[R-sig-phylo] NEWBIE - ANOVA and PGLS for categorical data

[Anna Rice]

Hello,

I'm relatively new to phylogeny analysis and have some questions.
Overall, my data consists of two groups and I'd like to compare ecological
attributes between these two groups. So here are my questions:

1. I've tried implementing both phylogenetic ANOVA using aov.phylo (Geiger
package) and regular ANOVA test, and got the same results. Do you have any
idea why this could happen?
my Phyl. ANOVA --> aov.phylo(attribute~category,phy,nsim=1000,test="Wilks")
my regular ANOVA --> aov(attribute~category,data=Data)

2. I've read some papers in which the authors used PGLS instead of ANOVA in
comparing two groups. What could be the reason for using PGLS when having a
categorical factor and not a continuous predictor factor?

Many thanks for your help,
Anna

[[alternative HTML version deleted]]

[Liam J. Revell]

Hi Anna.

"1. I've tried implementing both phylogenetic ANOVA using aov.phylo
(Geiger package) and regular ANOVA test, and got the same results...."

See Garland et al. (1993). The phylogenetic ANOVA just fits a regular
ANOVA model & then generates a null distribution for the test statistic
(F) via numerical simulation on the tree.

"2. I've read some papers in which the authors used PGLS instead of
ANOVA in comparing two groups...."

This can be done. The assumption of (P)GLS is that the residual error in
y|x is multivariate normal with a distribution that ~vcv.phylo (i.e.,
the correlation of related species is proportional to their shared
fraction of time since the root), or some variant thereof (such as
Pagel's lambda model). This may apply even if x is a factor (although it
should probably be tested). E.g., see
http://blog.phytools.org/2013/02/a-comment-on-distribution-of-residuals.html
for more information.

All the best, Liam

Liam J. Revell, Assistant Professor of Biology
University of Massachusetts Boston
web: http://faculty.umb.edu/liam.revell/
email: ***@umb.edu
blog: http://blog.phytools.org

Post by Anna Rice
Hello,
I'm relatively new to phylogeny analysis and have some questions.
Overall, my data consists of two groups and I'd like to compare ecological
1. I've tried implementing both phylogenetic ANOVA using aov.phylo (Geiger
package) and regular ANOVA test, and got the same results. Do you have any
idea why this could happen?
my Phyl. ANOVA --> aov.phylo(attribute~category,phy,nsim=1000,test="Wilks")
my regular ANOVA --> aov(attribute~category,data=Data)
2. I've read some papers in which the authors used PGLS instead of ANOVA in
comparing two groups. What could be the reason for using PGLS when having a
categorical factor and not a continuous predictor factor?
Many thanks for your help,
Anna
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo

[Marguerite Butler]

Hi Anna,

You may also want to try ouch. It allows you to fit models of adaptive evolution to continous data based on adaptation to selective regimes specified by the ecological factors. What you describe is exactly the situation that ouch was designed for.

Another method OUwie, is along the same vein but allows to you test even more complex models where the rate of adaptation and stochastic evolution can vary.

Let me know if you need help implementing ouch.

Butler M.A. and King A.A. (2004) Phylogenetic comparative analysis: a modeling approach for adaptive evolution. The American Naturalist 164(6):683-695. DOI: 10.1086/426002

Another example where the alternative evolutionary hypotheses are really interesting is here:

Scales J.A.**, King A.A., and Butler M.A. (2009) Running for your life or running for your dinner: What drives fiber-type evolution in lizard locomotor muscles? Am. Nat.173:543-553. DOI: 10.1086/597613

OUwie:
Beaulieu, J. M., D.-C. Jhwueng, C. Boettiger, and B. C. O’Meara, 2012. Modeling stabiliz- ing selection: expanding the Ornstein-Uhlenbeck model of adaptive evolution. Evolution 66:2369–2383.

Take care,
Marguerite

Post by Liam J. Revell
Hi Anna.
"1. I've tried implementing both phylogenetic ANOVA using aov.phylo (Geiger package) and regular ANOVA test, and got the same results...."
See Garland et al. (1993). The phylogenetic ANOVA just fits a regular ANOVA model & then generates a null distribution for the test statistic (F) via numerical simulation on the tree.
"2. I've read some papers in which the authors used PGLS instead of ANOVA in comparing two groups...."
This can be done. The assumption of (P)GLS is that the residual error in y|x is multivariate normal with a distribution that ~vcv.phylo (i.e., the correlation of related species is proportional to their shared fraction of time since the root), or some variant thereof (such as Pagel's lambda model). This may apply even if x is a factor (although it should probably be tested). E.g., see http://blog.phytools.org/2013/02/a-comment-on-distribution-of-residuals.html for more information.
All the best, Liam
Liam J. Revell, Assistant Professor of Biology
University of Massachusetts Boston
web: http://faculty.umb.edu/liam.revell/
blog: http://blog.phytools.org

