"Trying to define the undefinable": is iNat too focused on species?

There is a lot of discussion on the need to ID observations, with the ultimate goal of getting a nice green Research Grade stamp on as many as possible. But how many of the “species” we ID are real divisions in nature?

A recent article in the New Yorker includes a good overview of why this may be problematic:

“You Name It: Carl Linnaeus and the effort to label all of life” by Kathryn Schulz, August 21, 2023

Extended quote from article:
"What Linnaeus sought to do was organize nature according to its fundamental, intrinsic divisions–to carve it at the joints, in Plato’s famous formulation. But what he actually did, for the most part, was impose artificial categories on the natural world for the convenience of scientists.

This is not a retroactive assessment; Linnaeus himself knew full well the limitations of his classification method. To achieve a system completely in accordance with nature was, he wrote, “the first and last wish of botanists.” But the more closely you looked at her bounty the more difficult that prospect became–so, in the meantime, “artificial systems are absolutely necessary.”

In philosophy, this tension between intrinsic and imposed categories takes the form of a debate between nominalism and realism. Realists believe that nature is full of real and discrete categories, from 'amphibian" to “zinc,” and that the job of the scientist is to discern them accurately. Nominalists believe that nature lacks clearly defined categories, and that we simply impose those distinctions upon it–creating, as it were, the illusion of joints where none really exist.

This is not just the position of post-truth relativists. “I look at the term ‘species’ as one arbitrarily given, for the sake of convenience, to a set of individuals closely resembling each other”: that is Charles Darwin, in the second chapter of “On the Origin of Species.” That book, of course, trumpeted to the world a very large problem with the entire notion of a species. According to evolutionary theory, species are constantly changing–emerging, diverging, going extinct.

The very concept of a species is in radical flux, too, with more than twenty competing definitions in circulation. Choosing a definition is not just a matter of what goes in the dictionary under “species”; which one you use will determine how you divide up nature, such that a group of creatures that would be regarded as a species by one standard might not merit the label by another.

All this confusion comes, as Darwin wrote, “from trying to define the undefinable.” Yet committed realists continue to promulgate more and more definitions, in the belief that one of them will map perfectly onto some intrinsic and stable feature of nature. Darwin called that idea “laughable,” a word that captures the impossibility but not the gravity of arbitrarily imposing categories on living beings.


This is a fairly common argument:

  • we’re taught species are real, consistent, discrete things (usually ‘actually or potentially interbreeding popoulations distinct from other such groups’)
  • but there are (many!) examples of ‘species’ that don’t fit this definition
  • therefore the species is an artificial construct, and isn’t real

This is a real conundrum for non-taxonomists, because it seems to undermine a lot of our biological understanding. The truth, as taxonomists well know, is that nature is complicated and diverse, not just in the number of species, but in the number of ways of being a species. A definition that makes sense for apomictic dandelions won’t be useful for mountain gorillas. That doesn’t mean that mountain gorillas aren’t a real species. (it may raise questions about dandelions, but let’s leave that alone!).

The truth is, for the general public, ‘species’ is a useful simplification of what in reality is (in some cases) a very complex reality. The people this really matters to (taxonomists, evolutionary biologists, etc) will have to come to terms with the ugly details as they apply to the organisms they study. For the rest of us, it’s a good enough concept to get us started in our explorations.


The design of iNat seems to leave room for the ambiguities that arise in the real world. Even though the goal appears to be to obtain a species ID for an observation, there are a lot of places in each recorded observation where richer information may be represented that helps to give room for not only uncertainty in the ID, but also any underlying ambiguities with the definition of any particular species.

For example, if we have two defined species that intergrade, observations of organisms that are intermediate can have commentary attached to their IDs where someone could annotate the ID with caveats such as “Could also be considered species Y as those two species intergrade in this area.”

The problem posed by the original poster here would be more of an issue in the design of a ID app where the computer and the computer alone was the only source of an ID, and the ID was simply a name with no context. Then, I agree, you would have problems with gray areas or unclear species boundaries. The computer’s response might be misleading if it is not programmed very carefully. If it only gives one or the other species with no other qualifications, then it is a case in point of relying too much on a conceptual construct (namely “species”) and be guilty of reducing the complexity of nature in a way that diminishes it.


