Reference 4

Combining Plots with patchwork

Author

Byeong-Hak Choe

Published

March 31, 2026

Modified

March 31, 2026

1 Introduction to `patchwork`

When building a data story, we often need to place multiple plots side by side, stack them vertically, or arrange them in a custom grid. Base R’s par(mfrow = ...) can do this for base graphics, but it does not work with ggplot2.

patchwork is the modern solution. It extends ggplot2 with a simple, expressive syntax for combining plots — using everyday operators like +, |, and /.

“The goal of patchwork is to make it ridiculously simple to combine separate ggplots into the same graphic.” — Thomas Lin Pedersen (package author)

1.1 Installation and loading

install.packages("patchwork")
# remotes::install_github("thomasp85/patchwork")

library(ggplot2)
library(patchwork)

2 Building Blocks — Four `mtcars` Plots

We create four plots from mtcars that we will combine in different ways throughout this lecture. Think of these as the building blocks.

p1 <- ggplot(mtcars) +
  geom_point(aes(mpg, disp), color = "steelblue") +
  labs(title = "Plot 1", x = "Miles per gallon", y = "Displacement")

p2 <- ggplot(mtcars) +
  geom_boxplot(aes(factor(gear), disp, fill = factor(gear)),
               show.legend = FALSE) +
  labs(title = "Plot 2", x = "Gears", y = "Displacement")

p3 <- ggplot(mtcars) +
  geom_point(aes(hp, wt, color = mpg)) +
  scale_color_viridis_c() +
  labs(title = "Plot 3", x = "Horsepower", y = "Weight")

p4 <- ggplot(mtcars) +
  geom_bar(aes(factor(gear), fill = factor(gear)),
           show.legend = FALSE) +
  facet_wrap(~cyl) +
  labs(title = "Plot 4", x = "Gears", y = "Count")

Display them individually to understand what each one shows before combining.

p1

p2

p3

p4

3 Basic Use — the `+` Operator

The simplest way to combine two plots is +. Patchwork places them side by side in a single row.

p1 + p2

Note

The + operator is the same one used in ggplot2 to add layers, but when applied between two complete ggplot objects, patchwork intercepts it and creates a combined layout.

3.1 Additions flow to the last plot

When you add a ggplot2 element (such as labs()) after a patchwork, it is applied to the last plot in the composition:

p1 + p2 + labs(subtitle = "This subtitle appears in the last plot (p2)")

This is consistent with ggplot2’s own + semantics — keep it in mind to avoid accidental modifications.

4 Filling a Grid

Adding more than two plots fills a grid automatically. Patchwork tries to keep the grid as square as possible, filling in row-first order.

p1 + p2 + p3 + p4

4.1 Controlling rows and columns with `plot_layout()`

Use plot_layout() to override the default grid dimensions.

p1 + p2 + p3 + p4 +
  plot_layout(nrow = 3, byrow = FALSE)

Argument	Effect
`nrow`	Number of rows
`ncol`	Number of columns
`byrow`	Fill by row (`TRUE`) or column (`FALSE`)
`widths`	Relative widths of columns
`heights`	Relative heights of rows

4.1.1 Relative sizing

Control how much space each column or row gets:

p1 + p2 + p3 +
  plot_layout(ncol = 3, widths = c(2, 1, 1))

Here p1 gets twice as much horizontal space as p2 and p3.

5 Stacking and Placing Side by Side

patchwork provides two dedicated operators for the two most common layouts:

Operator	Layout
`\\|`	Place plots beside each other (same row)
`/`	Stack plots on top of each other (same column)

5.1 Side by side with `|`

p1 | p2

5.2 Stacked with `/`

p1 / p2

6 Nesting Layouts

Because | and / return patchwork objects themselves, they can be nested using parentheses to create complex multi-panel figures.

6.1 One column beside a stacked pair

p1 | (p2 / p3)

6.2 A stacked pair beside another plot

(p1 / p2) | p3

6.3 Three-level nesting

p1 | (p2 / (p3 | p4))

Tip

Reading nested patchwork expressions — evaluate the innermost parentheses first, just like arithmetic. p3 | p4 is a row, p2 / (p3 | p4) stacks p2 above that row, and finally p1 | ... places p1 to the left of the whole thing.