Post by Anna Rice
Hello,
I'm relatively new to phylogeny analysis and have some questions.
Overall, my data consists of two groups and I'd like to compare ecological
1. I've tried implementing both phylogenetic ANOVA using aov.phylo (Geiger
package) and regular ANOVA test, and got the same results. Do you have any
idea why this could happen?
my Phyl. ANOVA --> aov.phylo(attribute~category,phy,nsim=1000,test="Wilks")
my regular ANOVA --> aov(attribute~category,data=Data)
2. I've read some papers in which the authors used PGLS instead of ANOVA in
comparing two groups. What could be the reason for using PGLS when having a
categorical factor and not a continuous predictor factor?
Many thanks for your help,
Anna
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo

_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo

____________________________________________
Marguerite A. Butler
Associate Professor

Department of Biology
University of Hawaii
2450 Campus Rd., Dean Hall Rm. 2
Honolulu, HI 96822

Office: 808-956-4713
Dept: 808-956-8617
Lab: 808-956-5867
FAX: [phone]
http://www.hawaii.edu/zoology/faculty/butler.html
http://www2.hawaii.edu/~mbutler
http://www.hawaii.edu/zoology/








[[alternative HTML version deleted]]

[Anna Rice]

Hi,

Thank you all for your prompt replies.
I realized I haven't specified that I was not referring to the F value but
to the phylogenetic PV. I now understand that only when I ask for the test
summary itself I get this value and can see the difference between the two
implementations, otherwise I can't see any difference.
In any case, your suggestions and references are mostly helpful.

Best,
Anna

Post by Marguerite Butler
Hi Anna,
You may also want to try ouch. It allows you to fit models of adaptive
evolution to continous data based on adaptation to selective regimes
specified by the ecological factors. What you describe is exactly the
situation that ouch was designed for.
Another method OUwie, is along the same vein but allows to you test even
more complex models where the rate of adaptation and stochastic evolution
can vary.
Let me know if you need help implementing ouch.
*Butler M.A.* and King A.A. (2004) Phylogenetic comparative analysis: a
modeling approach for adaptive evolution. *The American Naturalist **164*(6):683-695.
DOI: 10.1086/426002
Another example where the alternative evolutionary hypotheses are really
Scales J.A.**, King A.A., and* Butler M.A.* (2009) Running for your life
or running for your dinner: What drives fiber-type evolution in lizard
locomotor muscles? *Am. Nat.**173:*543-553. DOI: 10.1086/597613
Beaulieu, J. M., D.-C. Jhwueng, C. Boettiger, and B. C. O’Meara, 2012.
Modeling stabiliz- ing selection: expanding the Ornstein-Uhlenbeck model of
adaptive evolution. Evolution 66:2369–2383.
Take care,
Marguerite
Hi Anna.
"1. I've tried implementing both phylogenetic ANOVA using aov.phylo
(Geiger package) and regular ANOVA test, and got the same results...."
See Garland et al. (1993). The phylogenetic ANOVA just fits a regular
ANOVA model & then generates a null distribution for the test statistic (F)
via numerical simulation on the tree.
"2. I've read some papers in which the authors used PGLS instead of ANOVA
in comparing two groups...."
This can be done. The assumption of (P)GLS is that the residual error in
y|x is multivariate normal with a distribution that ~vcv.phylo (i.e., the
correlation of related species is proportional to their shared fraction of
time since the root), or some variant thereof (such as Pagel's lambda
model). This may apply even if x is a factor (although it should probably
be tested). E.g., see
http://blog.phytools.org/2013/02/a-comment-on-distribution-of-residuals.htmlfor more information.
All the best, Liam
Liam J. Revell, Assistant Professor of Biology
University of Massachusetts Boston
web: http://faculty.umb.edu/liam.revell/
blog: http://blog.phytools.org
Hello,
I'm relatively new to phylogeny analysis and have some questions.
Overall, my data consists of two groups and I'd like to compare ecological
1. I've tried implementing both phylogenetic ANOVA using aov.phylo (Geiger
package) and regular ANOVA test, and got the same results. Do you have any
idea why this could happen?
my Phyl. ANOVA --> aov.phylo(attribute~category,phy,nsim=1000,test="Wilks")
my regular ANOVA --> aov(attribute~category,data=Data)
2. I've read some papers in which the authors used PGLS instead of ANOVA in
comparing two groups. What could be the reason for using PGLS when having a
categorical factor and not a continuous predictor factor?
Many thanks for your help,
Anna
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at
____________________________________________
Marguerite A. Butler
Associate Professor
Department of Biology
University of Hawaii
2450 Campus Rd., Dean Hall Rm. 2
Honolulu, HI 96822
Office: 808-956-4713
Dept: 808-956-8617
Lab: 808-956-5867
FAX: [phone]
http://www.hawaii.edu/zoology/faculty/butler.html
http://www2.hawaii.edu/~mbutler
http://www.hawaii.edu/zoology/

[[alternative HTML version deleted]]