Standard deviation in alpha rarefaction plots

(Marc T P) #1

Hi there,

I have data from several individual samples that group into 3 groups. When I see the alpha rarefaction plot with all the samples, I get:

And if I see them grouped:

The question I have is why I’m not seeing greater standard deviations in the grouped plot. Maybe it’s a silly question, but I expect greater SD comparing with the other plot.

Many thanks,

(Matthew Ryan Dillon) #2

Hey @marctormo!

I think this is an awesome question! It stumped @ebolyen and I for a few minutes…

First thing, I just want to point out, those box plots are not standard deviation — they are 7-number summaries of value distributions (see the “Help” box above the legend for more details).

So, the metrics are computed at each rarefaction depth a number of times (default 10) — the boxplots are representative of the selected diversity metric for that group, and show the spread of diversity values at that given rarefaction depth, for those trials. This is why you aren’t see thing spread that you anticipated, because the spread is of the 10 rarefaction trials, not of the individual samples. Does that make sense?

(Marc T P) #3

Thank you very much!
Now it’s clear for me.