There are 575 cicada observations for North America that are incorrectly IDed as “Genus Cicada,” a genus that does not occur in North America. Making things even worse, some of these IDs are naively confirmed by others and some of those people have become inactive. This makes it difficult to get any reasonable research grade observation since this ID conflicts with everything. Is there any solution for this mess? Can’t someone with power squash them somehow?
Like any ID on iNat, it relies on the community ID algorithm. No one’s ID has more weight than anyone else’s on iNat.
This seems not necessarily due to computer vision errors, but you could add this genus to the wiki here to call in some help: https://forum.inaturalist.org/t/computer-vision-clean-up-wiki/7281
Yep, we could organize some users to go through them and move them up. What taxonomic level would they be best to go to for someone who doesn’t know enough to get to genus, family Cicadidae?
Unfortunately getting things to research grade requires experience with the actual species involved, so that’s more difficult to solve as it requires experienced identifiers. For example, feel free to tag me to help with observations but I will need a brief ID explanation for observations outside of Ontario.
It seems that has already happened. There are no longer any observations of Genus Cicada anywhere i the Western Hemisphere. This won’t prevent future misidentification, but it seems @dan_johnson bringing attention to the problem has lead to a solution for those 575 observations.
Dan might be referring to observations that have been tagged at genus Cicada and are thus stuck out at family or superfamily level? I periodically look at the ones with community ID of genus Cicada and I haven’t seen any stuck with a community ID of Cicada for a bit. I know there is a way to search for things that have been tagged as x even if the community ID is y, but I can’t remember right now.
North American observations that contain an active ID for Cicada, do not have a community ID in the Cicada genus, and have a quality grade of “Needs ID”:
After research came out, a small group rounded up the North American Euphorbia esula observations and got them changed to Euphorbia virgata. We had to do this rather than taxon split because esula is still valid globally, and there are some limited occurrences in North America, but nearly all North American observations are virgata. Every now and then esula pops up somewhere because all the literature still uses that species name, so we kindly explain what’s up and give our ID. It’s easier now that the image recognition is trained to virgata.
I’d make a plan, round up a posse, and start doing that. A hyperlink to an explainer is helpful to save time.
I would beware of introducing any mechanism that prevents IDs of a certain group being made because that group ‘does not occur in ABC place’.
This can create a vicious cycle where, if a genuine observation were to be made, it might not be identified correctly. Many weeds would fall under this category.
Just for fun here’s an example of what we have to deal with:
We have an observation with a good photo, but we have “Genus Cicada” double confirmed. All identifiers of this taxon have long since departed inaturalist. So am I supposed to somehow round up six experts to get my ID of Neotibicen tibicen research grade? Or am I supposed to just get six people who trust my expertise and just confirm even though that don’t personally know? Or should we just consider this observation beyond repair and to be ignored from here on out?
I did a quick spot check of some of the 500 odd records in the link above. While I did not check all, everyone I checked the genus Cicada ID was not done by the computer vision.
What is likely happening here is users know it is a cicada, type that in and then see the genus in the drop down and choose it.
I dont know how you get away from that without adding a geo check to all identifications not just CV ones.
The same thing happens with Bushtit a Western NA bird whose banding code is BUSH, users have an unknown shrub, enter bush and ID it unwittingly as the bird.
Strangely if I type in “cicada” in the ID text box, superfamily Cicadoidea (correct) is the first option. Both “Cicada” and “Cicadas” have been added as common names for it. I wonder if it is also the first option on the apps, and which option would automatically be chosen if you put in “cicada” as a placeholder when using the app offline. Of course a lot of problems also come from people for some reason not choosing the first option…
I see your point. There are no observations in the Western Hemisphere with Community ID of Genus Cicada, but there are still 538 with at least one identification as Genus Cicada that still need identification. Anyone with the knowledge can help correctly ID them here: https://www.inaturalist.org/observations/identify?place_id=97394&ident_taxon_id=205175
Is there a system implemented yet in iNaturalist that will allow ranges of taxa to be input into a database and then inaturalist warns users if their identification in probably wrong based on location? If somehow I get the range for “Genus Cicada” loaded, will users be warned in the future?
Yes - range maps, checklists, atlases are 3 ways. I should note these are typically done for species, not genera, but at least 2 of the 3 (not sure about atlases) can be done for a genus.
How does inaturalist user range maps, checklists, and atlases to help reduce bad IDs? How can I check if a taxon has such information present?
you can check for range maps by going to the relevant taxa page and going to the map tab. If there is one it will show in pink (at least I think it is pink, I’m colour blind so I think that is what it is). Sometimes it is hard to see so you can click the stacked books icon in the corner of a map and turn off the other layers.
The same tab and map will show checklist statuses, green for on relevant checklist and has confirmed records, orange for on the checklist, but no records.
As you zoom into the map, it will gradually refine into showing more specific locations.
Checklists follow political boundaries, range maps are freeform.
Atlases are mainly a curator tool, at one point they were the great hope for maintaining distribution data but they seem to have stalled out in use. On the taxa page, if you click the curation dropdown there will be an Edit Atlas option. Please note I’m unsure what it says or does if you are not a curator though.
Just for everyone’s amusement, I just found this: https://www.inaturalist.org/observations/2885952 . Four “Genus Cicada” ids for one observation. Looks like we’ll need 13 good IDs to make this research grade.
I bet at least some of these erroneous IDs could be avoided by adding a common name to this genus indicating their geographic range. I’m not sure what would be an appropriate name for Cicada (hopefully some Europeans can chime in), but I imagine a name like “European Ash Cicadas” would make some North Americans think twice.
I used to see this a lot with locusts (plant genus) where the observer had clearly meant to select locusts (insects).
Also with foulbrood aka Paenibacillus larvae (a bacterium) when the user had typed larva into the species field.
Ams then there’s vas (a fly) vs vascular plants (I eventually favorited the fly genus just to keep an eye out for those)