Fig 1.
The current distributions of P-horse and Khulan in China.
Regions near the western part of Sino-Mongolian border were highlighted by break lines (a) and zoomed into (b). The three extant Khulan habitats along the border were colored in light gray in (b), while two current release sites for P-horses were marked by circled “X”s. (c) showed the coverage of Kalamaili Nature Reserve and the distributions of P-horse (dark gray) and Khulan (light gray) in it. The distribution of P-horses was estimated by the 70% kernel home range of three collared harems (2007~2010).
Fig 2.
The geographic locations of the four permanent water points (dark triangles) and the two ranger stations (black halo dots) in Qiaomuxibai.
The background was this area’s topographic map with contour lines.
Table 1.
The status of four permanent water points in Qiaomuxibai, with the measures of quality (salinity) and safety (distance to station), and the results from data collections in 2010~2011.
Fig 3.
The percentages (a) and frequencies (b) of average daily water visits by P-horse and Khulan at four water points.
The texts inside bars are the average number and percentage of water visits by each equid per day per water point. Error bars represented the standard error (SE).
Table 2.
Partial Pearson’s correlation tests on factors might affect the water point visits of the two targeted equid species.
Fig 4.
The circadian rhythm (by hour) of P-horse (dark gray) and Khulan (light gray).
Each bar represented the percentage of total trapping events of that species recorded in that hour of day. Background colors indicated the day (white), twilight (light gray) and night (dark gray).
Table 3.
Two equids’ average daily daytime (Nday) and nighttime (Nnight) visit frequencies at four water points and their average Normalized Daytime Water Visit Indices (NDWVI).
Fig 5.
The number of water visitations by P-horse negatively affected the circadian pattern of water use (NDWVI) by Khulan at W02, the water point disproportionally used by both species.
The grey-scaled scatter point represented each observation within 10-day sliding window. All scatter points were transparent so that overlapped points would be darker in the plot. A linear regression line (solid) was fit to the dataset with the 95% Confidence Intervals (dashed).
Table 4.
Contingency table of behavioral responses by two equids to each other during direct encounters at water points.