It’s only comes up first in the list when there are more observations on complex level than on the species.
There are data fields that document the status complex in the source and attribution to the species in GBIF instead of this
The data doesn’t get wrong with that. You would accept a wider error margin. That is of course the background of any of those species complex. Accept some error margin instead of just labelling the observation as an completly unrecognizsed member of the genus.
Just a heads-up, you can quote multiple posts in one post and thus reply to multiple posts in one post. That makes it for a much cleaner thread.
3 Likes
I’m not against using data on complexes, I just want to maintain a way to search for observations that still need species level IDs even if they are in a complex
You can specify the exact level or a range of levels to search and identify specifically let’s say only observations at complex level
As an example
https://www.inaturalist.org/observations/identify?hrank=complex&lrank=complex&place_id=any
I still want to be able to eliminate obs that others have marked as “good as can be” and to view obs that have one ID to species level but need another
2 Likes
This doesn’t make sense in the context of the statement I was replying to:
Essentially you are deciding on a different margin of error than the one set by the system, and not even applying it consistently across all complexes. Don’t identify to the main species of the complex unless you can actually rule out the other species of the complex. Otherwise you are claiming a much narrower margin of error than the one that is actually being represented and setting a new margin of error that no one else agreed to or is aware of.
1 Like
From the theoretical perspective you are of course right. I would also prefer to identify and report the complex to GBIF (without the extra DQA step). However, there are cases where the status of newly described micro species is not really settled and in addition those micro species are obviously undistinguishable even by a macro photo. Nobody would expect to get reliable data to those micro species from a citizen science project without looking into every observation that should be used. I think on the case of Portlacca oleracea. To distinguish the newly described species you have to do some specific preparation and then use a microscope (or a REM) to investigate the shape and surface of single epidermis cells. Among over 40 thousand observations less than one hundred include pictures of such kind and are proposed to be one of those. In this context for me it doesn’t make a relevant difference to identify P. oleracea. It simply doesn’t create confusion of a practical relevance.
On the other hand I would support to push the 40 thousands to the complex if they wouldn’t loose RG.
Edit:
I think I should emphasize that I changed my mind in parts back and forth during the discussion. There was a point where it seemed that observations of complexes with RG would be summarized under the genus in GBIF. That was the point where I thought it might be better to ID to species instead ( and forget my proposal to elevate them to RG in general). After a while I took it to a test. I looked for a handful of some random older observations what GBIF did in such cases. It turned out that GBIF summarized the complex under the species combined with some extra information about the alteration from complex in inat to species in GBIF. With that I changed my mind back. Complex observations with RG makes sense again and I will continue to use the DQA tick box. The Portulacca example is somewhat specific because of the still moving taxonomy.
Portulaca oleracea is almost a textbook case I guess… however, the story is not completely over. In this very peculiar instance, the iNat ‘complex’ did not prove extremely useful, RG or not. (Too) many obs still went to the “good old P. oleracea species”, despite its new very restricted meaning.
(In a nutshell: after the hypersplit “~20 microspecies” scheme of the 2010s, a change of heart is looming around the corner which would bring things down to ~2 good species only; few iNatters were made fully aware of the ‘taxon hypersplit’ in the first place, and so the new ‘complex’ went quite ignored. Not a big deal really, if everything is to be merged back and the complex discarded!)
I can’t wait to see it.
However it might mean that some 40 thousand observations should go up from species to complex and would loose RG.
1 Like
At the moment, all of the 46,399 observations located inside the ‘Complex P. oleracea’ would have to go [back] to P. oleracea species - it would not change much for the 99.7% of them already in the P. oleracea species, their RG would essentially be unaffected. A few of the current ‘microspecies’ would have to be ID’ed again or redirected to varieties though - that’s about a hundred obs to deal with. A matter of minutes really.
So? In a best case scenario, it is virtually painless to revert some unfortunate split and deactivate the unnecessary ‘complex’… on the condition that the split was ignored by a vast majority of users :)
Back on topic, I doubt the P. oleracea case in its current state is the most supportive example to this feature request - but it could have been.
Those 99 %are at species level because none of the micro species hat been described or widely recognized at the time. In the sense of the curation guidelines they should be changed to the complex as a taxon change. Almost none if any of those observations would meet the criteria the micro species. In other words: the current representation of the taxonomy would mean what was formerly the species is now the complex. I guess it didn’t happen because nobody dared to do that. It would probably the same if it might be consensus that the one species should be split in two instead off twenty.
And, as we had a data quality discussion: the actual data quality in a scientific perspective is of course only the complex though iNat reports them as RG for the species. And the same applies in practice for any other source of observations for this species.
1 Like
I just checked, it appears it is not always the case (but why then?)
E.g. this observation on iNat (Research-Grade, “Complex Portulaca oleracea”, near Rackwitz Germany).
As expected, it gets correctly exported to GBIF, but interpreted at the genus level “Portulaca” (cf. the field ‘GBIF Taxon Interpretation’, and big orange-coloured unaccepted name).
