Choosing not to identify subspecies

[sandboa]

There is a disagreement festering in the community of herp identifiers here on iNat at the moment that I would like some input about.

Do contributors have a right to their own taxonomic perspectives about the value/validity of subspecies?

In the herp ID community here there are some very hard working volunteers who go through and help to identify / verify identifications. We would be lost without them. But there are two sorts of “factions” out there.

• One group believes every animal can / should be identified down to subspecies based on morphology or range. If you are in a contact zone, you name it based on which one it looks more like.
• Another group believes that there are some animals that simply cannot be identified to subspecies because the location / morphology indicates that it should be an intergrade so neither taxon is correct.
• Another group believes that whether an animal should be identified to subspecies is a particular choice of the person who enters the record. Some people don’t like/agree with the subspecies concept so they just make all their IDs at species level. However, people from the “subspecies believers” group like to come in and identify it to subspecies even though the original person chose not to.

What I find most problematic is a trend towards a sort of “bully” taxonomy where they recruit other people (by flagging them) of their particular taxonomic bent to come in and confirm their IDs. It is getting almost “testy” between some parties.

I know the original data contributors can refuse to accept community IDs (if they are in group 3 for example), but I find that solution antithetic to the fundamental principles of the community of iNat.

Is there a way for a user to restrict the identification of their records to species level?

Chris

[charlie]

Welcome to the new forums!

There was a recent topic about identifying based on location (including subspecies) here: https://forum.inaturalist.org/t/id-based-on-range-or-appearance/524

I personally do NOT believe subspecies should be identified based on range, but there is no official policy so there isn’t a rule against it. If you can see the diagnostic features… you can do so if you want to, as long as you are confident in the ID. I personally don’t see any value in guessing subspecies if you can’t tell, at best it creates pointless data and at worst it creates downright wrong data.

I personally won’t ‘agree’ with IDs to subspecies on my observations based only on range, except in a few rare cases. Granted I mostly deal with plants. However, the community ID should not shift to subspecies without a lot of IDs anyway.

Mostly it’s up to the observer. if you encounter a thread where people are not being civil, feel free to email help@inaturalist.org to let the admins know about it and they can weigh in. But disagreeing with an ID doesn’t in and of itself count as not civil.

Personally, i’d say if you have an issue with it on your own observation, just reject community ID for that one observation. It isn’t ideal, but isn’t going to cause a huge issue either.

(also, if this thread becomes similar to or relevant to that last one i can move your post into there).

[tonyrebelo]

Part of the issue is that all are correct and all are wrong.

Some species and subspecies do have hard edges, and some have fuzzy edges, and some have extensive intermediate zones and in some the boundaries may change over time.

And then there are taxonomists who split and others who lump, and those that believe that subspecies do not exist - only species. And occasionally some maverick taxonomist comes along and makes all subspecies to species or a lot of species to subspecies. And with each generation the fashion appears to change: the pendulum over time swings from one extreme to the other.

On iNat the problem is that the ID tools (e.g. compare) do not even bother showing subspecies when they exist, so identifiers using the tools are unaware that their identifications might be “incomplete”.

I differ from Charlie. It is only by attempting to “force” IDs as finely as possible that one is able to discover the issues with that particular species’ subspecific status. It is only when you are prepared to put out your neck and make a call, that others will take you to task. And that is how we learn, and discover, and how the subspecies boundaries and morphological features are tightened up and nailed down - in those cases where it is possible.

Plant people have the additional complication of varieties, with the again added complication of different taxonomists meaning different things when designating subspecies versus varieties.

So long as matters stay cordial, does it matter? But there are always opinionated bullies. Like those that wont allow subspecies to be included in the compare, identotron and identify-suggestions tools?

[tiwane]

Hi Chris,

Thanks for posting. I think this is a tough situation. On iNaturalist, anyone can add an ID to an observation if that is what they genuinely believe it is, and the Community ID algorithm is supposed to make sense of the IDs added to an observation. Also, it is accepted practice that a user can tag other knowledgeable users to evaluate an observation and add an ID to it in order to get the ID moving in the right direction.

But as you say, this can be abused, although it is difficult to define where this practice crosses the line from helpful to harmful and “bullying”. This practice, on a larger scale, is sometimes called “vote brigading” and is illegal on sites like Reddit. Again, defining it is a judgement call, so it’s difficult to enforce. Happy to hear any suggestions for how to prevent brigading on iNat.

One thing you can do at the moment on a personal level is reach out to these users and ask them to not add IDs to your observations. Our Community Guidelines do ask that a user’s wish to be left alone is respected, so if they keep on adding IDs to your observations, you can report them to help@inaturalist.org.