In order to talk about things, even to think about things, humans need words that are neat and non-overlapping. The variation that we try to name exists in nature. It’s real. But it’s not neat. We taxonomists do the best we can to parse nature into real groups to which we apply names, but reality often doesn’t cooperate.

Sometimes the confusion is mostly semantic. For example, Island, California, Woodhouse’s, and Florida subspecies of Scrub-Jay all differ from each other despite an overall similarity. They don’t interbreed because their ranges are allopatric (non-overlapping). They’ve been called subspecies of one species and now they’re called separate species. What is right? What is true? The variation is real, but the names and rank are arbitrary. Keep in mind that the variation is real!

At least with Scrub-Jays, the reality is kind of neat. In some groups it’s not. For example, dandelions . . . let’s not go there. For larkspurs of the Pacific Northwest, there are three closely related groups. In some places, they all grow together and stay separate. In other places, species A and B interbreed but species C does not. In others, A stays separate while B and C interbreed. Etc. Where do we humans choose to apply our words? Sigh. These difficulties drive taxonomists crazy, sometimes literally.

I think we do best to realize that the question “Do species exist?” is the wrong question. Variation exists and a lot of it is important – worth having words to talk about. The question is, how do we apply our necessarily arbitrary words / categories to the variation?


Also note that iNaturalist has some species complexes – groups of similar and closely related species that can be treated as a unit. Both complexes and genera can reach Research Grade, so we don’t always need to agonize over species boundaries.


As with any attempt to categorise messy data, there will always be a degree of simplification in applying a taxonomy, but I think your suggestion that many species are not “real divisions in nature” is taking the idea too far. Sure, there are quibbles around whether something should be a species or subspecies, and many imperfections in the system, but I would argue the model is impressively close to reflecting reality in a way that is useful in 99% of applications.

To me, your post has the essence of the musings of a philosopher who has spent too long staring at a rainbow and, in frustration at not being able to find the exact boundaries between the colours, wonders if we ought to stop using colour words entirely. For all their imprecision in capturing exact reality, colour words serve us well, and we have additional vocabulary we can use to modify them when greater precision is necessary.


I don’t think iNat is too focused on species, it’s kind of the whole framework iNat is built on (like if you just want pretty photos of trees, without understanding what sort of tree, you’d go to Instagram with your photo instead). I think the concept of species is complicated, and the way taxonomists are taking it now is really problematic which makes iNat harder to use in the sense of identifying species. But, I don’t see how de emphasizing identification will help iNat. Maybe we need to be clearer that species are a cultural concept used to classify organisms but aren’t perfect and we need to recognize we need species units that are acutally useful, not following some scientific ideal only 0.01% of the population can actually use. If iNat stepped away from that stuff, the species concept would be more accessible to its intended audience. But to do that we’d have to break some social norms from within academia that many don’t feel willing to break, so we continue to struggle


I think this overstates the case a bit; the critique seem to apply more to common names than to binomial nomenclature.

Look, that iNat is teaching millions of people to use binomial nomenclature instead of common names is an incredible accomplishment!


In our day, because so much of what we think of as progress is science-driven, it creates the thought pattern that only science is right. This in turn reinforces the unwillingness of academics to break from academic ways of thinking.

That is not what I get out of this at all. What I get out of this is very much a critique of binomial nomenclature. A Linnaean binomial seeks to circumscribe an entity exactly, and when a subset of that entity doesn’t quite fit the tight circumscription, we feel that we must propose another one to circumscribe that subset separately. Common names do not have this difficulty.


Well, and taxonomy isn’t even ‘science’ in the strictest sense, it’s one interpretation of how to sort species. There’s no experiment that proves splitting is better than lumping, or for that matter classification by ancestry is better than by form or some other factor, it’s down to opinions.

i went back to using common names on iNat because the scientific names change so much i can’t use them any more. Really iNat is teaching people that scientific names aren’t consistent or constant, rather they change constantly and then change back for no obvious reason,


Love the above quote!


