scale_fill_brewer() - specify na.colour and include NA in legend

[Olivia]

Hi there,

I would like to use scale_fill_brewer and the option to colour NAs by a specific colour, but include NAs in the legend. Reproducible example below.  Couple of options I've tried are commented out. I know if you aren't using scale_fill_brewer there's a work around by changing NA to something else in the data (http://stackoverflow.com/questions/18970580/how-to-plot-na-bar-with-ggplot2), but that would defeat me trying to use a sequential scale from brewer along with a very different colour for NA.  The legend item doesn't need to be called "NA".

I've gone from integer to a factor in the example because that's what I do with my real data (I'm leaving justifications for doing what is pretty much a stacked histogram out as the principle applies more widely anyhow). 

Any thoughts?

Best wishes,

Olivia

#install.packages("vegan")

#install.packages("ggplot2")

require(vegan)

require(ggplot2)

data(mite.env) # dataset from mite

miteenv_na <- mite.env # not to pollute mite if already using it

miteenv_na$familiarity <- sample(1:5)

miteenv_na[ , 6][sample(seq(miteenv_na[ , 6]), 10)] <- NA #introducing NAs

is.na(miteenv_na[ , 6]) # checking

miteenv_na$familiarity_fac <- factor(miteenv_na$familiarity, exclude = NULL)

levels(miteenv_na$familiarity_fac)

fam_labs <- levels(miteenv_na$familiarity_fac)

ggplot(miteenv_na, aes(x = Topo))+

  geom_bar(position = position_fill(), 

           stat = "bin", 

           aes(fill = familiarity_fac)) +

  scale_fill_brewer("Familiarity", 

                    type = "seq", 

                    na.value = "orange",

                    #labels = levels(miteenv_na$familiarity_fac))+ #doesn't work 

                    #labels = fam_labs) + #also doesn't work

  labs(y = "proportion") +

  theme_bw()

[Olivia]

edit (commented out the changes but didn't fix the end of the scale_brewer():

#install.packages("vegan")

#install.packages("ggplot2")

require(vegan)

require(ggplot2)

data(mite.env) # dataset from mite

miteenv_na <- mite.env # not to pollute mite if already using it

miteenv_na$familiarity <- sample(1:5)

miteenv_na[ , 6][sample(seq(miteenv_na[ , 6]), 10)] <- NA #introducing NAs

is.na(miteenv_na[ , 6]) # checking

miteenv_na$familiarity_fac <- factor(miteenv_na$familiarity, exclude = NULL)

levels(miteenv_na$familiarity_fac)

fam_labs <- levels(miteenv_na$familiarity_fac)

ggplot(miteenv_na, aes(x = Topo))+

  geom_bar(position = position_fill(), 

           stat = "bin", 

           aes(fill = familiarity_fac)) +

  scale_fill_brewer("Familiarity", 

                    type = "seq", 

                    na.value = "orange") + #,