Module 5: Data Visualization with ggplot2

Introduction

One of the most useful — and fun — things we can do with R is create graphics.

Data visualization can help us:

understand our data,
identify patterns,
detect unusual values,
compare groups,
explore relationships,
and communicate results.

In this module, we will use the package ggplot2.

ggplot2 is part of the tidyverse and uses a consistent system for building graphics by adding different layers.

Download the example data

We will use a camera-trap record table throughout this module.

Download recordTable.csv

Let’s get started

Open your intro-r-course project and create a new R script.

Save it inside the scripts folder as:

module-05.R

First, load the tidyverse package:

library(tidyverse)

Then import the dataset:

recordTable <- read_csv(
  "data/recordTable.csv"
)

Remember that we should always inspect a dataset after importing it.

head(recordTable)

# A tibble: 6 × 25
  RootFolder       File  RelativePath Station Camera DateTime            Species
  <chr>            <chr> <chr>        <chr>   <chr>  <dttm>              <chr>  
1 Field Station 2… DSCF… "Control 0\… Contro… Contr… 2025-05-15 14:53:20 White-…
2 Field Station 2… DSCF… "Control 1\… Contro… Contr… 2025-06-05 04:25:19 Armadi…
3 Field Station 2… DSCF… "Control 1\… Contro… Contr… 2025-06-26 07:39:44 Armadi…
4 Field Station 2… DSCF… "Control 1\… Contro… Contr… 2025-06-26 08:07:20 Armadi…
5 Field Station 2… DSCF… "Control 1\… Contro… Contr… 2025-06-26 08:39:46 Armadi…
6 Field Station 2… DSCF… "Control 1\… Contro… Contr… 2025-07-20 01:11:11 Armadi…
# ℹ 18 more variables: N_individuals <dbl>, Sex <chr>, Age <chr>,
#   Identifier <chr>, Comments <chr>, Favorite <lgl>, DateTime_original <dttm>,
#   year <dbl>, date_shift <dbl>, n_images <dbl>, Date <date>, Time <time>,
#   delta.time.secs <dbl>, delta.time.mins <dbl>, delta.time.hours <dbl>,
#   delta.time.days <dbl>, DateTime2 <dttm>, Date2 <date>

dim(recordTable)

[1] 3898   25

names(recordTable)

 [1] "RootFolder"        "File"              "RelativePath"     
 [4] "Station"           "Camera"            "DateTime"         
 [7] "Species"           "N_individuals"     "Sex"              
[10] "Age"               "Identifier"        "Comments"         
[13] "Favorite"          "DateTime_original" "year"             
[16] "date_shift"        "n_images"          "Date"             
[19] "Time"              "delta.time.secs"   "delta.time.mins"  
[22] "delta.time.hours"  "delta.time.days"   "DateTime2"        
[25] "Date2"

str(recordTable)

