During the first visit of 2014, 3988 individual bee specimens from 23 genera were collected. These groups include species from all 5 families of bees in Alberta. Over half of these individuals were collected from the Grassland Region (n=33), while 1451 individuals were collected from the Parkland (n=33), 157 were from the Boreal (n=5), and 51 individuals were collected from 2 rangeland sites in the Rocky Mountain natural region. Two of the most abundant genera of bees were Lasioglossum and Agapostemon--both members of the Halictidae family (Figure 7). Both of these groups, but in particular Agapostemon, tended to occur at sites in the Grasslands Natural Region. Other abundant bee genera included: Bombus, Halictus, Andrena, Apis, Osmia, Hylaeus, and Hoplitis. There were also a large number of genera with fairly low overall occurrences. Many of these rarer species occurrences were from the Parkland natural region.
Overall, the total number of bee detections did not have a strong relationship with any of the weather variables measured. The strongest relationship occurred between bee abundance and cloud cover (Figure 8). Bee detection levels tended to be lower on cloudier days than on clear days. In particular, three sites were surveyed on fairly cloudy days and also had fairly low bee numbers. To avoid possible confounding effects of cloud cover, these sites were dropped from subsequent analyses. Regression of bee detections against wind speed and temperature showed no visible bias from either variables--both explained less than 2% of the total variation in bee abundance (Figures 9 and 10). The lack of strong effect is likely due to restrictions placed on field sampling; bee surveys were restricted to days where the temperature was above 16 degrees Celsius. By standardizing the weather requirements for sampling, much of the potential bias was already removed during field sampling. Based on the results for cloud cover, weather requirements for 2015 sampling will be improved to provide stronger requirements to avoid sampling on very cloudy days.
An additional factor for evaluation is sampling effort, measured as the number of hours that the pan traps were deployed at each site. Similar to temperature and wind speed, the relationship between bee detections and the length of time that traps were set up was weak (Figure 11). Guidelines were also provided for this aspect of field sampling, requiring that the traps be left out for a minimum of 5 hours at each site. This requirement is a likely explanation for the lack of stong bias from sampling effort. By evaluating and removing the effects of bias from cloud cover, wind speed, temperature, and sampling effort, we can have stronger confidence in our sampling methods and the accuracy of our bee community data. These results will also help refine our requirements for field sampling for bees moving forward.