Why specialists should do identifications etc

@spiphany on principle, your points are all correct. However, reality is this: many insects require dissection of genitalia, and the subtleties of external morphology are too great to easily communicate. Yes, there may indeed be descriptions another person could refer to but typically those are not yet available online and Wikipages are not yet written. Also, I read earlier this year, that it takes 100 observations with a minimum of 50 RG status, for the CV to understand that the species exists. In my group, the Donaciinae beetles, rather a precious few have attained that lofty position, and so the CV still does not render them available as suggestions when trying to upload obs or make identifications… I don’t recommend recruiting just ayone to confirm a specialist’s IDs - that person needs to have some expertise. In my case, someone who actually knows beetles rather well, knows of my publication, and can be confident that if I say it’s Donacia hypoleuca then it is in fact that and he can confidently Agree.

3 Likes

A mechanism exists, which is to explain difficult IDs in comments, follow your IDs (turn off notifications for agreements), and if necessary tag in other people if subsequent incorrect IDs appear. Adding information to your profile will also help everyone know you are a specialist. Even for neglected groups of beetles, there are enough skilled identifiers on iNat that incorrect IDs by non-specialists can be overturned relatively easily, once you get to know the community.

It’s rare for non-specialists to add wrong, disagreeing IDs. Much more common is when they add initial incorrect IDs that get confirmed by the observer, taking observations out of the Needs ID pool. For this reason it’s useful for specialists to review Research Grade observations as well (which is incidentally a more manageable task in obscure groups with fewer observations).

5 Likes

Yeah, it’s unsustainable for one person to keep up with all the observations of any taxon, it takes some effort but it helps a lot if they try to share their knowledge if they don’t want that.

When I started identifying hover flies in 2017, it was possible for me to keep up with all the observation of one genus in North America, but now I’m a lot busier and there are wayyy more observations being posted than there were then. I and others realized at the time that we needed to spread the knowledge so that other people could get involved in identifying, and there’s some documentation of that process here: Identifier Profile: Fly Identifiers

A couple related topics here: Identifier Mentoring / Research Collaboration: Method for supporting non-experts to label in ‘unpopular’ taxa

5 Likes

I think ingolfaskevold’s point was more that if you are a/the expert on a taxon on iNaturalist and are able to make or confirm IDs based on your detailed knowledge, you may have to actively recruit other experts who can check and confirm your IDs because there are few other people making specific IDs of those organisms.

To give a very trivial example, I’m the top identifier of silver-spotted skipper caterpillars on iNaturalist. This is a common species that is pretty easy to ID (at least in the US and Canada) and there are actually at least 20 other identifiers who regularly ID the caterpillars. But my 2 observations of this species are still at Needs ID because they were made outside the habitual geographic ranges of those identifiers. Since the species is common and widespread, I’m not particularly worried about getting these to Research Grade, but if I wanted to, I probably would need to explicitly ask one or more of those other identifiers to take a look at the observations.

If/when I get around to posting even a selection of my reared microlepidoptera caterpillars from my research in Costa Rica, I’ll definitely have to actively collaborate with regional and/or taxon specialists to confirm IDs

5 Likes

I hard disagree with this. iNat’s guidelines are very clear that identifiers should only agree to their own level of expertise, not based on their perception of any other user’s expertise. I routinely find observations that have been misidentified by users following this principle (including ones where identifiers have agreed with my ID, because I am an “expert” in a very narrow field…who also occasionally makes mistakes). I’d rather have a smaller dataset of RG observations that I had higher confidence in, than a larger set of RG observations with a larger proportion of wrong IDs. If a taxon specialist of an obscure taxon wants records from iNat, they can always just download them directly and based their data on their own or any other user’s IDs - no need to push things to RG without the requisite level of support.

9 Likes

Or he could check your publication and assess your ID himself, if he has the skill to interpret identification literature on beetles, no? i.e., not because you say so and he trusts you, but because he thinks the ID is correct.

Edit: I ID bees. I am well aware of the challenges and limitations of photo ID and the extreme shortage of experts. I think it is reasonable and even desirable for specialists to tag other IDers with relevant expertise to look at an observation; I see this sort of collaboration regularly. It often leads to discussion about the ID and how it was determined. Sometimes it results in the other person agreeing with the ID even if it is a taxon they weren’t previously familiar with – and sometimes it does not, because the other person decides that they are not currently confident confirming the ID. But knowledge has been exchanged and the other person has learned something that they may be able to use in the future.

What you are advocating, on the other hand, does not sound like collaboration.

4 Likes

This is not the case, according to the current explanation in the iNat help:

Which taxa are included in the computer vision suggestions?

This has changed over time and may change before this FAQ is updated again, as we are continually working on improving the training process. But basically, here’s what’s needed for a species to be included in the Computer Vision model:

  • There must be a [sic] least 100 photos of the species and 60 observations of the species, and we don’t choose more than 5 photos from an observation to train the model. Observations do not need to be Research Grade in order to be used in training, but observations with a matching Community ID will be prioritized.

  • Some photos that are not included in the training phase are used to test and validate the model. These must have a Community ID.

Because of this, not every species with at least 100 photos and 60 observations will meet the requirements to be included in a training run. It’s dependent on how many photos there are per observation, and whether the randomly chosen group of observations meets the requirements.

If no species within a broader taxon like genus or family meets the requirements, we may train the model on that genus or family, based on those photos.

(emphasis in original)

4 Likes

Thanks for soliciting feedback and for giving this presentation!

FWIW I believe Dr. Kavanaugh has taken at least one, maybe more, young beetle enthusiasts on iNat under his wing/elytra and helped them learn Carabid ID.

