There are many R packages on CRAN that can fetch data from Google Maps, Stamen, Bing, or OpenStreetMap—even some of the packages that we have previously used in this chapter, such as the ggmap package, can do this. Similarly, the dismo package also comes with both geo-coding and Google Maps API integration capabilities, and there are some other packages focused on that latter, such as the RgoogleMaps package.

Now we will use the OpenStreetMap package, mainly because it supports not only the awesome OpenStreetMap database back-end, but also a bunch of other formats as well. For example, we can render really nice terrain maps via Stamen:

> library(OpenStreetMap)
> map <- openmap(c(max(map_data$y, na.rm = TRUE),
+                  min(map_data$x, na.rm = TRUE)),
+                c(min(map_data$y, na.rm = TRUE),
+                  max(map_data$x, na.rm = TRUE)),
+                type = 'stamen-terrain')

So we defined the left upper and right lower corners of the map we need, and also specified the map style to be a satellite map. As the data by default arrives from the remote servers with the Mercator projections, we first have to transform that to WGS84 (we used this previously), so that we can render the points and polygons on the top of the fetched map:

> map <- openproj(map,
+   projection = '+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs')

And showtime at last:

> plot(map)
> plot(deldir(dt$lon, dt$lat), wlines = "tess", lwd = 2,
+   col = c('red', 'black'), pch = 19, cex = 0.5, add = TRUE)

This seems to be a lot better compared to the outline map we created previously. Now you can try some other map styles as well, such as mapquest-aerial, or some of the really nice-looking cloudMade designs.