[R-sig-eco] null models with continuous abundance data

[Jari Oksanen]

Howdy,

On 11/01/10 09:30 AM, "Peter Solymos" <solymos at ualberta.ca> wrote:

> Going back to your original problem, I just realized the huge lag
> between the public and the developmental versions of permatswap in
> vegan. 

I may go to details later (if I have time), but now I only tell you that the
gap between release and development versions of vegan was closed today:
vegan version 1.17-0 with new permutation routines was submitted to CRAN
today. Source code is there already, Windows version will be soon, and Mac
version in a few days. Propagation to all CRAN mirrors can take some days
longer.

> Actually, there are two more algorithms that implement some
> features which might be of interest. By using permatswap, you can
> achieve these constraint combinations:
> 
> 1) "swap"/"quasiswap": marginal totals and matrix fill fixed (but not
> col/row occurrences);
> 2) "swsh": row and column occurrences (not totals) fixed (thus fill, too);
> 3) "abuswap": row and column occurrences (and so the fill) fixed, and
> row OR column totals (this latter algorithm was published by Hardy
> 2008 J Ecol 96, 914-926).
> 
> Maybe this "abuswap" is the closest to the one you wanted.

Cheers, Jari Oksanen