Featuring a Desertas petrel that stunned scientists by chasing a tropical storm (see their incredible chase mapped here!), tiny nightingales that cross the Sahara twice a year, and Bewick's swans that have changed their stopovers and diet

Variable: e.g. Carrion Crow -ve, Raven +ve, Rook/Hoodie CIs overlap 1.

For the very common stuff, you start to run into detectability effects - particularly in the BirdTrack analysis - where species are likely to be the first-recorded regardless of preference, given how common and obvious they are.

A lot of common species are in that sweet spot. Skylark perhaps closest: 1.001 from the eBird analysis and 0.999 from the BirdTrack analysis. From a quick check through Table S1, other examples inc. Goldeneye, Blue Tit, Goldfinch, Little Egret, Kestrel, Fulmar, Bullfinch, Blackcap.

In spite of these preferences leading to the overrepresentation of certain species in the eBird & BirdTrack datasets, we found that this bias had limited impacts on actual applications of the data (occupancy models). Even so, we urge consideration of observer bias in analyses of #citizenscience.

Rare #birds were significantly more likely to drive survey initiation, with other species-specific factors like novelty and charisma also potentially contributing, especially among more common species.

These observer preferences were most prevalent on short-duration surveys (less than 5 min). ⤵️

Using #citizenscience data from eBird @cornellbirds.bsky.social and BirdTrack @birdtrack.bsky.social @btobirds.bsky.social, we explore what drives birders to initiate semi-structured surveys on these two large-scale citizen science platforms.

The biggest factor: species rarity! ⤵️