Type: Package
Title: Networks Characterising the Physical Configurations of Animal Habitats
Version: 0.1.0
Maintainer: Peng He <phe@ab.mpg.de>
Description: Functions for generating and visualising networks for characterising the physical attributes and spatial organisations of habitat components (i.e. habitat physical configurations). The network generating algorithm first determines the X and Y coordinates of N nodes within a rectangle with a side length of L and an area of A. Then it computes the pair-wise Euclidean distance Dij between node i and j, and then a complete network with 1/Dij as link weights is constructed. Then, the algorithm removes links from the complete network with the probability as shown in the function ahn_prob(). Such link removals can make the network disconnected whereas a connected network is wanted. In such cases, the algorithm rewires one network component to its spatially nearest neighbouring component and repeat doing this until the network is connected again. Finally, it outputs an undirected network (weighted or unweighted, connected or disconnected). This package came with a manuscript on modelling the physical configurations of animal habitats using networks (in preparation).
License: GPL-3
Imports: stats, igraph (≥ 1.2.4.1), ggplot2 (≥ 3.2.0)
RoxygenNote: 6.1.1
Encoding: UTF-8
NeedsCompilation: no
Packaged: 2019-11-15 14:30:04 UTC; Peng
Author: Peng He ORCID iD [aut, cre], Damien R. Farine ORCID iD [ths]
Repository: CRAN
Date/Publication: 2019-11-25 19:40:02 UTC

Generate networks characterising habitat physical configurations

Description

Generate undirected networks (weighted or unweighted, connected or disconnected) characterising the physical attributes and spatial organizations (or distributions) of habitat components (i.e. habitat configurations).

Usage

ahn_gen(N, L, mu, lamda, Connected = TRUE, Weighted = TRUE, eta = 1,
  A = 25, X = NULL, Y = NULL, U = NULL, V = NULL)

Arguments

N

The number of nodes

L

A side length of the rectangle landscape within which nodes are anchored

mu

the critical Dij (i.e. Euclidean distance between node i and j) at which the link removing probability curve P(Dij, mu, lamda) transits from concave to convex (see ahn_prob)

lamda

the steepness of the link removing probability curve P(Dij, mu, lamda), see ahn_prob

Connected

TRUE for connected while FALSE ignores whether the networks are connected or not

Weighted

TRUE for weighted while FALSE for unweighted networks

eta

mediates the weight, i.e. (Dij)^-eta, of the link rewiring node i from one network component and node j from another network component (i and j are with an Euclidean distance of Dij) when the network becomes disconnected after removing links from the initial complete network with the probability P(Dij, mu, lamda) = [1 + exp(-lamda(Dij - mu))]^-1 when both Connected = TRUE and Weighted = TRUE

A

The area of the rectangle landscape within which the network is defined

X

A vector of X coordinates for the N nodes (sampled from [0, L] uniformly at random if NULL)

Y

A vector of Y coordinates for the N nodes (sampled from [0, A/L] uniformly at random if NULL)

U

A vector with N elements specifying node attributes (qualitative or quantitive), by default NULL

V

A vector with N elements specifying node attributes (qualitative or quantitive), by default NULL

Value

Return an animal habitat network (an igraph object)

Examples

# generate a connected and weighted network
ahn_gen(N = 10, L = 5, mu = 1, lamda = 5)



N <- 10
x <- runif(N, 0, 5)
ql <- sample(LETTERS, N, replace = TRUE)
qn <- sample(1:20, N, replace = TRUE)

# specify the X coordinates, node attributes U and V for a connected and unweighted network
ahn_gen(N, L = 5, mu = 1, lamda = 5, Weighted = FALSE, X = x, U = ql, V = qn)

# specify the Y coordinates, node attributes U and V for a weighted network, no matter if the
# network will be connected or not
ahn_gen(N, L = 5, mu = 1, lamda = 5, Weighted = TRUE, Connected = FALSE, Y = x, U = ql, V = qn)

Plot networks

Description

Visualise networks generated by the function ahn_gen.

Usage

ahn_plot(ahn, NodeLabels = unname(V(ahn)), NodeColors = unname(V(ahn)),
  NodeSizes = rep(3, length(V(ahn))))

Arguments

ahn

Networks returned by ahn_gen

NodeLabels

The labels of nodes in ahn (node IDs by default)

NodeColors

The colors of nodes in ahn (each node has a unique color by default)

NodeSizes

The sizes of nodes in ahn (nodes are with the identical size of 3 by default)

Value

Return a plot of the network

Examples

# generate a weighted and connected network and plot it by default
N <- 10
x <- runif(N, 0, 5)
ahn <- ahn_gen(N, L = 5, mu = 1, lamda = 5, X = x)
ahn_plot(ahn)



# plot the network with specified colors, labels and sizes for nodes
ahn_plot(
ahn,
NodeColors = sample(4, N, replace = TRUE),
NodeLabels = letters[1:N],
NodeSizes = seq(1, 5, length.out = N))

Plot probability curves

Description

Plot the probability curve P(Dij, mu, lamda) for removing links from the initial complete network

Usage

ahn_prob(Dij = seq(0.05, 10, length.out = 30), mu = c(0.1, 2, 5, 10),
  lamda = c(1e-04, 0.15, 0.35, 0.75, 1.25, 5, 30))

Arguments

Dij

A vector of Euclidean distances between node i and j

mu

The concave-to-convex transition point of the probability curves P(Dij, mu, lamda) = [1 + exp(-lamda(Dij - mu))]^-1, where Dij is the Euclidean distance between node i and j

lamda

The steepness of the probability curves

Value

Return a plot with probability curves

Examples

# plot the probabilities for removing network links between node i and j with
# Euclidean distances Dij

dis <- seq(.05, 10, length.out = 20)
m <- c(.1, 2, 5, 10)
l <- c(.0001, .15, .35, .75, 1.25, 5, 30)
ahn_prob(dis, m, l)