7 Annotating a Composition

A combined figure often needs a shared title, subtitle, caption, or tagged subplot labels. Use plot_annotation() for this.

7.1 Shared title and caption

(p1 | (p2 / p3)) +
  plot_annotation(
    title   = "A story about `mtcars`",
    subtitle = "Exploring displacement, horsepower, weight, and fuel economy",
    caption  = "Source: Motor Trend Car Road Tests (1974)"
  )

7.2 Auto-tagging subplots

Patchwork can automatically label each panel so you can refer to them in text as (A), (B), etc.

p1 + p2 + p3 +
  plot_annotation(tag_levels = "A")

Available tag styles:

`tag_levels`	Labels
`"A"`	A, B, C, …
`"a"`	a, b, c, …
`"1"`	1, 2, 3, …
`"I"`	I, II, III, …
`"i"`	i, ii, iii, …

p1 + p2 + p3 + p4 +
  plot_annotation(tag_levels = "I")

7.3 Styling tags with `tag_prefix` and `tag_suffix`

p1 + p2 + p3 + p4 +
  plot_annotation(
    tag_levels = "A",
    tag_prefix = "(",
    tag_suffix = ")"
  )

8 Modifying Individual Panels in a Composition

After combining plots, you can still apply ggplot2 modifications to a specific panel using [[i]] indexing.

combined <- p1 + p2 + p3

# Change the theme of the second panel only
combined[[2]] <- combined[[2]] + theme_minimal()

combined

9 Collecting Legends

When multiple panels share the same aesthetic (e.g., the same color scale), their legends are redundant. Use plot_layout(guides = "collect") to merge them into one.

# Two plots that both use fill = factor(cyl)
q1 <- ggplot(mtcars) +
  geom_point(aes(mpg, disp, color = factor(cyl))) +
  labs(title = "mpg vs. disp", color = "Cylinders")

q2 <- ggplot(mtcars) +
  geom_point(aes(hp, wt, color = factor(cyl))) +
  labs(title = "hp vs. wt", color = "Cylinders")

q1 + q2 + plot_layout(guides = "collect")

Without guides = "collect", each panel carries its own legend — cluttering the figure. With it, the legend is deduplicated and placed at the right edge by default.

10 Adding Empty Space with `plot_spacer()`

Sometimes you want a blank cell in your grid (e.g., to shift a plot to the right or leave room for a manual annotation). Use plot_spacer():

p1 + plot_spacer() + p2

# Place p3 in the bottom-right of a 2x2 grid
plot_spacer() + plot_spacer() + plot_spacer() + p3

11 Insets with `inset_element()`

inset_element() places one plot on top of another as an inset — like a zoomed-in detail or a small map overlay.

p3 + inset_element(p1, left = 0.6, bottom = 0.6, right = 1, top = 1)

The left, right, bottom, and top arguments are proportions of the base plot’s area (0 to 1). Here the inset occupies the upper-right 40% × 40% of p3.

12 Putting It All Together — A Publication-Ready Figure

Combining all the tools we have learned: nested layouts, collected legends, auto-tags, and a shared annotation.

final_figure <-
  ((q1 | q2) / (p2 | p4)) +
  plot_layout(guides = "collect") +
  plot_annotation(
    title    = "Motor Trend Car Road Tests — `mtcars` Overview",
    subtitle = "Top row: continuous relationships by cylinder count | Bottom row: discrete summaries",
    caption  = "Data: Motor Trend (1974)",
    tag_levels = "A",
    tag_prefix = "(",
    tag_suffix = ")"
  ) &
  theme_minimal(base_size = 11)

final_figure

The & operator

& theme_minimal() applies a theme to all panels at once, unlike + which only targets the last panel. Use & whenever you want a global style change across every plot in a patchwork.

