Interpretation of gl.diversity

[Soraia Barbosa]

Hi everyone,

Thank you for developing what it's looking like a very comprehensive and integrative R package!

I have some questions regarding the interpretation of results from gl.diversity, which might just relate to my lack of understanding of the functions.

I calculated alpha and beta diversity for q=0, q=1 and q=2, and I'm now trying to interpret the data based on what is described in Sherwin et al. 2017.

I understand that for within population diversity, i.e. alpha values, q=0 represents allelic richness index, q=1 Shannon information index and q=2 heterozygosity index, and that these indices are represented by H values. These H values can then be converted to a common scale of effective numbers, D.

For between population comparisons of diversity, i.e. beta values, q=0 represents number of allelic types not shared between pops, q=1 measures the chance of a type of individuals providing info on the sampling locality, and q=2 measured the chance of two random individuals from different localities having different allelic types.

Please correct any of these if I'm wrong.

My question with beta diversity is that the supp material of Sherwin et al. 2017 provides many options to calculate each of these indices, and I'm not sure what the output from gl.diversity represents for each index. For example, Beta diversity for q=1 can represent the Mutual Information (I) or Shannon differentiation, but I'm not sure what is outputted. Do we get only calculations of H and D and we then need to calculate these measures a posteriori with a different function? Can we interpret H and D directly for both alpha and beta?

Any guidance would be most welcome.

Best,

Soraia

[Stephanie Todd]

Hi Soraia,

Did you get anywhere with this?

I'm was wondering the same thing re which measures were used for calculating beta indicies.

Also Im not 100% sure what the values in the top half of the matricies are... are the beta measures directional or are they some sort of error value? 

About to start digging through the source code to figure it out but was hoping you might be able to save me the trouble.

Cheers,

Stephanie

[David. Berman]

I draw your attention to a related issue, where dartR & gl.diversity are mentioned:

Konopiński MK. 2020. Shannon diversity index: a call to replace the original Shannon’s formula with unbiased estimator in the population genetics studies.

PeerJ 8:e9391 https://doi.org/10.7717/peerj.9391

Any chance of incorporating Zahl's formula as an option, to deal with small sample sizes Bernd?

David

[mm.j...@gmail.com]

Dear All, 

I am looking for answers to the same question. I am wondering if there is any feedback todate. 

With regards, 

Gati

[Rodrigo Monjaraz]

Hello everyone,

Did you find any answer to this question?

I'm using the q profile obtained by the function gl.report.diversity and I came with the same concern/question as all of you, I did check and compare a few measures of differentiation for beta diversity like Fst, Jost D, etc, but couldn't find any that actually match with the output of $two_H_beta so any help will be appreciated.

Best,

Rodrigo.