Performs leave-one-out deletion analysis for phylogenetic linear regression, and detects influential species, while taking into account potential interactions with intraspecific variability.
intra_influ_phylm(
formula,
data,
phy,
Vy = NULL,
Vx = NULL,
y.transf = NULL,
x.transf = NULL,
n.intra = 10,
distrib = "normal",
model = "lambda",
cutoff = 2,
track = TRUE,
...
)The model formula: response~predictor.
Data frame containing species traits with row names matching tips
in phy.
A phylogeny (class 'phylo') matching data.
Name of the column containing the standard deviation or the standard error of the response
variable. When information is not available for one taxon, the value can be 0 or NA.
Name of the column containing the standard deviation or the standard error of the predictor
variable. When information is not available for one taxon, the value can be 0 or NA
Transformation for the response variable (e.g. "log" or "sqrt"). Please use this
argument instead of transforming data in the formula directly (see also details below).
Transformation for the predictor variable (e.g. "log" or "sqrt"). Please use this
argument instead of transforming data in the formula directly (see also details below).
Number of datasets resimulated taking into account intraspecific variation (see: "intra_phylm")
A character string indicating which distribution to use to generate a random value for the response
and/or predictor variables. Default is normal distribution: "normal" (function rnorm).
Uniform distribution: "uniform" (runif)
Warning: we recommend to use normal distribution with Vx or Vy = standard deviation of the mean.
The phylogenetic model to use (see Details). Default is lambda.
The cutoff value used to identify for influential species (see Details)
Print a report tracking function progress (default = TRUE)
Further arguments to be passed to phylolm
The function intra_influ_phylm returns a list with the following
components:
cutoff: The value selected for cutoff
formula: The formula
full.model.estimates: Coefficients, aic and the optimised
value of the phylogenetic parameter (e.g. lambda) for the full model
without deleted species.
influential_species: List of influential species, both
based on standardised difference in intercept and in the slope of the
regression. Species are ordered from most influential to less influential and
only include species with a standardised difference > cutoff.
sensi.estimates: A data frame with all simulation
estimates. Each row represents a deleted clade for an iteration of resimulated
data. Columns report the calculated regression intercept (intercept),
difference between simulation intercept and full model intercept (DIFintercept),
the standardised difference (sDIFintercept), the percentage of change in intercept compared
to the full model (intercept.perc) and intercept p-value
(pval.intercept). All these parameters are also reported for the regression
slope (DIFestimate etc.). Additionally, model aic value (AIC) and
the optimised value (optpar) of the phylogenetic parameter
(e.g. kappa or lambda, depending on the phylogenetic model used) are
reported.
data: Original full dataset.
errors: Species where deletion resulted in errors.
When Vy or Vx exceed Y or X, respectively, negative (or null) values can be generated, this might cause problems for data transformation (e.g. log-transformation). In these cases, the function will skip the simulation.
Setting n.intra at high values can take a long time to execute, since the total number of iterations
equals n.intra * nrow(data).
This function fits a phylogenetic linear regression model using phylolm, and detects
influential species by sequentially deleting one at a time. The regression is repeated n.intra times for
simulated values of the dataset, taking into account intraspecific variation. At each iteration, the function
generates a random value for each row in the dataset using the standard deviation or errors supplied, and
detect the influential species within that iteration.
All phylogenetic models from phylolm can be used, i.e. BM,
OUfixedRoot, OUrandomRoot, lambda, kappa,
delta, EB and trend. See ?phylolm for details.
influ_phylm detects influential species based on the standardised
difference in intercept and/or slope when removing a given species compared
to the full model including all species. Species with a standardised difference
above the value of cutoff are identified as influential. The default
value for the cutoff is 2 standardised differences change.
Currently, this function can only implement simple linear models (i.e. \(trait~ predictor\)). In the future we will implement more complex models.
Output can be visualised using sensi_plot.
Paterno, G. B., Penone, C. Werner, G. D. A. sensiPhy: An r-package for sensitivity analysis in phylogenetic comparative methods. Methods in Ecology and Evolution 2018, 9(6):1461-1467.
Ho, L. S. T. and Ane, C. 2014. "A linear-time algorithm for Gaussian and non-Gaussian trait evolution models". Systematic Biology 63(3):397-408.
phylolm, intra_phylm,
influ_phylm,intra_influ_phyglm,sensi_plot.
# NOT RUN {
# Load data:
data(alien)
# run analysis:
intra_influ <- intra_influ_phylm(formula = gestaLen ~ adultMass, phy = alien$phy[[1]],
data=alien$data,model="lambda",y.transf = log,x.transf = NULL,Vy="SD_gesta",Vx=NULL,
n.intra=30,distrib = "normal")
summary(intra_influ)
sensi_plot(intra_influ)
# }
Run the code above in your browser using DataLab