Many thanks for the explanation!
Regards,
Ricardo

2009/5/13  <[email protected]>:
> Dear Ricardo,
>
>  > I am using the compana function in adehabitat to analyse habitat
>  > selection of a small mammal in an agricultural area, using a study
>  > design III (used and available measured for each animal). I have some
>  > missing values both in used and available habitats for some animals. I
>  > replaced null proportions of used habitats by a small value and a mean
>  > Wilks $,1ry (B lambda was computed using the randomization of log-ratio
>  > values. I got a profile of preferences (ranking) of habitat types but
>  > results don ,A4 (Bt show me the log-ratio matrices of used and available
>  > habitats. Is there any way to look at the log-ratio matrices?
>
> Actually, the function compana does not return the matrix of log
> ratio, as it is quite easy to compute it yourself. Let a be the matrix
> containing the available habitat (animals in rows and habitats in
> columns) and u the matrix containing the use (supposing that you have
> replaced null values by very small values). Then the log-ratios
> matrices for the use can be computed by:
>
> li <- lapply(1:ncol(u), function(i) log(u[,-i]/u[,i]))
>
> Each element of the list contains the log-ratios based on a column of
> the matrix (i.e. if there are 5 columns in u, the first element of the
> li is a logratio based on the first column of u as the reference, the
> second element uses the second column as the reference, etc.) Note
> that because you have missing values in your matrix describing the
> available habitat, the results of compositional analysis depends on
> the choice of the reference for the computation of the log-ratio. So
> that it is essential to consider simultaneously all the matrices...
>
> And similarly,for the available habitat:
>
> li <- lapply(1:ncol(a), function(i) log(a[,-i]/a[,i]))
>
> HTH,
>
>
> Cl ,Ai (Bment Calenge
> --
> Cl ,Ai (Bment CALENGE
> Office national de la chasse et de la faune sauvage
> Saint Benoist - 78610 Auffargis
> tel. (33) 01.30.46.54.14
>
> _______________________________________________
> AniMov mailing list
> [email protected]
> http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov
>
>
_______________________________________________
AniMov mailing list
[email protected]
http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov

Reply via email to