how to make polygons clipped rather than vanish in ggplot

Question

I'm trying to create a choropleth map with my own polygons. The polygons tile the entire world but I only want to display the area within the US. What I want is to clip the polygons that go outside of a bounding box covering the US, but what happens when I specify the bounding box as limits is that all polygons that are partly outside the bounding box aren't drawn at all.

I have data that looks like this, specifying rectangles:

         lat          long rank group
1   36.56792 -80.260138889    1     1
2   37.97319 -80.260138889    1     1
3   37.97319 -78.866805556    1     1
4   36.56792 -78.866805556    1     1
5  -45.00000  -0.008333334    2     2
6  -22.00000  -0.008333334    2     2
7  -22.00000  22.075000000    2     2
8  -45.00000  22.075000000    2     2
9  -44.99500 -67.441666667    3     3
10 -33.60000 -67.441666667    3     3
11 -33.60000 -45.700000000    3     3
12 -44.99500 -45.700000000    3     3
...

I'm using code like this:

library(ggplot2)
library(maps)
tab = read.table("hiergrid.level1.penn-ave.dat", header = TRUE)
data = data.frame(tab)
mapstates = map_data("state")
ggplot(data, aes(long, lat, group=group)) +
  geom_polygon(aes(fill=rank)) +
  ylim(25,50) +
  xlim(-125,-60) +
  coord_map(project="polyconic") +
  geom_path(data = mapstates, color = "white", size = .75)

I get output like this:

As can be seen, the polygons outside the bounding box aren't drawn at all, but I want them drawn and clipped. Any help?

Answer 1

OK, it looks like this can be fixed by specifying limits to coord_map(). But I still have to specify xlim and ylim:

library(ggplot2)
library(maps)
tab = read.table("hiergrid.level1.penn-ave.dat", header = TRUE)
data = data.frame(tab)
mapstates = map_data("state")
ggplot(data, aes(long, lat, group=group)) +
  ylim(0,90) +
  xlim(-180,-20) +
  geom_polygon(aes(fill=rank)) +
  coord_map(project="mercator", ylim=c(24,50), xlim=c(-126,-66)) +
  geom_path(data = mapstates, color = "white", size = .75)

Output is like this:

Note, however, if I leave out xlim and ylim, or if they're too wide in the X axis, I get very weird output, like this:

I'm not sure why this is happening.