ggplot - plot bars as points

A plain R script with the code is here: plot.R

Test data

Create test dataframe for stations

id <- c(1,2,3,4) # station id
lon <- c( 6.1,  8.7,  9.4, 11.9)
lat <- c(55.9, 55.2, 57.1, 54.9)
df_stns <- data.frame(id=id, lon=lon, lat=lat)

Create test dataframe for bar chart data

set.seed(99)

Species <- c("a","b","c","d","e")
Category <- c("x","y","z")

df_bar_data <- merge(data.frame(Species=factor(Species,levels=Species)),
                 data.frame(Category=factor(Category, levels=Category)),
                 all=T)

we include station id with the test data to give a postion to each bar chart

df_bar_data <- merge(data.frame(id=id),
                     df_bar_data,
                 all=T)

add random values for our y-variable “n”

df_bar_data <- df_bar_data %>%
  mutate(n=rnorm(nrow(df_bar_data), mean=10, sd=4)) %>%
  mutate(n=ifelse(n<0,0,n))

normalise values to as fractions of the total within each category/station combination

df_bar_data <- df_bar_data %>%
  group_by(id, Category) %>%
  mutate(n=n/sum(n,na.rm=T)) %>%
  ungroup()

Function to make individual bar charts

bar_chart <- function(selected_id=NA,
                      df,
                      xvar="Category",
                      yvar="n", 
                      group="Species",
                      imgdir="img/",
                      size_x=100, 
                      size_y=200){
  
  df <- df %>%
    filter(id==selected_id)
  
  p <- ggplot(df, aes(x=!!as.name(quo_name(xvar)), 
                      y=!!as.name(quo_name(yvar)), 
                      fill=!!as.name(quo_name(group)))) +
    geom_bar(stat = "identity", position="stack") +
    theme_void(base_size=7) +
    scale_fill_discrete(guide=NULL) +
    theme(axis.text.x = element_text(vjust=3))

  return(p)
}

Make the bar charts

Make the bar chart for each station and save them locally

id_list <- unique(df_bar_data$id)
plot_list <-  purrr::map(id_list, bar_chart, df=df_bar_data) 

Add the list of image files to the station data frame

df_stns$bar_img <- plot_list

Define positions for the images so that they are slightly offset from the station positions.

I haven’t done it here but We could also specify separate sizes for each bar chart.

df_stns <- df_stns %>%
  mutate(x_img = lon+0.3, y_img=lat+0.3)

Make the plot

Get a shape file for the Danish EEZ

eez <- terra::vect("shp/EEZ_polygon.shp")
terra::crs(eez) <-  terra::crs("EPSG:4326") # add the CRS

Simple map with EEZ and points

ggplot() +
  geom_spatvector(data=eez) +
  geom_point(data=df_stns,aes(x=lon,y=lat), colour="#FF0000") +
  geom_text(data=df_stns,aes(x=lon,y=lat,label=id),
            hjust=1, vjust=1)

Map with bar charts

The bars are introduced to the ggplot as images

We need to find another way to add the legend for the bars to the map.

We can do this by plotting a layer which uses the same fill variable as the bar charts (Species). We will use a geom_rect layer - this plots rectangles with fill according to species. Because we use the same x-values for xmin and xmax and the same y-values for ymin and ymax, then the rectangles will have zero height and width. Therefore they should not be visible but they will trigger the creation of a legend with species colours.

df_legend <- df_bar_data %>%
  distinct(Species) %>%
  mutate(n=1)

df_legend <- df_stns %>% 
  merge(df_legend, all=T)

And here is the plot:

ggplot() +
  geom_rect(data=df_legend,
            aes(xmin=lon, xmax=lon, ymin=lat, ymax=lat, fill=Species)) +
  geom_spatvector(data=eez) +
  geom_point(data=df_stns,aes(x=lon,y=lat)) +
  geom_text(data=df_stns,aes(x=lon,y=lat,label=id),
            hjust=1, vjust=1) +
  geom_subview(data=df_stns,aes(x=x_img,y=y_img,subview=bar_img),
               width=0.5, height=0.5)

You will need to mess around with the themes and formatting in the bar chart plot function and with the size and positioning of the bars.