3 (www.digikam.org), we extracted information through a process using the R package “camtrapR” (Niedballa et al. 2016). We obtained a relative abundance index by calculating the number of photo-captures obtained for each species within a period of 100 trap days. To understand the diel activity periods of this phasianid, we used the R package “overlap” to visualize its activity pattern.
To examine the habitat use of the Tibetan Snowcock, we divided the entire monitoring period into consecutive 5-day segments. Then, as described by Mackenzie et al. (2002), we set up a Tibetan Snowcock detection matrix, which would meet the two assumptions of this model as outlined in the introduction. Each element in the matrix represented one segment at one camera trap sampling site. We used 1 to represent that the Tibetan Snowcock was detected in this segment, used 0 to represent no detection, and used NA to represent data missing. Detection probability of the Tibetan Snowcock was assessed in relation to two detection covariates, and ten site covariates were considered to be potentially influential for its habitat use (Table 1). The elevation, slope, and aspect data were recorded by the field staff. The EVI (enhanced vegetation index) data were acquired from the Geospatial Data Cloud of the Chinese Academy of Sciences (http://www.gscloud.cn). The administration of the Gongga Mountain National Nature Reserve provided the raw data on rivers, settlements, and roads. Therefore, the other six site variables (distance to the nearest river, distance to the nearest settlement, distance to the nearest road, river density, settlement density, road density) were extracted using geographic information systems (GIS) and the raw data.
Pearson’s correlation test was used to identify collinearity between all continuous site covariates (Additional file 1: Table S1). Any combination of covariates with r > |0.6| was considered correlated (Tan et al. 2017). Distance to the nearest road and distance to the nearest settlement were removed, because they were correlated with road density and settlement density respectively. Since road density and settlement density could represent different (lines and points) human impact, we retained these two covariates in the analysis. We also retained elevation and EVI, because the two site covariates should affect the use of habitat in different ways, e.g., physiology versus food richness.
Then, we used the R package “Unmarked” (Fiske et al. 2011) and called on the occupancy model (Mackenzie et al. 2002) to estimate the occupancy rate and detection probability of Tibetan Snowcock. We modelled detection probability (p) by allowing the site covariates to remain constant. The significant contributing detection covariates were retained and used to model habitat use probability in relation to the site covariates (Long et al. 2011; Tan et al. 2017). We used R package “MuMIn” to run and list all the potential models. Akaike’s information criterion corrected (AICc) values were then used to rank the occupancy models (Sugiura 1978; Hurvich and Tsai 1991). All models with ΔAICc ≤ 2 were considered as competing models. The sum model weight of each covariate in these competing models was used to determine the most influential variables for the habitat use of the Tibetan Snowcock.