Etiquette for ID of species with no visual differences

That’s a pretty big assumption to apply to all of us.
I am one such amateur looking for a connection with
nature and

  1. “Most likely” isn’t good enough for me. When someone IDs one of my observations that I thought wasn’t distinctive to species I will comment to ask how they identified it before I agree with them. **
  2. I am fine with higher level talons when that’s the best I can get with the evidence I supplied.

And, while I know you didn’t mean it that way, I am a little insulted that you think us amateurs can’t be satisfied with the ambiguity of a higher level ID.

** See the following examples, and I don’t think I am unique:


The guideline is provided by iNat when there is an implied disagreement:

"Is the evidence provided enough to confirm this is [taxon]?"

This is the question iNat expects us to ask ourselves when we make any ID (and IMO this is perfectly reasonable way of handling it). Adding a finer ID without being confident there is enough evidence can only lead to finer yet less-certain identifications. The point is for the community to form a consensus. If this must be at a level coarser than what you would personally like then so be it. The aim of iNat isn’t to get everything to species level. Its to form a community consensus on IDs of observations.


the positive benefit of deferring to the observer first in such situations, in my opinion, is that it shows a little added respect to the observer and helps to foster a friendly and supportive community. i believe that iNaturalist is a community first and a scientific tool second. a disagreement to genus without some other interaction first, i think, can sometimes come off as “i know better than you even though i have only secondary evidence and you were actually there observing in person”. if the observer feels bad as a result of that kind of interaction, that seems to be a case where the community mission is eroded for the sake of scientific purity and/or expediency, which seems backward to me.

for example, in my area, we have the potential for both Eastern Meadowlarks (Sturnella magna) and Western Meadowlarks (S. neglecta). sometimes it’s hard to visually tell the difference between the two even with decent photos, but it’s relatively easier to tell between the two based on call. so if someone puts up an observation and calls it S. magna, but based on the photo i think i can only call it Sturnella, i don’t think i’m in the right if i put a Sturnella ID and explicitly disagree with S. magna. i think the better process in this situation would be to put a comment with or without a non-disagreeing Sturnella ID, and say that the photo might not be enough to distinguish between S. magna and S. neglecta and why. that gives the observer a chance to say, “hey, i didn’t know that. thanks for teaching me something” and change the ID on his own, or else clarify with something like “even though i didn’t record it, i heard the call, and i’ve been birding in the area for 50 years, and it was definitely S. magna.” or maybe the observer might point out some visual cue that you missed altogether…


Hm. I agree with the general tenor of your comment. I feel like a lot of this thread is trying to reduce some serious difficulties of interpretation to pure procedure and rule: click this set of buttons according to this absolute prescription, and you will be doing the right thing! But in practice, having the sort of conversation you describe is a necessity–it can’t be reduced to a mere exchange of IDs and agree/disagree.

I think I would say that it’s not necessarily a breach of etiquette to introduce a disagreeing higher ID at first interaction, as long as you include that explanation of why you don’t think it’s identifiable to species level. If we regard iNat as a long-term repository of biodiversity data, I think it’s not unreasonable to expect observers to make legible additional information that led them to make an identification, beyond what’s available to everyone from photo or sound, date/time, and location. (If you’ve, say, measured fern spores to determine ploidy and hence species, you don’t have to post the micrograph…but you should explain that you’ve done so in your comments when making the observation.)

I think similar constraints should apply to the “can the Community Taxon still be confirmed or improved” checkbox. I use that very rarely, because I always have some doubt as to whether some expert out there might be able to consider other factors of habitat etc. that I’m unfamiliar with (and because knocking things out of the identification stream gets people very stirred up, cf. the cultivated/introduced discussion), but perhaps I (and others) should use it more. On the other hand, if that’s checked above species level, I think the person checking it should explain why, at least tersely.


The goal, for me, of this thread was to get a reasonable shared approach to species which are not visually distinguishable - even if now I feel is drifting a bit off topic.
The usual case by case approach does not work when you are reviewing like I sometimes do hundreds of obs of a specific species to bring some consistency.
But I understand the issue seems to be considered minor, probably because few people try to do some taxa wide cleaning up and that’s why in some cases things looks that spotty. To me, at least.

If I am understanding correctly, this is something that is done often for groups that are difficult to identify and often misidentified, such as these. Those are a major issue because the computer vision takes observations that have been identified to species (often when they shouldn’t have been) and suggests the species for other observations, leading more people to thinking that they’re identifiable. Eupeodes americanus is an example that I mentioned above. Generally the process is the same; make a disagreeing ID to push it back to genus (or whatever is appropriate) and include a comment explaining that the species options are indistinguishable.

1 Like

I didn’t mean to imply that you were engaged in reductionism for its own sake–I’m a fairly high-volume identifier in my own taxon of interest (ferns) and it is great that we can handle a lot of data just by button-pushing.

But I don’t think there is a fully generalizable approach for all non-visually-distinguishable species. I would agree that it’s better to identify to genus level than to pick one of several equally plausible species. However, if you’re disagreeing with other species-level identifications that are (probably) too precise, or checking the box that allows the genus-level ID to become research grade, I do think this has to be done more slowly and in small batches.

If I were dealing with some of these taxa, I might start with a journal post giving some background on the genus, why it’s not usually identifiable to species (and what characters an expert would have to use to do so), and “@” some of the top identifiers onto the post for comment. If no one seems to object, I would write a short message like this: “You’ve identified this as X, but species Y and Z are also found in this area and are visually indistinguishable. I’ve identified it as [genus] only because I don’t think it can be shown to be X based on the evidence here. Please reply if you disagree.” I’d add genus-level IDs, cut-and-pasting that message in the comments, to a reasonable number of observations, and sit back and wait a reasonable amount of time to see people’s responses. If everything was going well, and the number of conversations I had to reply to wasn’t unmanageable, I would scale up and continue processing observations until the genus was cleaned out.

