Thread by @ali__johnston

I’ve been privileged to work with #CitizenScience data in #ornithology for many years at @_BTO and @CornellBirds. I’ve seen some of the amazing opportunities and the tricky challenges that come with citizen science data.

1/15 #BOUasm22 #Sesh1

#Ornithology has a long history of #CitizenScience, enabling us to calculate population status for birds over many decades. For example, @_BTO’s data for Herons goes back to 1928!

2/15 #BOUasm22 #Sesh1

One of the most important #CitizenScience projects in the UK, the Breeding Bird Survey @BBS_birds has allowed us to monitor which bird populations are decreasing and increasing over time.

3/15 #BOUasm22 #Sesh1

#CitizenScience projects lie on a spectrum, from structured surveys with standardised protocols to opportunistic observations. In between these, we have semi-structured data that collect lists of birds and survey effort.

Paper: doi.org/10.1093/biosci…

4/15 #BOUasm22 #Sesh1

About 300 bird watching hours were contributed each day during the 2021 @BBS_birds season, whilst semi-structured surveys contributed in the UK in 2021 on average at least 3000-daily bird watching hours of data throughout the entire year.

5/15 #BOUasm22 #Sesh1

.@Team_eBird provides a huge amount of bird observations, covering every species and every country in the world. These data allow us to estimate bird distributions and relative abundance over large spatial scales.

science.ebird.org/en/status-and-…

6/15 #BOUasm22 #Sesh1

Semi-structured surveys provide us with lots of data and increased engagement, as they allow non-expert bird watchers who may be excluded from structured surveys to be involved, making #ornithology much more accessible.

7/15 #BOUasm22 #Sesh1

Is there a trade-off between data quality and quantity when it comes to semi-structured surveys? Observer behaviour does lead to some challenges with data from semi-structured surveys, which need to be considered.

Read more here: doi.org/10.1111/2041-2…

8/15 #BOUasm22 #Sesh1

We found two types of survey sites, distinguished by distance from home and species rarity. Observers travel further to see rarer birds. This information allowed us to improve the species distribution models.

Paper: doi.org/10.1016/j.ecol…

9/15 #BOUasm22 #Sesh1

Observers vary in their experience and equipment. Accounting for these differences increased the accuracy of species models.

10/15 #BOUasm22 #Sesh1

Our paper: doi.org/10.1111/2041-2… | #ornithology #CitizenScience

Some species are more appealing or popular than others. We found that rare species are more popular on incomplete checklists. Ignoring this could lead to biased results.

Paper: doi.org/10.1111/2041-2…

11/15 #BOUasm22 #Sesh1

A recent paper published in @IBIS_journal looked at misidentification of bird species in photos finding high accuracy and also identifying pairs and groups of similar-looking species that are most often misidentified.

Paper: doi.org/10.1111/ibi.13…

12/15 #BOUasm22 #Sesh1

One action we can take to start to address all these biases, is to carefully select the data that we use. By refining our data selection and accounting for effort, we can improve our models.

Paper: doi.org/10.1111/ddi.13… | #ornithology

13/15 #BOUasm22 #Sesh1

If we can use eBird data to estimate population trends, do we still need structured surveys? Yes! Models more correctly predict species presence when combining data from structure & unstructured surveys.

Paper: doi.org/10.1111/ddi.13…

14/15 #BOUasm22 #Sesh1

Semi-structured data have huge potential to contribute and improve our understanding of bird species, providing data for more species, more places and engaging more people along the way.

Thank you for reading!

15/15 #BOUasm22 #Sesh1

Huge thanks to @IBIS_journal for the opportunity to talk.

Massive thanks to @Nat_B_Zielonka who put together this thread after I ran out of time from being sick last week. Much better than I could have done it myself!

16/15 #BOUasm22 #Sesh1