13 Quick Reference

Task	Code
Place side by side	`p1 \\| p2`
Stack vertically	`p1 / p2`
Fill a 2×2 grid	`p1 + p2 + p3 + p4`
Control rows/cols	`+ plot_layout(nrow = 2)`
Relative widths	`+ plot_layout(widths = c(2, 1))`
Shared title/caption	`+ plot_annotation(title = "...")`
Auto-tag panels	`+ plot_annotation(tag_levels = "A")`
Collect legends	`+ plot_layout(guides = "collect")`
Blank cell	`plot_spacer()`
Inset plot	`+ inset_element(p, left, bottom, right, top)`
Style all panels	`& theme_minimal()`
Modify one panel	`combined[[2]] <- combined[[2]] + ...`

14 Further Reading

patchwork documentation
Plot Assembly guide — wrap_plots(), wrap_elements(), nesting details
Controlling Layouts — area(), complex grid specifications
Annotation and Style — per-level tagging, theme control
Multi-page Alignment — aligning panels across grid and cowplot pages

--- title: "Reference 4" subtitle: "Combining Plots with `patchwork`" author: "Byeong-Hak Choe" date: last-modified format: html: toc: true toc-depth: 6 number-sections: true theme: cosmo code-tools: true execute: echo: true warning: false message: false fig-width: 10 fig-height: 8 --- ```{r setup} #| include: false library(tidyverse) library(patchwork) library(hrbrthemes) library(ggthemes) theme_set( theme_ipsum() + theme( strip.background = element_rect(fill = "lightgray"), axis.title.x = element_text( angle = 0, size = rel(1.5), margin = margin(10, 0, 0, 0) ), axis.title.y = element_text( # angle = 0, size = rel(1.5), margin = margin(0, 10, 0, 0) ), axis.text.x = element_text(size = rel(1.25)), axis.text.y = element_text(size = rel(1.25)), strip.text = element_text(size = rel(1.3)), legend.position = "top", legend.text = element_text(size = rel(1.2)), legend.title = element_text(size = rel(1.2)) ) ) scale_colour_discrete <- function(...) scale_colour_tableau(...) scale_fill_discrete <- function(...) scale_fill_tableau(...) ``` --- # Introduction to `patchwork` When building a data story, we often need to place multiple plots side by side, stack them vertically, or arrange them in a custom grid. Base R's `par(mfrow = ...)` can do this for base graphics, but it does not work with `ggplot2`. **`patchwork`** is the modern solution. It extends `ggplot2` with a simple, expressive syntax for combining plots — using everyday operators like `+`, `|`, and `/`. > "The goal of patchwork is to make it ridiculously simple to combine separate ggplots into the same graphic." > — Thomas Lin Pedersen (package author) ## Installation and loading ```{r} #| eval: false install.packages("patchwork") # remotes::install_github("thomasp85/patchwork") ``` ```{r} library(ggplot2) library(patchwork) ``` --- # Building Blocks — Four `mtcars` Plots We create four plots from `mtcars` that we will combine in different ways throughout this lecture. Think of these as the building blocks. ```{r} p1 <- ggplot(mtcars) + geom_point(aes(mpg, disp), color = "steelblue") + labs(title = "Plot 1", x = "Miles per gallon", y = "Displacement") p2 <- ggplot(mtcars) + geom_boxplot(aes(factor(gear), disp, fill = factor(gear)), show.legend = FALSE) + labs(title = "Plot 2", x = "Gears", y = "Displacement") p3 <- ggplot(mtcars) + geom_point(aes(hp, wt, color = mpg)) + scale_color_viridis_c() + labs(title = "Plot 3", x = "Horsepower", y = "Weight") p4 <- ggplot(mtcars) + geom_bar(aes(factor(gear), fill = factor(gear)), show.legend = FALSE) + facet_wrap(~cyl) + labs(title = "Plot 4", x = "Gears", y = "Count") ``` Display them individually to understand what each one shows before combining. ```{r} p1 ``` ```{r} p2 ``` ```{r} p3 ``` ```{r} p4 ``` --- # Basic Use — the `+` Operator The simplest way to combine two plots is `+`. Patchwork places them side by side in a single row. ```{r} p1 + p2 ``` ::: {.callout-note} The `+` operator is the same one used in `ggplot2` to add layers, but when applied *between two complete `ggplot` objects*, `patchwork` intercepts it and creates a combined layout. ::: ## Additions flow to the last plot When you add a `ggplot2` element (such as `labs()`) after a `patchwork`, it is applied to the **last** plot in the composition: ```{r} p1 + p2 + labs(subtitle = "This subtitle appears in the last plot (p2)") ``` This is consistent with `ggplot2`'s own `+` semantics — keep it in mind to avoid accidental modifications. --- # Filling a Grid Adding more than two plots fills a grid automatically. Patchwork tries to keep the grid as square as possible, filling in **row-first** order. ```{r} p1 + p2 + p3 + p4 ``` ## Controlling rows and columns with `plot_layout()` Use `plot_layout()` to override the default grid dimensions. ```{r} p1 + p2 + p3 + p4 + plot_layout(nrow = 3, byrow = FALSE) ``` | Argument | Effect | |----------|--------| | `nrow` | Number of rows | | `ncol` | Number of columns | | `byrow` | Fill by row (`TRUE`) or column (`FALSE`) | | `widths` | Relative widths of columns | | `heights` | Relative heights of rows | ### Relative sizing Control how much space each column or row gets: ```{r} p1 + p2 + p3 + plot_layout(ncol = 3, widths = c(2, 1, 1)) ``` Here `p1` gets twice as much horizontal space as `p2` and `p3`. --- # Stacking and Placing Side by Side `patchwork` provides two dedicated operators for the two most common layouts: | Operator | Layout | |----------|--------| | `\|` | Place plots **beside** each other (same row) | | `/` | **Stack** plots on top of each other (same column) | ## Side by side with `|` ```{r} p1 | p2 ``` ## Stacked with `/` ```{r} p1 / p2 ``` --- # Nesting Layouts Because `|` and `/` return patchwork objects themselves, they can be **nested** using parentheses to create complex multi-panel figures. ## One column beside a stacked pair ```{r} p1 | (p2 / p3) ``` ## A stacked pair beside another plot ```{r} (p1 / p2) | p3 ``` ## Three-level nesting ```{r} p1 | (p2 / (p3 | p4)) ``` ::: {.callout-tip} **Reading nested patchwork expressions** — evaluate the innermost parentheses first, just like arithmetic. `p3 | p4` is a row, `p2 / (p3 | p4)` stacks `p2` above that row, and finally `p1 | ...