MPSEM (version 0.3-6)

PEM-class: Class and methods for Phylogenetic Eigenvector Maps (PEM)

Description

Class and methods to handle Phylogenetic Eigenvector Maps (PEM).

Usage

# S3 method for PEM
print(x, ...)
  # S3 method for PEM
as.data.frame(x, row.names = NULL, optional = FALSE, ...)
  # S3 method for PEM
predict(object, targets, lmobject, newdata, 
  interval = c("none", "confidence", "prediction"), level = 0.95, ...)

Arguments

x

A PEM-class object containing a Phylogenetic Eigenvector Map.

row.names

Included for method consistency reason; ignored.

optional

Included for method consistency reason; ignored.

object

A PEM-class object.

targets

Output of getGraphLocations.

lmobject

An object of class ‘lm’ (see lm for details).

newdata

auxiliary trait values

interval

The kind of limits (confidence or prediction) to return with the predictions. interval="none": do not return a confidence interval.

level

Probability of the confidence of prediction interval.

...

Further parameters to be passed to other functions or methods (currently ignored).

Value

A PEM-class object contains:

x

the graph-class object that was used to build the PEM (see PEM.build),

sp

a logical vector specifying which vertex is a tip,

B

the influence matrix for those vertices that are tips,

ne

the number of edges,

nsp

the number of tips,

Bc

the column-centred influence matrix,

means

the column means of B

dist

edge lengths,

a

the steepness parameter (see PEM.build for details),

psi

the relative evolution rate along the edges (see PEM.build for details),

w

edge weights,

BcW

the weighted and column-centred influence matrix,

d

the singular values of BcW,

u

the eigenvectors (left singular vectors) of BcW, and

vt

the right singular vectors of BcW.

In addition to these standard component, function, PEM.fitSimple and PEM.forcedSimple add the following members, which are necessary to make predictions:

S2

the variance(s) of the response(s),

y

a copy of the response(s), and

opt

the list returned by optim,

as well as a copy of the estimated weighting parameters as edge properties.

Details

The print method provides the number of eigenvectors, the number of observations these vectors are spanning, and their associated eigenvalues.

The as.data.frame method extracts the eigenvectors from the object and allows one to use PEM-class objects as data parameter in function such as lm and glm.

The predict object is a barebone interface meant to make predictions. It must be given species locations with respect to the phylogenetic graph (target), which are provided by function getGraphLocations and a linear model in the form of an object from lm. The user must provide auxiliary trait values if lmobject involves such trait.

References

Gu<U+00E9>nard, G., Legendre, P., and Peres-Neto, P. 2013. Phylogenetic eigenvector maps (PEM): a framework to model and predict species traits. Meth. Ecol. Evol. In press.