There are some species of birds that visually identical and can only be distinguished in the field by hearing the bird vocalize. An example in the U.S. would be Willow vs. Alder Flycatcher. But while ID’ing observations, I see a bird specifically identified as an Alder Flycatcher or as a Willow Flycatcher. As far as I know, there is no way to confirm such an observation. What is the appropriate way to handle that situation? Do you just mark it “reviewed” and move on? Or for every such observation, do you ask the observer if they heard the bird?
1 Like
I recommend doing this, and if they don’t respond, add a disagreement at genus. Same thing applies to other taxa and ranks. For organisms requiring microscopy, and no microscope image is supplied, you can add the disagreement first.
I recommend making copypastas. Something along the lines of:
Microscopy is required to differentiate species A and species b
Or
The call of the birds is needed for ID. Did you hear the bird?
12 Likes
Sometimes location and habitat and time of year observed can be important clues to narrow down the species ID. A bird on its breeding grounds might be IDable to species that way, especially if singing. Info the observer can provide in notes — if the photo itself is not diagnostic— can be valuable.
5 Likes
Yeah, these are important points. Some species that overlap in appearance but not in range.
I’d also add that if the ID is stuck at the genus and can’t be improved, then it should be marked as such in DQA and bumped up to Research Grade.
5 Likes
I usually just make a genus-level ID noting that more information is needed to confirm the species and then check the “as good as it can be” box. The specific example I’m thinking of is the shark eye and false shark eye (moon snails). West of Florida both species are in range and you need a picture of the underside to differentiate them. https://olram9.wixsite.com/letstalkseashells/copy-of-template-326?lightbox=dataItem-j9fkd0dp
2 Likes
One can revert to a genus level ID, and provide a reason, such as “Genitalia dissection is needed to determine species”, which seems to be quite a common thing for insect, spider or other terrestrial arthropod, or “Need photo of this view to distinguish at species level”.
2 Likes
The example was two species in the Genus Empidonax. Empids may all look alike, but they have so much diversity in nesting habitats that a genus-level ID is not useful. Returning to the OP’s specific example, the Peterson Guide says,
Alder Flycatcher: Alder swamps, wet thickets, usually near water.
Willow Flycatcher: Wet and dry thickets, brushy pastures, old orchards, willows.
Evolutionarily, I suspect that they first speciated by habitat, and secondarily developed different songs.
Is there anything below the genus that could be used? I don’t see keeping these observations as ‘Needs ID’ forever any more useful.
2 Likes
I’ve occasionally, where one species is more probable than the other but both are possible, let them go to genus by adding an ID for the more probable species when the less probable one was the initial ID, or leaving one to stand where someone else suggested the alternate option, along with making a comment describing my understanding of the situation and discouraging any confirming ID to be more specific than genus without a good explanation.
Particularly where the genus is large and those are the only two species found in that range.
I could ID to genus and leave a comment about the possible options - but where doing the above works as planned it seems like a reasonable way to indicate that the community thinks it’s definitely one or the other but we can’t be sure which, with nice links directly to the taxa pages of the species involved.
Not sure I’d recommend it as a Best Practice, but it’s an experiment that so far doesn’t seem to have backfired badly.
I have no idea whether it would be technically feasible, but I’ve often wished for a “cf” status for these cases – i.e., the ability to ID the observation to genus but have a note under the observation ID “cf. species” that could be included in searches for the species if desired. This seems like it would both allow for the observations to be recorded at a level that is appropriate to the available evidence, but also satisfy observers’ desire to acknowledge the species that it probably is.
The problem with this (one species ID that nobody will confirm or disagree with) is that observations end up in a sort of limbo, never becoming RG, and it becomes very difficult for IDers to sort observations to find the ones that can be confirmed vs. the ones that can’t but nobody has pushed back to genus.
As a specific example as to why this isn’t just a theoretical problem created by IDers with overly strict standards of certainty: I am grappling with the challenge of how to ID Xylocopas in parts of Europe where only one species is widely established but a second species is expanding its range. The habitats are similar (they have even been found sharing a nesting site) and only the males are easy to distinguish from photos.
It will obscure important data about what we know (or don’t know) about the occurrence and relative frequency of the two species if unidentifiable females are simply ID’d as the more common species, so there are strong arguments for leaving such observations at genus.
On the other hand, it also feels rather absurd to disagree with species IDs in areas where the likelihood that it is the less common species is close to zero – but the problem is, if we decide it is OK to ID the most likely species at this location, where do we draw the line? How far away does it have to be from the nearest documented occurrence of the second species? Absent any clear geographical barriers, it inevitably will result in things like people IDing the species at location x but not 50 or 20 km further south, which is pretty arbitrary, particularly given that the known ranges are currently very dynamic.
5 Likes
If the species has not already been ID’d, I will just identify as much as precisely as I can. If the species has been ID’d more than I’d be comfortable, I will just leave it alone. I’m personally not the kind of iNaturalist user to put too much effort into identifications I’m not already familiar with, so if the ID was like this I wouldn’t engage. If I had to leave an ID, I would just put whatever I feel comfortable with and explain my confidence level if asked. It doesn’t hurt to ask, but usually people present all the information they have, so I’m not one to go out of my way to ask for the sound or any other information, unless they asked me about my ID.
