…also, the concepts of “method detection limit” and “instrumentation sensitivity” might be measures worth considering.
These are my people. 
4 Likes
Thought this recent paper might be of interest to this group: Deriving indicators of biodiversity change from unstructured community-contributed data. It’s pretty much about how to overcome the biases and “messiness” of iNat data to find patterns and see change through time. I apologize that it’s not open-access - Wiley’s OA fees are prohibitive, and since this was grant-funded work we couldn’t afford it. Our co-author did a nice tweet-summary of it where you can see some of the figures describing the work. I’m also more than happy to send the pdf to anyone who’s interested in reading the whole thing!
7 Likes
Could you please show screenshots of that tweet? My computer has complicated relationships with twitter and site don’t want to show anything without my phone number.
2 Likes
Thank you!
I have only dipped in and out of this discussion, so apologies if this has been discussed already. But what is meant by biases? What would be an example of a species that is not biased, neither under- nor over-represented? Or is it assumed that everything is biased and it is just a matter of deciding which of the two categories applies in each case?
I understand under-representation of e.g. difficult to photograph common species. But for the charge of over-representation to apply, it suggests to me there would need to be some falsity in the sightings, e.g. if chiff-chaff was over-represented because too few people were considering willow warbler.
Overrepresenation in this case is either in comparison with other species or, if we talk about one sp. only, it’s geographical biases, so species is overrepresented in cities and underrepresented in countryside while actual distribution can be reverse. Overrepresenation doesn’t actually mean we need to stop observing it or observe less.
Not-biased species would be one that is observed everywhere as often as it actually, well, exists there, it’s possible if all parts of its areal are heavily studied by users and, realistically, it’s an easy to recognise species. It’s hard to achieve, e.g. mallard is heavily presented in cities, while it actually lives pretty much in every other possible water bodies across its range. It’s also needed to be noticed that when we usually refer to unbiased set we talk about some works on pretty small areas, where sampling was done with methods that eliminate human factor as much as possible. When we go iNatting there will always be biases unless we get to the point of getting tons of data from every corner possible.
1 Like
Some of us are already doing that. But I would add a few more.
As a corollary to
How 'bout: make a point of learning to identify the above! There are observations of particularly fungi and invertebrates which have sat for years at very broad identifications. Learning how to refine these can in many cases help even more than adding yet more observations of the same. Instead of hoping that a specialist will help with your observation, pick one of these taxa to BECOME a specialist in.
2 Likes
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.