Location: Pollinating Insect-Biology, Management, Systematics Research
Project Number: 2080-21000-019-065-I
Project Type: Interagency Reimbursable Agreement
Start Date: Jul 3, 2023
End Date: Jul 2, 2027
Objective:
The objectives of this research is to 1) Produce a DNA library of known bee species that are sympatric in the Sacramento National Wildlife Refuge Complex (SNWRC) with the Cooperator; 2) Produce a library of bee images to train a deep learning model that could enable automated, imaged-based identifications of bee species in the SNWRC with the Cooperator; and 3) Identify wild bee fauna of the SNWRC through non-lethal survey techniques (DNA barcoding, eDNA, computer vision).
Approach:
Objective 1: We aim to target at least 124 known bee species documented in the Sacramento Valley from previous studies. Two publications have provided a good list of the expected species in the northern central valley. We will acquire specimens for the natural history collections for DNA analysis. We aim to generate DNA barcode data for up to 20 specimens per species for 2,480 specimens. However, we also recognize that 20 specimens may not be available for some species due to their rarity in collections. The data will be generated using a combination of approaches. For specimens collected within the last 5-10 years, we will employ targeted PCR methods combined with nanopore sequencing). For older or more degraded samples, which we expect to constitute a minority of samples, we will use alternative, ‘museomic’ methods that do not rely upon targeted PCR. For both methods, we aim to obtain sequences of two loci, cytochrome oxidase I (COI), the so-called barcoding gene, and 16s RNA (n = 2,480 x 2 = 4,960).
Objective 2: In Year 1, we aim to conduct non-lethal surveys across 3 mixed riparian plant communities paired with 3 restored riparian communities on USFWS property that has been historically sampled (Williams 2011). At each field sites, bees will be netted by hand for 1 collector hour in the morning (8:45 – 11:45) and 1 collector hour in the afternoon (12:00 – 14:30). Only bees actively visiting flowers will be netted. Bees will be placed in a small insect vial and placed in a cooler on ice to induce chill coma. At the completion of the collection period the dorsum and venter of each bee will be firmly swabbed to obtain surface DNA and loose tissues, like setae, a unicellular process of the insect integument. The swab will be placed immediately into a 1.5 mL microtube or 96-well plate and placed in a cooler. After tissue (e.g., setae) has been non-lethally sampled, the bees will be imaged in the field with a Macropod Pro DSLR camera-system, specifically developed for field imaging. Images will be used too support generic/species identification using taxonomic keys and image-based deep learning techniques currently being developed in at the USDA (see Objective 3). All bees will be released at the completion of processing. In addition to non-lethal sampling of tissue from bees in the field, we aim to conduct a pilot study using sentinel flowers equipped to passively sample flower-visiting insects for DNA-based species identifications. The expectation is that bees will be attracted to brightly colored sentinel flower to extract resources from a wick that secretes artificial nectar (akin to a nectary gland of a true flower). Fifteen sterile sentinel flowers will be distributed systematically at each field site. DNA extracted and sequenced from the flowers.
Objective 3: We aim to initiate the development of an image library of museum specimens representing the 124 bee species documented in the SNWRC. These images will be used to train CNN models to make image-based species identifications.
