r - Turning off some legends in a ggplot

Suppose I have a ggplot with more than one legend.

mov <- subset(movies, length != "")
(p0 <- ggplot(mov, aes(year, rating, colour = length, shape = mpaa)) +
  geom_point()
)

I can turn off the display of all the legends like this:

(p1 <- p0 + theme(legend.position = "none"))

Passing show_guide = FALSE to geom_point (as per this question) turns off the shape legend.

(p2 <- ggplot(mov, aes(year, rating, colour = length, shape = mpaa)) +
  geom_point(show_guide = FALSE)
)

But what if I want to turn off the colour legend instead? There doesn't seem to be a way of telling show_guide which legend to apply its behaviour to. And there is no show_guide argument for scales or aesthetics.

(p3 <- ggplot(mov, aes(year, rating, colour = length, shape = mpaa)) +
  scale_colour_discrete(show_guide = FALSE) +
  geom_point()
)
# Error in discrete_scale


(p4 <- ggplot(mov, aes(year, rating, shape = mpaa)) +
  aes(colour = length, show_guide = FALSE) +
  geom_point()
)
#draws both legends

This question suggests that the modern (since ggplot2 v0.9.2) way of controlling legends is with the guides function.

I want to be able to do something like

p0 + guides(
  colour = guide_legend(show = FALSE) 
)

but guide_legend doesn't have a show argument.

How do I specify which legends get displayed?

Answer

You can use guide=FALSE in scale_..._...() to suppress legend.

For your example you should use scale_colour_continuous() because length is continuous variable (not discrete).

(p3 <- ggplot(mov, aes(year, rating, colour = length, shape = mpaa)) +
   scale_colour_continuous(guide = FALSE) +
   geom_point()
)

Or using function guides() you should set FALSE for that element/aesthetic that you don't want to appear as legend, for example, fill, shape, colour.

p0 <- ggplot(mov, aes(year, rating, colour = length, shape = mpaa)) +
  geom_point()    
p0+guides(colour=FALSE)

UPDATE

Both provided solutions work in new ggplot2 version 2.0.0 but movies dataset is no longer present in this library. Instead you have to use new package ggplot2movies to check those solutions.

library(ggplot2movies)
data(movies)
mov <- subset(movies, length != "")