Barrambin Bioblitz impact

In 24 hours, we went from:

• 794 to 1026 observations (+232)
• 309 to 339 species (+30)
• 1443 to 1680 identifications (+237)
• 150 to 164 observers (+14)
• 8 to 17 project members (+9)

But the effort didn’t stop on the 23rd of May: at the time of writing, there are 1069 observations in the project.

barr_obs$created_at <- as.Date(barr_obs$created_at)
library(dplyr)
cumulative <- barr_obs |> 
    mutate(rg = quality_grade == "research") |> 
    count(created_at, rg) |> 
    arrange(created_at) |> 
    group_by(rg) |> 
    mutate(cumulative_obs = cumsum(n)) |> 
    ungroup()
library(ggplot2)
theme_set(theme_bw())
ggplot(cumulative, aes(x = created_at, y = cumulative_obs, fill = rg)) +
    geom_vline(
    xintercept = as.Date("2026-05-23"),
    linetype = "dashed"
    ) +
    geom_area() +
    geom_point(x = max(barr_obs$created_at), y = nrow(barr_obs), inherit.aes = FALSE) +
    annotate("label", x = max(barr_obs$created_at) - 80, y = nrow(barr_obs),
    label = paste("Total:", nrow(barr_obs), "observations."), hjust = "right") +
    scale_fill_discrete(labels = c("Others", "Research")) +
    labs(
        x = "Date",
        y = "Cumulative observations",
        fill = "Quality grade")

So what new species were added recently? Looking at observations that have a species-level ID, and including observations recorded during and after the Bioblitz, we added 45 to the list:

library(stringr)
species <- barr_obs |> 
    filter(taxon.rank == "species") |> 
    select(created_at, taxon.common_name.name, taxon.name, taxon.id) |> 
    # mark species that were seen after cutoff date but not before
    mutate(
        before = created_at < "2026-05-22",
        after = created_at >= "2026-05-22") |> 
    group_by(taxon.id) |> 
    mutate(only_after = !any(before) & any(after)) |> 
    ungroup()
# only keep the new ones, one row per species
new_species <- species |> 
    filter(only_after) |> 
    select(taxon.name, taxon.common_name.name, taxon.id) |> 
    unique() |> 
    # tidy up
    rename(
        scientific.name = taxon.name,
        common.name = taxon.common_name.name
        ) |> 
    mutate(common.name = str_to_sentence(common.name)) |> 
    arrange(scientific.name) |> 
    # construct link to iNat taxon
    mutate(taxon.id = paste0(
        "<a href='https://www.inaturalist.org/taxa/",
        taxon.id,
        "'>",
        taxon.id,
        "</a>"))
# print table
library(gt)
new_species |> 
    tibble::rowid_to_column() |> 
    gt(rowname_col = "rowid") |> 
    fmt_markdown(taxon.id) # ro render HTML

Keep in mind that the newly-spotted species listed above are not necessarily all confirmed with several identifications. Here are some other places to explore what has been recorded at Barrambin:

• The full list of taxa observed at this location, starting with the most commonly seen
• A list of all the “Research Grade” observations
• The project’s ‘Identify’ page, on which you can help refine the data by adding your own identifications

More biodiversity data on the ALA

For a more exhaustive list of species recorded at Barrambin, the Atlas of Living Australia (ALA) collates observations from various sources – including the “Research Grade” observations from iNaturalist. For the same area used in our Barrambin project, at the time of writing, the ALA lists 497 taxa, including 21 “threatened species”.

Remember: by contributing more observations to the project, and identifying existing observations, we can continue building up a more accurate and exhaustive record of biodiversity at Barrambin, which eventually feeds into authoritative databases like the ALA.

Protecting Barrambin

Here are other ways you can continue protecting Barrambin:

• Sign the community submission to support the Section 10 Applications, before 5 pm, 29th of May
• Donate to the Goori Camp Embassy, join the camp between 29th of May and 2nd of June, and follow them on Instagram
• Join the protest on Sunday 31st of May, 12 pm, outside the Herston Road entrance.
• Join actions and learn more with Save Victoria Park

iNaturalist: what is it?

iNat is a network of enthusiasts recording sightings of anything alive. No need to have any background in ecology, taxonomy or botany. All you need is something to take photos, and some interest in the natural world that surrounds us.

