Results
A total of 1356 individual bumble bees belonging to 10 different species were recorded in the meadow during the two years (Table S1).Bombus lepidus, B. friseanus and B. festivus , the most abundant species, represented 32.22%, 44.25% and 13.42% of the total individuals, respectively. The proportion of pollen collectors out of all visits recorded in the two years was 7.95%, 14.65% and 56.13% forB. lepidus, B. friseanus and B. festivus , respectively. In addition, the length of proboscis for workers of B. lepidus, B. friseanus and B. festivus was 6.46 ± 0.08 mm, 8.07 ± 0.08 mm, and 9.52 ± 0.19 mm, respectively (Table S1). The two bumble bees with higher abundance, B. lepidus and B. friseanus , differed by about 10 days in peak of abundance (Fig. 1).
In total, 55 flowering plant species were visited by the three bumble bee species (Table S2). Bombus friseanus , B. lepidus andB. festivus visited 29, 33, and 24 species, respectively (Table S2). Flowers of the most dominant species, Polygonum macrophyllum , attracted the most visits among all of the flowering plants for the two mainly nectar-seeking bumble bee species, B. lepidus and B. friseanus . B. lepidus visitedGeranium refractum second most often, followed by Trollius farreri , while B. friseanus , visited Phlomis atropurpureaand Pedicularis densispica in sequence. For the mainly pollen-seeking bee, B. festivus , the foraging preference wasPedicularis cephalantha followed by P. rhinanthoides(Table S2; see also Table S3). The flower preferences of the bumble bees did not significantly differ in all of the species pairs in the two years even though the number of visits of the bees varied in the two years (Table 1). Moreover, for each of the bumble bees, the relative frequency of visits to a plant varied highly across the sampling dates (Table S3).
The niche overlaps between species pairs of the three bumble bee species varied widely across the sampling dates (Table 2). Moreover, Pearson correlation analysis indicated that the abundance of each bee species (primarily workers, with relatively few males, and the four queens were not included) was not correlated with the value of niche overlap between species pairs for all the three bees (Fig. 2).
Our results indicated that the bee abundance, the proportion of nectar collectors out of total visits by a bee species, and the composite flower depth of plants visited by a bee species varied widely across the sampling dates (Table 3). Moreover, for B. festivus , the mainly pollen-seeking bee with the lowest abundance, the proportion of nectar collectors out of all recorded visits was positively correlated with its niche overlap to bothB. lepidus (R = 0.67, P = 0.007) and B. friseanu (R = 0.83, P = 0.001; Fig. 3). For the other two mainly nectar-seeking bees with higher abundance, B. lepidus and B. friseanus , we failed to find a significant relationship between niche overlap and foraging preference of nectar vs. pollen. Furthermore, the niche overlap between these two species was positively correlated with the composite flower depth of plants visited by B. lepidus (R = 0.60, P = 0.039; Fig. 4), but negatively correlated with the composite flower depth of plants visited by B. friseanus (R = 0.898, P < 0.001; Fig. 4). However, forB. festivus , there was no significant relationship between niche overlap and composite flower depth of plants visited by the mainly pollen-seeking bee.