The default isn’t publication-ready

Built-in themes: a quick win

Add a theme_*() layer and the whole look changes.

Further customization with theme()

A theme controls everything that isn’t data: Fonts, sizes, colours, grid lines, spacing, legends, …

Inheritance of theme elements

text  ──┬── axis.text  ──┬── axis.text.x
        │                └── axis.text.y
        ├── plot.title
        ├── plot.subtitle
        └── legend.text

line  ──┬── axis.line   ──┬── axis.line.x
        │                 └── axis.line.y
        ├── axis.ticks
        └── panel.grid   ─── panel.grid.major / minor

rect  ──┬── plot.background
        └── panel.background

• Set a parent and it affects all its children
• E.g. Change text once and every piece of text updates

How to change a theme

p_bubble +
  theme_light(base_size = 18) +
  theme(
    legend.text = element_text(size = rel(0.85)),
    panel.grid.minor = element_blank()
  )

Live demo

Open demo/01_themes.R

Pick intuitive colours

Use colours your reader already associates with the thing.

Two things that matter

For scientific figures:

1. Intuitive colours that match what your reader expects
2. Colourblind-safe colours

Three kinds of palettes

Okabe-Ito (base R)

A colourblind-safe qualitative palette, built into base R.

• 8 distinct hues plus black and grey, separable under all common CVD types
• access with palette.colors(palette = "Okabe-Ito")

See here for details

viridis (built into ggplot2)

Perceptually uniform, colourblind- and greyscale-safe. Best for ordered or continuous data.

• access with scale_colour_viridis_d() / _c()

See here for details

scico R package

Perceptually uniform and colourblind-safe. Made for scientific data and designed to be fair, readable, and citable

• access with scale_colour_scico_d() / _c()

scico R package, the colour maps by Fabio Crameri

Live demo

Open demo/02_color.R.

Common multi-panel layouts

The patchwork package

• Compose plots
• Collect shared legends and axes
• Tag panels
• Apply additional plot layers to every panel
• Place plots inside plots as insets

The patchwork documentation is excellent

Live demo

Open demo/03_patchwork.R.

Same plot, different canvas

Rule of thumb

Starting points for Canvas and font sizes:

Live demo

Open demo/04_export.R.

gghighlight

gghighlight to fade data and highlight specific elements.

library(gghighlight)
p_bubble +
  scale_color_scico_d(palette = "batlow") +
  theme_workshop() +
  gghighlight(
    continent %in% c("Europe", "Africa"),
    use_direct_label = FALSE
  )

ggrepel + ggtext

ggrepel and ggtext for labels and markdown text

library(ggrepel)
library(ggtext)
ggplot(gap_continent, aes(year, mean_lifeExp, color = continent)) +
  geom_line(linewidth = 1) +
  geom_text_repel(
    data = gap_continent |> filter(year == max(year)),
    aes(label = continent),
    size = 6,
    nudge_x = 2,
    hjust = 0,
    direction = "y",
    seed = 1
  ) +
  scale_x_continuous(expand = expansion(mult = c(0.02, 0.15))) +
  scale_color_scico_d(palette = "batlow") +
  # use markdown in the title (see theme layer)
  labs(
    title = "*Life expectancy* by **continent**",
    x = "Year",
    y = "Life expectancy"
  ) +
  theme_workshop() +
  theme(
    panel.grid.major.x = element_blank(),
    plot.title = element_markdown(),
    legend.position = "none"
  )

ggdist raincloud plots

ggdist to show distributions and uncertainty -> Barplot alternatives

library(ggdist)
ggplot(
  gap_2007 |> filter(continent != "Oceania"),
  aes(continent, gdpPercap, fill = continent)
) +
  stat_halfeye(
    adjust = 0.5,
    width = 0.6,
    .width = 0,
    justification = -0.3,
    point_colour = NA
  ) +
  geom_boxplot(width = 0.15, outlier.shape = NA, alpha = 0.5) +
  stat_dots(side = "left", justification = 1.1, dotsize = 0.4) +
  scale_y_log10(labels = label_dollar(accuracy = 1)) +
  scale_fill_scico_d(palette = "batlow") +
  theme_workshop() +
  theme(legend.position = "none")

ggridges

ggridges for ridgeline distribution plots

library(ggridges)
ggplot(gap_2007 |> filter(continent != "Oceania"), aes(lifeExp, continent, fill = continent)) +
  geom_density_ridges(alpha = 0.8, scale = 1.1) +
  scale_fill_scico_d(palette = "batlow") +
  theme_workshop() +
  theme(legend.position = "none") +
  labs(x = "Life expectancy (years)", y = NULL)

Takeaways

What you can take back to your own plots:

1. Custom themes: write one theme_*() function, source it, theme_set() it everywhere it’s needed
2. Colour with intent: pick intuitive, colourblind-safe palettes (Okabe-Ito, viridis, scico), and check with cvdPlot()
3. Multi-panel layouts: compose plots with patchwork, collect shared legends, add tag panels, …
4. Export: pick the canvas size, preview with ggview::canvas(), tweak base_size and geoms, and save with ragg::ragg_png or cairo_pdf.

Next lecture/workshop

Topic tba

📅 18.06.2026 🕓 4-5 p.m. 📍 Webex

🔔 Subscribe to the mailing list

📧 For topic suggestions and/or feedback send me an email