GGPLOT2 : geom_area with ordered character variable as x axis

Question

I have a dataset like the following :

dat <- data.frame(sp = c("a", "a", "b", "b", "b", "c", "c"),
nb = c(5, 44, 32, 56, 10, 1, 43),
gp = c("ds1", "ds2", "ds1", "ds2", "ds3", "ds1", "ds3"))

With sp = species ; nb = nb occurrences ; gp = sampling group

I want to make a geom_area graph where values for species (sp) are displayed on y axis, with species grouped on x axis and ordered by descending order based on their total sum.

Up to now I only managed to do that :

ggplot(dat, aes(x=as.numeric(factor(sp)), y=nb, fill=gp, colour = gp)) +
geom_area()

Which gives this output (please don't laugh ;))

Could you help me to sort the x axis on descending order of the sum of stacked values ? And to fill the empty area ?

E.g. I try to do something like that (here in ascending order, but it no matters) :

Answer 1

Try this. The gaps in your plot could be filled by filling the df with the missing combinations of gp and sp using tidyr::complete. To reorder the levels of sp I make use of forcats::fct_reorder:

library(ggplot2)
library(dplyr)
library(tidyr)
library(forcats)

dat <- data.frame(sp = c("a", "a", "b", "b", "b", "c", "c"),
                  nb = c(5, 44, 32, 56, 10, 1, 43),
                  gp = c("ds1", "ds2", "ds1", "ds2", "ds3", "ds1", "ds3"))

dat1 <- dat %>% 
  # Fill with missing combinations of gp and sp
  tidyr::complete(gp, sp, fill = list(nb = 0)) %>% 
  # Reorder according to sum of nb
  mutate(sp = forcats::fct_reorder(sp, nb, sum, .desc = TRUE),
         sp_num = as.numeric(sp))

ggplot(dat1, aes(x=sp_num, y=nb, fill=gp, colour = gp)) +
  geom_area()