2.1 Study area
The study was carried out at Tukchak in Gyama valley, Meltrokongkar county, Tibet (29°41.095’N, 91°40.423´E) in 2005, 2009 and 2010. Here, the average annual temperature was 6.5°C and annual precipitation was about 600 mm in the period 2000–2007 (data from the Meltrokongkar county meteorological station). The study area is situated at an altitude of approx. 4100–5000 m and consists of a valley with two major hill sides facing towards north and south. The south-facing slope is steeper, dryer and warmer than the north-facing slope. In between were some smaller slopes facing in a more easterly or westerly direction. Vegetation in these slopes largely resembled that of the north-facing slope. A representative part of the north-facing slope where we located most nests is shown in Figure 1.
In June 2005, before most nests were found, we carried out a basic vegetation survey in the study area in an attempt to get a general overview of the plant species that grew there and their density. This was done in 12 squares in each of the south- and north-facing slopes (i.e. a total of 24 squares). The vegetation survey followed two transect lines situated at the approximate centre-lines of the slopes and stretching from the bottom of the valley and up towards the highest point. The surveyed squares were distributed along these transect lines with neighbouring points being separated by approximately 100 meters of altitude, as indicated by a Garmin eTrex® GPS. Within each square we measured densities of shrub and herbs within areas of 5 × 5 m and 1 × 1 m, respectively. Six species of shrub were recorded on the north-facing slope, of which the most common were Rhododendron primulaeflorum, R. nivale, Salix oritrepha and Potentilla fruticosa. This differed markedly from the south-facing slope where we recorded seven species of shrub and where the species Spiraea tibetica, Potentilla fruticosa, and Sabina pingii dominated in numbers. Bush density (number of bushes/25 m2) was higher on the north-facing slope than on the south facing slope (north-facing slope: 12.58 ± 6.3, south-facing slope: 9.07 ± 5.90; two-sample t-test: t = 2.34, df = 22, p = 0.03). Thorny bush species were mainly distributed on the south-facing slope and included Rosa sericea, Berberis hemleyana, Caragana maximovicziana and C. jubata. In total 31 species of herbs were recorded in the north-facing slope and 28 species in the south-facing slope, of which 16 species occurred in both slopes. There were no significant differences in herb densities (number of individual plants/m2) between north-facing and south-facing slopes (north-facing slope: 30.99 ± 122.71, south-facing slope: 14.10 ± 52.5; two-sample t-test: t = 0.91, df = 22, P = 0.37).
Gyama is one of the most important animal husbandry regions for Lhasa Prefecture and the study area has probably been used by grazing livestock for a very long time. Major livestock today are horses, yak, sheep, goat and cattle. Some wild mammals are also commonly grazing in the area, for instance Blue Sheep (Pseudois nayaur) and White-lipped Deer (Cervus albirostris) which both have considerable populations in the region. The lowermost parts of the study site mainly consisted of farming areas in which major crops included barley, spring wheat, bean, colza, potato, and some vegetables. Mammalian predators in the study area included Siberian Weasel (Mustela sibirica), Red Fox (Vulpes vulpes), Tibetan Sand Fox (V. ferrilata), and Lynx (Felis lynx). Common avian predators are Magpie (Pica pica), Raven (Corvus corax), Golden Eagle (Aquila chrysaetos) and Sparrowhawk (Accipiter nisus).
2.2 Field procedures
We searched for nests from late April to late June each year. Except in 2005, when field work was less intensive, we searched systematically for nests across the whole study area with the assistance of 5–10 local people. Nest searches were carried out twice a week. We divided the study area between hillsides east and west of the main river (Gyama puchu) and scheduled two days on the east side and one day on the west side (with farmland) to cover the complete area. From the foot of slopes, or edges of farmland, nest searchers walked through the terrain with approx. 100–200 m distance (depending on the number of participants) and searched carefully for nests up to the top of slopes (or to the opposite edge of the farming area). Nests were found by flushing incubating females or by using sticks to lift vegetation and uncover nests which were left unattended. Presence of Tibetan Partridges was revealed by calling males in early spring and gave an indication of where nests would be situated. We did not observe Tibetan Partridges outside nest areas which would have indicated that some nests were overlooked. Nevertheless, it should be noted that, despite the large nest-searching effort, we cannot exclude the possibility that some nests were left undetected by us. The nest densities and nearest neighbour distances reported in this paper should therefore be considered minimum and maximum estimates, respectively. Nests were marked with small pieces of coloured plastic or cloth at a distance of 10 m. In order to minimize disturbance, all nest-site characteristics were recorded after a clutch was completed or following depredation or desertion. We recorded the nest’s geographical location and altitude with the GPS. Slope aspect (direction) and degrees of inclination were recorded using a compass.
