Why identifying ‘by range’ is a problem

General
1

Usually location data is instrumental in helping identify organisms. However, recently I’ve increasingly noticed a problem when the location is entered as a country, state, or province without more precise location information. The default in situations like this is iNat puts the pin wherever on the map that name for country, state, or province happens to be with an error bubble spanning the size of that country, state, or province. This may or may not actually be in the range of that thing. The same happens any time location is just a place name (city name forexample). Smaller than province, this probably doesn’t affect much, but still occurs. So I have several suggestions: one set for observers, one set for identifiers, one set for researchers.

Observers: try not to list location by name, but try to place the pin as close to the actual location as possible. This may seem obvious, but please recognize saying something is generally in that area with an error bubble larger than, say, several km, could very well affect the identification of that observation as well as the usefulness of that observation.

Identifiers: be aware that when the text below the map just lists a country, state, or province name AND the error bubble is the size of that country, state, or province then location might not be trustworthy enough to identify to a specific level. Meaning identifying solely ‘by range’ is probably the exact wrong thing to do. Solutions to this include marking observations as location inaccurate, though this seems heavy handed. Rather, I often settle for identifying to a less specific level, say, genus, and call it good. However, you could also engage with the observer to get a more accurate location.

Researchers: if the thing your researching has a very fine range distinction with different species occurring on either side of a supposed range boundary, then be very careful to exclude observations with an error bubble large enough to introduce a misidentification error. Personally, I would feel comfortable with error bubbles of, say, 20 km, or about the size of a city, but some species boundaries can approach that level of error. At very least, having an understanding of this issue should allow you to avoid including observations that are very likely mis-identified because of this issue.

13 Likes
2

There are actually many reasons why IDing based off of range is very problematic, of which this is just one. Species wander to places they “shouldn’t” be all the time, by a variety of natural or human-driven processes.

2 Likes
3

Ah, but what would identifiers do all day if they couldn’t knock your observations back to genus with the comment “_____ does not occur here”?

2 Likes
4

I don’t see those comments as too big a problem when they’re in response to the Computer Vision’s poor ID suggestions, like recommending an American species in Europe. I tend to not even leave a comment in such cases.

When it’s a person making the mistaken ID however, I do agree that more information other than range should be included in a comment. I point to specific field marks when that happens.

7 Likes
5

That assumes that it was “the Computer Vision’s suggestion.” When it comes to mushrooms especially, it seems like half the species in the field guide suddenly “only occur in Europe.” So we have a situation where the initial ID was based on morphology and field marks in a published source, and the disagreement is “by range.” Those of us who find that problematic seem to be in a shrinking minority.

2 Likes
6

Other threads on this forum have discussed at length the fact that anyone with training in research would know better than to use crowdsourced data without examining indicators of data precision and quality. The risk on that front is low.

The much larger risk, as @alexbinck points out, is that we may miss range shifts or migrants because we identify by range.

7 Likes
7

And yet . . . Many plants really don’t live on a different continent. If the observation is from the “wrong” area, I think it needs to be held to a fairly high standard. If that observation lacks the traits needed for confident identification, it should be marked as wrong based on range.

Also, fairly often there are subtle clues that let one know that this observation isn’t the plant in question, but explaining would be difficult. The range thing can be mentioned efficiently.

Of course, sometimes plants do show up where they “shouldn’t.” When traits needed for identification are well displayed in the observation, those are fun.

42 Likes
8

I don’t mind as long as they are not a jerk about it. I’ve had a couple that were insulting when they did it. I have gotten the initial ID wrong because of the CV, which is nice to have corrected but being polite is the best way to go about.

8 Likes
11

I can’t speak for plants, but for moths, if range were ignored when making IDs, it’s not an exaggeration to say that many if not most of the moths on inat should not be ID’d to the species level. Name virtually any moth species in your area, and a cryptic species that looks identical to it in photos very well may exist somewhere else in the world. Even big “obvious” ones like the silk and sphinx moths- luna moths and Mexican moon moths are basically identical except for size, the Syssphinx spp in Arizona/west Mexico each have a virtually identical sister species in TX/east Mexico, the big obvious Manduca in temperate North America are pretty much identical to tropical congeners… and the little brown ones are even worse. The common Nearctic “armyworm moth” is identical to a load of other Mythimna from Asia and Australia, the Lucerne moth of America and the Rush Veneer of Europe are essentially identical… the corn earworm has eastern and western hemisphere cryptic species, etc., etc… of course they all differ by DNA and ecology and genitalia; but if all you have is a single snapshot of a live moth, you can’t consider any of those factors. When identifying any of these past genus, the only factor that can possibly allow for a species ID is the location the moth was observed. It’s pretty much the central conceit of any attempts to identify moths from live photographs- that cryptic species found in other parts of the world are eliminated based on location. If you take that away, the whole idea of live moth photograph ID becomes virtually impossible in many cases.

There absolutely are hitchhikers and new invasions of introduced species that occur, but in cases when a cryptic species appears in a new part of the world, the only way to even know this is happening is to be taking DNA samples and/or dissecting specimens to detect the new introduction. See, for example, the 2013 paper “Shared but overlooked: 30 species of Holarctic Microlepidoptera revealed by DNA barcodes and morphology”, which added 30 species to the North American fauna that had been overlooked due to confusion with their externally-identical congeners.

So yes, I will continue to disagree with IDs of wildly-out-of-range species that hypothetically could be correct were it not for the range. Because if known ranges can’t be considered when making IDs, we should be putting many more of our live moth photos at the genus level, at best. I’ll start by kicking the 45,000 Luna moth observations back to genus Actias with the comment “A 2,000+ mile out-of-range Actias truncatipennis cannot technically be ruled out based on this photo.” /s

20 Likes
12

I think these kinds of IDs need to be made on a taxon-by-taxon basis. You highlight a good example of a class of cases where it’s more justified, but for many other taxa it may be more typically much less necessary or useful. So I wouldn’t just say that ID by range is just generally good practice on iNat.

