I disagree. I have found that photographing an organism such as a butterfly and then using the CV to ID it, on numerous occasions over time, does in fact work quite well to teach me what it is and how to recognize it visually. It is like using flash cards with small children. I often don’t find that I need to “start reading books and learning from other people” in order to learn to identify that species well. And when I do want to know more about a species, I often just go to the Taxon page for that species here on iNat and peruse the info and links on that page.
12 Likes
I could learn to recognise butterflies just as well from a pictorial pocket guide - but that wouldn’t make me an expert. Such resources do not include every relevant detail about every species that you might encounter. An expert is someone who has learnt the truth of that the hard way, and decided to do something about it.
While I will not diminish the value of detailed taxonomic guides, necessitating that level of investigation is a high bar to impose on individuals making observations in iNat. I think there’s the possibility of becoming a “regional expert” on some species through repeat photographic-documented observations. In fact, I think repeated exposure to the same organism, even without the associated photography, is exactly how people become experienced, subsequently leading to expertise.
7 Likes
I suppose it depends on what you mean by “expert.” For iNaturalist purposes, it seems to me that being able to identify a species consistently, with an awareness of potential pitfalls like range or similar species, makes one an “expert.” I’ve learned some species (in certain places) by seeing them repeatedly on other people’s observations. Some by reading books. Some from other people. Some by researching the topic through many articles. We can learn some species from the CV.
8 Likes
Why hard way? If I know there’s only one Leiobunum species locally I can id it all the way I want to and be a local expert, you don’t need to be a professor of Opiliones to id them and be called expert.
4 Likes
This sounds rather defeatist. If the more detailed references are too difficult for everyone to understand, we should set ourselves the task of making them more comprehensible to all, rather than claiming the bar is forever set too high.
Assuming it’s the same organism is just begging the question. Visually identical species are an ever-present possiblity. People that aren’t aware of that are doomed to repeat the same mistakes over and over again. If the only tool they have is a hammer, everything looks like a nail.
1 Like
It depends what you mean by “learn”. The CV is a very useful search tool. I often use it for moths, because it’s a quick and convenient way to narrow down the possibilities. But it only ever suggests the “what”, and never the “how”. To gain expertise (know-how), I have to consult external resources that will suggest ways for me to confirm for myself what can or cannot be reliably identified, and why. The CV might help people to develop better species recognition (know-what) - but I would say that is not the same thing as developing expertise.
You can never guarantee that things will always stay the same. New species can arrive, or ones thought to be locally extinct can be re-discovered. The effects of climate change have greatly increased the likelihood of such things in recent years.
Up until 2015, everyone thought the common and widespread species Dicranopalpus ramosus was easy to ID because of its conspicuously forked pedipalps. Then a couple of authorities on Opiliones showed that a previous synonym of that species - Dicranopalpus caudatus - was in fact a valid species. The two species can occur together in some parts of their combined ranges, and microscopic examination is required to reliably separate them.
So it only proves the point? If nobody knows there’re hidden species then everyone will id wrong, no matter how big of an expert they’re? And then everyone will be able to id complex.
2 Likes
I can see what you’re saying, and yet this seems a weak objection. Having decent pattern recognition skills, I provide a tentative “how.” I see other examples of the same species. If I’m using CV, it gives me another “what,” evidence that strengthens or weakens my understanding of “how.” Repeatedly seeing CV suggestions and the reactions of other users to the identification, I can learn from the CV to be very, very good at identifying some things.
I’ve done this kind of thing by identifying plants for certain people doing many, many surveys in similar habitats. Eventually I stop skipping some species I didn’t know and start IDing them. I learned Rubus pedatus and Fomitopsis mouncei, among others, this way, seeing them repeatedly and figuring out what those observations had in common. (Unfortunately I didn’t learn how to spell that last one.)
Right now, in the herbarium, I’m sorting out specimens of a widespread, variable species, seeing the “what” and trying to figure out the subtle “how” well enough to communicate it to others and to decide if the taxon should be split.
By the way, I am by anybody’s definition an expert on identification of certain taxa. I’m reasonably good at many others. I’m a real dunce at others, including some I thought I understood, as repeated explanations from other iNaturalist identifiers have taught me. So it goes.
8 Likes
Another aspect of practicing by looking at computer vision suggestions is that in certain taxa I think it is all but certain that the computer vision is learning features that the writers of field guides and keys either did not know about or did not include. This may be particularly common in plants, if guides are built off herbarium specimens which may look different compared to when they were alive in the field. Hopefully some day it will be possible for the computer vision to open up the black box and tell us what the features it is using are; the technology is still pretty new.
2 Likes
The computer vision system uses licensed, proprietary computer code. Unless the owners of that code decide to make it open source, it can never be made public (legally).
I didn’t mean the current computer vision model specifically. I just meant that the technology has only been a part of the site for 4 years and has improved so much in just the last 5-10 years, so it seems well within the realm of possibility that some day a model that can spit out its insights in a nice comprehensible format will exist and be trained on public inat data, even if it isn’t the current one or even one trained by the site staff.
3 Likes
Not trying to sound defeatist, just realistic. You seem to be proposing that everyone is an expert in every discipline and I just think that is not a reasonable expectation. I’m a taxonomist of marine benthic invertebrates in my day job and we have a whole bookcase of dichotomous keys as well as filing cabinets full of research papers documenting changes to the dichotomous keys. It takes months, if not years, to learn the terminology and glean the distinguishing features between some organisms with other taxonomists guiding you through. And this is on a professional level, and for really only a minute subsection of all organisms. I think expecting this of recreational users is, again, unrealistic.
