I recently needed to install a R environment with several geospatial dependencies. There were conflicts with the base operating system’s configuration of geospatial libraries, so I used the environment manager Miniconda in order to install R into a clean, completely separated environment.

The procedure is as follows:

# Create a new conda environment named r-gis
conda create -n r-gis
# Activate it
source activate r-gis
# Install R and geospatial dependencies
conda install -c conda-forge r-essentials geos r-rgeos r-rgdal

Note that it is important to install all the packages from the conda-forge channel in order for the dependencies to be resolved. Note also that the package geos has to be installed as well as r-rgeos. This is because r-rgeos only provides an interface to an existing geos library installation.

With the main dependencies installed via conda, you can now start an R shell an install any remaining geospatial dependencies such as raster and GISTools:

install.packages('raster')
install.packages('GISTools', dep=T)