To measure areas of the study site we first used the ruler tool in Google Earth to outline the whole study area and each habitat category. From the resulting aerial photo we used the software Adobe Photoshop CS5 Extended to measure areas. Distance was calibrated from the scale automatically generated by Google Earth. In this way we found the total study area to cover 22.9 km2 (10.2 km2 of north-facing slopes, 11.4 km2 of south-facing slopes, 1.2 km2 of farming area and 0.05 km2 of two small villages). In calculations of nest density we considered a marginal area of 2.6 km2 of the south-facing slope to be unsuitable habitat for the Tibetan partridge because there were almost no bushes growing there. In line with this assumption we did not record any nest in this area. The villages were also considered unsuitable breeding habitat for the species. Consequently, when calculating nest densities we use an area of 20.3 km2.
We sampled data on the general habitat within a 5 × 5 m square around each nest, with the nest located in the centre. Within each of these squares we estimated percentage of bare ground and counted the number of plant species and the number of yak dung. We used the latter as a crude proxy for grazing intensity in our analyses. Both fresh and old dung was present in the nest areas but since we did not separate between them it is possible that the density recorded by us differed from the density experienced by the partridges prior to nest building. Yet, if trampling risk is important in nest site selection, the partridges may well use other cues than dung density to assess grazing intensity. It is therefore unknown whether varying dung age would bias the results in any way. The proportion of vegetation cover within each square was estimated by sight. Vegetation height (cm) in the nest area was defined as the average of five measurements: in the centre and in each corner of the 5 × 5 m sample areas. For each nest we recorded the species of plant that covered it and measured the plant’s height and basal area. The latter was defined as length × width of the widest part of the foliage. For each nest we estimated the proportion that was covered by the foliage when viewing the nest from above (hereafter ‘foliage cover’). We also measured the distance between nests and the nearest path or track. These were typically made by grazing animals and were distributed fairly evenly throughout the whole study area. To see if nest placement was non-randomly associated with variables of the nest habitat, we compared nest site information with similar data from control sites located 100 m from the nests in both a southern and a northern direction. At these two sites we located the nearest plant of the same species as the one which covered the focal nest and used it as a control plant. To get a comparable measure for foliage cover we here estimated how much of the nest would have been covered by the control plant if the nest was located at an equal position as for the nest plant, both with regard to distance from plant stem and direction. The control plant was also used as the centre point for a 5 × 5 m control plot. We used the mean value of the two control measurements in analyses, making sample sizes equal among nest samples and control samples.
Distances between nests were measured to the nearest 1 m using a Nikon Laser 1200 range finder. A FLUKE® digital thermometer was used to record soil temperature at a depth of seven cm from the surface within five cm from the rim of the nest cup, or from the stems of the control plants. We measured soil surface temperatures in the shade by placing the temperature sensor directly on the surface, whereas air temperature and relative air humidity was measured in the shade one meter above ground. For the latter we used a CEM DT-616 CT professional temperature-humidity meter. Soil humidity was measured from samples collected at random 10 cm away from the nest or at the same distance from the stem of the control plant. We weighed soil samples immediately upon sampling and again after being dried at 40°C for 50 hours. Soil humidity is defined as the percentage of mass lost from the original sample during drying. Since sampling of microclimatic data was not done at the same time these data could be biased by seasonal progress. Therefore we only analyse microclimatic data by paired comparisons in which the pairs of data from nest and control areas were taken at the same time. Measurements and recordings of all data were done by the same person (TD) in all three breeding seasons.
2.3 Data analyses
Statistical analyses were conducted using R version 2.11.1 (R Core Team [2010]). To explore if a bush species was over- or under-represented as nest cover we used a modified version of Ivlev’s electivity index (Ivlev, [1961]):
Here E
i
is the electivity index for species i, r
i
indicate percentage of species i in sample of nest bushes, and n
i
indicate percentage of bush species i available in the study area. The latter was estimated from the 24 samples collected in the general vegetation survey in the study area (see above). The value of E
i
varies between −1 and +1. Positive values indicate a preference for the plant in question, whereas negative values indicate avoidance.
We here limited our investigation to focusing on bush species covering the nests and therefore excluded three nests placed below herbs in the calculation of electivity indices. In addition, two nests were placed below a mixture of R. primulaeflorum and P. fruticosa but where the former was the dominant species. Hence, we treated these nests as being covered by R. primulaeflorum. Similarly, one nest found under a mixture of R. sericea and S. pingii was treated as covered by R. sericea.
We also tested statistically if a given plant species was over- or under-represented as nest cover. We then compared the proportion of nests covered by that plant with its availability in the study area (the estimated proportion of all bushes from the general vegetation survey; see above. In these cases we used a Chi-square test with Yates’ continuity correction or, when any of the cells in the 2 × 2 table were smaller than 5 (Siegel and Castellan, [1988]), Fisher's exact test. Paired t-tests were used to test for differences between nest samples and control samples. Averages are given with standard deviations and all tests are two-tailed with alpha = 0.05.
The study adhered to international guidelines for the treatment of animals in behavioural research (ASAB 2012).