Adding multiple mean bars to ridgeline plot.

[tjsamo]

I was able to add mean bars to my ridgeline plots using the code below. However, I'd like to differentiate the line type to distinguish between the means for NU and U on each line. I've attached an image of another plot type to show what I would like the lines to look like.

fig2 <- ggplot(combined_sla, aes(x=SLA,y=Data,fill=Soil)) +
  geom_density_ridges(scale = 1, show.legend = TRUE, alpha = 0.4,
                      quantile_lines=TRUE, quantile_fun=function(x,...)mean(x),
                      ) + theme_ridges() +
  scale_y_discrete(expand = c(0.1, 0)) +
  scale_x_continuous(expand = c(0.1, 0)) +
  labs(x = expression(paste("SLA (", mm^{2},mg^-1,")")),y = "", fill="") +
  xlim(0,42) +
  scale_fill_manual(values = c('lightblue','gray45','lemonchiffon1')) +
  theme_classic2()
print(fig2)

What I'd like mean lines to look like (one solid, one dashed):

Thanks!

[AlexisW]

Very partial answer.

Checking the source code of geom_density_ridge(), there is an entry for:

# vline aesthetics, all inherited
    vline_colour = NULL, #vline_color = NULL,
    vline_size = NULL, vline_linetype = NULL

And indeed, it seems to work:

library(ggridges)

combined_sla <- tibble(SLA = c(rnorm(20,15,20), rnorm(20,20,20),
                               rnorm(20,10,20), rnorm(20,22,20)),
                       Data = rep(c("California","South Africa"), each = 40),
                       Soil = rep(c("A","B"), each = 20) |> rep(times=2))


ggplot(combined_sla, aes(x=SLA,y=Data,fill=Soil, vline_linetype = Soil)) +
  geom_density_ridges(scale = 1, show.legend = TRUE, alpha = 0.4,
                      quantile_lines=TRUE, quantile_fun=function(x,...)mean(x),
  ) + theme_ridges() +
  scale_y_discrete(expand = c(0.1, 0)) +
  scale_x_continuous(expand = c(0.1, 0)) +
  labs(x = expression(paste("SLA (", mm^{2},mg^-1,")")),y = "", fill="") +
  xlim(0,42) +
  scale_fill_manual(values = c('lightblue','gray45','lemonchiffon1'))

However:

• I'm not sure the package authors mean to expose this to users (as it's not in the documentation), so it could stop working in future versions
• I can't think of a way to change the linetype scale (to use e.g. dashed and dotted)

Another way is defining these lines in a different stat/geom:

combined_sla_means <- combined_sla |>
  group_by(Data,Soil) |>
  summarize(mean = mean(SLA),
            .groups = "drop")

ggplot(combined_sla) +
  geom_density_ridges(aes(x=SLA,y=Data,fill=Soil),
                      scale = 1, show.legend = TRUE, alpha = 0.4,
                      quantile_lines=FALSE) +
  geom_vline(aes(xintercept = mean, group = Data, linetype = Soil),
             data = combined_sla_means) +
  theme_ridges() +
  scale_y_discrete(expand = c(0.1, 0)) +
  scale_x_continuous(expand = c(0.1, 0)) +
  labs(x = expression(paste("SLA (", mm^{2},mg^-1,")")),y = "", fill="") +
  xlim(0,42) +
  scale_fill_manual(values = c('lightblue','gray45','lemonchiffon1'))

But the vertical lines span all the Data locations, so probably not what you want.

[tjsamo]

Awesome! That works pretty well. The only issue is there are these two keys. Can you think of a way to combine the right line types with the right fill key? See below:

[AlexisW]

Actually at this point I feel it's easier to just forgo geom_density_ridge() and use a standard geom_density() with the appropriate faceting:

library(tidyverse)

combined_sla <- tibble(SLA = c(rnorm(20,15,20), rnorm(20,20,20), rnorm(20,20,20),
                               rnorm(20,15,20), rnorm(20,20,20), rnorm(20,20,20),
                               rnorm(20,10,20), rnorm(20,22,20), rnorm(20,20,20)),
                       Data = rep(c("California","South Africa","Puerto_Rico"), each = 60),
                       Soil = rep(c("BIEN","NU","U"), each = 20) |> rep(times=3))


combined_sla_means <- combined_sla |>
  group_by(Data,Soil) |>
  summarize(mean = mean(SLA),
            .groups = "drop")

ggplot(combined_sla) +
  geom_density(aes(x=SLA,fill=Soil),
                      alpha = 0.4) +
  geom_vline(aes(xintercept = mean, group = Data, linetype = Soil),
             data = combined_sla_means) +
  facet_grid(rows = vars(Data)) +
  theme_minimal()

^{Created on 2022-07-06 by the reprex package (v2.0.1)}