Personally I don’t ID to subspecies unless a) there is clear phylogenetic evidence that the organism is a subspecies and b) if IDing it to the subspecies level will obscure its location because it’s an endangered taxon (eg Thamnophis sirtalis tetrataenia). I think it’s the latter situation that would make it difficult for iNat to allow users the option of species-level ID and above for their observations.

[kiwifergus]

The ones I “like” are where a genus level ID is made, you put species (or subspecies), and the previous IDer withdraws their ID and remakes it with explicit disagreement, simply because they believe CID should be at genus. When challenged on it, they admit it could be the species, but there is no way to tell from a photo! Not on, in my book, we are each entitled to call it as we see fit, and we should leave CID to work itself out.

I will tag in others, and it’s not about forcing a CID change. It is as much about bringing in as many opinions as possible, which should help CID to figure out the right position to settle at. It will usually be people of like mind that I tag in, and I would fully expect (and encourage) those holding the other positions to do the same.

[charlie]

@kiwifergus

The ones I “like” are where a genus level ID is made, you put species (or subspecies), and the previous IDer withdraws their ID and remakes it with explicit disagreement, simply because they believe CID should be at genus. When challenged on it, they admit it could be the species, but there is no way to tell from a photo! Not on, in my book, we are each entitled to call it as we see fit, and we should leave CID to work itself out.

i generally don’t think we should push disagreement if there’s a good chance the species in the ID is correct, even if the identifier isn’t sure. There’s lots of field context one can’t infer from a photo, and knocking someone back to genus (or whatever) without a good reason (way out of range/habitat or looks wrong) seems iffy to me. I know others disagree on that point… but if you do it to me and don’t have a convincing argument why i might be wrong, i will turn off community ID for that observation and keep it where i thought it was. In the least, if it’s an active user, start a conversation first. Often they will say ‘oh, i didn’t consider it could also be species Y’ and either add info that helps you verify it is species X, or bump it to genus themselves. Granted that gets messier with ‘fled’ users who don’t use the site any more.

[reuvenm]

I think this should be seen as normal, and only results from a quirk in the identification system - you can’t explicitly disagree until somebody has entered the identification to explicitly disagree with.

As an example, say person A wants to identify as House Sparrow, and person B and C want to identify as Eurasian Tree Sparrow. If all three put their ID in, the CID will be at the genus level because there isn’t over 2/3 agreement.

On the other hand let’s say person A thinks the identification should remain at the genus level. If they are the first person there, they can’t use explicit disagreement, because there is nothing to disagree with. If person B and C then come and identify as Eurasian Tree Sparrow, the CID becomes Eurasian Tree Sparrow - person A’s identification is disregarded! So they have to re-add the ID, or else their input is not incorporated into the CID.

I think in most cases you describe, the person would have explicitly disagreed in the first place if that was an option.

[kiwifergus]

@reuvenm

I think in most cases you describe, the person would have explicitly disagreed in the first place if that was an option.

no… they would not have! Read carefully…

@kiwifergus

When challenged on it, they admit it could be the species, but there is no way to tell from a photo!

as I clearly stated, they are re-identifying with explicit disagreement because they feel that CID should be at genus… and when challenged they fully admit that there is no reason why it couldn’t be the species ID that you have proposed. We have a spider genus here in New Zealand, of which we have only two species. I first noticed this explicit forcing when an IDer forced genus against my species ID, which to my mind boldly stated they thought it should be at genus level, because if they actually disagreed with my species ID, then surely it could only be the other, so why did they not ID as such. When I asked, I was told that it is not possible to ID these to species without dissection and micro on the genitalia. I respect their position, but to explicitly disagree that it is species B is saying that it is NOT SPECIES B!

I’ll take another example… another species of spider that we have here in NZ was described a long time ago, and a 2010 publication listing the spiders known to be present in New Zealand made reference to the family as being seriously in need of revision:

From Paquin et al. (2010) p92
“New Zealand Zoropsidae are badly in need of taxonomic attention. They have been recently transferred from Miturgidae to Zoropsidae by Raven & Stumkat (2003). The three known species are not recognisable based on published descriptions. Another 40 undescribed species are suspected (R.J. Raven, pers. comm.) and may represent several distinct genera. The males of Uliodon albopunctatus and Uliodon cervinus are undescribed.”

