Fig 1.
Flowchart of the wild bee identification processing steps for the morphological (nondestructive to specimen) and molecular (partially destructive to specimen) approaches.
Fig 2.
Difference in the number of bee genera detected per site visit with two identification methods (morphological minus metabarcoding).
Comparisons were made for 47 site visits where bees were captured. Most site visits included both a large and a small composite vial. Positive values along the x-axis indicated how many additional genera were identified using morphological identification compared to metabarcoding.
Fig 3.
Number of vials in which a bee genus was detected using DNA metabarcoding (blue) and morphological identification (orange).
Bees were morphologically identified using images, and a leg from each bee was composited. Vials contained composite bee legs from one or more genera, and genera richness was assessed in each sample that had a genetic detection (n = 84).
Fig 4.
Proportion of bee legs within a composite sample for genera that were detected and not-detected using DNA metabarcoding.
Box and whisker plots depict the median (horizontal line within the box), mean (x within the box), first quartile (bottom line of box), third quartile (top line of box), minimum value (bottom whisker), and maximum value (top whisker) unless the values are outliers (greater than 1.5 times the interquartile range above the third quartile, shown as points above top whisker).
Table 1.
Taxonomic assignments for cytochrome c oxidase I (CO1) barcode sequences obtained from single specimens compared to their image-based morphological identifications.
Table 2.
Advantages and disadvantages of image-based morphological identification and DNA metabarcoding for wild bee identification.