A recent published paper in the Journal of Mammalogy (Which mammals can be identified from camera traps and crowdsourced photographs? Kays et al. 2022) provides a consensus from experts on how easily distinguishable mammals are from one another. I thought this might be of interest to some folks on here. I highly recommend anyone interested giving this a read! It’s not too long or jargon-filled.
Takeaway for everyday iNat use: Species comparisons with range overlaps that are numbered a “4” in the supplemental excel file should be left to Genus for research grade. For species comparisons with range overlaps numbered a “3”, you should take many high-quality pictures to expect an accurate species identification.
Journal Page Link
Short Summary: The researchers compared 335 species of nonvolant mammals (sorry, bats) to each other and rated them 1-4 on how readily distinguishable they are from one another: (1) always, (2) usually, (3) rarely, or (4) never. 96.5% of pairwise comparisons were readily distinguishable. This includes mostly super obvious species to distinguish such as comparing a photo of a red fox to a deer mouse. In all, out of 55,000+ possible species comparions, 2000 species can rarely or never be distinguished from photographs. When taking ranges into account, this number drastically decreases and depends on the area (U.S. Southwest proved the most difficult due to high small mammal biodiversity). All of this information is available in the excel file in the supplemental info for the paper.
From the paper: The color of branch tips indicates whether a taxon should be always (dark blue), usually (light blue), rarely (pink), or never (red) distinguishable in typical photographs. The large groups of red tips indicate species groups that are practically impossible to tell apart in photographs, whereas the single blue branch tips on the bottom right represent one-of-a-kind species that are unmistakable. Color of nontip branches shows to what extent higher taxonomic groups can be distinguished.
Not surprisingly it appears from the figure that the hardest species to tell apart based on camera trap images are also the hardest to tell apart when you have the animal in hand. Mammals are tough. I’ve often gotten Peromyscus and Neotoma on camera traps in the Southwest US and wouldn’t even try to ID those to species. They can be tricky even as prepared specimens. I’ll have to read this paper when I get the reprint. Thanks.
From the abstract:
We evaluated all pairwise comparisons of species and judged, based on professional opinion, whether they are visually distinguishable in typical pictures from camera traps or the iNaturalist crowdsourced platform on a 4-point scale: (1) always, (2) usually, (3) rarely, or (4) never. Most (96.5%) of the 55,944 pairwise comparisons were ranked as always or usually distinguishable in a photograph, leaving exactly 2,000 pairs of species that can rarely or never be distinguished from typical pictures, primarily within clades such as shrews and small-bodied rodents.
Alas, the article is paywalled. Is there a preprint version of it somewhere?
For those who can read it, I have two questions:
- Did the researchers use any measure to quantify image quality for these comparisons? If so, what measures were applied or included?
- Did the researchers take into account any ancillary observer data (i.e. notes for details which were not captured within the photo), besides geography and time, provided by the observers in reaching their expert ID conclusions?
I can send you the PDF if you’d like. Are the links behind a paywall for you?
I have access to the PDF if you’d like it (not sure if that’s against copyright to share? I really have no idea…)
To answer your questions:
- I don’t believe so, it was more of a subjective consensus. Basically 2+ experts on a group of organisms got together and rated 1-4 of whether two species could be distinguished. So if these experts know that skeletal features or genetic features are the only reliable ways to tell apart 2 species, they rated it a 4.
"We did not systematically evaluate photographs, but assessed the overall
physical similarity of species pairs using a regional field guide
(Kays and Wilson 2009), and based on expert opinions for each
taxonomic group from the coauthors on this paper based on
their experience working with living animals, with museum
specimens, and with photographs. Each author evaluated the
species groups for which they had expertise, with each group
having at least two experts. "
- They didn’t look at this, only if it was possible to distinguish species based on photographs and/or location/range.
Ah, okay. The way that the article was written made it sound dependent on the quality of the images. I have a bit of experience with quantitative image analysis, and I was curious how whether the distinction was made based in relation to some image-dependent metrics. I take it that their use of “typical pictures” was more in contrast to photos that show particular diagnostic features.
For a lot of mammals, especially rodents, body measurements and examination of skull, teeth, and/or baculum characteristics are required to nail down a species ID. Obviously that information isn’t available in a photo of a live animal. Color patterns might work for a few but if the camera trap image is a black & white night shot (which it often is), that information will also be lacking. A genus-level ID is often the best one can do based on a camera trap photo and sometimes not even that. Geographic location and habitat can provide important clues as to what particular species you’d expect, but that is separate from what the photo tells you.
Looking at the figure, my mental summary is, “bigger than a ground squirrel? Probably okay.”
The experts’ version of - I saw a panther! - Nah it’s a puddytat …
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.