spc_tbl_ [3,898 × 25] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
 $ RootFolder       : chr [1:3898] "Field Station 2025" "Field Station 2025" "Field Station 2025" "Field Station 2025" ...
 $ File             : chr [1:3898] "DSCF0007.JPG" "DSCF0134.JPG" "DSCF0425.JPG" "DSCF0429.JPG" ...
 $ RelativePath     : chr [1:3898] "Control 0\\Control 0 A" "Control 1\\Control 1 B" "Control 1\\Control 1 B" "Control 1\\Control 1 B" ...
 $ Station          : chr [1:3898] "Control 0" "Control 1" "Control 1" "Control 1" ...
 $ Camera           : chr [1:3898] "Control 0 A" "Control 1 B" "Control 1 B" "Control 1 B" ...
 $ DateTime         : POSIXct[1:3898], format: "2025-05-15 14:53:20" "2025-06-05 04:25:19" ...
 $ Species          : chr [1:3898] "White-tailed deer" "Armadillo" "Armadillo" "Armadillo" ...
 $ N_individuals    : num [1:3898] 0 1 2 1 1 1 1 1 1 1 ...
 $ Sex              : chr [1:3898] NA "Unknown" "Unknown" "Unknown" ...
 $ Age              : chr [1:3898] "Unknown" "Adult" "Unknown" "Unknown" ...
 $ Identifier       : chr [1:3898] NA NA NA NA ...
 $ Comments         : chr [1:3898] NA NA NA NA ...
 $ Favorite         : logi [1:3898] FALSE FALSE FALSE FALSE FALSE FALSE ...
 $ DateTime_original: POSIXct[1:3898], format: "2025-05-15 14:53:20" "2025-03-08 04:25:19" ...
 $ year             : num [1:3898] 2025 2025 2025 2025 2025 ...
 $ date_shift       : num [1:3898] NA 89 89 89 89 89 89 89 89 89 ...
 $ n_images         : num [1:3898] 1 2 2 1 1 2 1 1 2 1 ...
 $ Date             : Date[1:3898], format: "2025-05-15" "2025-06-05" ...
 $ Time             : 'hms' num [1:3898] 14:53:20 04:25:19 07:39:44 08:07:20 ...
  ..- attr(*, "units")= chr "secs"
 $ delta.time.secs  : num [1:3898] 0 0 1826065 1656 1946 ...
 $ delta.time.mins  : num [1:3898] 0 0 30434.4 27.6 32.4 ...
 $ delta.time.hours : num [1:3898] 0 0 507.2 0.5 0.5 ...
 $ delta.time.days  : num [1:3898] 0 0 21.1 0 0 23.7 0 0 27 0 ...
 $ DateTime2        : POSIXct[1:3898], format: "2025-05-15 14:53:20" "2025-06-05 04:25:19" ...
 $ Date2            : Date[1:3898], format: "2025-05-15" "2025-06-05" ...
 - attr(*, "spec")=
  .. cols(
  ..   RootFolder = col_character(),
  ..   File = col_character(),
  ..   RelativePath = col_character(),
  ..   Station = col_character(),
  ..   Camera = col_character(),
  ..   DateTime = col_datetime(format = ""),
  ..   Species = col_character(),
  ..   N_individuals = col_double(),
  ..   Sex = col_character(),
  ..   Age = col_character(),
  ..   Identifier = col_character(),
  ..   Comments = col_character(),
  ..   Favorite = col_logical(),
  ..   DateTime_original = col_datetime(format = ""),
  ..   year = col_double(),
  ..   date_shift = col_double(),
  ..   n_images = col_double(),
  ..   Date = col_date(format = ""),
  ..   Time = col_time(format = ""),
  ..   delta.time.secs = col_double(),
  ..   delta.time.mins = col_double(),
  ..   delta.time.hours = col_double(),
  ..   delta.time.days = col_double(),
  ..   DateTime2 = col_datetime(format = ""),
  ..   Date2 = col_date(format = "")
  .. )
 - attr(*, "problems")=<pointer: 0x000001f2ca6ffd20>

For this module, some of the variables that will be especially useful are:

Station
Species
N_individuals
Age
n_images
Date
Time

The logic of ggplot2

Most graphics in ggplot2 follow the same general structure:

data
  +
aesthetics
  +
geometry
  +
additional layers

In R, this looks like:

ggplot(
  data = data,
  aes(x = variable)
) +
  geom_something()

The function:

ggplot()

creates the graphical object.

Inside:

aes()

we specify which variables will be represented visually.

Then we add a geometry, or geom, that determines how those variables will be displayed.

For example:

geom_bar()
geom_histogram()
geom_point()
geom_boxplot()
geom_violin()

Each geometry answers a slightly different type of question.

Geometry	Useful for
`geom_bar()`	Counts among categories
`geom_histogram()`	Distribution of a numeric variable
`geom_point()`	Relationship between two variables
`geom_boxplot()`	Comparing numeric distributions among groups
`geom_violin()`	Visualizing distributions among groups

I’ll show you the basics, but you can get creative

There are many more options depending on what you want to explore or communicate with your data.

