In luismurao/bam: Species Distribution Models in the light of the BAM theory

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

The bam package is an R package designed to create and operate on large (tens of millions of cells) matrices regarding to each element of the $\textbf{BAM}$ scheme, for example, the adjacency matrix (connectivity matrix), and the niche suitability matrices.

The following vignette presents the basic functions and methods of the bam package. The package has three main functionalities related to each element of the BAM scheme.

library(bam)

A functions

Convert niche models to sparse matrices

The package uses sparse matrices to represent those objects which allows it to optimize computers memory and make important speed ups in computation times by eliminating operations on zero elements. The basic function of the package is model2sparse which converts a raster model to an setA class

model_path <- system.file("extdata/Lepus_californicus_cont.tif",
                          package = "bam")
model <- raster::raster(model_path) 
# binary model
model_bin <- model > 0.7
sparse_mod <- bam::model2sparse(model = model_bin)

The slots of the object are

sparse_mod

M functions

Connectivity matrix

The function to estimate connecivity between pixels of a given raster is adj_mat, (generally is the M area but can be any matrix); it uses sparse adjancency matrices.

 # Adjacency matrix from a niche model


adj_mod <- adj_mat(sparse_mod,ngbs=1,eigen_sys = T)
adj_mod

The map of the connectivity matrix adj_mod is

model_eig <- model
model_eig[sparse_mod@cellIDs] <- abs(adj_mod@eigen_vec)
raster::plot(model_eig)

AM objects

These are objects that combine SetA and SetM .

Connectivity Suitability diagram

Estimates the connectivity suitability and dispersal diagram. It shows the number of geographic clusters as a function of dispersal and suitability.

clustersin <- bam::bam_clusters(model=sparse_mod,
                                ngbs=1,plot_model=FALSE)

The function returns the following slots

clustersin

Lets see the geogrpahic clusters as function of suitability and dispersal

clustersin@interactive_map

raster::plot(clustersin@raster_map)

The CSD diagram

The CSD diagram is a simple but useful tool to estimate the number of dispersal steps that a species needs to travel per unit time in order to occupay its potential area of distribution. In this example the approximate number of dispersal steps to archive all suitable patches is 30.

csd_plot <- bam::csd_estimate(sparse_mod,
                              dispersal_steps=c(2,4,8,16,32,40))

luismurao/bam documentation built on Nov. 28, 2022, 3:02 p.m.