If you add a non-disagreeing identification of lower specificity (let’s say, genus) to an observation with only one other ID (in this example at species level) and then check “Community taxon cannot be confirmed or improved”, you get a research grade observation where community taxon (CT) and and observation taxon (OT) don’t mach up.
If you then looked up research grade observations of the initial, more specific ID, this observation shows up, even though the “research grade” should apply to the CT, not the OT.
I’m wondering how this data is handled. If a researcher were to use the data, would this observation count as a RG observation for the species, or only as a RG observation for the genus?
If not this could be a potential cause of some faulty data. I am sure it is a very rare thing to happen, but perhaps should be prevented (perhaps by adding another requirement for RG that CT and OT must match?).
Of course nobody should click the “cannot be improved/confirmed button” in this case. I just discovered this out of curiosity.
It’s certainly not common, but you do get cases like you described. It’s more likely that an RG observation that was marked “cannot be improved” at the genus level (for example) gets a single refining species ID by a new person. When that happens, I don’t believe the “cannot be improved” box gets automatically unchecked. Accordingly, you end up with an observation that’s research grade because of its genus-level CT but is listed under the suggested species because that’s the OT. As far as I know, it’s the OT and not the CT that gets exported, which as you say can be misleading.
I’ve brought this up on the forums before, so this is a known behavior, but has not been changed. On the one hand, I find that very frustrating. On the other hand, the frequency of this is much lower than the number of misidentified observations that reach RG in the ‘normal’ way, so I understand why it hasn’t been a priority.
Another twist is that the Observation Taxon, not the Community Taxon, is what is sent to GBIF, which is another longstanding issue that is very frustrating.
So if someone disagrees to genus and uses the “Can’t be improved” box to send something to RG at the genus level, and then another user comes along and adds a species ID, that species ID will be sent to GBIF, even though it only has one “vote” (and may not be supported).
I’ve also seen observations where there was an initial CV genus ID from the observer and then a single species level ID from an IDer. The IDer ticked “cannot be improved” and the species level ID went to GBIF. Since RG observations get much less scrutiny from IDers, observations like this will get little oversight.
Most observation data exports, such as CSVs and GBIF, list the observation taxon, not the community taxon. The community taxon is available via the API as community_taxon_id.
Is there a way to search for observations in this category? Some specific URL that can be used to bring them up? I’d add this to my saved list of URL searches to bring up these observations when I’m identifying.
you can have the system filter for higher-than-species research grade observations, but you’d probably have to then get those observations via the API and run your own secondary filter on your end to get cases where the community ID is different from the observation ID or something like that, and maybe where IDs at the observation ID are not disagreements (and possibly also doing some special filtering for opt outs).
there are also probably cases where potential subspecies IDs could have been stuck at the species level becaue of the box, but these would probably be harder to find, i think. i’d have to see exactly what the system is doing with subspecies cases to see if there’s a reasonable way for these to be found efficiently.
if you have a particular case you’d like to search for, i can probably walk you through how i’d do it.
I agree that this can be a problem and should be addressed sooner or later. However, I’m glad iNaturalist works this way when something is ID’d to subspecies (or variety) and I don’t know or really care if it’s correct at the subspecies level but I do agree with the species. The observation goes to RG with the subspecies name and that’s fine with me because I think it is correct; I just don’t want to take the time to figure that out.
True, but I think subspecies are already handled separately. I believe that a single subspecies ID will change neither the CT nor the OT, so at least those two stay the same. Though I do not know whether this is true also when someone has checked “can be improved”