Etiquette for ID of species with no visual differences

Is there a shared etiquette/guideline for ID of species within the genus where there is no visual clue for the correct identification (precise determination possible only dissecting the specimen for instance)?
I see differen viewpoints, sometimes people keep the ID at genus level, other times it goes to species level. Which is also annoying some observers as this is a bit randomic.
My choice is going to species level as the ID even if potentially incorrect, helps to differentiate from other species in the same genus with different visual aspect/features.
Moreover, since it cannot be demonstrated as wrong, it is not misleading iNatters with other identifications.
Things even more complex when one species is known to be (in literature) quite more rare than the other.
Your thoughts?
By the way, is there a mechanism to link two species to warn for the (according to the current literature) impossibility to discern?

EDIT: I have started working out a workflow based on these comments and my (of course arbitrary) conclusions about them.
You can find it here in case it might be useful to anyone. Of course it is a work in progress and will be modified as required and test goes on.
If you feel like adding comments, please do that in this thread, not on the journal entries, so that everyone can read. Thanks (PS: no trolling or flaming please)

4 Likes

Can you give us examples please? Are those species share the same localities or not? How the people decide which species it is.

2 Likes

I’d disagree with this statement, even though I find it very frustrating when I’m unable to get down to a species level.

I’m not a taxonomist, but I believe this is the purpose of Species Complexes and/or Subgenera. However there are a lot of groups of species that aren’t in complexes. They also don’t apply to species in the same genus who look very similar, but may not actually be as closely related as they are to some of the others in their genus.

An example of this would be e.g. photos of an insect that you must look at under a microscope to ID at the species level, even though you can confidently narrow it down to 3-4 species within its genus of 50 from the photo alone.

12 Likes

This might be a cool feature request!

16 Likes

lates example I run into is Empoasca decipiens/pteridis, both present in France and Italy, even if pteridis is way more rare.
EDIT: just read it’s quite the opposite in France at least :flushed:
https://www.britishbugs.org.uk/homoptera/Cicadellidae/Empoasca_decipiens.html

The same occurred in the past with Araniella opisthographa/cucurbitina when opisthographa has 4 dots instead of the 5 canonic ones. We are talking of rarities, but this ignites discussions sometimes.
I stick to the most likely ID concept, but a lot of people think they should read the mind of Universe instead.

1 Like

One potential issue is that overly/unsupportedly specific IDs may cause issues for AI training.

One other issue is that this could have serious negative consequences if unsupportedly specific information is used in GBIF. Bad data could compromise science based on it (ie, garbage in, garbage out).

I personally feel that it is better to leave IDs at a genus level when it is impossible to ID to species level based on the information presented. It is important to be “humble” about what we don’t know. Sometimes people seem to have an expectation that every observation can be identified to a specific species, when this isn’t the case. Observations only identified to genus with an explanation of why this is so can help people make more detailed or better observations in the future. They also serve as an important lesson about the nature of science: that it is just about knowing things, but it is also important to acknowledge what we don’t know.

22 Likes

well, but if there is no visual clue AI cannot be blamed either. There should be a warning though.
I have thought of this matter quite a bit, and my idea is that the ID should be reasonable. The only impact might be for scientific studies, but I suppose a scholar knowing the known issue would have no issue in throwing the data in the same bucket

well, if there is no visual clue, there no making more detailed observations to help with ID.

2 Likes

It’s also been discussed a few times–as far as I know, without a widely-accepted resolution–that it can be a bad idea to use rarity or location to ID between similar-looking taxa, because you may be missing an ongoing shift in the range, etc. For example, deciding that some animal is a certain subspecies based purely on where you saw it.

13 Likes

Definitely. In this case, the ID just becomes an example of confirmation bias.

6 Likes

i agree with the “humble” part above. i also tend to think if you apply humility to the prior sentence, that also means that if you find an observation that is identified to species when you think it can only be identified to genus, that the best approach there is to make a genus level ID without an explicit disagreement and note to the observer why you think the genus level ID is better. let the observer change the original ID to genus if in agreement.

12 Likes

there have also been multiple suggestions for popups or warnings for these cryptic species but as it would be pretty tricky and sometimes subjective to implement, they haven’t gone anywhere as of yet.

I personally think the best route is speceis complexes or subgenus groups, but as mentioned above, sometimes they aren’t possible to create while being genetically correct, which some people don’t want to do, and to be fair does create some significant potentials for rabbit holes of too many self defined groups.

4 Likes

Well, if we exclude location as a criteria I think it would be a real mess in iNat. We would have species and whole genera popping up worldwide.
Take the case of Frontinella which is proposed as a ID by AI in Europe (where only Frontinellina genus is listed). You cannot drop location from the ID process in my view, even if the temptation to “discover a new species” can seem tempting. Usually is just a wrong ID.

