Monday, July 2, 2018

Things You Can Cite Meaghan's Newest Article For!

Meaghan had a new article come out recently in Palaeo-Electronica. It's great, and you should read it!

And the authors are VERY TRUSTWORTHY and super good at SCIENCE
It's also very long! With many tables! So in lieu of you having to read the whole thing, here are the top take-away points that Meaghan thinks you should probably cite her & Edward for:

1) Agisoft Photoscan Measurements are Comparable to Digital Calipers!
That's right, the 3D software that everyone is using for their measurements is, in fact, good for measurements. Huzzah! The measurements are not significantly different from digital calipers so you can mix techniques, and also throw this citation in the face of the one reviewer who might ask you about it if you use Photoscan. BUT Agisoft isn't as good on small measurements as Digital Calipers are - there's a bit higher uncertainty on Photoscan measurements once you get down into the 1 cm range. So Photoscan is great, but not for tiny measurements (at least, not in its current iteration, and it may relate more to user problems than software issues).

This person shouldn't be grumpy, cuz Agisoft is pretty good.

2) Tooth Variation of Some Artiodactyls Is High
Like... really high. Like, elephants-sized high. This could be an important point for a variety of reasons - it means that if you have some big artiodactyls and small artiodactyls, they might be the same species of artiodactyl! It makes species identification in artiodactyls a little harder than we maybe thought it was previously - or at least, harder when using teeth measurements.

Variably high, and invariably uncute.

3) Tooth Variation is Higher in Some Artiodactyls Than Others

Camels are crazy variable, and duikers are a lot less so. This probably has something to do with the shapes of their teeth, but it means that if you're looking at using a modern animal to understand your fossil animal's variation, you should probably pick one that is as closely related as possible... and has similar teeth styles. So don't compare peccaries to oreodonts, please - the different types of teeth make them non-identical.

WE ARE ALL UNIQUE (well kinda... at least compared to other species)

4) Tooth Variation is Higher in Bigger Animals Than Smaller... Even as a Coefficient of Variation
So paleontologists have this way of trying to make the variation of big animals be the same as the variation of small animals, so we can compare them. After all, a big mouse and a small mouse have maybe 10 grams of weight difference, but a big elephant versus a small elephant is going to have MANY MORE GRAMS OF WEIGHT DIFFERENCE. To scale that, people use this thing called the coefficient of variation which is basically the variation divided by the size of the measurement. Boom, it's unit-less and now elephants and mice are comparable, right?

Um, no. Our study confirms the findings of some others that CV might not entirely eliminate the effect of size. We think that might be because bigger things are more variable than they should be, but it's also possible that small things are more variable than they should be. It's complicated - but it means that if you want to understand a fossil organism, you need to compare it to something of the same size, too. So cite this paper when you want to talk about picking a comparable modern artiodactyl.

we r different too also mouse waaaaay not to scale

5) Finite Mixture Analysis Doesn't Always Pick Up Sexual Dimorphism or Multiple Species
There was a paper that came out a few years on dinosaurs and how Finite Mixture Analysis was the miracle that would identify all sexual dimorphism. We had some sexually dimorphic animals in our sample, and FMA couldn't find it in all of them. It could find it in some, though, so it's not terrible, just not a be-all end-all (duh). Also, it couldn't figure out how many species of duikers there were. But hey, if you've written a paper where FMA didn't say there were two sexes or species, but you're confident there are two sexes or species... cite this paper and say that there could be!

One of these things is not like the other, unless you do Finite Mixture Analysis and then, naaah... they're the same.

6) Discriminant Function Analysis and RandomForest Analysis Don't Always Pick Up Different Species Either
So we tried to use the maths to identify different species of duikers using their teeth. It worked sometimes, but not 100% of the time. Also, Discriminant Function Analysis worked better than RandomForest Analysis did. So if you're doing DFA or RFA and want to support DFA over RFA, you can cite this paper!
Here is a random forest. It is not very good at identifying duikers.

7) Artiodactyl Teeth Sometimes Change Size Through Wear
 And this affects both widths AND lengths... in completely opposite manners. Widths get larger through wear (probably there's some hidden eruption in artiodactyls that we didn't pick up), and lengths get shorter. We measured our stuff at the "widest" and "longest" part - which, if you're trying to be consistent, may not be the best way to do it! If you want to avoid wear, try measuring at the base of the tooth always, not the crown - which we know some of you do already, but some of you don't. Cite this paper as a reason for why!

Cuz it turns out that teeth that are wider at the top and experience wear, get smaller! Weird!
8) Also, Oreodonts are probably oversplit!
Though to be honest, we are far from the first paper to say that. But we ARE the most recent?

So go forth and cite this paper for many, many things! You don't EVEN have to read it now, isn't that nice of us?

Emery-Wetherell, Meaghan M., and Edward Byrd Davis. "Dental measurements do not diagnose modern artiodactyl species: Implications for the systematics of Merycoidodontoidea." Palaeontologia Electronica 21.2 (2018): 1-28.