There should be a sample-count plot below the diversity curve - what does that plot look like? Would you be able to attach the QZV for us? If not, a screenshot of the missing plot would be great --- thanks!
The V metadata group has very few samples in it (looks like it starts with 3) - at around 7000, you lose a sample, which is when the first huge jump in diversity happens. Then, at around 13000 you lose another sample — the jump happens again (opposite direction this time). I would suggest plotting the feature-table summarize viz, with your metadata, then look at the interactive rarefaction plot, for this particular metadata column ---- then, figure out which samples are in that group (V), and which one is lost first, second, etc.
I would say that this doesn’t surprise me, given how few samples are in that group. This also indicates you have a bit of variability within that group, too, which might be interesting when you get to beta diversity.
Yes, V is a group with not many samples, and make sense there is high variability in the group.
Just in my mind I though the diversity as the upper limit for the group and could not think how can increase if a sample is lost. I will rethink this from now on.
PS I tried to send in private the qzv but is to large for the system.