-------- Original Message --------
Subject: Re: correction of phylogenetic effects for comparing allometric
patterns
Date: Fri, 28 Jan 2011 14:01:04 -0500
From: Dean Adams <[email protected]>
To: [email protected]
Bruno,
Actually, Revell's 2009 method won't do what you want, as it is
implemented for a single X and Y value for each species. What that
method does is account for phylogenetic non-independence in allometric
trends across species (i.e. interspecific allometry). Thus, one cannot
compare allometric patterns within species, since each species is
represented by a single specimen rather than its allometric trend.
Conceptually, what you want to do is a 'phylogenetic' ANCOVA (or
MANCOVA). The ANCOVA allows the statistical comparison of allometric
trends among groups, but you wish to do so while accounting for
phylogenetic non-independence.
From an implementation perspective, I see two ways to go. Both require
a bit of coding in R (as I am unaware of existing routines for this),
but neither should be too terrible.
1: Phylogenetic simulation: Garland et al. 1993 (Syst. Biol.) described
a simulation method for evaluating ANOVA in a phylogenetic context. The
basic idea is to obtain the observed ANOVA results, and then simulate
data on the phylogeny (based on parameters of the observed data), and
compare ANOVA results from the simulated data to those of the observed
data. For your case, you would do the same, only for a MANCOVA design.
Luke Harmon's 'geiger' package has a function called 'phy.manova' that
will do a single-factor MANOVA + phylogenetic simulation. That code
could be altered for a MANCOVA design.
2: PGLS + within species allometric covariation: Implement a MANCOVA,
but account for phylogenetic non-independence. This is accomplished
using phylogenetic generalized least squares (PGLS) where the error
covariance term is not independently distributed, but rather is
distributed according to the phylogenetic relationships among taxa.
However, the wrinkle here is that you must include multiple individuals
per species, as the within-species allometric effect requires more than
one specimen per species, but your phylogeny likely includes one
representative per species. One solution to this is use Felsenstein's
2008 (Am. Nat) generalization, which can incorporate within-species
variation into comparative analyses. An alternative (though more clunky)
implementation is to add very short branches for each individual within
species as a polytomy for each species, then proceed using PGLS (early
example: Ruber and Adams 2001: J. Evol. Biol.).
Hope this helps,
Dean
--
Dr. Dean C. Adams
Associate Professor
Department of Ecology, Evolution, and Organismal Biology
Department of Statistics
Iowa State University
Ames, Iowa
50011
www.public.iastate.edu/~dcadams/
phone: 515-294-3834
On 1/28/2011 11:55 AM, morphmet wrote:
-------- Original Message --------
Subject: Re: correction of phylogenetic effects for comparing allometric
patterns
Date: Fri, 28 Jan 2011 10:11:43 -0500
From: Fabio de Andrade Machado<[email protected]>
To: [email protected]
I think that this article is right on the spot:
Revell. Size-correction and principal components for interspecific
comparative studies. Evolution (2009) vol. 63 (12) pp. 3258-3268
It also contains a R and Matlab script to perform the analysis.
You should also check:
McCoy et al. Size correction: comparing morphological traits among
populations and environments. Oecologia (2006) vol. 148 (4) pp. 547-554
for a brief discussion about differences in allometry. It also
contains a R code
best,
Fabio de Andrade Machado
Citando morphmet<[email protected]>:
-------- Original Message --------
Subject: correction of phylogenetic effects for comparing allometric
patterns
Date: Thu, 27 Jan 2011 16:32:55 -0500
From: Bruno Frédérich<[email protected]>
To: [email protected]
Dear morphometricians,
I am comparing allometric vectors among eight closely-related fish species.
However, I would like to correct the data to remove phylogenetic effects
(e.g. using standard comparative methods).
Has anyone already done a multivariate comparisons of allometries,
taking phylogeny into account?
If so, could you give me information about your methods?
Could you suggest me some relevant papers about this topic?
Many thanks in advance!
All the best,
Bruno
--
Bruno Frédérich PhD
Laboratory of functional and evolutionary morphology
University of Liège
Institut de Chimie, B6c
4000 Liège (Sart-Tilman)
Belgium
http://www.ulg.ac.be/morfonct/
Tél.: +32 (0)4 366 51 33
Fax.: +32 (0)4 366 37 15
Email:[email protected]
