I am trying to create a tile plot with ggplot in R with a multilevel label on the Y axis, and with only one copy of the x axis (e.g. no facets). The desired outcome is something like this (just imagine that the tiles are all there):

Here is what I have so far:

dat<- as.data.frame(cbind(Species = rep(c("a", "b", "c", "d", "e", "f", "g"),each = 3 ), 
                            threats = rep(c("1", "2", "3"), 7),
                            elev_grp = c(rep("Low", 9), rep("Mid", 9), rep("High", 3)), 
                            match = sample(1:3, size = 21, replace = T)))
dat$match.labs<- ifelse(dat$match == 1, "Both",
                        ifelse(dat$match == 2, "Neither", "One"))

ggplot(dat, aes(reorder(threats, match), Species, fill = match.labs)) 
  geom_tile() 
  scale_fill_manual(values = c("#bd421f",
                               "#3fadaf",
                               "#ffb030"))

That gives me this:

Faceting is inappropriate because every species has a "match" value for every level of "threats", and there are dozens of levels in my actual dataset, which makes repeating the x-axis over different facets of "elev_grp" impractical. I'd also considered stacked facets, but couldn't figure out how to force them to share one x axis so there aren't huge grey areas in the plot...

Any help would be appreciated. Thanks for your time!

CodePudding user response：

You can do the old "facet that doesn't look like a facet" trick:

dat$elev_grp <- factor(dat$elev_grp, c("High", "Mid", "Low"))

ggplot(dat, aes(reorder(threats, match), Species, fill = match.labs)) 
  geom_tile() 
  scale_fill_manual(values = c("#bd421f",
                               "#3fadaf",
                               "#ffb030"))  
  scale_y_discrete(drop = TRUE, expand = c(0, 0))  
  facet_grid(elev_grp~., scales = "free", space = "free_y", switch = "y")  
  theme(strip.placement = "outside",
        panel.spacing = unit(0, "in"),
        strip.background.y = element_rect(fill = "white", color = "gray75"))