People upload “observations” into iNat, usually using a phone app. In most cases, it’s a picture of what you saw, along with its location and when it was recorded (but you can also submit a sound, e.g. a bird or frog call). The app gives you a suggestion of what the species might be (based on the pictures).

Once it’s on iNat, others can see it and contribute identifications. The data becomes particularly useful once several people agree on which species was observed - that’s the kind of data scientists end up using for studying biodiversity, population dynamics…

So far, more than 300 million observations have been uploaded to iNat! That’s more than 550,000 different species.

iNat is a non-profit organisation, the apps are open source, and the data is open data anyone can make use of.

Getting started with iNat

1. Download the app on your phone (from the App Store on iOS, or from Google Play on Android)
2. Open the app and follow the steps to create an account
3. Log in

Once logged in, you can start recording what you spot:

1. Tap the green “new observation” button
2. Take a photo (or more)
3. The location and time are automatically added to your observation (but you can double-check that the location is accurate enough)
4. Tap the “What did you see?” area so you can pick a suggestion from the list (or search for the common name / species if you already know it)
5. Save and upload the observation
6. Repeat!

As an alternative (or complement) to the mobile app, you can use a more serious camera to take photos - particularly useful if you want to get good photos of birds, or do some macro photography of tiny organisms. In this case, you can upload your photos from your computer later, on the iNat website - but make sure you record the location of each observation if your camera doesn’t automatically attach it to your photos!

Barrambin Bioblitz

We want to record as many species as we can at Barrambin / Victoria Park! In particular, we want to have good coverage of the area GIICA is planning to fence off on the 1st of June 2026.

You can record anything that’s alive. That means: mammals, insects, birds, spiders, fish, fungi, lichen, mould, trees, shrubs, grasses, weeds… just don’t record other humans!

To get good coverage, and potentially spot more species, remember that crowd-sourced biodiversity can be biased in various ways. Most observations are:

• Close to where people live and travel (in cities, along roads…)
• Made during the day
• Depicting easy to approach, easy to spot organisms. For example: the ubiquitous Water Dragon, Brush Turkey and Australian Ibis.
• Depicting particularly noticeable, or “beautiful”, or rare, or popular organisms. For example: a colourful lorikeet; a cuddly koala; a showy flower.

So if you are able to, try also recording:

• Nocturnal animals (come back in the evening with a good torch) and early risers too
• The “weeds” and the “boring” ones we see everywhere
• In hard-to-reach areas, or inside patches of denser vegetation
• Tree-dwelling and aquatic organisms
• Tiny animals and plants your wouldn’t notice without stopping and looking very closely. Do you maybe have a camera that takes good macro photos?

We have an iNaturalist project that automatically collects the observations from the Barrambin area. You can “join” the project right now to get the latest news, find observations that still need IDs, or to just keep an eye on the stats to see how we are progressing!

Beyond the blitz

You can contribute to this project at any time, you don’t have to wait for the event, nor stop when it’s over!

• Do you have old wildlife photos taken at Barrambin? You can create observations from them on your computer, as long as you have an approximate time and location.
• Before the event, feel free to go to the park and contribute observations.
• After the event, please continue to build this valuable dataset! Record what you see every time you go through the park.

Thank you for your invaluable contributions! They can make a huge difference in sending a strong message about what is at stake, and to keep GIICA and BCC accountable.

Other ways to save Barrambin

• Sign the community submission to support the Section 10 Applications
• Donate to the Goori Camp Embassy and join the camp between 29th of May and 2nd of June
• Join actions and learn more with Save Victoria Park

Setting up

To follow this tutorial, you will have to have a number of packages available, which can be best sorted out with the following command:

install.packages(c("dplyr", "ggplot2", "sfnetworks", "tmap", "osmdata"))

If you run into issues installing and running sf (which relies on spatial libraries external to R), please refer to their installation instructions.

sfnetworks

The main package this tutorial centres around is the sfnetworks package (Lucas van der Meer, Lorena Abad et al.), which is the result of joining what the tidygraph package does for networks with what the sf package does for spatial vector data.

