Tutorial Contributors: Travis Seaborn, Erin Landguth
Version 0.1, Last updated July 15th 2021
Welcome to the tutorial on creating a pair-wise distance matrix to serve as input for a CDMetaPOP simulation. We are going to cover a one distance option in R, the least-cost path. Now, CDMetaPOP can handle many connectivity matrices, including more complicated connectivity models (such as omnidirectional resistance layers from Circuitscape. The goal of this tutorial is not to give a robust introduction to connectivity modeling.
We are going to build a very simple input, where any location not near one of our sites will have a high value, and everywhere around our sites will have a low value.
First, we will bring in our patch file and create the buffer around our points
library(raster)
library(gdistance)
library(geosphere)
library(rgdal)
library(tidyverse)
setwd("./data/")
points <- read.csv("./otherfiles/tut_patches.csv")
points_spat <- points
coordinates(points_spat) <- ~x+y
#set appropriate crs
proj4string(points_spat) <- CRS("+proj=utm +zone=11 +datum=WGS84")
#create raster/study extent. we are in meters because these are UTMS
ext <- buffer(points_spat, 2500)
# define the circles as polygons to dissolve after converting. Note: rgeos package is required to dissolve.
ext_poly <- aggregate(ext)
We will then next create a raster around these points, where the value of raster is 1 for the points and 1000 for areas outside of the buffer we created around the points.
#create a raster around this polygon
r <- raster(extent(ext_poly)+10000)
#set raster resolution
res(r) <- 200
#make known locations low resistance...
r[ext_poly,] <- 1
#and everything else high resistance
r[is.na(r[])] <- 1000
plot(r)
points(points_spat)
Last, we will calculate the least-cost path across all points and then write the matrix to a csv. Note that the input file does not have patch number or ID as the first row or column.
#first create the transition object
#create transition object
tr1 <- transition(1/r, transitionFunction=mean, directions=8)
tr1 <- geoCorrection(tr1, type="c") #type here should be changed if using random walks
#calculate least-cost paths
distance_matrix <- costDistance(tr1, as.matrix(points))
write.csv(as.matrix(distance_matrix), "./cdmats/tut_create_input.csv", row.names = FALSE, col.names = FALSE)
We can also do some simple visualization or our path
#next, going to do some simple visualization
#we can look at each path
plot(shortestPath(tr1, as.matrix(points[1,]), as.matrix(points[25,]), output="SpatialLines"), add=TRUE)
#or we can view all paths from point 1. this will create an animation in your plot window
for (i in 1:41) {
plot(shortestPath(tr1, as.matrix(points[1,]), as.matrix(points[i,]), output="SpatialLines"), add=TRUE)
}