Thanks for pointing out the typo, it’s been fixed. If you find an issue with a help doc and have a moment, there’s a feedback option at the bottom of the page.

4 Likes

This could go under “Why specialists should do identifications outside their home ranges.” If you know that a given taxon’s range is wider than your area, there is no inherent reason why you can’t broaden your geographic search for that taxon.

It’s risky. If I know a genus for Canada and the US, and I try expanding south into Mexico and beyond, I’m probably either going to be making a lot of mistakes or be overwhelmed by the increased diversity and decreased literature coverage. That’s a dramatic example but even expanding out of your home state/province involves a responsibility of looking up what other species options you need to learn. But yeah, sometimes a species is distinctive everywhere and it would be nice if people who know that, identified it everywhere.

3 Likes

I think most of these folks are doing IDs of distinctive caterpillars in their or neighboring states and/or have other lep families that they specialize in more. AFAIK, I’m the only person who really “specializes” on North American skipper caterpillars as a taxon. (In quotes because a lot of the caterpillars aren’t super distinctive or are lacking really good references. Which is probably part of the reason why I’m the only one.)

Ascertaining that there are no lookalikes in a region where there are different environmental conditions/greater biodiversity than the area one knows often requires finding checklists or guides for that region.

At least in Europe, there is a fairly high probability that this information is only available in the language(s) of the country or countries of that region. Now, I am fairly fit at doing internet research, and if I find resources I can generally figure out the gist of the content, using an automatic translator if necessary – however, one has to find the resources in the first place, and it is often not easy to figure out what keywords will provide meaningful results if one does not know the language at all.

So yes, in theory, knowledge about a particular organism can be applied to other regions. But in practice it is not always simple or feasible.

4 Likes

Actually, I have found that applying the name I think it is to an iNaturalist observation, when I shouldn’t have done it because there is a look-alike in a new area, gets me educated pretty quickly. Other identifiers correct the ID and some even explain why. I should have avoided making this kind of mistake, of course, but at least I don’t make it for very long.

2 Likes

Ultimately that obs has reached a finer ID, thanks to you. Just needed one more step notaproblem.

1 Like

Or the observer might agree and the observation might be RG for months or years until someone familiar with the local taxa manages to look at it. It really depends. In taxa and regions where there is a major shortage of expertise, adding speculative IDs and hoping to be corrected can be a risky strategy.

I mean, I do think that, in the long run, wrong IDs tend to be caught and corrected because observations are continually being looked at and reviewed by people. And inevitably we will all make mistakes now and again, whether because of a human lapse or gaps in one’s knowledge - not realizing that there was some highly localized endemic species, not being aware of a recent introduction in the region, etc. There isn’t a lot we can do about unknown unknowns, except being alert while IDing for things that don’t make sense.

But I think that where there are known unknowns (a region with which one is unfamiliar, with environmental conditions that are different or geographical barriers that would affect the movement of species), it is a good idea to proceed with caution and do some research before IDing and maybe refrain from providing species-level IDs if it seems likely that there could be other possibilities.

2 Likes

I agree (except with the lepidoptera/sawfly larvae confusion). However, I have with all good intentions identified what I thought was a species I knew, only to find out that the observation was a relative I didn’t know about. I’m glad I did get corrected. There may be ones that got to RG uncorrected, and that’s a problem. It’s hard to find the right balance of optimism vs. caution.

2 Likes

As I noted, I don’t think there is any way we can completely prevent mistakes due to not knowing that there is a similar relative, except for being alert to the possibility and doing more checking if something doesn’t seem right. But I would consider a good faith ID (believing that it is that species) a bit different than a speculative one (thinking it might be that species but being aware that one does not know enough to be certain there are no other possibilities).

I should clarify that I think the standards of reasonable certainty are different depending on whether one is adding a species-level ID or a broad ID, simply because the consequences of being wrong are very different.

A broad ID that seems plausible but is out of one’s area of expertise will move the ID forward and make it more likely to be looked at by an expert. Even if it turns out to be wrong, the experts more often than not recognize the common mix-ups (e.g. lepidoptera and sawfly larvae) and can suggest the correct order. There is a small amount of risk here (the observer may agree to your ID and then become inactive and never change it; a wrong ID may cause delay and result in the observation getting missed by experts who mostly look at recent observations), but I think the advantages of adding such an ID far outweigh potential problems.

5 Likes

IDs outside your geographic comfort zone can be tricky whenever there are lookalikes that are not represented in your comfort zone. You need to study the taxon also for the focus region.

I can reliably ID Eupeodes corollae in southern Africa, but not elsewhere with more species in that genus.

1 Like

As a non-expert I would like to reinforce one of the points made here: the more experts there are ID’ing a certain taxon, the more people will post observations of that taxon.
I think it’s an important point to have in stock for if an expert you’re trying to recruit says something like “well, there aren’t enough observations of what I’m studying to make it worthwhile.” Although sometimes there is no good rebuttal to this (the organism is microscopic or very cryptic) oftentimes it would be good to point out that feedback loop. Many taxa are hard to ID if you aren’t an academic, or don’t yet have people focused on them. This discourages observers as it feels like when they post an observation of those organisms they aren’t contributing anything, but if an expert starts ID’ing that taxon people’s impressions will change and they will start posting more observations.

6 Likes

I was doing fieldwork a few months ago and found an Argiope aurantia right on a traverse point (I’m a land surveyor, not a biologist), recognized it as Argiope in the field, took a photo, knocked the web because I had to put my tripod on it, and posted the photo. I’m in North Carolina.

Some time later I found a photo of an Argiope aurantia in Guerrero. I asked in Spanish if they look different in Guerrero than here. I got an answer “yes” in English from someone in Turkey.

4 Likes