Un-sung Pollination Hero

Half the size of your average bee, named for their curious hovering flight, and master’s of Batesian mimicry, hoverflies are gentle bee look-alikes and essential pollinators. Globally, there are thousands of hoverfly species all helping to pollinate over 70% of wild plant species, and are second only to wild bees in pollination efficiency.¹ They don’t sting or bite, and their larvae eat aphids and other small garden pests!

Perhaps the most striking is their long-distance pollen transfer. Unlike bees, which rarely travel farther than 1-2km from their home nests, hoverflies are known to be highly migratory, some even travelling distances comparable to monarch butterfly migration. Studies based in the UK have shown that hoverflies may cover hundreds of kilometers in a single day, even over open water, facilitating gene flow between plant populations that may otherwise remain unconnected.²

However, despite these crucial ecosystem services, they are often underappreciated when compared to other pollinator species such as bees and wasps.³

Using long term plant-pollinator interaction data from the HJ Andrews Experimental Forest, I wanted to showcase hoverflies while comparing them to a big time pollination star, bumble bees. To do so, I created an infographic that addresses the following questions:

• How do hoverflies compare to bumble bees as pollinators?
  • Other than bees, who are the busiest pollinators?
  • How do hoverflies shine in ways that bees don’t?
  • What meadow species do bumble bees and hoverflies pollinate the most?

You can check out the full data here!

My final infographic is made up of 3 plots, all of which were built using a combination of ggplot and Affinity Designer. Let’s break it down and walk through my decision making for each aspect of the vizualization!

Petal shaped pie chart

When I chose this data set, I was excited by the idea of making cute, flowery viz. I thought it would be so fun to make a flower shaped pie chart where each petal was sized congruently with its value. As we’ve learned throughout the quarter, volume can be tricky to interpret, so in order for that plot to work, there needed to be a big enough difference between my groups. The main takeaway from this plot is that hoverflies are the biggest group, and luckily that is clear in the visualization.

In ggplot, I used a totally void theme with the plan to use annotations and body text in place of traditional axes and titles. I also intentionally ordered the petals from largest to smallest, starting with the largest petal facing about 1 o’clock, and moving clockwise from there. This draws the readers eye to the largest petal first, centering hoverflies.

Coming out of ggplot, I had a very bare visualization, ready to be annotated:

Next, I moved into Affinity Designer, where I added all of my text and annotations. Annotating each petal with the pollinator type, the number of times they were observed pollinating, and using bold and highlighted text to emphasize petals representing hover flies, I aimed to make it super easy to see that hover flies were the most observed pollinator.

Here is that polished plot, annotated in Affinity:

Map

Next, I wanted to make a super simple map that would both contextualize where the data came from, and how far hoverflies can carry pollen. For this, I used tmap rather than ggplot. Starting with a super simple outline of Oregon, I layered on an X marking the data collection sites, and a 100km buffer surrounding that point representing the distance that hoverflies can travel.

Here is what I came up with using tmap:

I finished up this map in Affinity Designer. Again, aiming to keep this one super simple, I added two annotations to label the meadow sites and the hover fly radius. This, combined with the body text in the infographic, conveys a clear message that hoverflies can pollinate over impressive distances, while giving some context to the data itself.

Here is a look at the final map:

Favorite flowers lollipop chart

Finally, my favorite piece of the infographic. I wanted to show side by side that hoverflies and bumble bees prefer different plants. Starting in ggplot, I created two lollipop charts, showing the species that each pollinator visited the most ordered from #1 to #3. For this plot, I used a mostly blank theme, but kept the axis labels and horizontal grid lines to keep the chart feel and provide context to what I mean by “favorite”.

