The purpose of this exercise is to tour the table and graphics capabilities of R, and to explore methods for displaying patterns in data. If you need help with some of the commands, check the “Graphs & Tables” tab at the R tips page.
You can draw plots using commands in base R or use
ggplot()
from the tidyverse. Each option has passionate
adherents and it is pointless to argue. Try both and see which you find
most appealing and easiest to use.
These data were published as a data paper in Ecology and deposited in the Ecological Archives (F. A. Smith, S. K. Lyons, S. K. M. Ernest, K. E. Jones, D. M. Kaufman, T. Dayan, P. A. Marquet, J. H. Brown, and J. P. Haskell. 2003. Body mass of late Quaternary mammals. Ecology 84: 3403.) See the metadata for a description.
Most of the variables are categorical, with multiple named
categories. continent
includes mammals on islands
(“Insular” category) whereas “Oceanic” refers to marine mammals. Body
mass (in grams) is the sole numeric variable. The status
variable indicates whether species is currently present in the wild
(extant), extinct as of late Pleistocene (extinct), extinct within the
last 300 years (historical), or an introduced species
(introduction).
The original data were saved in mammals.csv file on our server here. Download the file to your computer and open in a spreadsheet program (e.g., Excel, Calc) to have a look at it.
Start R and read the contents of the file to a data frame. You will
need to modify the default argument that identifies missing data to
na.strings=““
(or na=”“
if you are using
read_csv()
from the readr
package) because in
this data file”NA” is used to symbolize North America in the
continent
column rather than missing data (don’t do this in
your own data).
Use the head()
function to view the first few lines of
the data frame on the screen. You’ll see that every row represents the
data for a different mammal species.
Which continent has the greatest number of mammal species? Which has the least? Make a table of the frequency of cases on each continent (remember that the category “NA” in continent stands for North America, not missing data).
You’ll notice in the frequency table for the variable
continent
that there’s a typo in the data. One case is
shown as having the continent “Af” rather than “AF”. Fix this using the
command line in R and recalculate the frequency table.
How many extinct mammals are recorded in the data file? Use a frequency table to find out.
Create a two-way frequency table (contingency table) showing the status of mammal species on each continent.
Judging by eye, which continent has the greatest number of extinctions relative to the number of extant species?
All lines below beginning with double hashes are R output
# Load the packages you might need
library(readr)
library(dplyr)
# Read and inspect the data
mammals <- read.csv(url("https://www.zoology.ubc.ca/~bio501/R/data/mammals.csv"),
na.strings="", stringsAsFactors = FALSE)
head(mammals)
## continent status order family genus species mass.grams
## 1 AF extant Artiodactyla Bovidae Addax nasomaculatus 70000.3
## 2 AF extant Artiodactyla Bovidae Aepyceros melampus 52500.1
## 3 AF extant Artiodactyla Bovidae Alcelaphus buselaphus 171001.5
## 4 AF extant Artiodactyla Bovidae Ammodorcas clarkei 28049.8
## 5 AF extant Artiodactyla Bovidae Ammotragus lervia 48000.0
## 6 AF extant Artiodactyla Bovidae Antidorcas marsupialis 39049.9
You could also use read_csv()
from the readr package
(not run).
mammals <- read_csv(url("https://www.zoology.ubc.ca/~bio501/R/data/mammals.csv"),
na = c(""))
Number of mammal species on each continent
table(mammals$continent)
##
## Af AF AUS EA Insular NA Oceanic SA
## 1 1033 346 1033 1484 779 78 977
# Fix "Af"
which(mammals$continent=="Af")
## [1] 322
mammals$continent[322]<-"AF"
table(mammals$continent)
##
## AF AUS EA Insular NA Oceanic SA
## 1034 346 1033 1484 779 78 977
How many extinct mammals? The table shows that 242 species of mammal are listed as extinct.
z <- table(mammals$status)
z
##
## extant extinct historical introduction
## 5388 242 84 17
Extinction status by continent (contingency table). Australia (AUS) seems to have the greatest number of extinct species relative to the total number. This might be easier to see if the row sums are included in the table.