` places `p1` to the left of the whole thing. ::: --- # Annotating a Composition A combined figure often needs a shared title, subtitle, caption, or tagged subplot labels. Use `plot_annotation()` for this. ## Shared title and caption ```{r} (p1 | (p2 / p3)) + plot_annotation( title = "A story about `mtcars`", subtitle = "Exploring displacement, horsepower, weight, and fuel economy", caption = "Source: Motor Trend Car Road Tests (1974)" ) ``` ## Auto-tagging subplots Patchwork can automatically label each panel so you can refer to them in text as (A), (B), etc. ```{r} p1 + p2 + p3 + plot_annotation(tag_levels = "A") ``` Available tag styles: | `tag_levels` | Labels | |--------------|--------| | `"A"` | A, B, C, ... | | `"a"` | a, b, c, ... | | `"1"` | 1, 2, 3, ... | | `"I"` | I, II, III, ... | | `"i"` | i, ii, iii, ... | ```{r} p1 + p2 + p3 + p4 + plot_annotation(tag_levels = "I") ``` ## Styling tags with `tag_prefix` and `tag_suffix` ```{r} p1 + p2 + p3 + p4 + plot_annotation( tag_levels = "A", tag_prefix = "(", tag_suffix = ")" ) ``` --- # Modifying Individual Panels in a Composition After combining plots, you can still apply ggplot2 modifications to a specific panel using `[[i]]` indexing. ```{r} combined <- p1 + p2 + p3 # Change the theme of the second panel only combined[[2]] <- combined[[2]] + theme_minimal() combined ``` --- # Collecting Legends When multiple panels share the same aesthetic (e.g., the same color scale), their legends are redundant. Use `plot_layout(guides = "collect")` to merge them into one. ```{r} # Two plots that both use fill = factor(cyl) q1 <- ggplot(mtcars) + geom_point(aes(mpg, disp, color = factor(cyl))) + labs(title = "mpg vs. disp", color = "Cylinders") q2 <- ggplot(mtcars) + geom_point(aes(hp, wt, color = factor(cyl))) + labs(title = "hp vs. wt", color = "Cylinders") q1 + q2 + plot_layout(guides = "collect") ``` Without `guides = "collect"`, each panel carries its own legend — cluttering the figure. With it, the legend is deduplicated and placed at the right edge by default. --- # Adding Empty Space with `plot_spacer()` Sometimes you want a blank cell in your grid (e.g., to shift a plot to the right or leave room for a manual annotation). Use `plot_spacer()`: ```{r} p1 + plot_spacer() + p2 ``` ```{r} # Place p3 in the bottom-right of a 2x2 grid plot_spacer() + plot_spacer() + plot_spacer() + p3 ``` --- # Insets with `inset_element()` `inset_element()` places one plot on top of another as an inset — like a zoomed-in detail or a small map overlay. ```{r} p3 + inset_element(p1, left = 0.6, bottom = 0.6, right = 1, top = 1) ``` The `left`, `right`, `bottom`, and `top` arguments are proportions of the base plot's area (0 to 1). Here the inset occupies the upper-right 40% × 40% of `p3`. --- # Putting It All Together — A Publication-Ready Figure Combining all the tools we have learned: nested layouts, collected legends, auto-tags, and a shared annotation. ```{r} #| fig-width: 10 #| fig-height: 8 final_figure <- ((q1 | q2) / (p2 | p4)) + plot_layout(guides = "collect") + plot_annotation( title = "Motor Trend Car Road Tests — `mtcars` Overview", subtitle = "Top row: continuous relationships by cylinder count | Bottom row: discrete summaries", caption = "Data: Motor Trend (1974)", tag_levels = "A", tag_prefix = "(", tag_suffix = ")" ) & theme_minimal(base_size = 11) final_figure ``` ::: {.callout-note} ## The `&` operator `& theme_minimal()` applies a theme to **all panels at once**, unlike `+` which only targets the last panel. Use `&` whenever you want a global style change across every plot in a patchwork. ::: --- # Quick Reference | Task | Code | |------|------| | Place side by side | `p1 \| p2` | | Stack vertically | `p1 / p2` | | Fill a 2×2 grid | `p1 + p2 + p3 + p4` | | Control rows/cols | `+ plot_layout(nrow = 2)` | | Relative widths | `+ plot_layout(widths = c(2, 1))` | | Shared title/caption | `+ plot_annotation(title = "...")` | | Auto-tag panels | `+ plot_annotation(tag_levels = "A")` | | Collect legends | `+ plot_layout(guides = "collect")` | | Blank cell | `plot_spacer()` | | Inset plot | `+ inset_element(p, left, bottom, right, top)` | | Style all panels | `& theme_minimal()` | | Modify one panel | `combined[[2]] <- combined[[2]] + ...` | --- # Further Reading - [patchwork documentation](https://patchwork.data-imaginist.com/) - [Plot Assembly guide](https://patchwork.data-imaginist.com/articles/guides/assembly.html) — `wrap_plots()`, `wrap_elements()`, nesting details - [Controlling Layouts](https://patchwork.data-imaginist.com/articles/guides/layout.html) — `area()`, complex grid specifications - [Annotation and Style](https://patchwork.data-imaginist.com/articles/guides/annotation.html) — per-level tagging, theme control - [Multi-page Alignment](https://patchwork.data-imaginist.com/articles/guides/multipage.html) — aligning panels across `grid` and `cowplot` pages