If you want to go further, I recommend visiting these two websites:

R Charts & R Graph Gallery

We will explore each of these using our camera-trap data.

Bar plots

Let’s begin with a simple question:

How many records do we have for each species?

Our variable Species contains the species identified in each record.

We can create a bar plot using:

ggplot(
  data = recordTable, # data
  aes(x = Species) #x axis variabe
) +
  geom_bar() # geometry

geom_bar() automatically counts how many observations occur in each category.

Because we have several species, the labels are difficult to read.

We can rotate the plot using:

ggplot(
  data = recordTable,
  aes(x = Species)
) +
  geom_bar() +
  coord_flip() # Rotate the axis

Now each bar represents the number of records for one species.

Adding color

We can change the appearance of the bars by adding a fill argument:

ggplot(
  data = recordTable,
  aes(x = Species)
) +
  geom_bar(fill = "steelblue") +
  coord_flip()

In this case:

fill = "steelblue"

does not represent a variable.

We are simply telling ggplot2:

Make all the bars this color.

Labels

A graph should tell the reader what they are looking at.

We can add labels using labs():

ggplot(
  data = recordTable,
  aes(x = Species)
) +
  geom_bar(fill = "steelblue") +
  coord_flip() +
  labs(
    title = "Camera-trap records by species",
    x = "Species",
    y = "Number of records"
  )

We can also add a theme:

ggplot(
  data = recordTable,
  aes(x = Species)
) +
  geom_bar(fill = "steelblue") +
  coord_flip() +
  labs(
    title = "Camera-trap records by species",
    x = "Species",
    y = "Number of records"
  ) +
  theme_bw()

Themes

Themes control the general appearance of a plot.

For example:

theme_bw()

theme_classic()

theme_minimal()

Try replacing theme_bw() with the other themes and see which one you prefer.

Tip

There is no single correct theme.

The most important thing is that the graph is clear and easy to interpret.

Histograms

A histogram helps us visualize the distribution of a numeric variable.

Our dataset contains:

n_images

which represents the number of images associated with a record.

Let’s look at its distribution:

ggplot(
  data = recordTable,
  aes(x = n_images)
) +
  geom_histogram()

`stat_bin()` using `bins = 30`. Pick better value `binwidth`.

By default, ggplot2 chooses the number of bins automatically.

We can control this ourselves.

For example:

ggplot(
  data = recordTable,
  aes(x = n_images)
) +
  geom_histogram(
    binwidth = 1,
    fill = "steelblue",
    color = "white"
  ) +
  labs(
    title = "Number of images per record",
    x = "Number of images",
    y = "Frequency"
  ) +
  theme_bw()

The dataset contains a few records with many images, so most observations are concentrated on the left side of the plot.

We can zoom into the first 20 images:

ggplot(
  data = recordTable,
  aes(x = n_images)
) +
  geom_histogram(
    binwidth = 1,
    fill = "steelblue",
    color = "white"
  ) +
  coord_cartesian(
    xlim = c(1, 20)
  ) +
  labs(
    title = "Number of images per record",
    x = "Number of images",
    y = "Frequency"
  ) +
  theme_bw()

Note

Changing the number or width of the bins can change how a histogram looks.

Always experiment with different values before interpreting the pattern.

Aesthetics: mapping versus setting

One of the most important ideas in ggplot2 is the difference between mapping an aesthetic to a variable and setting an aesthetic manually.

For example:

geom_point(color = "steelblue")

means:

Make every point steel blue.

But:

geom_point(aes(color = Species))

means:

Use different colors to represent different species.

The first is a setting.

The second is a mapping.

The same idea applies to:

color
fill
shape
size
alpha

Selecting some species

Our dataset contains many species.

For some graphics, it will be easier to focus on a few common species.

Let’s create a vector containing five species:

common_species <- c(
  "White-tailed deer",
  "Armadillo",
  "Racoon",
  "Hog",
  "Virginia opossum"
)

Remember what we learned about logical operators and selecting rows?

