Camera trapping survey data at the Yarlung Zangbo Grand Canyon National Nature Reserve in the eastern Himalayas

The study was carried out inside the Yarlung Zangbo Grand Canyon National Nature Reserve (29°05′–30°02′ N, 94°39′–96°6′ E) in Nyingchi City in the southeast Tibet Autonomous Region of China. Camera trap detections of terrestrial mammals were collected during the dry season from November 2020 to April 2021 and November 2021 to April 2022 to avoid the heavy rainy seasons in the region. The mean trapping efforts were 134 days (89 to 147 days range). We used Yianws L720 camera traps to conduct the survey. To determine how anthropogenic factors shape spatiotemporal associations among terrestrial mammals, we set up camera trapping stations within the nature reserve based on intensity of human activities and distance to nearest human settlement. We deployed 350 camera stations in the reserve with different degrees of anthropogenic disturbance, maintaining at least 800 m between them. We affixed camera traps to trees between 80 and 100 cm off the ground, and they were not baited. Camera sensitivity was set to “low” to reduce false detections triggered by nonanimal movements. We set cameras to take three photos per detection event, with 3 s delay between subsequent detections. Camera trap photos were later identified to species when possible. We combined all human presence photos into a single “Human” categorization representing a variety of human activities detected around a camera station (e.g. resource gathering, livestock grazing, recreation etc.). All photos of the same species (including humans) at the same camera station were considered independent detections if separated by at least 1 h. Several camera stations were invalid due to camera malfunctions or camera loss. The total valid sampling effort was 43,163 camera days from 322 camera stations that operated effectively. The target species consisted of ground-dwelling mammal species observed in more than 10 camera stations and weighing more than 1 kg. Thus, the occurrence dataset consists of presence-absence information on 17 mammal species at 322 camera stations. For each camera trapping station, we derived human modification and percent of forest cover with a buffer radius of 500 m. We explored the degree of habitat modification based on the Human modification (HM) map metric (Kennedy et al., 2019). We derived percent of forest cover based on the 250 m Moderate Resolution Imaging Spectroradiometer (MODIS) imagery (MOD44B Vegetation Continuous Fields (VCF) yearly product) of the study area for the period of 2021. We also quantified the level of acute direct human presence around each camera trapping station by calculating the independent detections of human-related activities (e.g., livestock grazing, forest resource collection and tourism) per 100 camera-trap days.本研究在中国西藏自治区东南部林芝市的雅鲁藏布大峡谷国家级自然保护区（Yarlung Zangbo Grand Canyon National Nature Reserve）内开展，地理坐标为29°05′–30°02′ N，94°39′–96°6′ E。为避开该区域的强降雨季节，于2020年11月至2021年4月、2021年11月至2022年4月两个旱季收集陆生哺乳动物的红外相机陷阱（camera trap）监测数据，平均诱捕时长为134天（区间为89至147天），采用Yianws L720型红外相机陷阱开展调查。为探究人为因素如何塑造陆生哺乳动物的时空关联，本研究依据人类活动强度及距最近人类定居点的距离，在自然保护区内布设红外相机监测站点：共在保护区内布设350个红外相机监测站点，各站点间至少保持800米的间距，涵盖不同程度的人为干扰梯度。将红外相机固定于距地面80至100厘米的树干上，未设置诱饵；将相机灵敏度调至"低"档，以减少非动物活动引发的误检测。设置每检测到一次活动即拍摄3张照片，相邻两次检测之间的间隔为3秒。后续对采集到的红外相机照片尽可能进行物种鉴定：将所有包含人类活动的照片统一归类为"人类（Human）"类别，涵盖相机站点周边检测到的各类人类活动（如资源采集、牲畜放牧、休闲娱乐等）。同一相机站点内，同一物种（包含人类）的照片若间隔至少1小时，则视为独立检测事件。部分相机站点因设备故障或遗失被判定为无效，最终有效采样时长为43163相机日，涉及322个正常运行的监测站点。本研究的目标物种为在10个以上相机站点被观测到、体重超过1千克的陆生哺乳动物物种，因此本出现-缺失数据集包含322个监测站点上17种哺乳动物的分布数据。针对每个红外相机监测站点，以500米为缓冲半径，提取人类改造（Human modification, HM）指数以及森林覆盖率：本研究基于Kennedy等人2019年提出的HM指数量化生境改造程度；基于研究区域2021年的250米分辨率中分辨率成像光谱仪（Moderate Resolution Imaging Spectroradiometer, MODIS）影像的MOD44B植被连续场（Vegetation Continuous Fields, VCF）年度产品，提取森林覆盖率。此外，通过计算每100个相机日中与人类活动相关的独立检测次数（如放牧、森林资源采集及旅游等），量化每个相机站点周边的直接人类活动强度。