Insect-flower network
In the order-level network structure between flower-visiting insects and
73 flowering species, almost all plant species had linkages with
dipteran and/or hymenopteran insects (Fig. 5a). Dipteran insects visited
most diverse flowers (H’ = 4.81) and showed the largest linkages
with plant species among four orders. Hymenopteran insects also visited
many flowers (H’ = 4.45) but the linkages with several plant
species, that have strong linkages with dipteran insects, were limited.
The diversity of foraging flowers of lepidopteran and coleopteran
insects was moderate (H’ = 4.12 and 3.20, respectively). Species
compositions of foraging flowers of lepidopteran and coleopteran insects
were highly overlapping with those of hymenopteran and dipteran insects
(0.88–0.95 in similarity), but the similarity of foraging flowers
between lepidopteran and coleopteran insects was moderate (0.69). When
the foraging patterns of dipteran insects and hymenopteran insects were
compared, the niche overlap was relatively low (0.34). Thus, the flowers
targeted by dipteran and hymenopteran insects were considerably
different.
In the network structure between three major insect groups and 70
flowering species (Fig. 5b), syrphid flies, non-syrphid flies, and
bumble bees linked to 62 (89%), 65 (93%), and 50 plant species (71%),
respectively. The values of diversity index for visiting flowers were
similarly high (H’ = 4.31–4.54). Similarities of foraging
flowers were moderate between bumble bees and non-syrphid flies (0.72)
and between bumble bees and syrphid flies (0.65), while the similarity
between non-syrphid and syrphid flies was relatively low (0.57). Niche
overlap among major insect groups was 0.303, suggesting that different
insect groups tend to forage on different plant species irrespective of
their wide foraging behavior.
GLM results conducted for the relationship between visitor frequency and
the number of flowering species in each term revealed that the number of
bee-visited species at the flowering stage was positively related to the
abundance of hymenopteran insects (z = 2.39, p = 0.017)
(Fig. 6a), whereas the number of fly-visited species at the flowering
stage was independent of the abundance of dipteran insects (z =
–0.53, p = 0.60) (Fig. 6b).