2 Likes

let’s take the example of Araniella cucurbitina vs opisthographa with 4 dots (usually has got 5).
When I make IDs I have chosen to identify Araniella with 4 dots as cucurbitina, keeping in mind that this would push a new picture reference to the database, which can not be demonstrated to be wrong.
Most A. cucurbitina, if you consider this possible 4 dot exception of the very less common opisthographa, should be IDed as Araniella species, And basically most Araniella should be IDed as Araniella sp then, for the grid of exception possible for each species.

This is just an example, please don’t focus on this genus but try to understand the concept of what I’m trying to convey
.

ID to the level you can be confident of from the evidence supplied. If range can reliably be used as characteristic, then use it! If you feel the need to qualify your ID then add a comment! Put the full name of the possibilities in the comment, so that it might show up in searches. Anyone looking for data on a species that has a look-alike can search the genus and review for themselves what they want to use, and in fact I think all data users would be advised to do so.

11 Likes

I think it is misleading since it implies that a species can be identified from your photos, and that your individual is of that species. People may use this as evidence to identify other observations.

Generally the etiquette is to identify to the lowest common taxon, usually genus but sometimes further up. If a subgenus or complex taxon exists that’s good, otherwise observation fields can work if you want to mark it as a species pair or other arbitrary subgeneric group. (there are other discussions and feature requests on the forum about improving this aspect)

What I’m unsure about is using location to identify difficult to distinguish species. Toxomerus geminatus and Toxomerus occidentalis look pretty much identical, but their range is well know and on opposite sides of the continent so I feel comfortable identifying them despite that. On the other hand I’ve been researching the millipede genus Brachyiulus and am not sure what to do. Only one species (pusillus) is documented for eastern Canada, but another species (lusitanus) has been recorded in North Carolina, California, and maybe Ohio. So is it safe to call mine pusillus or should I be careful because Ohio isn’t too far away? And then you get cases like these where both are possible and probability is used…

8 Likes

yeah the range one is slippery. On the one hand, it’s absurd to consider literally every pine species on the planet every time someone posts a pine observation in Vermont where there are really only a few pine species that ever occur here. On the other had, going down to subspecies level of everything based on location when there’s no differentiating characters is circular logic and doesn’t really serve any purpose, as discussed at length in that other thread. But where is the line exactly? Hard to say.Range, habitat, etc all make a difference and are valid in considering ID, but i don’t see the point of people who go through and change all my royal ferns to the North American subspecies based on the fact that I am in North America when there’s no other visible difference in photos… what’s the point? And if the European one ever somehow got introduced no one would ever know because they are only mapped by location…

7 Likes

I definitely agree that location should be a part of the IDing process; the potential problem comes from overlapping ranges and the like.

In the case where the observation shows what is almost certainly X, but some Y might somehow have an identical observation–personally, I think this will come down to a subjective decision that won’t necessarily be agreed upon. No one, not even experts, are IDing at 100% confidence. It just becomes a question of how uncertain should you be before backing off to the higher taxon.

For an expert who’s a stickler for accuracy, that might be 99.9% sure, for others, maybe 95%, for a beginner, it’s probably a lot lower–and the hope is that a good enough mix of users will all ID and a meaningful taxon will coalesce out of that. It might mean people will disagree with your species-level ID though.

3 Likes

If things were this way, I think iNat should not allow picture galleries for the affected species, don’t you think so?
If they are allowed I assume there is a thought behind (no possibility to challenge the ID).
Leaving the ID at genus level without sharing link which document the potential mixup, just let more people uselessly looking around for a solution to the riddle.
Moreover, in case only a couple of species are listed out of many, having a visual reference visually correct I think is a better option anyway.
Think of cases where the local documented distribution of species is limited to one of the two involved by the ambiguity.

well, I assume that in these cases a study need to engage some expert in the field, instead of being based on ID by iNat… with no offence meant to the IDer (me included) who just are doing their best.
But I think that there should be an etiquette guideline discussed thoroughly and maybe put to vote as a canvas suggestion to IDers and which is just a recommendation, non mandatory.
Just to share all the point of views in a thread and have a people read it and take their stance based on their understanding and view of iNat.

If it is possibly 1 of 2 species in a genus of say 10, and there is no basis to say it is one over the other, then why are you picking one? Genus is a valid id!

As for photo, if it doesn’t show you what you need in order to determine, then don’t. But if someone who makes an observation was able to look at the specimen with more detail say with a field loupe, then THEY could put an id to species. Some with a great level of experience with the taxa might see secondary characters that might suffice in tipping a “could be either” id into a “95% sure it’s this”.

It really doesn’t have to be any more complicated than “id as you can confidently call it yourself”, and either ignore what others put or have a conversation with them about it. The only “right answer” on an id is what the CID algorithm determines it to be, and that relies on YOU (meaning each of us) being HONEST about what you can confidently call it.

[Edit] I’ll add that if you are making a courser ID, please only explicitly disagree if you can see evidence (range included) that it can’t be the current CID. This is an ongoing issue until they get around to re-wording that danged dialog!

9 Likes