An issue with a legend in ggplot2

Question

I am asking you because I have a problem with a legend in ggplot2. Here's a simplified case

library(ggplot2)
library(datasets)

bp <- ggplot(data=PlantGrowth, aes(x=group, y=weight, fill=group, color=group)) + geom_boxplot()

bp +scale_fill_manual(values=c("#999999", "#E69F00", "#56B4E9"), 
                      name="Experimental\nCondition",
                      breaks=c("ctrl", "trt1", "trt2"),
                      labels=c("legend 1", "legend 2", "legend 3"))  +
    scale_color_manual(values=c("black", "red", "green"), 
                   name="Experimental\nCondition",
                   breaks=c("ctrl", "trt1", "trt2"),
                   labels=c("legend 1", "legend 2", "legend 3"))

In this case my legend labels are : "legend 1", "legend 2" and "legend 3"

Now, I would like to have the same labels for my legend : "legend 1" (with still three legend boxes). So my code is :

bp <- ggplot(data=PlantGrowth, aes(x=group, y=weight, fill=group, color=group)) + geom_boxplot()

bp +scale_fill_manual(values=c("#999999", "#E69F00", "#56B4E9"), 
                      name="Experimental\nCondition",
                      breaks=c("ctrl", "trt1", "trt2"),
                      labels=c("legend 1", "legend 1", "legend 1"))  +
    scale_color_manual(values=c("black", "red", "green"), 
                   name="Experimental\nCondition",
                   breaks=c("ctrl", "trt1", "trt2"),
                   labels=c("legend 1", "legend 1", "legend 1"))

But, clearly this is not what I want because from now on, I have 9 legend boxes. Therefore my question is : how in such a case, can I have a legend with three boxes (each one for each boxplot) but with labels for the legend which could be the same ? Besides, why ggplot2 treats my code like this ?

You would help me a lot, if you could help me ! Thanks for your help. Excuse me for my english mistakes, actually I am a non-native.

Answer 1

The problem arises, I think, from how ggplot combines legends together. If they have the same title and labels, they can be combined. I'm guessing that there is a merge involved which, when given duplicate labels, effectively creates an outer join (all possible mixes of the combinations). I'm not sure if there is a right way to fix this, but here is a hack that works:

bp <- ggplot(data=PlantGrowth, aes(x=group, y=weight, fill=group, color=group)) + geom_boxplot()
bp +scale_fill_manual(values=c("#999999", "#E69F00", "#56B4E9"), 
                      name="Experimental\nCondition",
                      breaks=c("ctrl", "trt1", "trt2"),
                      labels=c("legend 1", "legend 1 ", "legend 1  "))  +
    scale_color_manual(values=c("black", "red", "green"), 
                   name="Experimental\nCondition",
                   breaks=c("ctrl", "trt1", "trt2"),
                   labels=c("legend 1", "legend 1 ", "legend 1  "))

So you have labels which are not the same, but which look the same.