facet_wrap changes counts to percentage?

[MDaumal]

Hi,
I am new to facet wraps and managed to make a nice distribution plot with our fish size data.

But, even though ggplot shows me a combined histogram with the counts (max 25 individuals),
facet_wrap converts it to percentages?

Can someone help me understand and fix this? The different "scales" options did not fix it.

My code is (latin is the species):

p<- ggplot(data, aes(x = length_mm, fill = latin)) +
geom_histogram( binwidth=0.5, color="black", fill="black")

p+ facet_wrap(~latin , scales = "fixed")+
ylab(NULL)+ # remove the word "count"
theme( strip.background = element_blank(),
strip.text.x = element_text(face = "italic"),
legend.position="none"
)
Thank you!!
~MD

[AlexisW]

No, it's not percentages. It's just that the number of observation in one group is ~1/16th of the total number, so if p had an axis of 1-50, with the facet wrap you'd have an axis of ~1-3 (actually a bit more since the observations are not totally uniform).

A way to get a feeling for that is to change N in the reprex code below:

library(tidyverse)

N <- 50

data = tibble(length_mm = rpois(16*N, lambda = 50),
              latin = rep(LETTERS[1:16], each = N))


p <- ggplot(data, aes(x = length_mm, fill = latin)) +
  geom_histogram( binwidth=0.5, color="black", fill="black")

p

p+ facet_wrap(~latin , scales = "fixed")+
  ylab(NULL)+ # remove the word "count"
  theme( strip.background = element_blank(),
         strip.text.x = element_text(face = "italic"),
         legend.position="none"
  )

What happens for N <- 5000?

Also you might be interested in geom_density() if you want to plot all of these on the same graph with different colors:

p <- ggplot(data, aes(x = length_mm, fill = latin)) +
  geom_density(alpha = .2)

p

Still hard to read with 16 groups, but if there is a grouping that makes sense (e.g. by Genus) it could be better to have, say, 3 facets with 5 species each.

And since I'm already writing too much, I'm sure you're aware, but I would be very wary of a histogram with 1 single observation (as seems to be the case with A anguilla or S lucioperca).