The real problem comes from the fact that nature is inherently “untidy”, whereas the human mind (generally) hates messiness. Taxonomy represents a never-ending attempt to tie irredeemably loose ends and to divide a continuum into portions able to fit into our system of finite boxes. Darwin is right, it is “laughable”. So why do we go on doing it?
Simply because it is the only way we have of communicating. At the end of the day, scientific names are simply linguistic shorthand for describing a set of characteristics, traits, features (call them what you will) representing the concept of that particular taxon and its possible near or distant relationship with other taxon. It is a label on a box and we need it to quickly and simply communicate the contents of that box to someone who may be even half a planet away. The contents, even the size, of that box may change over time, but without that label, however approximate, we would have no way of communicating. And humans need to communicate.

iNaturalist is of course all about IDing observations and the only way to do this is to apply to that observation the currently accepted label (taxon name) which best describes that particular life form. The Research Grade stamp just means that more than one person agrees on that label and yes, it is gratifying to be able to file that organism away in our nice tidy scheme of things (my whole photographic archive is taxonomy based, so yes, I feel good when I get an ID and I can put an image away in its “rightful” place). Research Grade is, however, not just about personal observer gratification. It’s also about removing that observation from the ever-growing pile of observations needing an ID and without that, iNaturalist and its relatively few identifiers would risk suffocating in a turbid “Needs ID” soup, without light at the end of the tunnel (to mix my metaphors).
BUT it’s important to never lose sight of the fact that the “boxes” and the “labels” are our invention, our way of trying to make sense of nature’s complexity and to communicate it via a (more or less) universally accepted, even if approximate, shorthand. All is imperfect, all is in a state of flux, but it’s the best we have.


Yes, it’s true that scientific names are not engraved in stone and may change. But generally speaking they change consistently across the entire planet. Common names on the other hand can be insidious, unless you limit your activities to a circumscribed local area. Apart from the fact that they are obviously limited to a particular language, even within that language, the same name may be used for a range of species even in different families, while different names may be applied to the same species in even neighbouring regions of the same country.


That isn’t even remotely close to my own experience, and i work with vascular plants mostly, which are less complex taxonomically than most taxa. The system is completely in disarray and even professional organizations all use different names. No one can keep up, and increasingly field ecologists can’t even identify the split species anyway. Even if they can the splitting damages the integrity of older data unless subgroups are created very carefully.

It’s true common names are a mess too! There aren’t good options for applied field ecology any more. Which is a real bummer. It couldn’t come at a worse time.

and if you are a linguist - the word blue - opens up a fresh field of discussion

across many languages

For the linguist - it’s a dandelion! What is your problem?

1 Like

Personally, I find arguments about the “continuum” of biotic diversity terribly oversimplified. They obfuscate the reality that there are innumerable recognizable and categorizable discontinuities in genomes, genetic diversity, phenotypic diversity, etc. Yes, there is variance in the genomes of things we’ve boxed up and labeled as “species” and yes, there is hybridization, introgression, etc. But there are real evolutionary explanations why the discontinuities arrise over time and space (as well as why they are not discrete under various circumstances). These discontinuities–however messy in the larger picture–give us a real framework for what we (the scientific community) have labeled “species”. And yes, we have a myriad of definitions of “species” but for the most part they all hang their taxonomic hats on the structure that is recognizable in those genotypic and phenotypic discontinuities.


Colour words are slippery characters over time. Especially blue. One of the most astonishing things about the English word blue is that, etymologically, it shares its root with the words yellow, gold and blonde – despite the fact that none of them seem to sound alike anymore, all ultimately derive from the Proto-Indo-European root bhel/ghel, meaning ‘shining’.


Too often, humans seem unable to hold two somewhat conflicting ideas in their minds without insisting on only one or the other. Diversity in nature, at any given time (e.g. now) is NOT continuous. Very definitely not. Humans are not chimps and neither species is an elephant, for example. The odd African plant Welwitschia is not much like anything else on the planet. On the other hand, many of the current end products of evolution, which we want to call species, may not be distinct now, or may be similar in groups (e.g. Scrub-Jays) where we could reasonably rank the smaller groups or the larger one (Scrub-Jay) as species. The variation is real and very lumpy, more like a watery stew than a rainbow.


Now that is a quote worth borrowing if ever there was!


Reminds me of that big debate in the media a few years ago as to whether that dress was blue or gold, which nicely illustrated how different individual humans perceive the world visually. Not a perfect analogy, but taxonomists are equally variable when it comes to perception of differences between groups of organisms.