Methods

Site Selection

I used 2023 highway traffic data provided by Alberta Transportation and Economic Corridors [6] to find sections of highway that i) had a daily traffic load of 4-7 thousand vehicles per day, ii) travelled north-south or east-west, iii) were at least 1.5 km from the nearest town [7], iv) had no side roads or railways running parallel, v) were adjacent to cropland on both sides, and vi) were adjacent to herbaceous crop borders (defined as the absence of trees) or one side was herbaceous and the other treed (Figure 3). A list of potential sites were then ground-truthed which left me with a total of 22 sites. Half of the sites had herbaceous crop borders (Figure 4 green pins) with the other half having a treed crop border (Figure 4 blue pins). Sites with cereal crops om both sides of the highway were prioritized to avoid unwanted effects associated with mass-flowering crops. This was not possible for every site, however, so crop type and flowering times were recorded when relevant.

Figure 3. Satellite image of a site next to a treed crop border. Credit: Google Maps.

Figure 4. Geographical distribution of my sites. Blue pins are sites next to treed crop borders and green pins are sites next to herbaceous crop borders. Credit: Google Earth Pro.

Data Collection & Processing

Figure 5. Blue vane trap.

Bees were collected passively using blue vane traps (Figure 5). Blue vane traps were chosen because they are good at catching many different bee species [8] and can be left in the field for longer periods of time compared to pan traps [9]. Two blue vane traps were installed approximately 60 from each other, 1 m above the ground, 9 m from the highway, and varying distances from the crop and filled with propylene glycol (Figure 6; Figure 7) [10]. I chose to install two traps per site to add redundancy in case one were to fail which happened on two occasions. The blue vane trap were collected every two weeks three times (six weeks total). Blue vane trap samples were stored in 70% ethanol. The exact date and time of blue vane trap collection was recorded to account for differences in trap hours. Samples were brought back to the lab where they were sorted, pinned, and identified to species using identification guides [11, 12]. Currently only Honey bees (Apis) and Bumble bees (Bombus) have been identified from two of the three sample rounds.

Figure 6. Graphical representation of blue vane traps along a transect. The blue vane traps were installed 22 m from the transect ends and 60 m apart from each other.

Figure 7. Graphical representation of a transect. Each transect was approximately parallel 9 m from the highway. Floral surveys were conducted along a 30 m by 2 m sub-transect within the red square where the highest floral abundance was visible.

Floral surveys were conducted by identifying where the highest floral abundance was within the area between the transect and the crop (Figure 7 red box). Once identified, a 30 m by 2 m sub-transect was measured out (Figure 6 blue box). Within this sub-transect the flower species were identified to the lowest taxonomic rank using several plant identifying guides [13, 14, 15]. Then, the number of inflorescences (interpreted as floral abundance) were either counted or estimated for each species. Inflorescences were estimated when it was unreasonable to count by hand. This was done by averaging the number of inflorescences on three individuals then multiplying by the number of individuals along the sub-transect. Floral surveys were conducted every two weeks.

Alongside the blue vane traps, bees were also caught actively using hand netting. This was done for twenty minutes per site along the 30 by 2 m sub-transect every two weeks. The timer was stopped each time a bee was caught to account for processing time. Hand netting was done because it is recommended to do alongside passive sampling methods to increase abundance estimate accuracy[16] and to gather direct bee-plant interaction data. As of March 2026, the hand netting data has not been processed so was be excluded from the data, results, and discussion of this website but will be included in future analysis.

Myself conducting a floral survey.

Statistical Analysis

All statistics were run using R version 4.5.2 [17], with the packages vegan [18], mixOmics [19], MASS [20], CCA [21], ggplot2 [22] , and ggrepel [23]. The presented results are preliminary as only two of the three sample rounds have been processed in the laboratory and not all of the bee species have been identified. For the purpose of this website, only bumble bee data was analyzed from the first and third sample rounds. Honey bees and unidentified bees were excluded for simplicity.

Page updated

Report abuse