Use function spplot

You can do this quite easily with base graphics, I don't see a real need for spplot or ggplot:

Sample data: Spain and a random raster:

require(raster)
es = getData("GADM", country="ESP",level=1)
es =es[-14,] # drop canary islands
r = raster(extent(es),ncol=200,nrow=200)
r[]=runif(200*200)

Plot the raster and add the polygons - the alternative clips the raster to the Spanish polygons using the mask function:

plot(r)  # or try: plot(mask(r,es))
plot(es, add=TRUE)

Here's the clipped version:

If you want to stick with spplot, you could e.g. use layer from latticeExtra to add 'Spatial*' objects to an existing plot. Based on the code provided by @Spacedman, this would look as follows.

## load package
library(latticeExtra)

## plot raster and add polygons
spplot(r, scales = list(draw = TRUE), 
       col.regions = terrain.colors(100), at = seq(0, 1, .01)) + 
  layer(sp.polygons(es, lwd = 2))

Using only ssplot with sp.layout, without latticeExtra gives the same result:

## Define how the polygon should appear
borders <- list("sp.polygons", es, fill = "transparent", col = 'black', first = FALSE)

## plot both raster and polygon
spplot(r,
       sp.layout = borders,
       col.regions = terrain.colors(100), at = seq(0, 1, .01),
       scales = list(draw = TRUE))