Fig. 1. The study area in the North-Western part of the Koksoy river valley. A: Overview of the study area, location of burrows, search area and sampling sites. B: Part of the aerial image with the location of the analysed burrows. The green channel is enhanced, so vegetation is visible. Note, that this is not an exact visualisation of plant occurrence, as it was not manually corrected and can be biased by varying lighting conditions in different parts of the image. C: Example of sampling spots of plant biomass around a burrow with 20 m buffer. D: Group of 20 m buffers around burrows and the type of relationship between the distance from the closest burrow entrance and the plant cover (see Fig. 3). The aerial image is edited like in C. E: Location of the study area in Tajikistan.

Fig. 2. The distribution of parameters within burrows: a) Mean distance from the closest burrow entrance of vegetation hexagons and bare soil hexagons. For clarity, SD is only shown for vegetation hexagons (error bars) b) Total plant cover in buffers c) Adjusted R²of the linear binomial model of the relationship between the distance from the closest burrow entrance and plant cover. The number of burrows is smaller than in other parameters, as models were only done for burrows with total plant cover above 2%. d) The number of types of relationship between the distance from the closest burrow entrance and plant cover, based on the shape of the trend line. For plots a-c, the values are sorted from smallest to largest. Colours indicate the type of relationship between the distance from the closest burrow entrance and plant cover (see Fig. 3)

Fig. 3. Examples of trend lines between the distance from the closest burrow entrance and plant cover based on binomial linear models, assigned to different types of relationship.

Fig. 4. The relationship between the distance from the burrow entrance and parameters of plant samples. Only pairs with significant relationship are shown. Plant species and Spearman’s rho are given. The dashed lines in the N:P plot indicate values 14, below which N limitation can be assumed and 16, above which P limitation can be assumed. See discussion for details on the constraints of this inference.

Fig. 5. Stable nitrogen isotope ratios of plants and marmot faeces collected on each burrow. Asterisks indicate significant differences between δ¹⁵N values of faeces and plant species sampled from one burrow, based on the Wilcoxon test (p<0.05). The colour of the asterisk indicates the plant species.

Appendix 1. Comparison of nutrient content and stable isotope ratios between the low and high cluster within plant species that were sampled in both clusters. Pairs that are bold differ significantly (p<0.05). TS – test statistic value, t for t -test, W for Wilcoxon test. * - Welch’s t -test for unequal variances