We can use:

common_records <- recordTable[
  recordTable$Species %in% common_species,
]

Now check:

unique(common_records$Species)

[1] "White-tailed deer" "Armadillo"         "Hog"              
[4] "Racoon"            "Virginia opossum"

We have created a smaller dataset containing only those species.

Note

We are using base R selection here because we already learned how logical selection works.

Later in the course, we will learn easier ways to filter datasets using dplyr.

Scatterplots

Scatterplots are useful for exploring the relationship between two numeric variables.

For example, we can ask:

Is the number of individuals in a record related to the number of images taken?

We can visualize this using:

ggplot(
  data = common_records,
  aes(
    x = N_individuals,
    y = n_images
  )
) +
  geom_point()

Now let’s map species to color:

ggplot(
  data = common_records,
  aes(
    x = N_individuals,
    y = n_images,
    color = Species
  )
) +
  geom_point() +
  labs(
    title = "Number of individuals and images per record",
    x = "Number of individuals",
    y = "Number of images",
    color = "Species"
  ) +
  theme_bw()

We can make points partially transparent using alpha:

ggplot(
  data = common_records,
  aes(
    x = N_individuals,
    y = n_images,
    color = Species
  )
) +
  geom_point(
    alpha = 0.4,
    size = 2
  ) +
  labs(
    title = "Number of individuals and images per record",
    x = "Number of individuals",
    y = "Number of images",
    color = "Species"
  ) +
  theme_bw()

Lower values of alpha make points more transparent.

This can be useful when many observations overlap.

Shapes

Categorical variables can also be represented using different shapes.

For example:

ggplot(
  data = common_records,
  aes(
    x = N_individuals,
    y = n_images,
    shape = Age
  )
) +
  geom_point(
    size = 3,
    alpha = 0.5
  ) +
  labs(
    title = "Number of individuals and images per record",
    x = "Number of individuals",
    y = "Number of images",
    shape = "Age"
  ) +
  theme_bw()

Warning: Removed 45 rows containing missing values or values outside the scale range
(`geom_point()`).

Here:

shape = Age

is inside aes() because the shape of each point depends on a variable.

Boxplots

A boxplot is useful for comparing the distribution of a numeric variable among groups.

For this example, let’s first calculate the number of records of each species at each camera station:

sp_counts <- recordTable |> 
  count(Station, Species) |> 
  filter(Species %in% common_species)

Now we can compare how the number of records per station varies among species:

ggplot(
  data = sp_counts,
  aes(
    x = Species,
    y = n
  )
) +
  geom_boxplot() +
  scale_y_log10() +
  coord_flip()

Because the number of records varies considerably among stations, we are displaying the y-axis on a logarithmic scale.

The original values in sp_counts$n are not modified; we are only changing how they are displayed in the graph.

We can also map species to fill:

ggplot(
  data = sp_counts,
  aes(
    x = Species,
    y = n,
    fill = Species
  )
) +
  geom_boxplot(
    show.legend = FALSE
  ) +
  scale_y_log10() +
  coord_flip() +
  labs(
    title = "Camera-trap records per station",
    x = "Species",
    y = "Number of records per station"
  ) +
  theme_bw()

Violin plots

A violin plot is another way of visualizing the distribution of a numeric variable among groups.

ggplot(
  data = sp_counts,
  aes(
    x = Species,
    y = n,
    fill = Species
  )
) +
  geom_violin(
    show.legend = FALSE
  ) +
  scale_y_log10() +
  coord_flip() +
  labs(
    title = "Distribution of records among camera stations",
    x = "Species",
    y = "Number of records per station"
  ) +
  theme_bw()

The width of the violin represents where observations are more concentrated.

Combining geometries

One of the most powerful features of ggplot2 is that we can add multiple geometries as layers.

