Profile cover photo
Profile photo
Graeme Lloyd
200 followers
200 followers
About
Graeme's posts

Post has attachment
Hi Dave,

Thanks again for another stimulating post.

BTW, I like that you got a bunch of these numbers with a ruler and still provide answers to several decimal places. You are truly a palaeontologist!

The sampling thing is definitely an interesting problem and I'm not sure I can say anything to enlighten you. Just from pure speculation I would expect things like Cenozoic birds and Palaeo- and Mesozoic tetrapods to be worse than any of the groups in your list. Having said that, Cenozoic bats are the kinds of things you'd also expect to be terrible, so maybe they are in a similar ball park.

One question: What does it actually mean to have a sampling rate of >1 in terms of cal3? Does this mean we have all the ADRs and hence there would be no tree with hypothetical ancestors?

G

Post has attachment
Hi Dave,

Thanks for doing this, I hope it helps people!

A few comments/questions for you:

1. I might have asked this before, but is your starting tree a strict consensus? If so, could your function theoretically take a bunch of equally optimal trees instead? (Assuming that in practice every possible bifurcating resolution of a consensus is a larger set than the number of equally optimal trees used to produce that consensus, so the latter are really more informative.)
2. I like that you pointed out the 'time of observation' issue. I think this is under-discussed in the literature, but obviously has a big effect on topology. (Also, I have been updating date.phylo() and also included an add.terminal option.)
3. Another thought that occurred to me - could you use taxa not in your tree to estimate sampling rate? I.e., most of the time (and for various reasons) authors will not include every taxon that probably belongs to that clade in a phylogenetic analysis. On a similar note, does including an outgroup cause an issue? Should this be pruned out?
4. Do you think we'll ever be able to deal with changes in sampling probability across time, space and phylogeny? I guess we would expect this to occur in reality, but it seems like it would be a major headache to deal with in practice.

G
Wait while more posts are being loaded