iNaturalist's earthworm problem and how to fix it

Yes, I was definitely planning on having region-specific sections; that’s already how I make my own IDs (ie. Alaska, only options are Lumbricidae and introduced Chilean acanthodrilid; India, anything is possible). I’ll probably make some sort of directory for the various journals in a section of the general earthworm photography guide that I link in my IDs. Thanks for the suggestions!

Is the forum a good place to “advertise” earthworm ID guides? I get some but not enough interactions with my guides through commenting on obseravtions.

6 Likes

I’ll start off by saying sorry for not focusing much on worms for the last 6-7months! You were a great help with identifying some of mine and pushed me towards finding a few new species of them. I expect to mostly find the same boring species around my town, but if I’m ever in the spot to take some half decent photos, I’ll see what I can do.

I really do think you are doing important work. I’ve been focusing almost solely on Collembola ID this year, and in conjunction with a few other key observers/identifiers, we have added a few dozen species to the CV as well as a fair number of iNat firsts. The CV is finally getting “good” (in NA at least) at recognizing springtail species, as many of them are turning out to be quite unique in habitus… and I think that’s one of the major problems with worms.

The main issues are threefold:

  1. To the average observer, worms are pretty gross. Even if they COULD get better pictures, they simply don’t want to, assuming they take any in the first place.
  2. Worms all look extraordinarily similar to the untrained eye. Even if people know that there are X number of species in their area, seeing a potentially new worm isn’t as obvious and exciting as seeing a shiny new beetle, and so they are overlooked. This is something that I’m guilty of. I’ve been poking around the lower Appalachian mountains a lot recently, but I haven’t put much thought into what I’m assuming are the same worms I see everywhere. Though I am still on the lookout for some of the especially strange species that are rarely seen in the area.
  3. Worms require a lot more work on the part of the observer to get an accurate ID on. If someone only has so much time to make observations, they may be inclined (even subconsciously) to focus on more “reliable” taxa, because they want to feel like they’re contributing more to the platform (achieving research grade observations). So they spend 1 minute uploading a lady bug and the flower it’s on instead of 5 minutes trying to get the right angle on a worm that might not be identifiable anyway.

Unfortunately, I don’t think these are issues that are easy to fix in the short term.

For 1, well, there will always be some people who enjoy worms. Theoretically the number of them that wind up on iNat should grow over time. So, in time, this should theoretically get solved - ideally there will be a “critical mass” of worm identifiers and eventually the CV will improve to the point that they can handle incoming observations without too much fuss.

For 2, you are doing the best thing you can here. The easier worms are to distinguish (via associating microscopy with good habitus shots and restricting these species to their actual ranges rather than “Common Earthworm” Hell) the more likely it is that people will learn some of the more general traits of different worms and want to photograph new ones. But again, this is gated by time.

For 3, the fix is again going to be time. You need to wait until the subset of worm-posters who are capable of taking photos of distinguishing features grows to the point that there are enough good photos to train the CV on. This is sort of happening on its own as camera technology in phones gets better. Give it a decade, maybe there will be enough power in the average phone camera to pick up these minute features without the observer even needing to try. And hey, maybe enough users will pick up a cheap stereo microscope or macro photography setup in the meantime.

To summarize, this is something that can only be solved with time and effort, and you are already putting a lot of that in. Eventually, enough people are going to get interested (whether that’s a few people who fall in love with it or a few dozen who decide they can spend twenty minutes a week with a worm) and the changes will start happening that much faster.

8 Likes

More people have actually expressed concern for the earthworms (stress from handling) than earthworms being gross to touch but I’m sure both are factors! One does get rather filthy catching, cleaning, and handling; plus several groups can spray or ooze musty-smelling fluids when harassed and nearly all will defecate defensively too. Unfortunately handling is pretty much necessary for getting worm ID, even if worms are already on the surface during rain, but it is probably more agreeable for the worms than being put in ethanol for specimens like most keys assume the users will do…

Yeah, this is a big one. Not much you can do about that other than observe everything!

Perhaps; phone camera users are at a severe disadvantage here. A good digital camera with a macro lens or decent zoom can capture all the needed things quite quickly due to the shutter speed.

6 Likes

Vote on this feature request which would allow for taxa such as earthworms to be opted out of the CV and thus reducing the AI-slop identifications on iNat: https://forum.inaturalist.org/t/force-computer-vision-to-back-off-on-the-specificity-of-suggested-ids-in-regions-with-cryptic-or-hard-to-identify-species/70021

5 Likes

Done! I also voted in both other threads suggesting similar things sometime last year or earlier this year.

2 Likes

The forum attracts people who are very invested in iNat and/or have the spare time for keeping up with topics here. So you are more likely to find people who do a lot of observing or identifying here than on random observations, since most random observations are probably from newer more casual users. However it’s still the case that most people who do a lot of observing and identifying aren’t very active on the forum.

My experience with helping grow syphid fly ID support (that effort was already mentioned a couple times above) was that there will be a small handful of observers/identifiers out there somewhere on iNat who are a bit curious about the taxon and have already have put out some tentative curious observations or IDs, but don’t have the confidence with their current knowledge to go further. A well-placed helpful comment (even if it’s one you copy-paste to every ID) might be enough to excite their curiosity again, and you just need to find them on random observations repeatedly in a way that leads them to go to your resources and feel comfortable asking for advice (see also this thread). It sounds a bit weird to describe it strategically like this but I’ve experienced it from both sides. Generally we’re all here because we’re curious and want to learn more about biodiversity, it’s just a numbers game to find the people who are particularly curious about the same taxa.