I think “Visually identical species” are rare and, if they are encountered, I’d think genetic testing would be the only definitive way to distinguish them. Are you suggesting that at the iNaturalist level?
I agree that visually similar organisms can be challenging and that it is important to learn what features distinguish them. Adding comments along with an identification can help observes know what to look for (and hopefully document with a photo) in future observations. Yes, someone will have needed to spend the time doing the initial investigation, but that doesn’t mean that this information can’t be then passed on.
I guess, ideally in iNat, CV would provide something like that, e.g. Lesser Blue Crab, visually similar to the Atlantic Blue Crab but distinguished from it by 6 teeth between the occular lobes rather than 4." Something like that really would be a helpful learning tool in my opinion (and possibly even lead toward expertise). In retrospect that may be what you, @bazwal, were suggesting all along.
I think CV has a long way to go though before it’s at that level. Including reasonable ranges would be a necessity first, but that is another discussion (and thread).
6 Likes
Most insects consist of similar for regular user species, many groups are not ided without specimen to species, sometimes male only, but still cv can learn the grup and propose it, sometimes learn the species and propose it, while not possible to id from single photograph, so it’s not rare at all, but if user is getting seriously to iding, they will learn there’re many species, and thus becoming more knowledgable about the group and in the end learning how to id it, regionally or worldwide, so surely you can be an expert by learing that way.
1 Like
That depends on what taxon we are considering. If it looks like the field guide’s picture of a Harlequin Duck, I’m not going to leave it as Genus Histrionicus. Frankly, I find the lack of confidence of many IDers confusing – leaving something at genus that really doesn’t have any look-alike species.
And sometimes, that creates problems. Over and over again, I have come across Thorybes dorantes bumped back to Family Hesperiidae because one person, lacking confidence, hedged and IDed it as Genus Urbanus – which it was at the time – and then, after the taxon split, someone else IDed it to species. If the original IDer had gone ahead and said Urbanus dorantes, then the taxon split would have moved it to the correct new genus, and it would be at research grade. So, which other species is visually identical to Thorybes dorantes?
1 Like
I think bazwal’s comment about “visually identical species” was not referring to photographs but actual observed organisms. If, upon examination, one species can’t be visually distinguished from another (which I think is rare), I don’t know of another way to distinguish them.
Sure, though from how I read that comment, it was more of “generally identical”, like, there’re some genera looking identical from a single pic you take, you see cv suggestion and yeah, it looks right like it, and you think it’s right, while it can be not. As argument about dna-based species doesn’t support any side of this conversation, if they’re known to science, you will see them from cv eventually (or read when googling one of species).
1 Like
this strikes me as judgmental. We all learn differently. I am a very visual learner and while i can use a dichotomous key, i don’t learn very well that way. I learn by seeing a plant several times in different settings, or sometimes by sketching the plant. For me, observing a species a bunch of times and using the AI and other tools like that will work better than repeatedly keying it out, and will be more accurate too. We are all different, and learn differently, so the more tools we have the better.
10 Likes
You claimed that it’s possible to be an expert by knowing that there’s only one local species. But you didn’t explain how you would get to know this in the first place, nor why you would remain so certain of it. Right now, there are people starting out with the same kind of assumptions who could avoid making wrong IDs if only they were aware of all the relevant technical literature.
Well, to do the original research, you clearly do. I think it’s best to reserve the term “expert” for people who are recognised authorities in some sense. We can’t all be experts, but we can still develop expertise by learning from them.
Something I haven’t seen mentioned by other users is location. Simply put, where you live is going to influence how many observations you can make, simply because of local biodiversity.
Noticed this myself when it came to observations: I live around a rather nice urban-rural gradient in the forested part of the Midwestern United States, and I noticed that certain areas simply had more species in them than others. At the absolute worst you had urban parks where the only species around were the typical urban bird species (rock doves, starlings), fox squirrels, and a number of pollinators (bees, wasps, Chauliognathus, and a couple of butterfly taxa). On the other hand there are some really remote parks that I found I could get >60 different species recorded in a day, easily.
This also applies to abundance of organisms, so the number of observations isn’t just tied to how many species that were seen.
As a result, the ability of someone to produce observations is going to depend on where they live and how easily they can travel from one area to another (i.e., gas cost). Someone in an urban area with little money won’t be able to produce as many observations, unless they have a good local park that supports biodiversity.
This goes beyond the urban-rural gradient. Some areas are just plain less biodiverse than others. I went to visit family in the “corn belt” that extends from the eastern Dakotas to western Ohio and there was absolutely nothing around. Even the local parks seemed stripped of insect biodiversity. The only wildlife I saw were various bird species (mourning doves, hawks, etc.) perched on electrical wires.
Similarly, out in the American West (specifically, east of the Cascades and north of Arizona/New Mexico) there are just plain less species around than in other parts of the United States, largely because it’s more arid (but still very cold in the winter) and with regards to freshwater species few have swum that far upstream. This applies to both species diversity AND number of organism seen. If you live somewhere like Wyoming, it will be very difficult to get a high number of observations simply because there aren’t a lot of species around (even Yellowstone is fairly depauperate compared to, say, the temperate rainforests of the Northwest or the swamps of the Southeast), and you can’t travel to a more biodiverse area simply because it takes hours to drive anywhere.
4 Likes