Not very speedy or efficient, but it may be what’s necessary to help educate users about these taxa and find out what knowledge about them they already have.


This is indeed a great idea. If it only could be a common practice…
Would be a good step forward towards having curators at genus level -not meaning any kind of supreme court, just a person of reference. I have seen that just using the top identifier is quite ineffective, at least in my experience.
Yes, I’d say this approach would be sensible and has the advantage of creating places where refinements can be discussed. Usually these get lost in the comments to individual obs so everything gets Tantale supplice grade pretty fast and the quality of the dialogue is affected.

on the side theme of genus lookalikes + exceptions in habitus
Happened today the same on a larva of Trichoplusia ni (first determination) with green legs, green head, no black traces (as happens on Chrysodeixis). Asked a detailed picture of the thoracic area, discovered three black warts, therefore I had to move to Chrysodeixis chalcytes, unusual case very very green.
I know this may drive mad people who prefer generic IDs, but I think if you are strict about that, the more you see exceptions, the more you end up embracing a not determinable approach unless you examine DNA.
Which would make descriptions of habitus in literature a collection of useless literature.

This is why I tried to constrain the topic to the theme of species with no visible difference in literature.
But if you apply the same principle of confidence level (let’s say for me is 95% as a reference), in the case Empoasca pteridis/decipiens where a year long survey in Switzerland/north of Italy showed that in the total of the two species pteridis was 96% in frequency, why not identify it pteridis then?Still have doubts on this. A simple shared principle of perceived likelihood percentage would help though. I can imagine people OK with more than 50%

I don’t think disagreening id is a right thing, if you can’t idit to species you can’t say it not this species. So all you do is non-disagreeing genus id with your comment and add a mark for “it can be improved” so it won’t get rg if someone will agree with original identifier (if they won’t read your comment or they’re not on the site anymore).

I wonder if that happens with birds without identifiers being quite aware of it. i.e., species A is overwhelmingly common in the area but (visually indistinguishable) species B appears occasionally as a stray. So the tendency will be to identify individuals as A unless the individual happens to call or do something unusual that brings to light that it’s B.

1 Like

Speaking as someone who is firstly but not entirely a birder, you generally should know which birds are rare but expected vs. truly vagrant and exceptional.

For instance Eastern and Western Wood-Pewee are effectively indistinguishable visually. I once had a user knock back a sighting of one of mine to genus saying Western could not be eliminated. Well in 100+ years of organized birding in my province (for context the province I live in is +1 million square km in size, so they are not close by either), Western has been confirmed seen here twice. That was going too far. But someone questioning if something is a Willow or Alder Flycatcher, two effectively equally common birds here that can be separated only by voice is fine. If someone knocks back one of mine with those if I forget to note it was confirmed via voice, I’m fine with that. Knocking it back when I note it was voice identified, I’m less fine with. It is a spectrum.

For my 5 cents on the original question. I don’t want people guessing on ID’s they do on my observation, so I should not be guessing on theirs. If you know there are multiple reasonable possibilities from the evidence, you should not pick one.


It happens with every group. Different thrushes had rg as Fieldfare because identifiers were so used to it and didn’t notice it’s a different bird.


another example I run into right now: Emmelina
3.5k records of monodactyla, 0 of argoteles
In literature, they cannot be distinguished by habitus, so being rigorous all of Palearctic E. monodactyla should be reclassified as Emmelina sp.

1 Like

I think if someone would like to do that - a new info is always welcome, and a genus id (without strong disagreement) and copypasted comment about 2 species is a great thing to do in this situation.


agree, but there is the technical problem in sanitizing 3500 obs (no way you do that one by one :grinning:) + again , the right approach (nasty genus id+ holding bin+ not improvable?)

1 Like

That’s why I started with if, I’m not sure you want to check them all. :D
Yes, I think it’s optimal.

@edolis - I’m not trying to be difficult or argumentative here, but it is really difficult to follow what you are arguing for or think is appropriate through this thread. You start out with the statement I note above indicating you are perfectly comfortable doing ID’s that are not supportable with the evidence or cant be separated from something else, even if it is wrong.

Now you seem to be arguing that you can find examples of this going on and they need to be fixed.

Have you changed your mind on this in the past 2 days? Do you feel it is OK for you to do it, but unacceptable for others to do it?

Maybe I’m just not clued in enough to the discussion, but it seems really difficult to me at least to understand what you are advocating for here.


I opened this thread to discuss the best approach to handling this case (not to spread out a biblical truth), gathering the opinion of people who are purists in order to get to what I consider the best balance… Along the discussion (post 54), which is long I reckon, I have gathered elements which have pushed me towards the approach of applied likelihood or “nasty genus id+ holding bin+ not improvable”, so yes, I have changed my initial idea, which means this discussion has been constructive.
I’ll probably use this post as a storage of hints to refer to in the comments to obs as well.
And - I’ll soon modify the text of the topic trying to summarize the approach which is best balanced and I am going to apply for my ID activity.
In my world is gap analysis/VOC+ponder/find best workable tradeoff+standardization (good old PDCA?). A replacement for guidelines which are not specific.
Still have a couple of things to think over.


I agree. I’m frustrated, for example, with Bombus vosnesenskii. Bombus is a big genus, with lots of variation. It turns out that the bumblebees that I used to identify as B. vos are a complex of superficially identical species. I want to respect that . . . and yet, darn it, the bumblebees that look like B. vos are NOT dozens of other Bombus species. I want to restrict the possibilities. I think iNaturalist should be a lot faster to recognize complexes in these cases (and not worry too much if the complexes are actually clades).