Here were my charts coming out of ggplot, ready for Affinity magic:

Again, I moved over to Affinity for the finishing touches. For this I had big plans, wanting to make each line look like the species they represented. I was able to find beautiful illustrations online (check out the artists listed in the footer of my infographic!) and overlay the lines with the flowers themselves. I was careful to make sure that the heights of the flowers lined up with the heights of the bars. Next, I used arrows and annotations to point out that the species that bumble bees pollinate the most is the 3rd most common meadow flower, while the species that hover flies pollinate the most is the 18th most common!

I love the way this one turned out, giving a taste of our beautiful blooming Western Cascade meadows:

Putting it all together

Already settled on my graphic forms, text, and plot themes, it was time to think about colors, typography, general design, contextualizing my data, centering my primary message, and accessibility.

Colors & typography

I aimed to have the colors feel light and springy, while also matching colors typically found on a hoverfly. I made my own color palette, leaning into using pinks and oranges. Since I made my own palette, I made sure that the colors were color blind friendly:

The colors weren’t perfectly color blind friendly, but I was careful to make sure that I never relied solely on color to communicate any important messages.

I also knew that though I didn’t want a stark white background, I did want it to be quite light. I settled on a slight pink, which I think compliments the flower illustrations nicely.

For the first time, I found choosing a font easier than choosing colors! For all of the font families, I used Quicksand, which I found to feel quite flowery while still being simple and easy to read. I also chose to use bolded, highlighted text throughout my infographic, to help the reader focus on the key messages.

General design, contextulaizing the data, & centering my primary message

Designing the layout of my infographic to avoid information overload was the most challenging part of building this visualization. I tried to avoid this by keeping each of my individual plots quite simple. I also formatted all of the text absolutely necessary to read as titles, while anything extra I made a bit smaller, giving the reader the option to skim if need be.

I found ways to subtly contextualize the data throughout the infographic. The first and most obvious place is in the main subtitle, where I state that we are comparing hoverflies and bees “in Western Oregon meadows”. The next place I try and bring the data back into context is in the map, where I use the study sites to contectualize how far hoverflies can pollinate. Lastly, I use the two annotations in the last plot to say how common each pollinators favorite plant was in “our” meadows, subtly reminding the viewer that the data is coming from one place.

Finally, I wanted to find a way to help the reader move sequentially through each piece of the infographic. My primary messages, that hoverflies are busy and abundant pollinators that can travel long distances and pollinate different species than bees, are best communicated in that order. To do this, I have a hoverfly “flying” through the infographic, starting at the very top next to the main title and travelling down, making stops at each of my 3 plots. I felt that this was a fun way to guide the viewer, giving the eye something to follow and drawing attention to my key messages.

Accessibility and DEIJ

It’s always essential to consider accessibility and DEIJ when communicating about data. I already touched on ensuring that my color palette was color blind friendly, and in addition to this I embedded alt text to all of my images.

Explore the code!

Drop down this folded code chunk to see the code used to create these plots

# Load packages ----
library(tidyverse)
library(janitor)
library(here)
library(showtext)
library(tmap)
library(spData)
library(sf)

# Read in pollinator data ----
pp_interactions <- read_csv(here("data", "pp-interactions.csv")) |>
  clean_names()

################################ Petal Plot ################################

# Add a column with pollinator generic type to group hover flies together ----
pp_interactions <- pp_interactions |>
  mutate(pol_type = case_when(str_detect(vissp_type, "beetle") ~ "Beetle",
                              str_detect(vissp_type, "hover-fly") ~ "Hover Fly",
                              str_detect(vissp_type, " fly") ~ "Fly",
                              str_detect(vissp_type, "wasp") ~ "Wasp",
                              str_detect(vissp_type, "butterfly") ~ "Butterfly",
                              str_detect(vissp_type, "grasshopper") ~ "Grasshopper",
                              str_detect(vissp_type, "moth") ~ "Moth",
                              str_detect(vissp_type, "hummingbird") ~ "Hummingbird",
                              str_detect(vissp_type, "social") ~ "Social Bee",
                              str_detect(vissp_type, "solitary") ~ "Solitary Bee",
                              str_detect(vissp_type, "ant") ~ "Ant",
                              str_detect(vissp_type, "spider") ~ "Spider",
                              .default = "other"))