For example, we can combine a violin plot, a boxplot, and the individual observations:

ggplot(
  data = sp_counts,
  aes(
    x = Species,
    y = n,
    fill = Species
  )
) +
  geom_violin(
    alpha = 0.5,
    show.legend = FALSE
  ) +
  geom_boxplot(
    width = 0.15,
    show.legend = FALSE
  ) +
  geom_jitter(
    aes(color = Species),
    width = 0.15,
    alpha = 0.9,
    show.legend = FALSE
  ) +
  scale_y_log10() +
  coord_flip() +
  labs(
    title = "Distribution of records among camera stations",
    x = "Species",
    y = "Number of records per station"
  ) +
  theme_bw()

Each new + adds another layer to the graphic.

This is the basic logic behind ggplot2.

data
  +
aesthetics
  +
geometry
  +
geometry
  +
geometry
  +
labels
  +
theme

Which geometry should I use?

The graph should depend on the question and the type of variables you have.

One categorical variable

For example:

Species

A bar plot can show the number of observations:

geom_bar()

One numeric variable

For example:

n_images

A histogram can show its distribution:

geom_histogram()

Two numeric variables

For example:

N_individuals
n_images

A scatterplot can show their relationship:

geom_point()

One categorical and one numeric variable

For example:

Species
n_images

We can compare distributions using:

geom_boxplot()

or:

geom_violin()

Important

Do not choose a graph just because it looks nice.

Start by asking:

What variables do I have, and what do I want to show?

Saving a plot

We will often want to save a figure so that we can use it in a report, presentation, poster, or manuscript.

First, we can save the entire plot as an R object:

species_plot <- ggplot(
  data = recordTable,
  aes(x = Species)
) +
  geom_bar(
    fill = "steelblue"
  ) +
  coord_flip() +
  labs(
    title = "Camera-trap records by species",
    x = "Species",
    y = "Number of records"
  ) +
  theme_bw()

To display the plot:

species_plot

Then we can save it using ggsave():

ggsave(
  filename = "outputs/species_records.png",
  plot = species_plot,
  width = 8,
  height = 6,
  dpi = 300
)

This will create:

outputs/species_records.png

Because we are using our RStudio Project, the figure will be saved directly inside the outputs folder.

Tip

Saving a plot through code makes the figure reproducible.

If you later change your data or your graphic, you can simply run the code again.

The basic ggplot2 workflow

Most of the graphics we created followed the same logic:

ggplot(
  data = my_data,
  aes(
    x = x_variable,
    y = y_variable
  )
) +
  geom_something() +
  labs() +
  theme_something()

You do not need to memorize every geom, argument, color, or theme.

What matters is understanding how the pieces fit together.

Practice at home

Tip

This activity should take approximately 20–30 minutes.

Open your intro-r-course project and create a new script called:

practice-05.R

Save it inside the scripts folder.

Load tidyverse and import:

data/recordTable.csv

Part 1: Bar plot

Create a bar plot showing the number of records for each Species.

Your graph should include:

a title,
labels for both axes,
a color of your choice,
and a theme.

Try both the regular orientation and:

coord_flip()

Decide which version is easier to read.

Part 2: Histogram

Create a histogram of:

n_images

Try at least three different values for either:

bins

or:

binwidth

How does changing the bins affect the appearance of the distribution?

Part 3: Compare species

Create an object containing these three species:

practice_species <- c(
  "White-tailed deer",
  "Armadillo",
  "Hog"
)

Select only those records:

practice_records <- recordTable[
  recordTable$Species %in% practice_species,
]

Create either:

geom_boxplot()

or:

geom_violin()

to compare n_images among the three species.

Part 4: Scatterplot

Using practice_records and geom_points create a scatterplot with:

N_individuals

on the x-axis and:

n_images

on the y-axis.

Map:

Species

to either color or shape.

Experiment with:

size
alpha

until the plot is easy to read.

Part 5: Save a figure

Choose one of your plots, save it as an R object, and export it into:

outputs/

using:

ggsave()