stratified total evidence tip dating analysis

[evolucionario]

I'm setting up a stratified total evidence tip dating analysis using the FBD prior following Zhang et al. 2016:

prset samplestrat = diversity 2: 66 0, 23 0;

But MrBayes produces an error:

Setting 1 fossil sampling rate shift time to 66 for partition 1
Time of fossil sampling rate shift must be > 0.
Error when setting parameter "Samplestrat" (2)
There was an error on at least one processor
The error occurred when reading char. 37-37 on line 106
in the file 'stratified.nex'

This alternative code seems to work:
prset samplestrat = diversity 2: 66 23;

(so, not specifying rho_i = 0 as the manual indicates)
Is this correct?

Thanks!

• Operating system: MacOS Monterrey 12.0.1
• Version of MrBayes: MrBayes 3.2.7a x86_64

[zhangchicool]

The code has been changed since the paper was published. The new implementation allows d, r, s to shift at different times, and the order is s, d, r in the command.
If you use prset samplestrat = diversity 2: 66 23; it means extant sampling is diversified, fossil sampling proportion s changes at 66 and 33 Ma, net diversification d and turnover s are constant.
If you want d, r, s all shift at 66 and 33 Ma, use prset samplestrat = diversity 2: 66 23, 2: 66 23, 2: 66 23;
You can also use, for example, prset samplestrat = diversity 3: 66 50 23, 2: 66 23, 1: 55; so that s changes 3 times, d changes 2 times, and r changes 1 time.
The new implementation only sets up rho for extant taxa (prset sampleprob = ...), past rho_i's are all 0.
I will try to remember to update the manual.

[evolucionario]

Thanks for the clarification. It seems that I am specifying the stratification I wanted.

However the posterior looks a bit strange. Not only for s but also for d and r I'm getting a nearly binary posterior where two values are repeatedly revisited. See screenshots. And the values don't make much sense: fossilization in periods 1 and 3 should be near 0 as we don't have fossils there.