# Wrangle data for petal plot ---- 
petal_plot_data <- pp_interactions |> 
  
  # Remove 2021, as only bees were sampled in this year
  filter(year != 2021) |>
  
  # Filter out bees and "other" pollinators
  filter(pol_type != "other" & pol_type != "Social Bee" & pol_type != "Solitary Bee") |>
  
  # Count total visits per pollinator
  group_by(vissp_type) |>
  summarize(visits = sum(no_int, na.rm = TRUE)) |>
  ungroup() |>
  
  # Pull out the top 5 busiest pollinators
  slice_max(order_by = visits, n = 5)

# Font
font_add_google(name = "Quicksand", family = "quicksand")
showtext_auto()

# Define number of petals and petal angle
petals = 5
petal_angle = 360/petals

# Plot
petal_plot_data |>
  
  # Calculate angles and radii
  mutate(petal = row_number(),
         theta0 = petal * petal_angle) |>
  reframe(theta = theta0 + c(0, -petal_angle/2,  0, 
                             petal_angle/2, 0),
          r = visits * c(0, 0.6, 1, 0.6, 0), .by = c(vissp_type, visits, petal, theta0)) |> 
  
  # Plot theta and r and group by petal no. 
  ggplot(aes(theta, r, group = petal)) +
  ggforce::stat_bspline(geom = "area", n = 1000, fill = "#E195AB", ) +
  coord_radial() +
  theme_void() +
  theme(
    plot.title = element_text(family = "quicksand",
                              face = "bold", 
                              hjust = 0.5, 
                              size = 18),
    plot.subtitle = element_text(family = "quicksand",
                                 hjust = 0.5),
    plot.background = element_rect(fill = "white", color = "white")
  )

################################ Map ################################

# Grab Oregon basemap ----
oregon <- us_states |>
  filter(NAME == "Oregon")

# Define region with the coordinates from metadata (min_lat, min_lon, max_lat, max_lon) ----
min_lat <- 44.20308930 
min_lon <- -122.15208820    
max_lat <- 44.28020840    
max_lon <- -122.12605670 

# Create a box of the study region ----
bbox <- st_sfc(st_polygon(list(matrix(c(min_lon, min_lat,
                                        min_lon, max_lat,
                                        max_lon, max_lat,
                                        max_lon, min_lat,
                                        min_lon, min_lat), 
                                      ncol = 2, byrow = TRUE))), crs = st_crs(oregon))

# Find the centroid of the box ----
centroid <- st_centroid(bbox, crs = st_crs(oregon))

# Create buffers around centroid for bee and hover fly ----
# Transform to a projected crs
centroid_projected <- st_transform(centroid, crs = 26910)

# Create a 1km buffer (1000 meters)
bee_buffer <- st_buffer(centroid_projected, dist = 1000) |>
  st_transform(crs = st_crs(oregon))

# Create a 100km buffer (100000 meters)
fly_buffer <- st_buffer(centroid_projected, dist = 100000) |>
  st_transform(crs = st_crs(oregon))

# Map ----
tm_shape(oregon) +
  tm_fill(col = "#FDFAFB") +
  tm_borders(lwd = 5, col = "#666666") +
  tm_layout(frame = FALSE) +
  tm_shape(fly_buffer) +
  tm_borders(col = "#CB904A", lwd = 4) +
  tm_shape(centroid) +
  tm_dots(col = "#E195AB", shape = 4, size = 1, border.lwd = 5) +
  tm_shape(bee_buffer) +
  tm_fill(col = "black", border.lwd = 2)


################################ Favorite Flowers Plot ################################

