metricTester (version 1.3.6)

expectations: Generate expectations for null+metric combinations

Description

Will generate mean, SD and CI expectations for the desired metric + null combinations.

Usage

expectations(picante.cdm, tree, optional.dists = NULL,
  regional.abundance = NULL, distances.among = NULL, randomizations,
  cores, metrics, nulls, concat.by, output.raw = FALSE)

Arguments

picante.cdm

A picante-style community data matrix with sites as rows, and species as columns

tree

Phylo object

optional.dists

A symmetric distance matrix can be directly supplied. This option is experimental. Behavior depends on metric being used. If the metric in question relies on the phylogenetic distance matrix from a call to cophenetic(tree), then this optional distance matrix will be inserted instead.

regional.abundance

A character vector in the form "s1, s1, s1, s2, s2, s3, etc". Optional, will be generated from the input CDM if not provided.

distances.among

A symmetric distance matrix, summarizing the distances among all plots from the cdm. Optional, only used by some null models.

randomizations

The number of times the input CDM should be randomized and the metrics calculated across it.

cores

This function can run in parallel. In order to do so, the user must specify the desired number of cores to utilize.

metrics

Optional. If not provided, defines the metrics as all of those in defineMetrics. If only a subset of those is desired, then metrics should take the form of a character vector corresponding to named functions from defineMetrics. The available metrics can be determined by running names(defineMetrics()). Otherwise, if the user would like to define a new metric on the fly, the argument can take the form of a named list of new functions (metrics).

nulls

Optional. If not provided, defines the nulls as all of those in defineNulls. If only a subset of those is desired, then nulls should take the form of a character vector corresponding to named functions from defineNulls. The available nulls can be determined by running names(defineNulls()). Otherwise, if the user would like to define a new null on the fly, the argument nulls can take the form of a named list of new functions (nulls).

concat.by

Whether to concatenate the randomizations by richness, plot or both

output.raw

Default is FALSE. Set to TRUE if raw randomized values are preferred (as opposed to summarized mean, SD, CI, etc).

Value

A list of data frames, where each data frame corresponds to a given null model, and contains the mean, SD, and CI for each metric for that null model.

Details

Given a list of desired metrics (which should always include richness) and null models, will generate the expected mean, standard deviation and confidence intervals based on the number of specified randomizations. This function is flexible in that new metrics and nulls can be added and tested with it. By setting output.raw to TRUE, the function can also output raw randomized values as opposed to the summarized values.

References

Miller, E. T., D. R. Farine, and C. H. Trisos. 2016. Phylogenetic community structure metrics and null models: a review with new methods and software. Ecography DOI: 10.1111/ecog.02070

Examples

# NOT RUN {
#simulate tree with birth-death process
tree <- geiger::sim.bdtree(b=0.1, d=0, stop="taxa", n=50)

#simulate a log normal abundance distribution
sim.abundances <- round(rlnorm(5000, meanlog=2, sdlog=1)) + 1

#simulate a community of varying richness
cdm <- simulateComm(tree, richness.vector=10:13, abundances=sim.abundances)

#below not run for timing issues on CRAN
test <- expectations(picante.cdm=cdm, tree=tree, optional.dists=NULL,
regional.abundance=NULL, distances.among=NULL, randomizations=3, cores="seq",
nulls="richness", metrics=c("richness", "NAW_MPD"),
concat.by="both", output.raw=FALSE)

#an example of how to explore behavior of a new metric in the metricTester framework
#this "metric" simply calculates the richness of each plot in the CDM
exampleMetric <- function(metrics.input)
{
output <- apply(metrics.input$picante.cdm, 1, lengthNonZeros)
output
}

#test2 <- expectations(picante.cdm=cdm, tree=tree, optional.dists=NULL,
#regional.abundance=NULL, distances.among=NULL, randomizations=3, cores=1,
#nulls="frequency", 
#metrics=list("richness"=metricTester:::my_richness, "exampleMetric"=exampleMetric),
#concat.by="both", output.raw=FALSE)
# }