# base R
mytab <- table(mammals$continent, mammals$status)
addmargins(mytab, margin = c(1,2), FUN = sum, quiet = TRUE)
##
## extant extinct historical introduction sum
## AF 1017 13 4 0 1034
## AUS 261 45 23 17 346
## EA 1027 0 6 0 1033
## Insular 1405 29 50 0 1484
## NA 700 78 1 0 779
## Oceanic 78 0 0 0 78
## SA 900 77 0 0 977
## sum 5388 242 84 17 5731
You can also use dplyr
and tidyr
to make a
contingency table (not run). The last step is optional and converts NA
to zero.
# dplyr and tidyr method
library(tidyverse)
mytab <- summarize(group_by(mammals, continent, status), n = n())
mytab <- spread(mytab, status, n)
mutate(mytab, across(everything(), ~replace_na(., 0)))
## # A tibble: 7 × 5
## # Groups: continent [7]
## continent extant extinct historical introduction
## <chr> <int> <int> <int> <int>
## 1 AF 1017 13 4 0
## 2 AUS 261 45 23 17
## 3 EA 1027 0 6 0
## 4 Insular 1405 29 50 0
## 5 NA 700 78 1 0
## 6 Oceanic 78 0 0 0
## 7 SA 900 77 0 0
Plot the number of mammal species on each continent using a simple bar graph. Include a label for the y axis.
The plot categories are listed in alphabetical order by default, which is arbitrary and makes the visual display less efficient than other possibilities. Redo the bar graph with the continents appearing in order of decreasing numbers of species.
Generate a histogram of the body masses of mammal species. How informative is that?!
Create a new variable in the mammal data frame: the log (base 10) of body mass. (See “Transform” on the R tips “Data” page if you need help with this.)
Generate a histogram of log body mass. Is this more informative? Morphological data commonly require a log-transformation to analyze.
Redo the previous histogram but use a bin width of 2 units. How much detail is lost?
Redo the histogram but try a bin width of of 1; then try 0.5; and then 0.1. Which bin width is superior?
Redo the histogram, but display probability density instead of frequency.
How does the frequency distribution of log body mass depart from a normal distribution? Answer by visual examination of the histogram you just created. Now answer by examining a normal quantile plot instead. Which display is more informative? Do the data conform to a normal distribution?
Optional: redraw the histogram of log body mass and superimpose a normal density curve to help visualize deviations from normality. In what ways do the data depart from normality?
All lines below beginning with double hashes are R output
# Bar plot of mammal species by continent
barplot(table(mammals$continent), col="firebrick", cex.names=0.8,
ylim=c(0,1600), las = 1)
# Barplot sorted by frequency
barplot(sort(table(mammals$continent), decreasing=TRUE), col="firebrick",
cex.names=0.8, las = 1, ylim=c(0,1600), ylab="Frequency")
Alternatively, use ggplot()
instead of base R.
library(ggplot2)
ggplot(mammals, aes(x = continent)) +
geom_bar(stat = "count", fill = "firebrick") +
labs(x = "Continent", y = "Frequency") +
theme_classic()
# To order by category in ggplot, first make a new factor variable
mammals$continent_ordered <- factor(mammals$continent,
levels = names(sort(table(mammals$continent), decreasing = TRUE)) )
ggplot(mammals, aes(x = continent_ordered)) +
geom_bar(stat = "count", fill = "firebrick") +
labs(x = "Continent", y = "Frequency") +
theme_classic()
Histogram of body masses. Results with different bin widths not shown
hist(mammals$mass.grams, col="firebrick", right = FALSE, las = 1,
xlab = "Body mass (g)", main = "")
# Add a new variable, log10 of body mass
mammals$logmass <- log10(mammals$mass.grams)
hist(mammals$logmass, col="firebrick", right = FALSE, las = 1,
xlab = "Log10 body mass", main = "", breaks = seq(0, 8.5, by = 0.5))
Same but using ggplot()
. You’ll see a
Warning: Removed ?? rows containing non-finite values
.
These are rows with missing data for mass. Use the argument
na.rm = TRUE
in geom_histogram() to get rid of the
warning.
ggplot(mammals, aes(x = logmass)) +
geom_histogram(fill = "firebrick", col = "black", binwidth = 0.5, boundary = 0) +
labs(x = "log10 body mass", y = "Frequency") +
theme_classic()
## Warning: Removed 1372 rows containing non-finite values (`stat_bin()`).
Plot density instead
hist(mammals$logmass, col = "firebrick", right = FALSE, las = 1, prob = TRUE,
xlab = "Log10 body mass", main = "", breaks = seq(0, 8.5, by = 0.5))
or
ggplot(mammals, aes(x = logmass)) +
geom_histogram(fill = "firebrick", col = "black", binwidth = 0.5,
boundary = 0, aes(y = after_stat(density))) +
labs(x = "log10 body mass", y = "Density") +
theme_classic()
## Warning: Removed 1372 rows containing non-finite values (`stat_bin()`).
Normal quantile plot.
qqnorm(mammals$logmass)
qqline(mammals$logmass) # adds the straight line for comparison through 1st and 3rd quartiles
Histogram with best-fit normal curve superimposed.
