Study area
Shengjin Lake National Nature Reserve (30°15ʹ‒30°30ʹN; 116°55ʹ‒117°15ʹE), is situated south of the Yangtze River. It is a seasonally inundated freshwater lake of ca. 140 km2 and a shoreline of 156 km in winter (Fig. 1). During the wet season, water level increases to ca. 17 m a.s.l. and falls to lower than 10 m a.s.l. in the dry season. A sluice built in 1965 is used to regulate water flow between Shengjin Lake and the Yangtze River, allowing water retention in winter when river levels fall. Average annual precipitation is 1600 mm, with most rain during April to August. Average annual temperature is 16 °C and average January temperature is 4 °C.
Study sites
Data presented in these analyses were collected at Sites A and B at Shengjin Lake from mid-December 2008 to early February 2009 (Fig. 1). Site A (30°22′01.47″N; 117°04′10.99″E) was an open flat area covered by water until late November. After early winter water level recession, this site was used by Swan Geese during early December. From mid-December, this site was abandoned because of the combined effects of rapidly receding water levels and dry weather conditions, resulting in a larger area of unexploited habitat. Site B (30°22′01.47″N; 117°04′10.99″E) is a relatively enclosed bay ca. 80 m wide, heavily used by Swan Geese from mid-December 2008 to early February 2009. At both sites, the sediment type is clay and Vallisneria natans was the dominant aquatic plant over-wintering in the form of tubers buried in the muddy lake substrate.
Diet composition
In winter 2008/2009, Swan Goose faeces were collected at the two sites, separately stored in paper bags and oven-dried at 50 °C for 48 h using an oven (DHG-9053A, Shanghai). At least 20 faecal pellets from the same day and site were pooled and mixed thoroughly by hand. Vegetation samples were also collected at each site. Dicotyledonous plant epidermis patterns were derived by coating leaves with nail polish, the layer peeled off when dry and photographs of the impressions of cell patterns were taken under a microscope. Sub-samples of faeces were diluted with water in a petri dish, and drops taken from this suspension, placed on a glass slide, spread out evenly and covered with a cover slip. From each slide, 20 fields were identified using photomicrographs and area measured using a grid of 0.04 mm2 squares in the microscope eyepiece. However, no leaf epidermal fragments were found in the faeces, since all Swan Goose faeces contained fragments of Vallisneria tubers, readily identifiable by lack of cell structure and undifferentiated tissue, as well as compact dark flocculated fragments of tuber outer epidermis. A total of 100 fields were identified for each sample under the microscope (Owen 1975).
Determining Vallisneria tuber distribution
We observed the Swan Goose distribution every week at both Site A and Site B during the winter 2008/2009. The undisturbed distribution of buried Vallisneria tubers was determined in February 2009 at Site A. Seven 30 cm × 30 cm × 30 cm holes were excavated and tubers were extracted in situ within each of these quadrats within the substrate where they were unexploited by Swan Geese in these areas up to the sampling date. The quadrats were carefully excavated using a small shovel by removing mud layers to reveal each tuber. The vertical depth of each tuber from the substrate surface was measured to the nearest 1 cm from the surface level using a ruler. Tubers were put into separate paper bags. By contrast, Site B was subject to intensive foraging after the sampling date (Fig. 2). Hence, after prolonged use by feeding Swan Geese, tubers were again extracted at Site B following the same method from five 30 cm × 30 cm × 30 cm sample holes in February 2009. In the laboratory, all derived Vallisneria tubers were washed and remaining substrate was carefully removed using a toothbrush and dried at 50 °C for 48 h (Fig. 2). The dry weight of each tuber was determined to the nearest 0.001 g using an analytic balance (Sartorius, Germany).
Excavation pit measurements
After Swan Geese had exploited the two sites for feeding on buried tubers, both areas exhibited numerous tuber excavation pits created by the birds. The largest and smallest diameters and the greatest depths of the pits dug by Swan Geese were measured to the nearest centimetre using a ruler (site A: n = 50, site B: n = 28).
Estimated Swan Goose daily energy intake
We estimated daily energy intake of Swan Geese in each layer following the method described by Fox et al. (2011). Swan Goose intake rate was determined using focal sampling. In total, 28 focal samples of individuals were observed. Daily tuber intake was estimated based on the frequency of the observed goose found and swallowed the tuber. Time spent feeding was observed every 15 min using scan sampling based on the observation throughout daytime. Dropping rate was also recorded by focal sampling. Energy content for both tuber and dropping was analysed using a bomb calorimeter. In contrast to the earlier study, we further classified the tuber burial depth into three different layers (layer 1: 1‒10 cm, layer 2: 11‒20 cm and layer 3: 21‒30 cm) and separately calculated the mean tuber dry weight within each layer. Moreover, the observed intake rate (number of tubers consumed per minute) was the average value of three layers in the analysis of Fox et al. (2011). As tuber intake rate is closely linked to tuber density, we therefore adjusted the intake rate of each layer by multiplying the percent of tubers in each layer to determine the average intake rate in each layer. Finally, the daily energy intake was estimated by subtracting the energy content of daily output in droppings from that of daily food intake.
Statistical analysis
We calculated the average tuber dry weight density (g/m2) at each burial depth class based on the undisturbed data. To test if there was a relationship between tuber dry weight and tuber depth, a quadratic regression model was applied to ln-transformed tuber dry weight mass data from the unexploited area. A t-test was used to detect if there is a difference between undisturbed (original) and disturbed (giving-up) tuber total dry weight of each layer.
A t-test was also applied to test for differences in excavation pit greatest depth, largest and smallest diameters between Site A and Site B. The data were square root transformed to reach the assumption of normality. Statistical analyses were conducted in R 2.15.2 or SPSS 19.0.