In remi-daigle/marxanr: An R interface for Marxan

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

library(marxanr)
library(ggplot2)
library(dplyr)
library(sf)

This vignette provides step by step guidance to take spatial data layers, transform these into Marxan inputs, run Marxan, then load and interpret the results.

Loading data layers

Spatial data

data(pu)

# landmasses for plotting
data(land)

# x and y limits to make nicer plots
xlim <- st_bbox(pu)[c(1,3)]
ylim <- st_bbox(pu)[c(2,4)]
ggcoords <- coord_sf(xlim=xlim,
           ylim=ylim)

basemap <- ggplot(land)+
  geom_sf()

basemap+
  geom_sf(data=pu)+
  ggcoords

Marxan data

Data for this example, which are included in this package, are derived from Serdynska et al. 2021 will be used in this example. These data are the available bioregional-scale ecological and human use data used to inform draft Marine Protected Area (MPA) design for the Scotian Shelf-Bay of Fundy Bioregion. Please see the original technical report for further details.

While the input data are "real", the following example is NOT to be interpreted in any way as a suggestion for real MPA placement. The example is to show that you CAN run Marxan via R with real data. The example has not been properly paramaterized or optimized.

data("spec")
data("puvspr")
data("pu_dat")
data("bound")

head(spec)
head(puvspr)
head(pu_dat)
head(bound)

Creating a Marxan project folder

# name for new scenario
scen <- "MPAtest"

# create a folder for a new scenario
createMarxanFolder(scen=scen,dialog = FALSE)

# download marxan if necessary
downloadMarxan(path = scen)

# write marxan files to folder
write.csv(spec,file.path(scen,"input","spec.dat"),quote = FALSE,row.names = FALSE)
write.csv(puvspr,file.path(scen,"input","puvspr.dat"),quote = FALSE,row.names = FALSE)
write.csv(pu_dat,file.path(scen,"input","pu.dat"),quote = FALSE,row.names = FALSE)
write.csv(bound,file.path(scen,"input","bound.dat"),quote = FALSE,row.names = FALSE)

# create new input parameters
input <- newParams(BLM=10^5,NUMREPS = 50L,SCENNAME = scen)
writeParams(inputdat = input,file = file.path(scen,"input.dat"))

# run marxan
runMarxan(marxanpath = scen,
          inputdatfile = "input.dat")
# unlink(scen,recursive = TRUE) # if necessary

Plotting results

ssoln <- read.csv(file.path(scen,"output",paste0(scen,"_ssoln.csv"))) %>%
  mutate(planning_unit = as.character(planning_unit)) %>%
  right_join(pu,by=c("planning_unit"="id")) %>% 
  st_as_sf()
basemap+
  geom_sf(data = ssoln, aes(fill=number))+
  ggcoords

best <- read.csv(file.path(scen,"output",paste0(scen,"_best.csv"))) %>%
  mutate(PUID = as.character(PUID)) %>%
  right_join(pu,by=c("PUID"="id")) %>% 
  st_as_sf()
basemap+
  geom_sf(data = best, aes(fill=SOLUTION))+
  ggcoords