Methods

Figure 3. Map showing the locations of the 76 traps used to capture specimens covering nearly all of Alberta's natural sub regions.

The Alberta Native Bee council partnered with Alberta Forestry, parks and tourism and other organizations to set up traps to cover as much of Alberta as possible. The traps were primarily blue vain traps, however a few pan traps were also used. In total, there were 76 locations across Alberta that collected specimens (figure 3). The traps were set in natural or semi natural areas with lots of sun and no tall vegetation. Throughout June to September 2018, the traps were filled one third full with propylene glycol and left out to capture specimens for approximately 2 weeks at a time. After 2 weeks, the specimens were collected and placed in a sample bag containing glycol and stored in a cool, dry place. After two weeks off, some traps were reset to sample for another two weeks, the traps maintained this two on, two off cycle for the summer of 2018. Since the traps were set and maintained by generous volunteers there is some inconsistencies in how long they were set for and the number of times. To account for this, the bee community data was divided by the total number of days the trap was set for.

That following summer the captured insects were sorted; all native bees were cleaned and dried to be pinned for taxonomic analysis. For traps that captured over 15 native bees, DNA barcoding was used to determine the species rather than traditional taxonomic recording.

To determine the percent of land cover surrounding each trap I used ArcGIS pro to create 250m, 500m, 1000m and 2000m radius buffers around each trap location (figure 4). Using a wall to wall land cover layer from ABMI (ABMI, 2010) I used the Summarize Within tool to determine the proportion of land cover within each buffer size. In order to reduce the redundancy of the landcover data the larger buffers did not include the smaller buffers within them, resembling donut shapes. The summary of landcover types within the 2000m buffer surrounding each trap is shown in figure 5.

The land cover classes include:

Rock/Rubble
Exposed Land
Developed
Shrubland
Grassland
Agriculture
Coniferous Forest
Broadleaf Forest
Mixed leaf Forest

*Though water was originally part of the percent coverage calculation, the class was not included in the final analysis. Because including it means all land cover variables add up to 1 per spatial scale creating redundancy in statistical models.

Figure 4. example of the buffers used to calculate percent land cover class surrounding each trap location.

Figure 5. Bar graph showing the breakdown of landcover types at the 2000m scale surrounding each trap.

Climate data was derived from climateNA to get a 30 year average from 1961-1990 for 7 climatic variables at each trap location (Wang et al., 2016).

Climate variables used include:

MAT - Mean Annual Temperature
MWMT - Mean Warmest Month Temp
MCMT- Mean Coldest Month Temp
MAP- Mean Annual Precipitation
CMD- Climate Moisture Deficit
FFP- Frost Free Period
MSP- May to September Precipitation

In order to answer my three objectives I used a distance based redundancy analysis per spatial scale and per climate group which were determined using a cluster analysis.

Page updated

Report abuse