/bootdissim

R package for building turnover saturation curves via bootstrapping

Primary LanguageRMIT LicenseMIT

bootdissim

DOI

Overview

R package for building turnover saturation (rarefaction or accumulation) curves via bootstrapping. It offers a workflow to bootstrap dissimilarity metrics between ecological networks.

In the case of bipartite networks (e.g. plant - pollinator), such curves are informative about the level of effort still needed for collecting data to the point of having a trusted turnover estimate.

Installation

You can install bootdissim from GitHub with:

# install.packages("devtools")
devtools::install_github("valentinitnelav/bootdissim")

Examples

Given two networks, like vazarr and vazcer from the bipartite package. One can compute the bootstrapped saturation curve of turnover (dissimilarity) between these two networks when looking at their interactions.

library(bootdissim)
library(bipartite)
library(ggplot2)

# Get the character vectors of interactions from each network
net1 <- reshape_net(vazarr, seed = 1)
net2 <- reshape_net(vazcer, seed = 1)
vect <- get_interactions(net1, net2)

# Can also compute the observed dissimilarity between the two vectors of
interactions. Several indices are implemented for now:
abc <- get_abc(vect)

get_beta(abc, "whittaker")
# [1] 0.2846473
get_beta(abc, "jaccard")
# [1] 0.5568475
get_beta(abc, "sorensen")
# [1] 0.7153527

# Bootstrapp (resample) the dissimilarity:
dism_boot <- boot_dissim(lst_vect = vect,
                         index = "whittaker",
                         by = 50,
                         replace = FALSE,
                         size_short = TRUE,
                         n_boot = 100,
                         n_cpu = 3)

# Plot the saturation curve
p <- plot_dissim(get_stats_gg(dism_boot))
p <- p + labs(x = "Number of sampled interactions",
              y = "Total turnover (dissimilarity)") +
  theme_bw()
p

How to cite the package?

Valentin Ștefan, & Tiffany Marie Knight. (2021). bootdissim: R package for building turnover saturation curves via bootstrapping (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.7145381