Now, I see observations with IDs of male Uliodon albopunctatus fairly regularly, and I respond with a genus level ID (not explicitly disagreeing though), and I quote the above to explain my position. I don’t expect anyone to change their ID! I think they should though… Likewise, I am not going to re-make my ID just so that I can explicitly disagree. It could, after all, be a male Uliodon albopunctatus! I do wonder how, though, given that there is no description to base that on! The only reason I do not explicitly disagree (I mean, it has no description, come on!) is that perhaps it has been seen to be mating with the female of the species, and perhaps that female is that species, even though it is not recognisable from the description! Interestingly, there are a number of guides that show photos of this species and they are identified to species! It cracks me up to see a photo in a guide of a male Uliodon albopunctatus! After commenting (but not explicitly disagreeing) on one occassion I had the observer come back with a link to another observation, also IDd as U. albopunctatus, as the basis for her ID, and knock me down with a feather… it was IDd by the same author of that quote I give above!

I see observations often where there is an ID and a comment “can’t ID further from a photo”, which is a valid point of view for the identifier. I make such comments often myself. But to say “no-one else can either” is not for me to say!

I stand by my statement, forcing genus with an explicit ID when you don’t actually disagree with the finer ID is wrong. It is assuming that YOUR point of view is the right one. You make your ID, perhaps comment with why, and let CID sort itself out. If you feel strongly, tag others to give their point of view, but don’t game the system to override someone elses point of view, which is what re-identifying with disagreement does.

[reuvenm]

The phrasing used for explicit disagreement is

Is the evidence provided enough to confirm this is X?

It’s deliberately phrased that way because it is intended not just for cases where the ID is identifiably wrong, but where the identification is not possible to confirm. So:

but to explicitly disagree that it is species B is saying that it is NOT SPECIES B

Isn’t accurate, it just means the observation can’t be identified from the photos. Now this may or may not be factually correct for any particular observation, but that’s what the CID is for!

For your second example (Ulbion), I really think you should be explicitly disagreeing here. There are certainly judgement calls as to when something is unidentifiable - maybe someone else could identify it further. But from your description, it seems pretty clear that in most or all cases, this isn’t the case, and the identifiers don’t actually have a good reason for identifying to species.

You don’t have to use explicit disagreement this way if you don’t want to - but this is how the system was designed and it’s a legitimate use.

You make your ID, perhaps comment with why, and let CID sort itself out.

The point is that, until you explicitly disagree, your ID doesn’t get to contribute to the CID.

As an extreme example, let’s imagine one of the spiders you describe above. The observation is unidentifiable, so 100 experts have come along and all have identified at the genus level. Now two non-experts come along and identify to species based on inaccurate information. The CID is now at species, all the experts are disregarded! Should the experts really refrain from adding explicit disagreement? I don’t see why they should.

It is assuming that YOUR point of view is the right one.

Surely that’s what I’m doing with any identification? If I identify as House sparrow, I think that any other possible ID is inaccurate based on the evidence. If I identify as Passer species with explicit disagreement, I think that any other possible ID is inaccurate based on the evidence. I don’t really see the difference between the two situations.

[charlie]

@reuvenm

It’s deliberately phrased that way because it is intended not just for cases where the ID is identifiably wrong, but where the identification is not possible to confirm.

This isn’t firmly established and has been a point of contention. I side with Kiwifergus on this one for what it’s worth. I won’t explicitly disagree unless i think the ID is wrong (including being way out of range or habitat or whatever). The language isn’t necessarily consistent either. And the community ID algorithm itself does not say ‘this may not be a hosue sparrow, i can’t tell’. The algorithm explicitly reads your vote as ‘this is not a house sparrow’. I think in the least, we need some kind of demand from Ken-Ichi before we declare something part of the protocol. And at least in my reading of iNat etiquette, knocking an active users’s observation back to genus based on ‘i can’t tell the species’ without further discussion with them is considered rude.

More broadly, the question isn’t whether or not YOU can identify it based on the photo. Or else i could justify bumping all spider observations I see back to ‘arachnids’ because i don’t know spiders. The bar is VERY high - can ANYONE tell them apart based on the photo. There are very few cases where you can explicitly say not, a lot of times keys divide out based on obscure microscopic characters but in the end there are other factors that can also be used. For nearly all plant species, there are overall ‘gestalts’ that work very well once you see tons of each species. And often in the field you see more than the photo shows.

I think the response for these should be checking yes on this:
so that it does not reach research grade if you aren’t sure.

[reuvenm]

Yeah, I should have been clearer - it certainly shouldn’t be used if its just something you don’t know enough about to identify. But that’s the case for any ID you make.

Other than that case, some of this gets back to the “What is evidence” thread discussion. An observation of a Red Oak where you can’t distinguish Black Oak from the photo is obviously not useless. But I do think that should be firmly distinguished from observations of Red Oak where it is identifiable. Otherwise, you’re going to end up with a bunch of photos of Black Oaks that are marked as “Research Grade” for Red Oak. And when I browse photos of Red Oaks trying to learn the difference, I really don’t want any photos of Black Oaks in there! There will certainly be some misidentified, but we should be avoiding that as much as possible.

