A question from a new iNaturalist user. In ordinary life, when we observe an organism, we often make use of geographical and seasonal information in order to identify it: “X is common in this part of the world around this time, so that’s probably what it is” vs. “It can’t be a Y because they don’t occur in these parts now”. My question is to what extent this kind of reasoning is legitimate for iNaturalist IDs?
Here’s my problem. Presumably one of the functions of iNaturalist is to provide scientists with information with regard to the distribution of organisms in space in time. But then using such information to suggest IDs becomes circular. So for example, if the current consensus is that Ys aren’t found in these parts, but the consensus happens to be wrong (as it could quite easily be), then observing a Y here won’t yield a correction to the consensus, because, for geographical and seasonal reasons, it will never be identified as a Y — and so the faulty consensus will remain in place.
I imagine that in many cases, using geographical and seasonal information when offering IDs works fine. But I am worried about the potential for incorrect assumptions about organism distributions to be perpetuated in this way.
that’s a definite issue in ecology in general, but at least with plants it is rather minor or at least inevitable. For instance, when looking at a pine tree in Vermont, where there are maybe 6 pine species, i can’t consider every pine tree on the planet - around 115 species. I just cant, or i could never ID any pine ever. Other genuses are even bigger, and that’s assuming you even know the genus. Pretty much all keys and field guides use geographical information, because it’s almost impossible not to. If you find something really weird, that doesn’t fit the key or field guide, you probably documented something not known from the area and that’s when the international iNat community can help. But I think that’s all you can really do.
Worse is when you identify species only based on location, as with some subspecies. That gets very circular but there is already a thread (or several?) about that.
I do my best to identify based on the features of the organism alone, but when identifying from photographs on iNaturalist, location and season are definitely confidence-boosters (or confidence destroyers). And sometimes, when features aren’t very clear in the photos, location and season can really make or break an ID. Along the lines of, “nothing else (or too much else) that lives here could be making that particular blurred color and shape.”
Keeping in mind, of course, @charlie’s caveat that you could be looking at something new to or previously undocumented in a place.
iNat struggles to learn Cape Town’s fynbos, but is determined to offer a suggestion.
A fuzzy picture of Phylica gets IDed as a New Zealand plant.
Distribution map for that plant shows New Zealand obliterated by green dots … and a few unlikely outliers in Cape Town.
Yes, one should always use the Computer Vision suggestions with the understanding that it only knows what it knows, not what it doesn’t (yet) know, and will always offer the most similar suggestions among the things it does know. The only way to increase it’s knowledge is to keep adding and (correctly) identifying enough observations.
And as you suggest, it is always good practice to consider distribution maps and other information before accepting a Computer Vision suggestion.
I would suggest chiming in on some of the above topics if you are interested. I think this thread (correct me if I’m wrong) was intended to be more about human-generated IDs.
Yes, I was thinking about human-generated IDs, though I guess the same issues arise also with respect to automatic algorithms (I just hadn’t thought of that).
I think one way to look at it is that extra-ordinary claims require extra-ordinary evidence. It is reasonable to assume that it is the species commonly seen in the area unless there is evidence to the contrary. And I suppose iNaturalist is not really concerned with the odd vagrant. If it later turns out that there is more than one species present, old observations can be re-scrutinised and maybe demoted to genus if differentiation isn’t possible from the photos.
It also varies a lot by taxa. Based on your wording i am guessing you aren’t referring to plants. Different taxa really vary in mobility. A white pine tree won’t randomly fly to Europe but sometimes North American birds do. Etc…
For me location is more like supporting information to the morphological features of the organism. And in my humble opinion that important factor is much overlooked by many users, though I do use it with caution, because cosmopolitan species, cryptic species and all that etc.
Very interesting discussion. Like Charlie, I use geography to narrow options when keying organisms (aquatic invertebrates in my case). Psephenus herricki is the only true water penny I’m likely to find in Vermont so I go with that…even though I do not use a larval key to species. There’s probably only one false water penny also. I have hesitated to make the same assumptions on iNaturalist, but perhaps I should.
In most regions there are organizations watching for invasives so it is usually possible to find out if Chinese mystery snails or emerald ash borers are in a particular locality or expanding. It’s certainly important to be open to the possibilities.