Fig 1.
Satellite image of Sunny Ridge farm 38°25’25.2"S 144°53’33.8"E (perimeter indicated by dotted grey line).
Managed honeybee hives (yellow hexagons) with the number of hives at locations close to study sites indicated. Raspberry/strawberry polytunnels and the weed location used in the study are marked. Nearby areas of native bushland are indicated along with their ranges from the strawberry polytunnel for reference.
Fig 2.
(A) Strawberry polytunnel, (B) raspberry polytunnel, (C) open weed area. A blue, yellow, and white pan-trap were placed in each location to allow for known impacts of colour on insect counts of different species [36].
Fig 3.
(A) Raspberry Pi and camera setup in the strawberry fields. (B) A frame from the recorded video.
Fig 4.
(A) Strawberry polytunnel, (B) raspberry polytunnel, (C) polytunnel edges near the farm road, (D) strawberry polytunnel near remnant habitat, (E-F) weed locations.
Fig 5.
(A) Strawberry polytunnels and (B) raspberry polytunnels (refer to Fig 1 for site overview). Numbers/rectangles indicate locations of insect quadrat observations and pan-traps under the polytunnels #1–9, #11–12, and #10 in the weeds. The white-black dashed rectangle (location #9) additionally shows the location of the camera observations to record detailed insect- strawberry flower interactions. Managed honeybee hives (yellow hexagons) with the number of hives at locations close to study sites are indicated.
Fig 6.
Quadrat observations (blue) and pan-trap counts (yellow) of honeybees (HB) and other insects (OI).
Strawberry polytunnels (1–9), weeds (10–note axis scale), raspberry polytunnels (11–12)–see Fig 5 for polytunnel layout. Grey shading indicates the middle of the strawberry polytunnels, green shading indicates the weed area. Error bars represent the standard error of the mean.
Table 1.
Average insect count per quadrat along the polytunnels and in the middle of the uncovered weed with the mean and standard error of the mean (SEM) shown in brackets.
Table 2.
Average insect count per pan-trap location along the polytunnels with the mean and standard error of the mean (SEM) shown in brackets.
Fig 7.
Honeybee tracking data visualisation.
(A) Honeybee tracks (n = 51) extracted using the automated software and strawberry flowers (n = 11) within the field of view of the camera. Each coloured line represents an individual honeybee track. (B) Count and time of day of honeybee sightings.
Table 3.
Distribution of the durations of time spent within the field of view of the camera for all insects.
Fig 8.
Video-monitored insect flower visitation summary data.
(A) Mean number of flowers visited by each insect type (per insect) modelled with a negative binomial model. Bars represent the predicted mean number of visits by each pollinator type (per insect) and error bars at 95% confidence intervals for the predicted mean. (B) Median time spent on flowers for each insect type per insect. Error bars indicate the median absolute deviation.
Table 4.
Results of the automated software validation study.
Table 5.
Average insect count per location by pan-trap and by quadrat observation using only the middle of polytunnels and weed area with standard error of the mean (SEM) shown in brackets.