If people aren’t correcting me and saying, actually, this is unidentifiable, how am I going to learn what features I actually need to photograph for a proper ID? I am very thankful to get explicit disagreements on my observations and I’m a bit surprised that other people feel differently. Never a good idea to underestimate how different people are!

The algorithm explicitly reads your vote as ‘ this is not a house sparrow ’.

Good point, there is definitely some ambiguity in different places. It’s currently

Reuvenm disagrees this is X

But I think it should be, if we wanted to be completely rigorous about it:

Reuvenm disagrees this should be identified as X

Because that’s what we’re actually doing I feel - coming to a consensus or majority view on what the identification should be.

The bar is VERY high - can ANYONE tell them apart based on the photo.

I’m not sure this is a reasonable standard. I would say the criteria I use is: “Are the people who have already made IDs able to tell them apart”.

For the easiest example imagine that the observer left explicit comments explaining why they made the identification they did, and that I know the features they’ve used aren’t actually conclusive. The identification could be right, but only by chance.

Normally it’s not that easy, as things are way more subjective. But that’s the kind of thing I’m talking about. If I’m an intermediate-level identifier, I don’t think it’s harmful for me to correct beginners, even if an expert could determine that they sometimes actually got things right by chance [at least assuming that I can’t get an expert to provide their opinion]. Of course, extreme care should be taken that you are actually more knowledgable!

I think the response for these should be checking yes on this:
so that it does not reach research grade if you aren’t sure.

This works in some ways. The issues I see:

• It still allows the existing identifiers to counteract it by checking “no, it’s as good as it can be”. Which means something can still be research grade with under 2/3 of people thinking it should be. Not a big deal.
• More importantly, it raises the question of what is more valuable. It isn’t clear to me that an unidentifiable oak photo is more valuable as “Red Oak, needs ID” than “Oak species, needs ID” or ultimately “Oak species, research grade” if enough people agree.

I think we all agree that CID should be at the genus level if everyone agrees. But I’m not sure why the 2/3 standard used elsewhere wouldn’t apply.

[charlie]

@reuvenm

If people aren’t correcting me and saying, actually, this is unidentifiable, how am I going to learn what features I actually need to photograph for a proper ID?

I think the hard part here is the difference between ’ i added this oak within the range of both species but i’ve never heard of black oak and didn’t check, i just assumed it was red oak, maybe i’m wrong’ and ‘i know both species really well and am sure this is red oak, but somehow the camera didn’t really capture the bark well enough for you to be able to tell, but someone who knows the species really well still could’. And tons of grey area. Which is why i think for most cases it’s best to leave a comment asking first. When people do that with me, and i realize the former was true (i didn’t know about or check for a look-alike) then i will happily back off on ID. If someone forces the ID through without asking, and no, it’s actually in an area of Vermont where there’s no black oak and i am very familiar with these trees and they are all red oak, then i get annoyed. If that makes any sense.

It’s not an easy one at all. An d I admit i have bumped things back to genus the way you describe in the case of fled users who haven’t posted in years and the ID really seems questionable. I think there is a lot of grey area that is really hard to define.

[kiwifergus]

The absentee, or fled, users are a spanner in the works for any discussion around CID. On the positive side, they aren’t going to be around to continue throwing spanners in the works! In the case of museum or herbaria specimens, you can (assuming they are still alive) track them down and have discussions or ask questions directly, but then again that is a huge effort to go to…

I was thinking about the tagging in of others… and the reference earlier about brigading. I tag others in when “weight” is needed, particularly for an errant ID from an absentee identifier. Mostly I tag in if I think someone is likely to have a more comprehensive understanding of the taxon than I have. If I am tagged in by someone else, I will agree if I am of the same opinion.

However, I also consider it a huge flaw in the iNat system that there is no reputation system. I am not saying experts are never wrong, but it is bizarre that someone that has many years of experience working with a taxa carries the same ID weight as someone that has never even seen one in real life before. I know it is “not per the rules”, but I do ID to add weight to those I have come to respect. There are a number of long standing members that have been very active with identifying, and that I believe are cautious identifiers, who I would agree with even if I am not understanding directly of the taxa myself, simply because I have the confidence in their IDs. I only do this because I know I am around to withdraw my ID should a dialog ensue that warrants me doing so. If those identifiers tag me for support or weight on an ID, I give it (if I can see it’s to shift an absentee mis-ID for example, or if it’s an obscure taxa that only they are up to speed on, or perhaps even because they were the authors behind the taxa in the first place!), but if there are others weighing in with a dialog that has a different view to them, I will tend to not add my weight (or even withdraw it if already given), as I don’t fully understand the taxa enough to know who is right and who is wrong. I don’t see this form of tagging in as being “brigading”.

