Xosé López Goldar

Michigan State University | MSU · Department of Entomology

I am new with Bayestraits v3.0.1. Fortunately, I already set up the data, tree, parameters and command list to test for phylogenetic signal in several traits and I would like to analyze every single trait individually in a single step instead of changing the script for every trait before running. I work on Windows.

Any ideas?

I am performing PCA analysis using PROC FACTOR PROCEDURE in SAS 9.4. I use the same dataset all time, but I obtain different loading scores of my variables for each of the factors when I specify different NFACTORS to extract. The rest of the output remains invariable (eigenvalues, % variance explained...). The dataset comprises 60 variables and 500 observations (thanks Horst for pointing). This is my script:

PROC FACTOR DATA=raw

CORR

SIMPLE

METHOD=PRIN

PRIORS=ONE

SCREE

plots(vector)=all

ROTATE=VARIMAX

OUT=outputdata; ODS OUTPUT OrthRotFactPat=varimax eigenvalues=eigenvalues;

VAR var1 var2 var3 var4 ... var60

I use the same dataset and the only thing I change is NFACT, that is the number of factors that I want to extract. It seems like the calculations of eigenvalues and % variance explained is independent of the calculations of the loading scores. By reading some specific articles and websites it seems that if you request more factors to be extracted, your residual correlations get smaller and you will find more "fit-statistically" factors according to your variables (which does not mean more interpretability of the results).

However I feel a bit confused with this and I would really appreciate any comment clarifying this point. Maybe I am missing something in the script?

Thank you in advance!!

I am very new with this... I am trying to perform a SAS analysis on several sites about the incidence of an insect over time measuring damage inflicted on planted young pine seedlings in pine clearcuts. We have the following factors:

SITE, TIME, PINE SPECIES, BLOCKS (within SITES).

Site and block(site), species and time are fixed. But because the design implies measuring over time every plant placed within each block (5 plants per species per block) and each plant is independent of each other in other blocks and sites, I considered the following script:

proc mixed data=SITES covtest;

class SITE TIME BLOCK PLANT SPECIES;

model damage=TIME|SITE|SPECIES BLOCK(TIME*SITE);

repeated TIME /subject=PLANT*SPECIES(B*SITE);

run;

I remember that also provides the same output if /subject=PLANT(SPECIES*B*SITE)

Thank you very much.

I have 7 terpene external standards but I found nearly 60 terpene compounds in my pine samples which I do not have standards. In each sample I have also internal standard. I know that between terpene volatiles the differences in chemical structure are quite diverse, not like many of the diterpene resin acids.

What is the basis to assign an equivalence to a terpene without standard using another terpene which I have its standard?

- Closest terpene with standard in the chromatogram?

- Closest terpene using chemical structure?

- Using internal standard with those I do not have external standard?

Ex. Suppose I have Pinocarvone in my sample, but not its external standard. As external standards I have a-Pinene, b-Pinene, Limonene and b-Caryophyllene. Which one could be the best for that compound and why?

This is a question addressed more to chemists than ecologists (like me).

Thank you very much in advance.