[R-sig-eco] Rarefaction related queries

[Jari Oksanen]

On 29/04/2015, at 17:25 PM, Rajendra Mohan panda wrote:

> Dear All
> 
> I am struggling to rarefy my species data. If I use rarefy function in
> vegan, I get two columns for all the dataset (312 rows and 6220 columns)
> even if I change the sample value.
> 
> If I use rarefaction function of *Jenna Jacobs - email at jennajacobs.org
> <email at jennajacobs.org>, *I get different rows and columns for the same
> input data.
> 
> If I understand properly, the data with sample =5 should yield one-fifth of
> the total rows*columns of input data but that does not come.
> 
Dear Rajendra Mohan,

I have no idea what you mean with rarefaction. However, rarefy() function does what it is documented to do: it gives the rarefied number of species for each *row* (sampling unit) of your data. So you should get 312 rarefied species richness values for your data. You should only get two columns for each sampling unit if you specify se=TRUE in which case you also get what they call "standard error" of the rarefied number of species. The function assumes your data be counts (of individuals) so that you can meaningfully rarefy each sampling unit separately.

It sounds like you did not want to get this, but something completely different. It could be useful if you give a hint about your desired outcome.

There are some other functions in vegan that do something else that may be closer to what you want to get. Function rrarefy() returns one *random* rarefied community of your data -- each call returning a different realization of rarefaction. Function drarefy() gives probabilities for each species to occur in a rarefied sample of each sampling unit. Functions in specaccum() bunch sample subsets of sampling units instead of sampling individuals within sampling units. Several other packages have several other functions that do the same, almost the same, similar or completely different things loosely associated with rarefaction.

Cheers, Jari Oksanen