11 Likes

I would like to add my thank you for what you have done. It might not be much but here in South Australia we have a local iNaturalist facebook group and have added your photographing earthworms post to our files, so hopefully it will others.

4 Likes

I have no useful suggestions but I did, as a thank you, spend some quality time on your excellent observations (though only managed 2). I wish there were a way for selfless IDers to be featured more, rather than just occasionally. It might better promote the endeavor to those on the fence about getting involved. iNaturalist desperately needs more.

5 Likes

I have provided a link to the journal post now on all taxon pages of Lumbricidae and two other earthworm families - I hope this will guide more people towards this great resource.

Also, I’d wish more journal posts providing ID help would be linked on the taxon pages in general, this can be even narrowed down to certain countries or areas, in case these are not applicable worldwide

9 Likes

Agreed. I think a lot of people brought that up during the whole AI debacle a while ago, that user-written wiki posts or short ID guides would be a valuable addition to iNaturalist. Certainly would encourage a little more caution for tricky groups of organisms.

2 Likes

I’d settle for an observation tick box that indicated “Information about how the ID was made is contained in Notes and/or comments" Then people could navigate to those observations and use the information in making their own ID. Ideally, some observations might have “golden gems” (really amazing ID help), so some way to rank ID tips would be fantastic. I have the same desire for BugGuide … experts clearly have taken the time to make thousands of IDs but they rarely say why, and there’s no way to know which observation might have some of their thinking/tricks.

3 Likes

Topic is closed, but you might want to have a read here:
https://forum.inaturalist.org/t/proposal-for-a-new-filter-option-the-id-tips-button/61624

1 Like

Related, since underside pictures are mentioned a few posts up - how would I take a picture of the underside? I assume when I see an earth worm, it’s upper side is visible. But if I flip it over, how would I know which side it landed on, or whether it already put itself back to upright position?

2 Likes

This is the paragraph I copy-paste in comments of observations:

Most lumbricids and many other worms have a saddle-shaped clitellum which is smooth and swollen on top but the ventral side is still visibly segmented. Worms with an annular clitellum smooth all around will likely have some sort of genital structures on the underside which are distinctive. The number of segments plus various reproductive structures present there are important characters for identification. Many pigmented earthworms will have an underside that is paler than their dorsum. If you look at a few of my photos showing the dorsal and ventral sides of a worm I think it should be more obvious than you are imagining. They also tend not to want to be flipped over, and will actively try to right themselves if flipped; it can take some handling to get a worm to give you a photo that is useful.

https://www.inaturalist.org/observations/279371127

https://www.inaturalist.org/observations/206661431

https://www.inaturalist.org/observations/235537857

3 Likes

Every taxon page should get a section where it explains how to do the best pictures to ID that taxon.
There are often some important details to ID some life form and distinguish them from similar forms of life.

So on a taxon site there should always be such important information.
Even generated by a AI it would be better than nothing.

The Earthworm picturing tutorial here…
I dont understand why it is not on the taxon page of every worm where it matters?
It would be a favor to every one, just add a section “Picture tips” or “How to picture to ID” or some thing like that.

Then there you can write general stuff like
”Please take pictures from the lower side of the fungi”
”Please turn the leaf because there is an important detail that loos otherwise on this and that plant”
”Please turn all the worms on the back”
”Please take a picture that includes the host because it matters”

Just a whole page as needed per taxon.

Ask some AI
”Generate a text about organism X, write one section about how to distinguish this organism from similar looking organisms, write a section about the vegetative properties, write a section about the generative properties, make a list of all important details which help to identify the organism precisely”.

Then generate this for every taxon.
And if you want it better then write such a cool tutorial as we have here some where in a journal entry.

I think such information should be on every taxon page.
i understand that this is a impossible task in 1900, but at the end of 2025 an AI can do it in seconds
actually when i look at https://www.inaturalist.org/taxa/69758-Lumbricidae i just see an non english wiki.
If i want to know some thing about worms i can go and get my informations on wiki or the same on iNat.
But iNat wants good pictures to ID the worms, so from the perspective of iNat the wiki is not interesting, more interesting would be the worm picture tutorial.
People on iNat must first get informed about how to do good pictures and then how to ID stuff, and if they want to read a fairytale about some species they can open wiki them self.

3 Likes

There is a lot of prior discussion about this if you search on the forum…and a project in the works to surface this information using machine learning.

But better to act on the stuff we can control ourselves in the meantime.
For now we can add stuff to the wikipedia page.
Or another nice workaround is to add annotated photos :

https://www.inaturalist.org/taxa/321894-Lonchoptera-lutea

2 Likes

But it is! From family through every taxon down to (sub)species level, there is the link right there on the information page. Not visible in the apps, but readily accessible on the web version

9 Likes

A centralized resource is a far better use of identifiers’ limited time than expecting them to put that information on individual IDs that are probably going to get lost in the neverending stream of observations.

And speaking as an identifier, a huge portion of observers never respond to IDs, so taking some of your limited identifying time to write up a comment about how you IDed it, on the aggregate, feels like talking to a wall.

3 Likes

For the time being, I don’t find it too bad if you use keyboard shortcuts. Especially if it’s a controversial ID I need to apply to tons of observations like pushing up a common CV error species and I expect some pushback. Usually I’ll type out the first few letters of the taxon name for the ID which leaves my clipboard free to Ctrl+V the ID comment text in. Once the text is written I can just copy and paste from previous IDs.

However if I had different texts for multiple taxa then a text expander browser extension would be the way to go. This custom iNat extension also incorporates ID comments.

4 Likes

Oh, this might be helpful! Thanks for the tip. I’ve been copypasting from a word doc which works well but maybe the text expander will be even faster.

3 Likes