# Create a pollinator data frame that only includes bumble bees and hover flies ----
hvb <- pp_interactions |>
  filter(year != 2021) |>
  filter(str_detect(vissp_type, "hover-fly") | str_detect(vissp_name, "Bombus")) |>
  mutate(pollinator = case_when(str_detect(vissp_type, "hover-fly") ~ "Hover-fly",
                                str_detect(vissp_type, "bee") ~ "Bee")) |>
  group_by(pltsp_code, pollinator) |>
  summarise(visits = sum(no_int, na.rm = TRUE)) |>
  ungroup()

# Lollipop chart for bees ----
hvb |>
  filter(pollinator == "Bee") |>
  slice_max(order_by = visits, n = 3) |>
  mutate(rank = c("A", "B", "C")) |>
  ggplot(aes(x = rank, y = visits)) +
  geom_point(fill = "#ECCED7") +
  geom_linerange(aes(ymin = 0, ymax = visits)) +
  labs(y = "Number of flower visits") +
  scale_x_discrete(labels = c( "#1. Nuttail's Larkspur", 
                               "#2. Blue Thimble Flower", 
                               "#3. Mountain Owl's Clover")) +
  scale_y_continuous(labels = scales::label_comma()) +
  theme_minimal() +
  theme(
    # Axis labels
    axis.title.x = element_blank(),
    axis.title.y = element_text(family = "quicksand", margin = margin(10,15,0,10),
                                size = 40),
    axis.text.x = element_text(size = 40, family = "quicksand", face = "bold"),
    axis.text.y = element_text(size = 40, family = "quicksand"),
    
    # Grid lines
    panel.grid.major.x = element_blank(),
    panel.grid.minor = element_blank(),
    
    # Margins
    plot.margin = margin(10, 10, 10, 10),
    plot.background = element_rect(color = "white", fill = "white")
  )

# Lollipop chart for flies ----
hvb |>
  filter(pollinator == "Hover-fly") |>
  slice_max(order_by = visits, n = 3) |>
  mutate(rank = c("A", "B", "C")) |>
  ggplot(aes(x = rank, y = visits)) +
  geom_point(fill = "#ECCED7") +
  geom_linerange(aes(ymin = 0, ymax = visits)) +
  labs(y = "Number of flower visits") +
  scale_x_discrete(labels = c( "#1. Gray's Licorice Root",
                               "#2. Oregon Sunshine", 
                               "#3. Blue Thimble Flower" )) +
  scale_y_continuous(labels = scales::label_comma(),
                     breaks = c(0, 1000, 2000, 3000, 4000)) +
  theme_minimal() +
  theme(
    # Axis labels
    axis.title.x = element_blank(),
    axis.title.y = element_text(family = "quicksand", margin = margin(0,15,0,0),
                                size = 40),
    axis.text.x = element_text(size = 40, family = "quicksand", face = "bold"),
    axis.text.y = element_text(size = 40, family = "quicksand"),
    
    # Grid lines
    panel.grid.major.x = element_blank(),
    panel.grid.minor.x = element_blank(),
    
    # Margins
    plot.margin = margin(10, 10, 10, 10),
    plot.background = element_rect(color = "white", fill = "white")
  )

Footnotes

1. Doyle, T., Hawkes, W. L., Massy, R., Powney, G. D., Menz, M. H., & Wotton, K. R. 2020. Pollination by hoverflies in the Anthropocene. Proceedings of the Royal Society B: Biological Sciences, 287(1927), 20200508.

2. Wotton K. R., Gao B., Menz M. H. M., Morris R. K. A., Ball S. G., Lim K. S., Reynolds D. R., Hu G., Chapman J. W. 2019. Mass seasonal migrations of hoverflies provide extensive pollination and crop protection services. Curr. Biol. 29, 2167-2173.e5.

3. Chisausky, J. L., Soley, N. M., Kassim, L., Bryan, C. J., Miranda, G., Gage, K. L., and Sipes, S. D. 2020. Syrphidae of Southern Illinois: Diversity, floral associations, and preliminary assessment of their efficacy as pollinators. Biodiversity Data Journal, 8, e57331.