Hi Daniel,

It sounds like a Phylogenetic Generalized Least Squares (PGLS) analysis might be what you are looking for, but it is hard to know without knowing what question you are trying to ask and with what data.

PGLS uses an estimate of phylogenetic correlation, pagel's lambda, that varies between zero (complete phylogenetic independence) and one (indicating species' traits covary in direct proportion to the dregree of shared evolutionary history) and controls for this nonidependence due to phylogeny in a generalized least squares framework. I have used the R packages "ape" and "caper" for these analyses in the past, but it seems that there is now even a "PGLS" package, although I don't know anything about it.

There are some nice tutorials out there for doing PGLS analyses in R: the anthrotree website has a great introduction and walk through exercises (http://www.anthrotree.info/wiki/pages/w5j9m8/751_Running_PGLS_in_R_caper.html), the Queenborough Lab (http://www.simonqueenborough.com/R/specialist/phylogenetic.html) and the R-Phylo Wiki (http://www.r-phylo.org/wiki/Main_Page) to list a few.

I hope that helps, but if you have any additional questions, let me know!

Jan

Article Seasonal mortality patterns in non-human primates: Implicati...

Hi Daniel,

It sounds like a Phylogenetic Generalized Least Squares (PGLS) analysis might be what you are looking for, but it is hard to know without knowing what question you are trying to ask and with what data.

PGLS uses an estimate of phylogenetic correlation, pagel's lambda, that varies between zero (complete phylogenetic independence) and one (indicating species' traits covary in direct proportion to the dregree of shared evolutionary history) and controls for this nonidependence due to phylogeny in a generalized least squares framework. I have used the R packages "ape" and "caper" for these analyses in the past, but it seems that there is now even a "PGLS" package, although I don't know anything about it.

There are some nice tutorials out there for doing PGLS analyses in R: the anthrotree website has a great introduction and walk through exercises (http://www.anthrotree.info/wiki/pages/w5j9m8/751_Running_PGLS_in_R_caper.html), the Queenborough Lab (http://www.simonqueenborough.com/R/specialist/phylogenetic.html) and the R-Phylo Wiki (http://www.r-phylo.org/wiki/Main_Page) to list a few.

I hope that helps, but if you have any additional questions, let me know!

Jan

Article Seasonal mortality patterns in non-human primates: Implicati...

That's almost exactly what I needed, thanks a lor.

However, none of these tutorials explain how should you specify a dependent variable (for example, mortality at birth) and more than one predictor variable(body weight, mean anual temperature...); they only work with 2 variables.

I see from your paper you did both bi-variate and multivariate PGLS. Can you help me with this issue?

Hi Daniel,

The formula works the same as for a multiple regression in R; for example:

pgls(formula = y ~ x1+x2, data = yourdataframe, lambda = "ML")

For more details on syntax for setting up equations in R: http://faculty.chicagobooth.edu/richard.hahn/teaching/FormulaNotation.pdf

Hope that helps and best of luck,

Jan

You were incredibly helpful, that should be more than enough for me to make it work.

Thanks Jan

I have a problem with PGLS in APE. I am working with Y matrix of 20 variables, but the analysis only is working with the first variable. Does is it normal? Could someone help me?

Thanks for your collaboration

Victor