How to Control How Inset R Plot Legend Is

Question

I'm trying to force my legend to be in lower right corner AND flush against both axes. Right now, the legend defaults to being further inset into the plot than I want it to be (around the middle of the x-axis). I've poked around here and on R documentation and discovered that "inset=" is an optional argument that can be used in the legend function. I THINK this is how I can accomplish what I want. However, when I try only entering one numerical value (inset= 0), the legend ends up in the same place as the default. When I try to enter two numerical arguments (inset= c(0,-2)), the legend won't show up at all. I thought maybe I was using improper syntax and tried inset= 0:-2, but that didn't work either. I've tried playing around with higher values and that causes the legend to become further inset, so I know it's the right argument, but maybe I'm not giving it the right values? I've also tried changing the values for the "xjust=" and "yjust=" arguments and those don't seem to have any notable affect. Does anyone have advice on what I might be doing wrong? Attached are both my code and my current plot output (and arrows showing where I WANT the legend to be):

layout(matrix(c(2,1), 1, 2, byrow = FALSE))
par(pty="s", mar= c(1,3,1,3), mgp=c(1.5,0.5,0))

plot(x1, y1, 
     xlab="Total X1", 
     ylab="Total X2", 
     type= "n",
     ylim = c(-3,2),
     xlim = c(-2.5,2.5),
     cex.lab= 0.7,
     cex.axis= 0.7)
text(x1, y1, labels = SedCombined2$Site,
     col= c("#D55E00", "#000000", "#009E73", "#0072B2")[as.factor(SedCombined2$Event)],
     cex=0.7,
     font = 2)
legend("bottomright",
       border = "black",
       bty= "n",
       legend= c("November 2014", "April 2015","November 2015","April 2016"),
       text.col = c("#D55E00", "#000000", "#009E73", "#0072B2"),
       cex=0.6,
       x.intersp = 0.5,
       y.intersp = 0.5,
       inset = 0)

x2 <- fitSedCombinedISO$points[,1]
y2 <- fitSedCombinedISO$points[,2]
plot(x2, y2,
     xlab="Toxic X1", 
     ylab="Toxic X2", 
     type= "n",
     ylim = c(-3,2),
     xlim = c(-2.5,2.5),
     cex.lab= 0.7,
     cex.axis= 0.7)
text(x2, y2, labels = SedCombined2$Site,
     col= c("#D55E00", "#000000", "#009E73", "#0072B2")[as.factor(SedCombined2$Event)],
     cex=0.7,
     font = 2)
legend("bottomright",
       border = "black",
       bty= "n",
       legend= c("November 2014", "April 2015","November 2015","April 2016"),
       text.col = c("#D55E00", "#000000", "#009E73", "#0072B2"),
       cex=0.6,
       x.intersp = 0.5,
       y.intersp = 0.5,
       inset = 0)

Current Plot Output

Welcome to SO! Please read [ask] and https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example to make your example reproducible. — jogo, Sep 26 '18 at 19:38
Inset with an argument of -2 would be placing completely out of the plot area. It's supposed to be the fraction of the plotregion, so values above 1 or below -1 wouldn't leave it in the plot region. — IRTFM, Sep 26 '18 at 21:00
Yes, as @42 said: use a fractional value for the inset. I can't tell you which one to use, because your code isn't reproducible here, but start with something like `inset = c(-0.1, 0)`. — user2554330, Sep 26 '18 at 22:10
@jogo Apologies! I'll shortly rewrite the sample code so that it adheres to the "shortest code necessary" and so that it's more reproducible. — Christine K., Sep 27 '18 at 18:37

Josh Rumbut · Answer 1 · 2018-09-26T20:54:08.307

This requires a little hack, assign your legend to a variable, and then use that to create a text box:

temp <- legend("bottomright",
       border = "black",
       bty= "n",
       legend=c('','','',''),
       text.width=max(strwidth(c("November 2014", "April 2015","November 2015","April 2016"))),
       text.col = c("#D55E00", "#000000", "#009E73", "#0072B2"),     
       xjust=1, yjust=1)

text(
    temp$rect$left + temp$rect$w, 
    temp$text$y, 
    c("November 2014", "April 2015","November 2015","April 2016"), 
    pos=2, 
    c("#D55E00", "#000000", "#009E73", "#0072B2"))

As a general point, if this level of control over plot appearance is important, I would highly recommend ggplot.

How to Control How Inset R Plot Legend Is

1 Answers1