Rarefaction: diversity or richness?

I am perplexed by the qiime2 visualization tool for generating rarefaction curves from a microbiome dataset, and I'm hoping someone can explain! As near as I can tell, there are only two options in qiime2: either alpha-rarefaction or beta-rarefaction.

According to Legendre & Legendre's (Numerical Ecology) explanation of Hurlbert's (1971) method and formula for rarefaction curves is calculated on "true (untransformed) counts of individuals". Therefore, using alpha or beta diversity metrics (i.e. transformed counts) seems to violate this core assumption in the calculation of rarefaction curves.

Given that context, I have so many questions:

  • why does qiime2 lack an option to plot rarefaction curves as richness vs. read counts? (or am I just failing to find it?!) ...basically I'm looking for the qiime2 analog of the rarecurve() function in the vegan package for R...
  • How exactly does qiime2 calculate rarefaction curves? Does it follow Hurlbert's formula, or something else? Is there a method for correcting for diversity vs. richness when calculating a rarefaction curve?

Thank you!!

Good morning Monika,

First, welcome to the forums! :qiime2:

Secondly, have you found the Moving Pictures Tutorial?
This is a common starting point for learning Qiime2, and relates to many of the questions you ask.
And it has data we can use for examples!

This is the default option. From the Moving Pictures tutorial:

'richness' is number of 'observed_features'
'read counts' is 'Sequencing Depth'

This is an excellent question, as it perfectly highlights the difference between diversity metrics and normalization methods!

(I'm suggesting there is a mistake hidden in here. Can you find it?)


Recently I had this same doubt but can you please suggest this qiime2 alpha_rarefaction plots can be used as similar to vegan package rarefaction plot in R? @iamfisch Thanks

Yes, these plots are the same when made with Qiime2 and/or Vegan.
They both help you see the effect of sampling depth on alpha diversity.

I'm feeling like I may be confused or misunderstanding your questions... :thinking:

Is there something wrong with these graphs?