It’s 11th January 2018 as I write this, sadly my last day of work for the Blooms for Bees project (though I will stay in touch with the project and continue to support Gemma and Judith in any way I can). One of the big winter jobs for Judith and I was analysing the data you all sent in. As verifier, I was interested in the number of bumblebee species recorded and how well our citizen scientists were able to identify them. I was able to achieve both by comparing the provisional identifications with the photographs submitted by users. So here are some stats:
We received 4,200 bumblebee sightings, 31.5% of which were accurately identified, but because most records included supporting photographs, I was able to increase this accuracy rate to about 72%, which is pretty impressive. There is an important message that comes out of this – citizen science is great and can produce valuable data as long as expert verification is built in. It is possible to obtain a lot of records from a variety of locations, and for participants to learn stuff and have fun whilst doing it – then I get to see interesting data from those places too. And isn’t smartphone technology just amazing!
There were records for 23 bumblebee species prior to verification. Verification reduced this number to 15 confirmed species (plus the White-tailed/Buff-tailed Bumblebee category – grouped because workers of the two species are so difficult to differentiate). Within the verified data, the most frequent species were the Buff-tailed Bumblebee (1159 records), Common Carder Bee (751 records), Red-tailed Bumblebee (318 records), Early Bumblebee (219 records) and Tree Bumblebee (100 records). There were fewer than 100 records for each of the other species.
Identification accuracy rates varied substantially between species but were greatest for submissions of the Common Carder Bee (72.2%), Red-tailed Bumblebee (69.9%), Early Bumblebee (60.1%), Buff-tailed Bumblebee (57.2%) and Tree Bumblebee (56.7%). All submitted records for Barbut’s Cuckoo Bee, Gypsy Cuckoo Bee, Great Yellow Bumblebee, Heath Bumblebee, Brown-banded Carder Bee, Moss Carder Bee, Red-shanked Carder Bee, Larger Garden Bumblebee, Broken Belted Bumblebee, Short-haired Bumblebee and Shrill Carder Bee were incorrect. However, true records for the less common Heath Bumblebee, Moss Carder and Large Garden Bumblebee did arise from misidentifications of other species.
This analysis is really useful because it helps us better understand the planning and resources that are required when setting up a citizen science project. It will also help with identification advice we give to participants in the future. I have thoroughly enjoyed being involved with Blooms for Bees and am very grateful to all of you who have sent in records. The project will continue into 2018, so please keep doing the surveys and sending in your records.
(We are now looking at the bumblebee records verified by Steven in conjunction with the plant records, in order to investigate the floral preferences of different bee species.)