# The curve function is fussy about the name of the variable: must be "x"
x <- mammals$logmass
hist(x, col="firebrick", right = FALSE, las = 1, prob = TRUE,
xlab = "Log10 body mass", main = "", breaks = seq(0, 8.5, by = 0.5))
m <- mean(x, na.rm = TRUE)
s <- sd(x, na.rm = TRUE)
curve(dnorm(x, mean = m, sd = s), col="red", lwd = 2, add = TRUE)
Use a box plot to compare the distribution of body sizes (log scale most revealing) of mammals having different extinction status. Are extinct mammals similar to, larger than, or smaller than, extant mammals?
Examine the previous box plot. How do the shapes of the body size distributions compare between extinct and extant mammals?
Redo the previous box plot but make box width proportional to the square root of sample size. Add a title to the plot.
Optional: Draw a violin plot to compare the frequency distribution of log body sizes of mammals having different extinction status. Which do you find is more revealing about the shapes of the body size distributions: box plot or violin plot?
Use multiple histograms to compare the frequency distribution of log body sizes of mammals having different extinction status. Stack the panels one above the other. In this plot, how easy is it to visualize differences among treatments in the distributions compared to your previous plots?
Make a table of the median log body mass of each extinction-status group of mammals. Are the values consistent with the plotted distributions?
All lines below beginning with double hashes are R output
The graphs show that Extinct mammals tend to have large mass compared to extant mammals. The frequency distributions for these two groups also have opposite skew, with extinct mammals having left skew.
# Base R box plot to compare the distribution of body sizes
boxplot(logmass ~ status, data = mammals, ylab = "log10 body mass",
col = "goldenrod1", las = 1)
or
ggplot(mammals, aes(x = status, y = logmass)) +
geom_boxplot(fill = "goldenrod1", notch = FALSE) +
labs(x = "Status", y = "Log10 body mass") +
theme_classic()
## Warning: Removed 1372 rows containing non-finite values (`stat_boxplot()`).
# Violin plot
ggplot(mammals, aes(x = status, y = logmass)) +
geom_violin(fill = "goldenrod1") +
labs(x = "Status", y = "Log10 body mass") +
stat_summary(fun = mean, geom = "point", color = "black") +
theme_classic()
## Warning: Removed 1372 rows containing non-finite values (`stat_ydensity()`).
## Warning: Removed 1372 rows containing non-finite values (`stat_summary()`).
# Multiple histograms
ggplot(mammals, aes(x = logmass)) +
geom_histogram(fill = "goldenrod1", col = "black",
binwidth = 0.2, boundary = 0) +
labs(x = "log10 body mass", y = "Frequency") +
facet_wrap(~status, ncol = 1, scales = "free_y", strip.position = "right") +
theme_classic()
## Warning: Removed 1372 rows containing non-finite values (`stat_bin()`).
The data are from L. Partridge and M. Farquhar (1981), Sexual
activity and the lifespan of male fruitflies,
Download the file to your computer and open in a spreadsheet program to have a look at it. View the first few lines of the data frame on the screen, and familiarize yourself with the variable names.
Our goal here is to find a plot type that clearly and efficiently visualizes the patterns in the data, especially the differences among groups.
Read the data file into a new data frame.
Use a strip chart to examine the distribution of longevities in the treatment groups. Try the jitter method to reduce overlap between points. If needed, adjust the size or rotation of the treatment labels so that they all fit on the graph. What pattern of differences between treatments in longevity is revealed? Males from which treatments have the highest longevity? Which have the lowest longevity?
Compare the strip chart to a box plot of the same data. Is the pattern in the data as clear in both types of plot?
The variable thorax
stands for thorax length, which
was used as a measure of body size. The measurement was included in case
body size also affected longevity. Using ggplot()
, produce
a scatter plot of thorax length and longevity. Make
longevity
the response variable (i.e., plot it on the
vertical axis). Is there a relationship?
