When I prompt the computer vision for this observation of Buchnera america I get the following results, where the correct genus is suggested, but no species options in that genus are provided while a species in another genus is:
I suspect this is related to the unusual fact that the nearest non-obscured observation is some 600 km away and that the geographic filtering is functioning somewhat differently for the “we’re pretty sure” suggestion and the “top suggestions”. Obviously kind of a weird edge case but does not seem like intended behaviour to me.
I presume a Buchnera species would be shown then. This usually happens when the CV is fairly sure about a species but it isn’t predicted to occur in an area, so although the genus is shown, the species is hidden by default
This just means that the geomodel doesn’t predict the species in that specific area. It may change when version 2.15 of the CV comes out in about a month (Correction: the geomodel appears to be updated separately from the computer vision)
This seems like a bug perhaps due to all the nearby observations being obscured.
The resulting behaviour that seems like it is maybe not functioning as intended. If a species is being excluded from the “we’re pretty sure” for geographic reasons, I’m not sure why it would be desirable to instead suggest the genus which also has no nearby observations? It seems like it would make much more sense to just exclude the species entirely, and have Prairie Phlox as the top suggestion.
Interestingly there are no observations of the species from that coloured region! But maybe this is just an example of the model not dealing well with a species with several extremely isolated populations.
I can’t tell if the obscuration is due to OP choice or because of the species’ conservation status. I assume the latter. But does CV and it’s suggestions work off of obscured locations or the precise (hidden) locations? Note that the majority of obscured pins in the Port Franks areas show up on the lake. Is the geomodel summarily dropping American Bluehearts from it’s list of possibilities because it “knows” it’s not an aquatic taxon?
The vision score is 97 (max = 100) for Buchnera americana, but it has not been “seen nearby”. This might be the reason to suggest the genus (the correct ID could be a very similar species at this location, but not yet in the computer vision model ?). I agree that it would be better to suggest the species Buchnera americana, even if not “seen nearby”.
The vision score is 0 for Phlox pilosa, a very good reason not to suggest this species, even thought it has been “seen nearby”.
