Collecting Flower Visitation Data
During each visit to the six farms, we took 30-min video recordings of
pollinators visiting eight randomly selected male squash flowers per
site (N = 112, mean video length: 30.87 min [sd: 3.75 min]). Each
video was recorded between 07:30 AM and 12:00 PM on sunny, non-windy
days. Video recordings were watched to record data on the identity and
frequency of pollinator visitors to the flowers. Pollinators captured on
video were identified to genus where possible (e.g., Apis ,Bombus , Eucera ), or to morphospecies for species that
require close inspection and/or a key for accurate identification
(Appendix S1: Table S2). Honeybees and bumblebees were easy to identify
in the video recordings due to their relatively large body size and
distinctive coloration. The behaviors of all other pollinators observed,
including small green and olive halictids (e.g., Augochlora ,Augochlorella , Augochloropsis , Halictus , andLasioglossum genera), Melissodes spp., Eucera spp.,Triepeolus spp., Vespula wasp spp., and hover flies, were
grouped together into an ‘other pollinators’ category to compare to
honeybee and bumblebee behaviors in later analyses (see Statistics
section).
During each individual pollinator’s visit to the observed flower, we
recorded the duration (seconds) of each visitor’s interactions with
specific flower parts, including petals (petal-only), nectar
(nectar-only), pollen (pollen-only), and both pollen and nectar
simultaneously (pollen+nectar). Typically, large-bodied bees, including
honeybees and bumblebees, could not avoid contacting the stamen while
drinking nectar (pollen+nectar) and led to relatively few observations
of nectar-only interactions with flowers (Appendix 1: Table S3). For
this reason, the nectar-only interactions were not considered as a
substantial interaction type and were not included as a response
variable in our main analyses. For each flower observed, the total
duration of all types of interactions were summed for each pollinator
group (honeybees, bumblebees, or all other pollinators) and then divided
by the number of flower visits for the respective pollinator group to
generate the duration spent per visit by each pollinator group to each
flower. Finally, to test how each pollinator group’s visitation behavior
impacted V. ceranae prevalence, we averaged the calculated
visitation metrics for all flowers observed during the same site visit
for each pollinator group. We followed the same process to calculate the
average duration per visit of time spent on petal-only, pollen-only, and
pollen+nectar interactions for each pollinator group. The number of
visits for each pollinator group was the raw count of each type of
pollinator that visited each observed flower within the 30-min
observation period, which was then averaged for each of the two visits
to each site.
Evaluating the average duration bees spent per floral visit ensured that
the duration metrics accurately reflected the time bees spent
interacting with flowers without being skewed by the number of bee
visitors. Each additional bee visitor inherently increased the total
duration of time bees spent on flowers (r = 0.76, t = 11.52, df =
95, p < 0.001) but did not necessarily increase the duration
per visit time (r = 0.02, t = 0.21, df = 95, p = 0.84). We
predicted that bees that spent a greater amount of time per visit
interacting with flowers would have a greater likelihood of either
depositing or picking up V. ceranae spores on flowers and would
be correlated with higher V. ceranae prevalence.