If you look on the map of observations for Narceus gordanus, you will see a LOT of observations in the region around atlanta and in some nearer parts of surrounding states. Most, if not all, of them are wrong. There may be some correct ones there, and if so that would be exciting because they’re not so common out there, but all the ones I sampled so far have definitely been in the Narceus americanus complex (this is too much for me to try to fix right now). I looked, and there is not a single one in that area that is research grade from what I can tell. And if none of them are research grade, then it shouldn’t say seen nearby. But when I open the observations and wait for CV suggestions, it does give Narceus gordanus as the top suggestion (or one of) and marks it seen nearby. Why would this happen?
heres some examples were the CV was very confident it was the N gordanus above all else and claims it was seen nearby despite that not seeming to be true. not singling out these users, theyre using inaturalist the way lots of people do, but the computer suggestion has led them astray. there are so many more just like these
Taking a guess just browsing the taxonomy, but does CV ever recommend anything other than genus or genus and species? If so, maybe it’s because N. gordanus has more observations than N. americanus (a member of the Complex)?
that’s interesting. it looks like the code does provide for 2 potential paths for determining “nearby”: https://github.com/inaturalist/iNaturalistAPI/blob/main/lib/controllers/v1/computervision_controller.js#L238-L246.
the “else” path is the one that should look for research grade observations within 100km of the coordinates of the observation.
but the other path uses different logic for determining nearby.
it looks like there’s some sort of configuration key that can trigger the latter logic. i’m not sure if it’s possible for folks like use to see the configuration, but just based on your examples, it seems like maybe the latter logic is being used?
@pleary or another staff member might be able to tell better what’s going on here.
It does, but rarely chooses something unusual like a tribe, I think it even suggests complexes.
I frequently work thru weird distribution maps. CV only needs ONE person to offer ONE ID - to then offer that as ‘Seen Nearby’. Sometimes it was ‘forgot’ to mark it as Not Wild. Sometimes it was a, well, it could be, this … which spurs more and more. Wrong wrong still wrong.
beyond the way the CV works regarding what people have wrongly id’d nearby, the Narceus in that area sometimes have a bit of a gordanus-ey look to them no? (Though definitely not N. gordanus of course) More brownish/yellowish rings than some of the more northernly reddish populations. I know kyle from roachcrossing works with this locality that has some really strikingly yellow banding on the males
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.