If I knew that one species was the only possible species, I would default to that one. Usually there is some subtle detail or hint that isn’t diagnostic, but if I see it slightly lean towards the significantly more likely species, I’m fine calling it that one also. It’s definitely case-by-case whether enough of that information can be determined, and if the location leans heavily enough on that particular species. If it’s likely, but not convincing enough, I will just ID what I’m comfortable with then comment whatever I’m leaning towards.
Another thought I have is that with some instances, part of me doesn’t believe the species is unidentifiable. Especially if it’s something I’m not familiar with, I’m often under the impression that some mega expert out there might be comfortable with the ID. Maybe they know a secret ID feature that I don’t, or are just more familiar with subtle proportional patterns I can’t recognize. If it’s a rarity, maybe some forensic analyst can match up the individual’s unique plumage or injury or something. Again, I just ID what I am comfortable with, but you never know.
1 Like
I’d absolutely LOVE this. There is a “cf” observation field, but it’s laborious to access and use and not immediate either to view or to consult.
Thinking about plants in my area (geographic and of expertise), or a whole lot of arthropods, there are so many times when for distribution, season etc., it is possible to be 95% sure it is THIS common species, but probably no-one can be 5% sure it is not THAT much rarer species. A “cf” field would give satisfaction to the observer, make the observation accessible in searches and also enable really expert experts to find the observation and resolve the “cf” if possible. I wonder, would it be technically feasible?
1 Like
It just occurred to me that for birds that are visibly identical but have different vocalizations, it would be great if the iNaturalist observation form had a checkbox that said something like “seen and heard.” This is a commonly used phrase in eBird reports of such species, and something that I would always state if I was reporting an Alder Flycatcher in eBird.
It would be even better if such species had a flag in the database that basically said “a picture isn’t enough to identify the species”, so that people reporting such an observation would know right away that they need to provide additional info. Such a flag could also be used for other species, e.g. bees.
1 Like
That would be a very niche feature that would slow down the process of making observations and would only apply to maybe a dozen of taxa. Perhaps an observation field would be better
I agree, which is why I think that a database flag that says “a picture isn’t enough to identify the species” would be better and much more general. Then have a separate table that is joined on that flag that specifies what other observational data is required, such as vocalization for birds, genitalia for bees, etc.
The problem is, what is necessary to ID a species can vary quite a bit, depending on life stage, sex, location and various other factors.
There may be a few species that can never be identified without one single specific piece of evidence (song, genitalia, pheremones, chemical test, etc.). However, my experience (based mostly on IDing bees) is that this is the exception. It is far more common to have situations where the species can’t be distinguished from similar species the majority of the time, but in a small percentage of cases, given the right photos and a suitably skilled IDer, an ID is possible.
Perhaps I’m misunderstanding what you are proposing, but it seems like it would be very difficult to create such a flag that would list all relevant info for a species (and not put off observers who might see this requirement as a hurdle that they have to overcome before they can upload their observations).
Those two (Alder and Willow Flycatchers) are considered sister species, phylogenetically each other’s closest relatives. If there was such a rank in the official taxonomy, it could be used. As of now, iNaturalist has no species complexes in Empidonax.
1 Like
The problem with Willow/Alder is that even if the observer says they heard the correct song, there is no way for me to confirm that they know how to distinguish it. (I am more inclined to take their word for it if they say it’s a Least, that one is more distinctive. Willow and Alder have different songs, but not so different that it is simply obvious to everyone.) I don’t disagree with people who say that’s how they made the ID, but I’m not comfortable confirming it, either.
However, I absolutely have known birders who at least thought that they could separate Willow and Alder Flycatchers at least some of the time. (I think it involved, uh, primary projection? Secondary projection? Something with the proportions of certain feathers.) Even if these people were wrong, it still might turn out to be possible to distinguish at least some of these birds. So, personally, I wouldn’t mark those as impossible to improve beyond the genus level. I just leave them alone.
It would be nice if there was an equivalent of Traill’s Flycatcher. I understand why there isn’t, but it would still be nice. It seems to me that if you can narrow it down to only two of the Empids, you deserve to get something for your efforts that’s better than being stuck at Empidonax. 
…none of that answers your question, does it? Probably the Empids are stealing our car keys right about now.
This is why I see it appropriate to bring a genus-level community ID to RG. There are currently no other options in this case. If there was a complex, ID to that level and then raise the observation to RG.
Perhaps an observation field can be created to track observations where the species can’t be distinguished?
1 Like
If the observer believes they can confidently ID to species using cues they observed but that aren’t evidenced in what they uploaded, then it’s appropriate for them to ID to species.
If you can only confidently ID to genus from the evidence provided, but believe their ID is plausible, then it’s appropriate for you to ID to genus without disagreeing with their ID. You generally shouldn’t confirm it to species unless you have some first hand supporting evidence of your own.