Common unidentifiable species: chipmunks, and bats oh my!

Another problematic group are for example cultivated plants like roses and Bougainvilleas. Some people tend to ID them to species which is often impossible. A warning would be something very good.

6 Likes

Agreed. Range can be useful, but for many species we have not yet established the entire range and some animals disperse or end up in odd places. However, I think that providing a warning and resources justifying the warning could only improve the quality of species identification in the vast majority of cases

1 Like

And from the other side, sounds like it would be helpful to talk about how I arrived at my ID when I post an observation. I just put basic data (habitat, size, etc.) in the description box - thought people wouldn’t want to read any more.

But I can see where explaining, for example, why I ID’d my rabbits to genus instead of species (can’t rule out the less common New England cottontail, even though it’s almost certainly the eastern cottontail), would improve ID responses.

I’m just glad I don’t have to figure out how to encourage people to do that without cluttering up the observation page!

1 Like

If you are an active user who understands that you are posting an observation of a potentially difficult to identify species and the photo does not contain sufficient information for the ID to be made, I don’t think it is at all unreasonable to expect some additional information in the description or comments.

As an active identifier, it’s quite the burden to request for me to add an @ comment and try to carry out a dialogue in every one of these instances, especially when >90% of them are cases where the user will never respond or is unaware of the identification issues. I try to at least do that when I recognize an active or knowledgeable user before downgrading an ID to genus or species.

A lot of the situations I come across are observations that are incorrectly at Research Grade when they cannot be identified to species (similar to @dbachen’s examples). If I don’t bump them down to genus and instead just leave a comment, there is no easy or efficient way to follow up later on all of those observations if the observer does not respond. The result is that a lot of bad data stays in the database.

As I see it, it’s not a big deal if I downgrade something to genus and later reverse that - I’m an active user and will respond if the observer initiates dialog and provides sufficient additional information to confirm the species-level ID.

9 Likes

as a person who is very much both active as an identifier and an observer, i’m gonna get seriously annoyed if you bump stuff back up to genus without asking about it first. Maybe that’s wrong, but it’s definitely what is going to happen and would make me more likely to turn off community ID for that observation rather than engage further.

1 Like

As others have mentioned, I think a pre-post warning would be a great idea. If I upload a picture and call it ‘Bat A’, but it is known that Bat A shares range with and is almost visually indistinguishable from ‘Bat B’ a pop up with this information and the option to ‘Choose Bat A, I know what I’m talking about’ (maybe put less bluntly) or ‘choose genus’ could cut down on people choosing a species level based on the AI, or a potentially limited/outdated ID book. I often flick though ID books to identify what i find, but I know that the ID book doesn’t have everything in my area so i am basically pattern matching against an incomplete set.

Of course, I think these notices would need to be set up by knowledgeable curators on a case-by-case basis.

Another point I want to mention is the infamous pull of research grade. Perhaps for some taxa where is is unreasonable to expect a species-level ID from just photographs, Genus level identifications could be considered research grade.

3 Likes

Bumping observations back to genus when the photos don’t support the observation but don’t contradict it doesn’t make sense with plants. The observer may know additional information so they can be treated like an observation without a photo.

With insects this is necessary though, otherwise the data quickly degenerates into garbage. I am always bumping black Aphis or Uroleucon to genus because there are a half dozen species consistent with the photos. If I don’t do this than there will be hundreds of incorrect observations which someone will eventually agree with. Yes, there is always a chance someone pulled out a microscope and keyed it out, but I have yet to find a case of this.

This has become worse since the vision algorithm only knows a few dozen of the thousands of aphids. Any small green bug immediately becomes a potato aphid. Hopefully once a few dozen more are known people will face a block of identical looking suggestions and realize how hard this is.

7 Likes

@glmory, @charlie: What would you do in this situation: A population of a difficult to identify plant has been properly documented (with detailed imagery) in the past and reached Research Grade with appropriate concurrence by other expertise. Then a continuing (e.g. floristic) survey adds additional images of individuals in the same exact local population on subsequent dates but offers only moderate quality (i.e. non-diagnostic) images. Based on all circumstances and local knowledge, the additional observations are assuredly the same species (and sometimes even the same individual) but they cannot be idenfitied with certainty from the subsequent imagery. Should an observer feel obligated to obtain diagnostic (e.g. dissected) imagery for every observation of that species/population? Should an identifier downgrade every subsequent observation because of the lack of definitive detail in each one? or should the identifier take previous documentation into account and concur to (properly) get the subsequent observations to Research Grade?

This can occasionally happen when a 3rd party reviewer in good faith downgrades a single observation, not realizing that it may represent continuing documentation of a single population/individual.

