traitstrap (version 0.1.0)

trait_parametric_bootstrap: Bootstrap traits parametrically

Description

Function for parametric bootstrap resampling to calculate community weighted trait mean and higher moments.

Usage

trait_parametric_bootstrap(
  fitted_distributions,
  nrep = 100,
  sample_size = 200,
  raw = FALSE
)

Value

a tibble

Arguments

fitted_distributions

Fitted distribution object returned by trait_fit_distributions

nrep

number of bootstrap replicates

sample_size

bootstrap size

raw

logical; argument to extract the raw data of the trait distributions. The default is raw = FALSE. If raw = TRUE, nrep is restricted to 1 to avoid memory issues.

Details

trait_parametric_bootstrap() is a parametric analogue of the trait_np_bootstrap(). It randomly samples from among the fitted distributions proportionally to species abundance. The number of samples per replicated are drawn specified with the parameter sample_size, and the number of replicates is specified by the parameter nrep. From these distributions the function estimates the mean and the higher moments including variance, skewness and kurtosis.

The output of trait_parametric_bootstrap() can be summarized using trait_summarize_boot_moments().

Examples

library(dplyr)
data(community)
data(trait)

# Filter trait and community data to make example faster

community <- community |>
  filter(
    PlotID %in% c("A", "B"),
    Site == 1
  )

trait <- trait |>
  filter(Trait %in% c("Plant_Height_cm"))

filled_traits <- trait_fill(
  comm = community,
  traits = trait,
  scale_hierarchy = c("Site", "PlotID"),
  taxon_col = "Taxon", value_col = "Value",
  trait_col = "Trait", abundance_col = "Cover"
)

fitted_distributions <- trait_fit_distributions(
  filled_traits = filled_traits,
  distribution_type = "normal"
)

# Note that more replicates and a greater sample size are advisable
# Here we set them low to make the example run quickly
parametric_distributions <- trait_parametric_bootstrap(
  fitted_distributions = fitted_distributions,
  nrep = 5,
  sample_size = 100
)

moment_summary <- trait_summarise_boot_moments(
  bootstrap_moments = parametric_distributions,
  parametric = FALSE
)