R help - Mar 2012 - Boxplot Fill Pattern

Hello R Help!

I would like to make a legible boxplot of tree growth rates for each of
seven tree species at each of seven different sites. It's a lot of data to
put on one figure, I know. I made a beautiful, interpretable figure using
color, but my target journal can't deal with color figures. I can use seven
shades of grey to fill the boxes, but the figure then becomes
uninterpretable - the colors of adjacent boxes are too similar. I cannot
figure out how to add a pattern, hatching, or cross-hatching to the boxes.

I see that in 2003 Professor Ripley confirmed that one cannot hatch
boxplots:

http://tolstoy.newcastle.edu.au/R/help/03a/2622.html

Is this still true? If so, does anyone have a suggestion for how to make
this figure interpretable in black and white? Should I pick a different
target journal?

Many thanks,

Gabe

Functional code to make the color figure with fake data follows:
#
leg.txt = c("Abies grandis","Acer
macrophyllum","Calocedrus
decurrens","Pinus ponderosa","Pseudotsuga
meziensii","Quercus
garryana","Quercus kelloggii")
site.txt = c("Brownsville","Chip
Ross","Finley","Jim's
Creek","Lowell","Mount Pisgah","South
Eugene")
colors =
c("gray","red","white","blue","yellow","purple","orange")
# Fake data here:
site = rep(site.txt, each = 21)
sp = rep(rep(leg.txt, each = 3), times = 7)
ga = runif(147, 0, 20)
datnew.lo = data.frame(site,sp,ga)
# Now make the plot:
boxplot(ga~sp*site,data=datnew.lo, range = 1,
        col = colors,
        ylim = c(0,30),
        xaxt = "n",
        xlab = "Site",
        ylab = "Basal Area Growth Increment",
        main = "Basal Area Growth Increment by Site and Species")
axis(1, at = c(4,11,18,25,32,39,46),
     labels = site.txt,
     )
abline(v = 7.5, lty = 3)
abline(v = 14.5, lty = 3)
abline(v = 21.5, lty = 3)
abline(v = 28.5, lty = 3)
abline(v = 35.5, lty = 3)
abline(v = 42.5, lty = 3)
legend("topright", legend = leg.txt, fill = colors, bg =
"white")

	[[alternative HTML version deleted]]

On Thu, Mar 8, 2012 at 1:08 PM, Gabriel Yospin <yosping at gmail.com>
wrote:
> Hello R Help!
Hello Gabe Yospin! (I feel like I should start playing some arena rock
anthem now ;-) )
>
> I would like to make a legible boxplot of tree growth rates for each of
> seven tree species at each of seven different sites. It's a lot of data
to
> put on one figure, I know. I made a beautiful, interpretable figure using
> color, but my target journal can't deal with color figures. I can use
seven
> shades of grey to fill the boxes, but the figure then becomes
> uninterpretable - the colors of adjacent boxes are too similar. I cannot
> figure out how to add a pattern, hatching, or cross-hatching to the boxes.
>
> I see that in 2003 Professor Ripley confirmed that one cannot hatch
> boxplots:
>
> http://tolstoy.newcastle.edu.au/R/help/03a/2622.html
>
> Is this still true? If so, does anyone have a suggestion for how to make
> this figure interpretable in black and white? Should I pick a different
> target journal?
>
I don't think you'd need to change journals just for graphical styles:
do it for much less important things, like impact factor. It sounds
like the idea of small multiples might help here: I'm not a lattice
pro, but here's something you could do in ggplot2 (and I know it's
doable in lattice as well):

I'm gonna put new lines in all your species names since space will be
at a premium:

levels(datnew.lo$sp) <- gsub(" ", "\n",
levels(datnew.lo$sp))

library(ggplot2)
ggplot(datnew.lo, aes(x = sp, y = ga)) + geom_boxplot() +
facet_wrap(~site, nrow = 2) + opts(axis.text.x=theme_text(angle=45,
size = 7))

Others with more ggplot / lattice - fu than I can help you tweak this
but hopefully this is a start.

Michael

> Many thanks,
>
> Gabe
>
> Functional code to make the color figure with fake data follows:
> #
> leg.txt = c("Abies grandis","Acer
macrophyllum","Calocedrus
> decurrens","Pinus ponderosa","Pseudotsuga
meziensii","Quercus
> garryana","Quercus kelloggii")
> site.txt = c("Brownsville","Chip
Ross","Finley","Jim's
> Creek","Lowell","Mount Pisgah","South
Eugene")
> colors =
c("gray","red","white","blue","yellow","purple","orange")
> # Fake data here:
> site = rep(site.txt, each = 21)
> sp = rep(rep(leg.txt, each = 3), times = 7)
> ga = runif(147, 0, 20)
> datnew.lo = data.frame(site,sp,ga)
> # Now make the plot:
> boxplot(ga~sp*site,data=datnew.lo, range = 1,
> ? ? ? ?col = colors,
> ? ? ? ?ylim = c(0,30),
> ? ? ? ?xaxt = "n",
> ? ? ? ?xlab = "Site",
> ? ? ? ?ylab = "Basal Area Growth Increment",
> ? ? ? ?main = "Basal Area Growth Increment by Site and Species")
> axis(1, at = c(4,11,18,25,32,39,46),
> ? ? labels = site.txt,
> ? ? )
> abline(v = 7.5, lty = 3)
> abline(v = 14.5, lty = 3)
> abline(v = 21.5, lty = 3)
> abline(v = 28.5, lty = 3)
> abline(v = 35.5, lty = 3)
> abline(v = 42.5, lty = 3)
> legend("topright", legend = leg.txt, fill = colors, bg =
"white")
Thanks for the great reproducible example! Made this much easier and more fun.
>
> ? ? ? ?[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

On 3/8/2012 1:08 PM, Gabriel Yospin wrote:
> I would like to make a legible boxplot of tree growth rates for each of
> seven tree species at each of seven different sites. It's a lot of data
to
> put on one figure, I know. I made a beautiful, interpretable figure using
> color, but my target journal can't deal with color figures. I can use
seven
> shades of grey to fill the boxes, but the figure then becomes
If you print your original figure (with the legend) in B/W, you'll see 
that the shading levels are not too bad -- they are relatively 
distinguishable except for Acer and Pinus/

Thus, one thing I often do in this situation is design a graph so that
it will render reasonably well also in B/W and suggest to the journal
to make a color version available online (or I put it on my own web).

Your final version using ggplot2 is a far worse graph, IMHO
because the labels are illegible, and using no fill in the boxplots
makes it impossible to distinguish the same species across sites.
You could
make it better by
(a) only labeling the species in alternate site panels
(b) rotating those labels by 45^o
(c) using some fill for the boxes

My .05

-- 
Michael Friendly     Email: friendly AT yorku DOT ca
Professor, Psychology Dept.
York University      Voice: 416 736-5115 x66249 Fax: 416 736-5814
4700 Keele Street    Web:   http://www.datavis.ca
Toronto, ONT  M3J 1P3 CANADA