To me, brigading would be when there are two differing points of view being contested, and the weight is called for as a sort of tug of war over the ID. I kind of struck this with the recent physalia physalis debate. I became aware of the discussion, I formed my opinion/position, and I made some IDs accordingly, and then out of the blue I was getting tagged to add weight by others, and I doubt they had developed any sort of confidence in my ID skills other than being of like mind on that one issue!

When I first joined iNat, I encountered some “wrong IDs” that I believe were made because of a google search that encountered mis-labelled images. I remember getting somewhat upset over how they could be so blind, even when I pointed out the nature of their error, they still held their IDs in defiance of my showing them how they likely came to that wrong conclusion. I remember frantically tagging in others that might help “fix it”! I had this belief that the ID could only be one thing, and it was right or it was wrong. And if I was right, they must be wrong! I have shifted from that position to now where I am much more about IDing as I see it, even with the possibility that I might be wrong, and if others put IDs different to mine, I am not bothered by it. I will comment to support or explain my position or ask questions to try and understand theirs, but I much prefer now to let CID sort itself out.

[charlie]

do you want to write up a feature request for the reputation system? I’m surprised no one has done so yet. Like i said, i’m pretty ambivalent and opposed to an externally based on myself, but a lot of people want some sort of reputation system, so it seems worthy that someone puts in the request.

[kiwifergus]

It is a hard one, because as much as I see it as a flaw, I also see it as a strength! I’m trying to think of how it would have affected me as a new user if my IDs carried less weight because of a reputation system. Would I have been more inclined to offer IDs, knowing that I would not shift the CID so easily with a wrong ID?.. probably… Would I have been more hesitant to put IDs if I felt it would harm my growing reputation score?.. also probably! Woud we get the buy in from the expert community if they didn’t feel the need to correct all the mis-IDs? I think I don’t really understand the dynamics of the whole thing well enough to put forward a better option, but perhaps to start a dialog that could explore it, sure. Along the same lines, I have often thought a means to weight ones own identifications, with a certainty score perhaps, so that one could make IDs with appropriate weight. If I am 95% certain on my ID I will make it, but if I am only 70% certain I won’t. Before anyone asks how you ascertain that you are 70% vs 95% or even 75%, I fully admit it is quantitatively gauging a qualitative value, but then so too is the black and white way it happens now! I’ll play around with some starting ideas and if it looks sensible, I’ll put up a topic for it.

[tiwane]

@charlie

do you want to write up a feature request for the reputation system?

I don’t think there’s a need for a feature request - basically it’s just really a really complicated thing to do right, at least the way we hope to implement it. So it’s on the radar but just difficult.

[charlie]

Yeah that makes sense if it is already on the radar and just hard to figure out! I agree! I suppose i am just trying to figure out how to help nudge the conversations because so many other threads diverge into that topic. I guess a general topic on the reputation system could work. And maybe i am being overzealous in doing that sort of nudging anyway.

[kiwifergus]

We need more zealous participants, Charlie! Keep nudging when you see the need, we can nudge back if we don’t!

[jdmore]

@reuvenm

Reuvenm disagrees this is X

But I think it should be, if we wanted to be completely rigorous about it:

Reuvenm disagrees this should be identified as X

Because that’s what we’re actually doing I feel - coming to a consensus or majority view on what the identification should be.

Not sure I’m quite on board with the logic here. When I add an ID to an observation, I’m saying what I believe that organism to be (to the lowest rank the evidence allows me). And, if it happens to disagree with another ID already present, I am additionally saying either “it’s not that” or “I can’t tell if it’s that” – my option.

I leave it to iNaturalist to decide what the observation “should be identified as” based on the collection of IDs present. All I am contributing is my opinion that “this is X.” (And am, of course, open to persuasion via evidence that I am wrong, and it’s actually Y.)

A distinction without a difference? Maybe. I guess the point being that my first reason to ID is not to push the consensus in a particular direction. It might or might not have that effect. It’s just to add my opinion as to what the organism is (often with a comment as to the evidence I am seeing).

[sgene]

Did I violate some norm here by agreeing to the species? I did not explicitly disagree with the “variety” ID. My ID made the observation Research Grade. https://www.inaturalist.org/observations/7151570#activity_identification_48256652

Should I not agree to species ID in these situations?