Abstract: Plateau pika (Ochotona curzoniae) is a widely distributed rodent on the Qinghai-Tibet Plateau and one of the key species of the alpine grasslands. This study aimed to understand the spatial difference of potential distribution intensity of plateau pika in the source region of the Yellow River and its influencing factors. Field investigations and environmental data, combined with BIOMOD2 ecological niche simulation platform were used to carry out simulation and validation on the plateau pika distribution and disturbance intensity. The results showed that: (1) The model accuracy of generalized boosted regression models (GBM), random forest (RF) and maximum entropy model (MaxEnt) was obviously better than other models, while the RF had the best Kappa, true skill statistics (TSS), area under the curve (AUC) and critical success index (CSI). (2) After integration using the RF independent model with Kappa greater than 0.8, the Kappa, TSS, AUC and CSI values improved by 0.125, 0.129, 0.027, and 0.120, respectively, and the accuracy was remarkably improved. Those results indicated that the integrated model simulating the potential spatial distribution of plateau pika is reliable. (3) The number of pika holes had a positive linear correlation with the area of bald patches (R²=0.928), and the density of pika holes was higher where the bald patches were crowded. The more clustered the bald patches, the larger their area and the higher their aggregation. (4) The density of plateau pikas was higher in the north-central and central parts of the study area, and lower in the southeast and west. The area with moderate and light disturbance of plateau pika accounted for 51.61% of the total area, which mainly distributed in Maqu, Dari, and Jiuzhi counties, the central-eastern part of Machin County, the southern part of Tongde County, the western part of Maduo County, the northeastern part of Qumalai County and the northern part of Chengduo County. The area with severe and extreme disturbance accounted for 18.32%, and the extreme disturbance area was mainly distributed in Zeku County, Henan County, north of Xinghai County, southeast of Maqin County and north of Gande County. The severe disturbance area was mainly distributed in the periphery of extreme disturbance area, as well as the east of Tongde County, central part of Xinghai County, central and east parts of Maduo County, northwest of Dari County and central part of Gande County. (5) Temperature, vegetation, slope, rainfall, soil properties, and altitude were important factors affecting the distribution of plateau pika. The intensity of plateau pika disturbance was positively correlated with temperature difference, minimum temperature, ground temperature, normalized difference vegetation index (NDVI), seasonality of precipitation, and clay content, but negatively correlated with average altitude, sand content, and cation exchange capacity. (6) The minimum of NDVI in the extreme disturbance area was larger than other areas, and the difference of the maximum of NDVI in various disturbance areas was not significant. With increasing disturbance intensity, the maximum annual mean value of the ground temperature decreased, and clay content increased first and then decreased. The highest disturbance intensity of plateau pika concentrated in the area with altitude of 3100-4500 m.

Key words: plateau pika, potential distribution, disturbance intensity, alpine meadow, integrated model, source region of the Yellow River