1 Introduction to patchwork

1.1 Installation and loading

2 Building Blocks — Four mtcars Plots

3 Basic Use — the + Operator

3.1 Additions flow to the last plot

4 Filling a Grid

4.1 Controlling rows and columns with plot_layout()

4.1.1 Relative sizing

5 Stacking and Placing Side by Side

5.1 Side by side with |

5.2 Stacked with /

6 Nesting Layouts

6.1 One column beside a stacked pair

6.2 A stacked pair beside another plot

6.3 Three-level nesting

7 Annotating a Composition

7.1 Shared title and caption

7.2 Auto-tagging subplots

7.3 Styling tags with tag_prefix and tag_suffix

8 Modifying Individual Panels in a Composition

9 Collecting Legends

10 Adding Empty Space with plot_spacer()

11 Insets with inset_element()

12 Putting It All Together — A Publication-Ready Figure

13 Quick Reference

14 Further Reading

1 Introduction to `patchwork`

2 Building Blocks — Four `mtcars` Plots

3 Basic Use — the `+` Operator

4.1 Controlling rows and columns with `plot_layout()`

5.1 Side by side with `|`

5.2 Stacked with `/`

7.3 Styling tags with `tag_prefix` and `tag_suffix`

10 Adding Empty Space with `plot_spacer()`

11 Insets with `inset_element()`