2.1 ǀ Study site and animals
This study was conducted around Duoma (103.01°E, 33.5°N), a village approximately 8.5 km southwest of the town of Ruoergai County in the Zoige wetland, the biggest plateau peat bog in the world (Zhang et al., 2005). The Zoige wetland is located in the eastern Qinghai–Tibet Plateau, southwestern China. The study site is a mosaic of grasslands, ground frost heaves, rivers and wet and dry wetland patches (Guo et al., 2020), and according to some local elders, the village has been here for at least 80 years.
The three marmot populations have been the subject of a ongoing behavioral ecology study since 2017, they live in different locations around the village and share the same habitat type (dry, flat patches with short grass and few frost heaves in the wetland) (Guo et al., 2020; Supporting Information Figure S1), but suffer different degrees of anthropogenic disturbances. The marmots living in front of the village, hereafter the high disturbance habitat (HDH), are persistently disturbed by the daily activities of local residents including passing motor vehicles and stray dogs. This interference does not directly kill marmots and does not alter their preferred habitat type (i.e. expel them from selected habitat to other unsuitable habitats) (Guo et al., 2020; Figure S1). Marmots living behind the village in the low disturbance habitat (LDH) endure relatively fewer disturbances than those living in front of the village. A third population living to the west of the village lives in a minimally-disturbed natural habitat (NH) and serves as a control group (Figure 1). We had no direct interactions like a routine collection of blood or tissue samples or simulated behavioral experiments with them during the burrow-related data collection in 2019. To further exam the burrow diversification derived flee strategy under different human disturbances, some individual-based flight initiation distance (FID) data were collected in June 2020 as an additional experiment to explore how human activities act on their habitat utilization features.
2.2 ǀ Sampling method andstatistical analyses
During the marmots’ active period (not in hibernation) in 2019, we classified the intensity of human disturbance of each habitat based on the degree of pressure from human activity on the different groups recorded during behavioral observations in 2018. The three study groups were designated as living in the high disturbance habitat (HDH), the low disturbance habitat (LDH), and the natural habitat (NH). To quantify the amount of human activity in the area, we recorded how many automobiles, motorcycles and stray dogs passed by the marmot habitat every 15 days from April 20 to October 5, 2019.
In each of the three habitats, we recorded the coordinates of reproductive burrows (the most extensively used burrows) and temporary burrows (used only occasionally for shelter) of every breeding pair, the pair-specific burrows were determined according to behavioral observation. The natural feature where all burrows occurred (i.e. hummock/mound or flat ground) and the physical parameters (long diameter, short diameter and height) of the mounds were also recorded to calculate their volume where burrows are located (simplify mound into a cone). The locations of all burrows were mapped in Google Earth to find: (1) the distance between each (adjacent) reproductive burrow, and reproductive burrows with geographical connectivity only were included in the following analysis. For example, the distance between NH8 and NH5, LDH3 and LDH14 were excluded in the subsequent analysis because they are isolated by a ditch (Figures S4, S5). A criterion finally result in 51, 47 and 31 inter reproductive burrow distances for HDH, LDH and NH; (2) the distance between all burrows (distance between temporary burrows, temporary burrows and reproductive burrows and 50 distances were randomly selected in each habitat to conduct the subsequent analysis); and (3) the distance from some reproductive burrows to the nearest road (only burrows next to the road with no other reproductive burrows between them and the road like HDH17, HDH18, LDH1 were included) (for detail please see Figure S2 to S5 and sheet named “distance to road” in Table S1). We calculated the density of breeding pairs by linking the outermost burrows recorded to form a perimeter and measured number of pairs inside, and link the outermost burrows of each breeding pair to calculate pair-specific home range size. The same procedure was applied to all three populations. Specially for natural habitat, because there is no direct disturbance from motor vehicles in the site, the same as the criterion in two disturbed habitats, the distance from the outermost reproductive burrow (NH2) (Figure S5) to the nearest road was used as the standard distance (311 m) to the road for all burrows in the natural habitat.
One adult individual in several pairs from each habitat was randomly selected to measure the flight initiation distance. A field assistant held binoculars from a long distance to observe and record, while Zhou Shuailing approached the focal marmot at a speed of 1m/s until the marmot started to run. FID (the distance between Zhou and the start point of the flee) of the focal individual was then measured (Blumstein et al., 2004). Finally, 28 FID samples from HDH, 20 from LDH, and 20 from NH (three more individuals from other undisturbed pairs were also included in the analysis) were included in the following analysis.
A chi-square test was used to determine: (1) whether there was seasonal variation in different human activities, (2) diversification of breeding pair density between each habitat, and (3) the variation of reproductive den site location in three habitats. A t-test was used to determine: (1) whether the intensity of different human activities was significantly different among the three habitats, (2) whether differences in parameters such as the number of burrows per reproductive pair, the distance between reproductive burrows and between all burrows, the distance between reproductive burrows and the corresponding nearest road were significantly different by population, and (3) the diversification of mound measurement (volume) selected as reproductive burrow site among different habitats. Besides, a t-test was also used to test if there were diversification on the FID of individuals and pair-specific home range size in different populations. All statistics were conducted in SPSS 20.0.