1 Like
13

I understand that your experience with changing mushroom taxonomy and IDers who are not adequately sensitive to how new taxonomy contradicts with existing information sources is frustrating and upsetting. I agree that when there have been taxonomic revisions, it is a good idea for IDers to note this or provide a link to more info when correcting IDs to reflect the current state of scientific knowledge. But I don’t see how repeatedly bringing up this experience in the forum – rather than with the relevant IDers – will bring you or anyone else any resolution.

In my experience, at least with arthropods, in virtually all cases observations ID’d as a species that is only documented on a completely different continent are indeed the result of the user accepting a CV suggestion without limiting results to species expected nearby. In the few cases where it is a considered ID – the observer has a strong reason to believe it is some imported species – they typically include a note explaining their reasoning and the context of the observation.

12 Likes
14

Likewise with fish, there are many lookalikes which are accepted as different species, with different (sometimes widely separated, but occasionally overlapping) ranges, so from most photographs of those it’s often only possible to identify to species if the range is known, and not confidently better than genus when observed where the ranges are known to overlap.

2 Likes
15

I often get lazy and just write “[…] does not occur here” or “[…] is a species from […]”, especially as I don’t know whether anyone is going to ever read the comment.
Also, it can get very complicated to describe the exact morphological differences. For example: Harmonia axyridis f. conspicua observations in Europe misidentified as Coelophoria saucia I cannot think of a way to convey the difference other than essentially saying “it looks slightly different”.
Or, another example: I think on a H. axyridis observation IDed as Adalia decempunctata in the US I once left a comment along the lines of “if you imagine a line through the row of three spots on each elytron, the line is too curved to be Adalia decempunctata”. I do not know how helpful that comment was to anyone. It almost confused myself… Haha

On some species it’s just impossible to ID that far if the location is off, and personally, I’d rather have a more accurate, but less precise life-list, than a very-precise, but over-confident one.
For example, I do not know morphological differences between the two flat-footed flies Polyporivora ornata (European), and P. polypori (North-American). Due to their life cycle, I believe it is unlikely for them to be introduced in each other’s region, so I’m confident in IDing these based on location.
However, there is one odd observation of a fly that looks exactly like these two in Taiwan. IDing this as either P. ornata or P. polypori would be a mistake. Has it been introduced from Europe? From the US or Canada? Perhaps it is a different (undescribed?) species? Who knows…
So I agree with this:

9 Likes
16

I agree that not every taxon is an issue with using range. Since I posted this, I tried to notice when I was assuming ID by range and it remains an easy short-cut or necessity. I think my primary concern is when the error bubble is so large as to encompass multiple taxa and when the way location is added compounds that by mis-placing the pin. In some very few cases, this likely generates error in the ID process. Otherwise, location is essential for IDs in many many instance.

5 Likes
17

In my experience the area where a too large area bubble becomes problematic is the issue of now having to make sure that the observation is legit. As in finding a southern species in the north, and now having to determine if the picture is from somewhere else online, or misplaced pin, or too large an area.

Regardless, considering that Research Grade observations go to GBIF, I have no problem at all disagreeing to a higher taxon or setting the observation to casual if the observer is unwilling to clarify when asked regarding where the photograph was taken. Especially if it would be significantly out of range.

7 Likes
18

I’ve actually had kind of the opposite problem: people disagreeing with an ID that I’ve made on an observation, because “you shouldn’t rule out X, because it occurs in Colorado too!”

Thing is, there’s this great big tourist attraction running down the middle of the state: the Rocky Mountains. (Maybe you’ve heard of them?) The average elevation of the Continental Divide is some 10,000’ / 3050m; the highest point (Grey’s Peak) is 14,270’ or just a hair under 4350m. That’s a pretty significant barrier.

I live on the Eastern side: the Front Range. Every time I’ve dealt with this sort of dissenting opinion, though (or usually not dealt with it, because I have no idea how to respond), what’s proposed is a species that occurs on the Western Slope, often in limited areas, without a single record east of the Divide in 60+ years.

I get that things show up in places where they were previously not supposed to be, and it’s just going to get more common. iNatters are documenting these changes and effectively redrawing range maps in real time. All that being said, you still have to take geography into account.

14 Likes
19

One that I’ve encountered in CO is an eastern species that has a very small handful of historic records in CO, but a couple hundred misidentified historic records from visually similar species on GBIF.

Had a few discussions on whether or not CO is in that species range.

2 Likes
20

Similar with Chironomids. Without range, nearly none could be IDed.

2 Likes
21

Assume people mean well. Tone is difficult to convey online, and many people misinterpret another’s intentions. This is especially easy on iNat, which is a community with many autistic users who generally find social interactions and being understood difficult in any context. There’s a very good chance that person who you think was being a jerk was actually well-meaning and had no intention of being insulting to you.

7 Likes
23

Seems like most of the responses in this thread are reacting solely to the title, but the original post is talking pretty specifically about issues relating to location accuracy. I encounter the described issue pretty frequently when identifying: an observation with a pin somewhere deep in northern Canada, I think “that’s way too far north for this species”, but then zoom out in the map and see that the location accuracy is 4000km and encompasses all of North America.

I wonder if it should be harder or even impossible to specify such a huge location accuracy circle - under what circumstances could anyone be unsure of an observation’s location to such a wide extent? And if they are, can the location really be described as “accurate” in the DQA?

(deleted previous post as it was unintentionally replying to raymie’s post)

10 Likes