Manual legend (scale_colour_manual) missing in plot with several aesthetics

Question

I have just used ggplot to plot data from different data frames. Somehow, I cannot get the (manual) legend to show up.

The code sample below is a good summary of the issue. What's the mistake?

library(plyr)
library(ggplot2)

df <- data.frame(gp=factor(rep(letters[1:3], each=10)), y=rnorm(30))

ds <- ddply(df, .(gp), summarise, mean=mean(y), sd=sd(y))

ggplot() +
  geom_point(data=df, aes(colour='one', x=gp, y=y), colour='red') +
  geom_point(data=ds, aes(colour='two', x=gp, y=mean), colour='green') +
  geom_errorbar(data=ds, aes(colour='three', x=gp, y=mean, ymin=mean-sd, ymax=mean+sd), colour='blue') +
  scale_color_manual('', values=c('red', 'green', 'blue'))

Please do not suggest that I combine the data in a single data frame and then group it by a new variable. I know this could be an option but it is in fact not possible in my particular case for reasons which are out of the scope of this question.

Answer 1

You were almost there. The color variable inside the aes needs to be mapped to an actual colour, and the colour outside aes is unncessary.

ggplot() +
  geom_point(data=df, aes(colour='one', x=gp, y=y)) +
  geom_point(data=ds, aes(colour='two', x=gp, y=mean))+
  geom_errorbar(data=ds, aes(colour='three', x=gp, y=mean, ymin=mean-sd, ymax=mean+sd))+
  scale_color_manual(values=c(one='red', two='green', three='blue'),
                     breaks=c("one","two","three"))