Fig 1.
Map of Qinghai Province, China.
Provincial boundary indicated by bold line, county boundaries by fine lines, and the location of Gouli Township within Dulan County with an asterisk (*). Inset shows location of Qinghai Province within the People’s Republic of China.
Table 1.
Characteristics of 4 pastures on which experimental exclosures and removals were conducted.
Fig 2.
Schematic diagrams showing the experimental designs to exclude livestock and reduce pikas.
A. Experimental design, showing the fence (solid black line), the area within which paired, non-exclosure plots were located (wide green line), and the 30 m extending beyond the experiment on each side (orange-hatching) within which pikas were reduced by snap-trapping during 2010–2013, Gouli Township, Qinghai Province, China. B. Schematic of exclosure experiment, showing typical placement of 12 vegetation plots established in September 2009, Village Five, Gouli Township, Qinghai Province, China. Bold solid line represents the 100 m2 livestock exclusion fence. Within the inner 64 m2 of the exclosure (dashed line), 6 vegetation plots (0.5 m2 each; colored squares) were randomly distributed (see text). Six paired vegetation plots that were similar in vegetation composition and density to each of the random plots (colors represent pairing) were selected within the area extending 10m in each direction from (but not within 2 m, dashed-line, of) the exclusion fence.
Table 2.
Initial conditions at each of the 12 experiments.
Fig 3.
Experiment 3, shown in July 2010.
Visible are the larger amounts of litter, primarily of Stipa purpurea, within the livestock exclosure than outside. Also evident is the larger amount of live S. purpurea inside than outside. In this experiment, S. purpurea fresh biomass increased the first year following exclosure, but subsequently declined to levels similar to or below those outside the exclosure.
Table 3.
Pikas removed and observed at each experiment.
Table 4.
Mean winter diets of livestock and wildlife.
Fig 4.
Histograms showing effects of livestock exclosure on biophysical variables in experiments where significant (P < 0.025).
Shown are mean and 95% confidence intervals (error bars) for exclosure-caused differences in percent plot covered by litter (orange), live vegetation (dark green), erosion index (gray), and bare soil (brown). Values prior to livestock exclosure indicated by solid shading, values after livestock exclosure indicated by vertical hatching (experiments 1 and 2), horizontal hatching (experiments 3 and 4), sparse points (experiments 5 and 6), diagonal upper left to lower right hatching (experiments 7 and 8), diagonal upper right to lower left hatching (experiments 2 and 5), reverse color sparse points (experiments 6 through 11), bricks (experiment 9), and diamonds (experiment 12).
Fig 5.
Trends on time, showing effects of livestock exclosure on biophysical variables where significant (P < 0.025).
Shown are mean differences between within and outside livestock exclosure fences of litter (orange), live vegetation (dark green), bare soil (brown), and erosion index (gray), with all values calibrated to zero difference in 2009, the year exclosures were established. Individual experiments are indicated by solid (experiments 1 and 3), dashed (experiments 5 and 6), and dash-dot (experiment 3) lines.
Fig 6.
Histograms showing effects of livestock exclosure on vegetation in experiments where significant (P < 0.025).
Shown are mean and 95% confidence intervals (error bars) for exclosure-caused differences in fresh biomass (g/m2) of Stipa purpurea (light green), Kobresia (dark green), Heteropappus altaicus (red), and Cardamine tangutorum (orange). Values prior to livestock exclosure indicated by solid shading, values after livestock exclosure indicated by open bars (experiments indicated adjacent to bar pairs).
Fig 7.
Trends on time, showing effects of livestock exclosure on plant species where significant.
Stipa purpurea (light green), Heteropappus altaicus (lavender), Cardamine tangutorum (orange), Thermopsis (yellow), Potentilla bifurca (lime), and Oxytropis spp. (blue). Shown are mean differences between within and outside livestock exclosure fences, with all values calibrated to zero difference in 2009, the year exclosures were established. Individual experiments are indicated by slender solid (2), compound dash (3 and 4), short dash (4 only), long dash (7 and 8), compound short dash (9), dash-dot (10), and thick solid (12) lines.
Fig 8.
Mixed-effect linear model showing the pattern of paired-plot differences in Stipa purpurea fresh biomass in experiment 2.
Raw data are scaled so that paired-plot differences in the baseline year 2009 were all set to zero, and vegetation plots (n = 6, solid circles) were treated as random effects. Y-axis scale is paired-plot differences in g/0.5 m2. Values in 2010 (year 0) were significantly greater than zero (P = 0.0045; greater annual production of S. purpurea above-ground biomass within than outside the exclosure), but the dynamic reversed beginning in 2011 (β = -114.56 g/m2/yr, P = 0.0005).
Fig 9.
Patterns of biophysical variables where differences among comparable experiments were attributable to reductions in pika abundance (P < 0.025).
Trends indicated by red lines are mean values from experiments in which pikas were reduced; blue lines from control experiments. A) litter cover, paired experiments 3 and 4 (solid lines), and experiments from pastoralist K (dashed lines); B) percent cover consisting of live vegetation, experiments 3 and 4; C) percent bare soil, experiments 1, 3, and 4; and D) erosion index, experiments 7 and 8 (solid lines), and experiments 1, 3, and 4 (dashed lines).
Fig 10.
Patterns of vegetation biomass where differences among comparable experiments were attributable to reductions in pika abundance (interaction between experiment and period of pika reduction where implemented, P < 0.025).
Trend indicated by red lines are mean values from experiments in which pikas were reduced; blue lines from control experiments. A) Stipa purpurea, experiments in pastures of Pastoralist B (solid lines) and experiments 3 and 4 (dashed lines); B) S. purpurea in experiments in pastures of Pastoralist S (solid lines), and experiments 7 and 8 (dashed lines); C) Leymus secalinus, experiments in pastures of Pastoralist K (solid lines), and Pastoralist S (dashed lines); D) Carex spp., experiments 2 and 5; E) Kobresia spp., experiments 7 and 8; F) Heteropappus altaicus, experiments in pastures of Pastoralist B (solid lines), and 7 and 8 (dashed line); G) Potentilla bifurca, experiments 7 and 8 (solid lines) and in pastures of Pastoralist K (dashed lines); H) P. bifurca, experiments in pastures of Pastoralist B (solid lines), and Pastoralist S (dashed lines); I) Thermopsis lanceolata, experiments in pastures of Pastoralist S.
Table 5.
Summary of the effect of livestock presence on quantifiable response variables, by livestock exclosure experiments.
Fig 11.
Relationship of differences between excluded and grazed plots in annual rate of change of S. purpurea biomass production (2009–2013) with its initial abundance in 2009.
Shown are paired, mean differences of slopes on time.
Table 6.
Summary of the effect of pikas on quantifiable response variables.