Visualising status and trends using ggplot2

Previous steps

If you would like to return to information from the previous section, please click here.

Context

There are a number of ways to visualise trends of coral reef monitoring data, depending on the level of information contained in the data, variables, and spatial and temporal extent.

In this module, we will be exploring a number of ways to visualise data and provide the basic skills to make plots using ggplot2 and related tools.

The code for the “status” example can be found here: analysis_code/examples/visualisation/plot_percent_cover_acosa.R

Code for the “trends” can be found here: analysis_code/examples/visualisation/plot_kenya_cover.R

Visualising Status

From our previous example, we created some initial summaries of the data to reduce the complexity of the taxonomic groups and obtain quadrat means for each transect.

With these data will will start visualising:

  # create visual
    sessiles_dat_summary %>%
      dplyr::filter(!Grouping %>% is.na()) %>%
    ggplot(aes(Grouping, Value_mean)) +
      geom_boxplot(fill  = "blue",
                   alpha = 0.7) +
      theme_bw() +
      xlab("Functional Group") +
      ylab("Mean Percent Cover")

Which gives us something like:

We can see from this plot is that this region “ACOSA” is dominated by Macroalgae, Non-living subtrata, with some Live coral. Note that the y-axis is “Mean Percent Cover” at the transect level, so we will probably need to check why there are values higher than 100%. We will pick this up later as part of our Exercises, but in essence this figure has provided us with our first data quality check!

A few things to note in the code:

• We are piping our data summary to filter the <NA> taxa groups and piping the result to ggplot()
• ggplot() starts by defining an “aesthetic” (i.e. aes() which defines the x and y values). In this case it is the taxonomic Grouping and the mean Value.
• As part of the “Grammar of Graphics” we then add (i.e. +) a geom, which in this case is a boxplot. However, it could be a number of things (e.g. geom_point).
• In the geom, we then set fill values, transparency, and other aesthetics or mappings (if required). We will come back to this in later exercises.
• Following the geoms, we then add the theme_bw(), which simplifies the figure to a “black-and-white”
• Finally, we change the axis labels using xlab() and ylab()

As a way to standardising GCRMN coding projects, we try to standardise this “layering” as much as possible. It is helpful to have elements to be modified (e.g. data, axes, themes) at the beginning or end of a block of code so it is easier to find if modifications are needed.

Visualising Trends

As one of the key functions of the GCRMN is to report on the trends of coral reefs through time, we can apply the similar data aggregation and plotting for a time series from Kenya.

Note that in this example, we are filtering for groups_of_interest, which is “hard coral”.

  # visualise
    kenya_cover %>%
      dplyr::filter(level1_code %in% groups_of_interest) %>%
      group_by(Year,
               Site) %>%
      summarise(mean_cover = mean_cover %>% mean(na.rm = TRUE)) %>%
      group_by(Year) %>%
      summarise(sd         = mean_cover %>%   sd(na.rm = TRUE),
                mean_cover = mean_cover %>% mean(na.rm = TRUE)) %>%
      mutate(sd_upper = mean_cover + sd,
             sd_lower = mean_cover - sd) %>%
    ggplot(aes(Year, mean_cover)) +
      geom_line(alpha = 0.7) +
      geom_point(size = 2.0, alpha = 0.7) +
      geom_errorbar(aes(ymax = sd_upper,
                        ymin = sd_lower)) +
      theme_bw() +
      ylab("Mean percent cover") +
      ggtitle("Kenya") +
      theme(plot.title = element_text(hjust = 0.5))

The result looks something like this:

In this example, we are conducting a similar aggregation by Year and Site to get means of Stations and then taking the mean() by Year. In order to add the error bars, we use mutate() to create new columns for the upper and lower limits for the bars.

mutate() functions similar to creating a new column in Excel and putting an equation into the column that takes the difference from mean_cover.

Next Steps

It is unrealistic to teach all of the potential visualisations and options in ggplot2 in a single lesson. So, we can suggest course participants visit details as part of the tidyverse documentation, including this cheatsheet.

There are also a number of extensions that add to the core ggplot2 functionality that can be explored here.

Now that we have some basic skills to visualise data using ggplot(), we will now explore how to add complexity to the graphics by using additional aesthetics and faceting.