Comment by purplehat_
18 hours ago
Cool find and a very interesting analysis!
There's a lot more to morphology than just the shape of the shell, and indeed the shape can sometimes be misleading, in that very different species can have somewhat similar shells, and different individuals of the same species can have quite different shell shapes. You've got a gasteropod, so it would be good to pay special attention to the peristome and siphonal canal (based on the bio classes I took in the area, I'm no expert) but of course there's lots of features that could be helpful in an identification.
https://en.wikipedia.org/wiki/Gastropod_shell#Parts_of_the_s... is a good list, and maybe you've already done this but you would want to find a dichotomous key of gasteropod families native to the area to narrow it down. Good luck in figuring out your shell!
I'm a little confused about how significant of information can be derived from a 2d projection of the shell. This sort of mathematical modeling looks like phrenology.
Using PCA on 3d shapes is a proven method for identification. It's nothing like phrenology aside from both involving morphology. Former actually works, latter does not.
(I upvoted, seems a little weird that you were downvoted) So, I spent a little time looking up "Principal Component Analysis" and yeah, this method works well with these forms, and my comment comparing it to phrenology was definitely more out of ignorance!
I was most surprised that you could flatten a 3d structure down to 2d and not lose so much information that it would cause a very high rate of error. Someone else was skeptical enough to do a study as a critique on it, only to have it retracted. (funny in light of this post)
https://elifesciences.org/reviewed-preprints/94685v1/pdf