Redraw the scatter plot using ggplot()
but this time
use different symbols and/or colors for the different treatment groups.
Add a legend to identify the symbols. After controlling for differences
among males in size, males from which treatments have the highest
longevity on average? Which have the lowest longevity?
Redraw, adding regression lines to your figure, separately for each group.
You can see how it can be fiendishly difficult to build a clean
visual showing the pattern in the data when there are multiple groups.
Redraw the figure again using just one color and symbol, but this time
use facet_wrap()
to plot the data in multiple panels, one
per treatment group. Compare your results with those from (6). Which
method shows the differences between groups most clearly?
All lines below beginning with double hashes are R output
# Read and inspect data
x <- read.csv(url("https://www.zoology.ubc.ca/~bio501/R/data/fruitflies.csv"),
stringsAsFactors = FALSE)
head(x)
## Npartners treatment longevity.days thorax.mm
## 1 8 8 pregnant females 35 0.64
## 2 8 8 pregnant females 37 0.68
## 3 8 8 pregnant females 49 0.68
## 4 8 8 pregnant females 46 0.72
## 5 8 8 pregnant females 63 0.72
## 6 8 8 pregnant females 39 0.76
Strip chart in base R.
stripchart(longevity.days ~ treatment, data = x, vertical = TRUE,
method = "jitter", pch=16, col = "firebrick", cex.axis=0.7,
ylab="Longevity (days)")
or using ggplot()
# Strip chart using ggplot
ggplot(x, aes(x = treatment, y = longevity.days)) +
geom_jitter(color = "firebrick", size = 3, width = 0.15) +
labs(x = "Treatment", y = "Longevity (days)") +
theme_classic()
Box plot.
boxplot(longevity.days ~ treatment, data = x, cex.axis = 0.7,
ylab = "Longevity (days)", boxwex = 0.5, col = "goldenrod1")
or using ggplot()
ggplot(x, aes(x = treatment, y = longevity.days)) +
geom_boxplot(fill = "goldenrod1", width = 0.5) +
labs(x = "Treatment", y = "Longevity (days)") +
theme_classic()
Scatter plot in base R.
plot(longevity.days ~ thorax.mm, data = x, pch = 16, col = "firebrick", las = 1,
xlab = "Thorax length (mm)", ylab = "Longevity (days)")
or using ggplot()
# Scatter plot with ggplot
ggplot(x, aes(x = thorax.mm, y = longevity.days)) +
geom_point(size = 3, col = "firebrick") +
labs(x = "Thorax length (mm)", y = "Longevity (days)") +
theme_classic()
Scatter plot with separate colors for each group using
ggplot()
ggplot(x, aes(x = thorax.mm, y = longevity.days, colour = treatment,
shape = treatment)) +
geom_point(size = 2) +
labs(x = "Thorax length (mm)", y = "Longevity (days)") +
theme_classic()
# Add lines
ggplot(x, aes(x=thorax.mm, y=longevity.days, colour = treatment,
shape = treatment)) +
geom_point(size = 2) +
geom_smooth(method = lm, linewidth = 1, se = FALSE) +
labs(x = "Thorax length (mm)", y = "Longevity (days)") +
theme_classic()
## `geom_smooth()` using formula = 'y ~ x'
Using facet_wrap()
to plot treatment groups in separate
panels.
ggplot(x, aes(x = thorax.mm, y = longevity.days)) +
facet_wrap(~ treatment, strip.position = "top") +
geom_point(size = 2) +
geom_smooth(method = lm, linewidth = 1, se = FALSE) +
labs(x = "Thorax length (mm)", y = "Longevity (days)") +
theme_classic()
## `geom_smooth()` using formula = 'y ~ x'
Here’s how you would draw a scatter plot with separate colors and symbols for each group using base R.
plot(longevity.days ~ thorax.mm, data=x, pch=as.numeric(factor(treatment)),
col = as.numeric(factor(treatment)), las = 1,
xlab = "Thorax length (mm)", ylab = "Longevity (days)")
legend( locator(1), legend = as.character(levels(factor(x$treatment))),
pch=1:length(levels(factor(x$treatment))),
col=1:length(levels(factor(x$treatment))) )
© 2009-2024 Dolph Schluter