2 Likes

I think the piece that’s missing in this example is whether or not the situation is explained in the comments (especially with links to previous observations). If the new survey photos were added without any comments, it seems completely reasonable to question the identification.

It’s my take that every observation should stand on its own, and it shouldn’t be on the identifier to dig through other observations to find important context.

2 Likes

To be honest I barely ever explicitly disagree unless I think the id is probably wrong - looks wrong or way out of range and habitat. If someone says they saw white ash and has a photo where I can’t tell if it’s white ash or green ash, and both grow nearby/are in range, I leave it alone. Bumping it back bumps it to genus with a ton of other ash species and ruins the observation. Also sometimes new people come along who are especially good at those species. I understand doing it in the case of the aphids or maybe a really rare and difficult plant like a sedge you need a scope to identify that a middle school kid records in the playground and the photo is just a blurry blob. But I think mostly… better to leave them be. I do wish “no further Id possible” still worked on observations with only one ID though. (For instance some of my own where I was confident of ID but when I upload I see that the photo is insufficient for anyone to verify). But with plants if nothing else someone can usually go back and look for it later.

3 Likes

I would tend to only disagree if I can see evidence that it is not. If I thought it was 95% likely to be that species, can see no evidence that it is not, and that I know that there are two species similar, I might ID at species with a comment that I am not certain, stating in the comment the other taxa that it might be. If the observer responded that they couldn’t be sure either, then I would change my ID to genus non-explicit. I agree with Charlie, if it represented an out of range scenario then that would significantly change that 95% confidence, and I would probably hold that as enough reason to determine it not to be that species, and ID explicit disagreement.

There has been a lot of discussion around the explicit disagreements, and as best I can tell, we should only be doing so if we can see evidence that it is not that species/taxa, or if it is an absentee/unresponsive identifier and they are not responding to dialogue over the ID and the community [would] agree that an explicit disagreement to bump the ID back is appropriate. So on that basis, you should be initiating dialogue before making such an ID, or at least have been involved in such dialogues on past observations of that taxa/place such that you can confidently predict what the outcome of such a conversation would be in the current instance.

3 Likes

@fogartyf You recognized a detail that I purposefully left out–the need for subsequent comments to link observations together. We’re struggling with just such a case here in Austin, TX. A plant that was photographed, dissected, and documented (with diagnostic images) was subsequently photographed in a general way (by multiple parties) without linking comments. Now a very knowledgeable identifier is downgrading all subsequent observations to genus/tribe/family solely because he can’t ID the subsequent images, even after the previous documentation has been pointed out. In this case, the identifier is taking the concept that “observations must stand alone” to what is, IMHO, an unreasonable extreme. I want to be clear: The identifier is absolutely correct that the subsequent imagery/observations, by themselves, are not identifiable to species level. But this identifier is going back through and downgrading everything he can’t ID as stand-along observations. Just a bit frustrating.

4 Likes

Let that identifier know that they should be identifying to genus (or whatever appropriate level) WITHOUT explicit disagreement

https://forum.inaturalist.org/t/change-wording-used-by-the-system-when-downgrading-an-observation-to-an-higher-level-taxa/3862/122

4 Likes

also:
https://forum.inaturalist.org/t/using-the-field-similar-observation-set-for-linking-observations-of-lepidoptera-when-raising-on/1018
might be useful for “linking” observations, especially note the section on using the link to the set in comments…

also:
https://forum.inaturalist.org/t/useful-html-tags-for-inaturalist-comments-and-other-text/6198

1 Like

I’ve encountered this too. Awfully frustrating, indeed!

2 Likes

This can happen now. It just requires an extra step. If someone answers “Based on the evidence, can the Community Taxon still be confirmed or improved?” with “No, it’s as good as it can be” then a genus level identification will become research grade. It still needs to meet the other criteria of research grade of course. It is documented but I’m not sure it’s well know.

2 Likes

Actually this isn’t correct. It just needs two IDs, they don’t need to both be at genus level. A family level ID + a genus level ID would do it for instance.

I believe this means it needs a date, location, media etc which is true.

3 posts were split to a new topic: Help identify a scorpion

Back to the original post and question: the examples of western Neotamias chipmunks and Myotis bats are good ones in that they often cannot be IDed to species based on photos, unless the photos are close-ups of certain key characteristics, and that’s usually unlikely. A number of small mammals are like this and can be challenging to ID even with a preserved skin + skull/skeleton specimen.

I usually just add a comment to the record pointing out it could be Species A (the ID provided by the submitter) but based on location and what’s visible in the photo it could also be Species B or C, etc. Sometimes if there is information on habitat and elevation you can lean towards one species or another, but I tend to not provide an ID, just a comment. And I don’t downgrade the record to genus; I figure that’s up to the submitter to address.

2 Likes