Practically, it means that anyone searching GBIF for ‘Portulaca oleracea’ data will not find this observation; it is neither available in the search results table, nor shown on the map just west of Rackwitz.
The same holds true for other RG Complex observations in other taxa, e.g. RG ‘Complex Achillea millefolium’ near Berson France treated as genus ‘Achillea’ on GBIF
Thank you for looking into it. I looked into your examples and some others. To my surprise I didn’t find one like I had seen earlier.
The more typical result seems to be: the the observation bears the name of the main species of the complex the expert taxon refers to the genus.
My examples might have been:observation as species, identification as complex DQA setting to change to research grade and may after the first transfer to GBIF the observer agreed to complex and with that added the „alteration“ entry.
Thought the result is not as good as I expected, I still plead for my initial proposal because the actual complex name is still visible.
Here are two examples of the typical result:
1 Like
Still, I respectfully disagree with the Feature Request.
As already said, on iNat the ‘complexes’ cover a broad variety of situations, not just undistinguishable microspecies. One of these situations whereby observations are stuck at ‘Complex’ level, is for organisms that are perfectly identifiable to species or subspecies - but are not yet, due to the lack of a second knowledgeable identifier to fix an initial mistake by the observer (usually as a result of Computer Vision shortcomings). Making these obs RG at Complex Level immediately after someone fixes the mistake would essentially a) export them needlessly to GBIF at genus level, and b) remove them from the ‘needs ID’ pool, leaving little chance for a second identifier to notice it and confirm it at the expected species or subspecies rank.
1 Like
Do you have examples for those species?
My experiance is different.
First: I mostly ignore the needs ID filter and use the unreviewed. There are alway some errors in the RG observations (1 in 1000 in what I go through)
Second: The observations at complex level are by far not identifiable because they are either not identifyable by photo at all or the evidence provided is just insufficient. And that is the idea of complexes. Make something of data as best as can get them rather than ignoring.
Third: I do not see many observations at complex level whithout a good reason from the beginning. They don’t get stuck at complex because of the lack of very experienced IDers but simply because those observations are not good enough.
Examples from my area are e.g. the ‘complex Dianthus caryophyllus’, the ‘complex Senecio leucanthemifolius’, the ‘complex Colchicum autumnale’, the ‘complex Ornithogalum umbellatum’, the ‘complex Juniperus phoenicea’, etc.
Most of the time, these observations get at the complex rank because visitors originating from mainland/Northern Europe spend their holidays at some remote Mediterranean island; but because they are unaware of the regional flora, they keep identifying local plants as if species from their homeland. I step in, fix their honest mistake by identifying to the (intensely studied, unambiguous) local species, and voilà: it is now at Complex rank, waiting for a second knowledgeable identifier to offer an easy species-level ID.
According to the Feature Request, it should immediately get removed from the ‘needs ID’ pool, leaving little chance for a second identifier to notice it and add a species ID. Moreover, it would get exported to GBIF in a rather unrefined state; for instance, as some ‘genus Ornithogalum’ data point (not very helpful), displayed in orange as ‘Ornithogalum umbellatum’ (misleading, for the eponymous species is not found here).
As already said, on iNaturalist, the ‘Complexes’ harvest not only hard-to-identify sister species, but also easy-to-identify mistaken species – if only temporarily. A third case I’ve personally experienced (whoops) is the observer (me) clicking the ‘Complex’ rank in the Computer Vision proposals when I really wanted the same-named species… and then a second friendly user ‘agrees’ (without much thought, just to be kind) with the wrongly-picked Complex-rank ID; had I not noticed my wrong click, it could become RG at complex rank for no reason.
3 Likes
Tried this example with Sicily, IT
There is no observation with RG and I didn’t find one with a second opinion at all and none with the complex but randomly one of two species which are hard to distinguish. All would be qualified for the complex.
Similar the 5 observations on the Baleares two species hard to distinguish three of them with RG. I would not be able to agree at species level because the observations don’t show many details.
Cyprus 18 observations one at complex level the others one of two species, hard to distinguish, none at RG.
With all respect I don’t see a major issue if they would all get research grade at complex level.
Edit
I have just created an umbrella project including the ten largest islands in the Mediterranean.
With that I filtered to observations at complex level. Of roughly 700000 observations about 3000 are at complex level in 184 complexes.
Many of these observations have a second or third opinion. Here and there identifiers explain why photo identification is not possible. I don’t see signs that that identifiers are inexperienced in this region or taxon groups.
What I see is that many would be worth to go to GBIF but don’t have the DQA tick mark.
Link to the filtered identify module
https://www.inaturalist.org/observations/identify?project_id=islands-in-the-mediterranean-umbrella&hrank=complex&lrank=complex&place_id=any
Screenshots
In my view the result support my proposal.
Why not try some Senecios instead? I mentioned several taxa on purpose - not for people to cherrypick.
Or some juniper trees maybe?
How about pretty carnations?
etc. etc.