adj_mat: adj_mat: Function to compute the adjacency matrix of an area.

In bamm: Species Distribution Models as a Function of Biotic, Abiotic and Movement Factors (BAM)

adj_mat: Function to compute the adjacency matrix of an area.

Description

Creates an adjacency matrix of an area of interest. This could be the accessible area (M) of a species or any geographic region of interest.

Usage

adj_mat(modelsparse, ngbs = 1, eigen_sys = FALSE, which_eigs = 1)

Arguments

modelsparse	A setA object returned by the function model2sparse.
ngbs	Numeric. Number of neighbors (see details).
eigen_sys	Logical. If TRUE the eigen analyses of the adjacency matrix will be returned.
which_eigs	Numeric. Which eigen value and eigen vector will be returned.

Details

The model is a raster object of the area where the dispersal process will occur.

The function creates an adjacency matrix where cells are considered connected based on:

• The specified neighborhood size ('ngbs')

• The spatial resolution of the input raster

• Only non-NA cells in the original model

When 'eigen_sys = TRUE', the function performs spectral decomposition using eigs, which is particularly useful for:

• Analyzing network connectivity properties

• Identifying clusters or communities in the landscape

• Modeling dispersal processes using spectral graph theory

The number of neighbors depends on the dispersal abilities of the species and the spatial resolution of the niche model; for example, a species's with big dispersal abilities will move throughout more than 1 km^2 per day, so the idea is to give an approximate number of moving neighbors (pixels) per unit of time.

Value

Returns an object of class setM with 7 slots. The first contains the adjacency matrix. A n x n sparse matrix (n=number of non-NA cells of the niche model) where connected cells are represented by 1. The second slot has the adjacency list. It is a list of matrices with four columns (FromRasCell -from cell ID of the raster-, -to cell ID of the raster-, -from non-NA cell-, -to non-NA cell-). Other slots contain information about initial coordinates where dispersal occurs (initial_points), number of cells used to define the neighborhood (ngbs), non-NA coordinates (coordinates), and a matrix of eigen vectors (eigen_vec).

Author(s)

Luis Osorio-Olvera & Jorge Soberón

References

\insertRef

SoberonOsoriobamm.

See Also

• model2sparse for creating the input 'setA' object

• setM for the output class structure

• eigs for the eigen decomposition implementation

For more information about see adjacency matrices in the context of the theory of area of distribution (Soberon and Osorio-Olvera, 2022).

Examples

x_coord <- c(-106.5699, -111.3737,-113.9332,
             -110.8913, -106.4262, -106.5699)
y_coord <- c(16.62661, 17.72373, 19.87618,
             22.50763, 21.37728, 16.62661)
xy <- cbind(x_coord, y_coord)
p <- sp::Polygon(xy)
ps <- sp::Polygons(list(p),1)
sps <- sp::SpatialPolygons(list(ps))
mx_grid <- bamm::shape2Grid(sps,resolution = 0.25,ones = TRUE)
mx_sparse <- bamm::model2sparse(model=mx_grid, threshold = 0.1)
adj_mx <- bamm::adj_mat(modelsparse=mx_sparse,
                        ngbs=1,eigen_sys=TRUE,which_eigs=1)
print(adj_mx)
mx_grid_eigen <- mx_grid
mx_grid_eigen[mx_sparse@cellIDs] <- adj_mx@eigen_vec
raster::plot(mx_grid_eigen)

bamm documentation built on Feb. 4, 2026, 1:07 a.m.