The data structure central to this package is a combination of two spatial objects, one describing the nodes (or points) and another describing the edges (or lines connecting the points). In the words of the developers:

“A close approximation of tidyness for relational geospatial data is two sf objects, one describing the node data and one describing the edge data.”

In other words, an sfnetwork object is made up of one sf object for nodes and one sf object for edges. (As opposed to two simple data.frames for a tidygraph object.)

In the edges component of a network, it is also possible (or necessary) to specify where the edge starts and finishes, using the from and to columns. Such information can be interpreted as the direction of the edge in a “directed” network, or simply as its two extremities in the case of an “undirected” network.

Finally, another important characteristic of the edges is their weight. This can relate to any kind of information, but if we are interested in distances between points, it will typically be the length of the edges.

From OSM data to an sfnetwork

Although the structure of an sfnetwork seems to be quite complex, it is possible – and useful! – to use a shortcut and create one from a simple collection of linestrings.

We can download download data from a spatial vector database like OpenStreetMap (OSM), focus on the lines (or “ways”) and convert that sf object into an sfnetwork that will use the lines as edges, and their endpoints as nodes.

Let’s first download from OSM all the ways that are primarily used by pedestrians in the suburb of West End. (To understand how these features are tagged in the database, a good starting point is the Map Features page on the OSM wiki.)

library(osmdata)
library(dplyr)

# build the query
we_foot <- opq("west end, meanjin") %>% 
  add_osm_features(features = c('"highway"="footway"',
                                '"highway"="steps"',
                                '"foot"="yes"',
                                '"highway"="living_street"')) %>% 
  osmdata_sf()

The osmdata package allows building Overpass queries to download OSM vector data that matches our criteria:

• Features tagged as highway=footway, including footpaths, crossings…
• Features tagged as highway=steps, for stairs
• Features tagged as foot=yes, for example bikeways on which foot traffic is allowed (see this shared bikeway for an example)
• Features tagged as highway=living_street, a shared road on which pedestrians often have right of way

Finding the right combination of OSM tags to use is always an iterative process, refining the selection at each step.

We choose to return an sf object, which might contain points, lines and polygons. We are only interested in the lines:

names(we_foot)
we_foot_lines <- we_foot$osm_lines

We can visualise the resulting object with the sf package:

library(sf) # load sf for st_geometry, but also before using any dplyr methods 
# for sf objects (like filter() and select())
# see https://github.com/ropensci/osmdata/issues/357#issuecomment-2962230662
we_foot_lines %>%
  st_geometry() %>% 
  plot()

It seems to have kept a ferry route, which was tagged with foot=yes and route=ferry on OSM. We can remove it:

we_foot_lines <- we_foot_lines %>% 
  filter(is.na(route))

Another caveat in our process is that some footpaths are mapped as closed ways on OSM, and are therefore returned as a polygon:

we_foot$osm_polygons %>% 
  st_geometry() %>% 
  plot()

The smaller ones are proper pedestrian areas, but the bigger ones are probably footpaths encircling a whole residential block. It is possible to “cast” these polygons as ways and include them in the network:

# cast polygons to lines
poly_to_lines <- st_cast(we_foot$osm_polygons, "LINESTRING")
# bind all lines together
we_foot_lines <- bind_rows(we_foot_lines, poly_to_lines)
# remove columns with problematic names
# (there would be an error when we use to_spatial_smooth)
we_foot_lines <- select(we_foot_lines, !contains(":"))

This is what we are left with:

# plot it
we_foot_lines %>% 
  st_geometry() %>% 
  plot()

We can now convert the object to an sfnetwork object, making sure we set it as undirected (the default is directed = TRUE):

library(sfnetworks)
foot_net <- as_sfnetwork(we_foot_lines, directed = FALSE)
plot(foot_net)

Prepare the network

st_crs(foot_net)

foot_net <- st_transform(foot_net, 7856)

library(tidygraph)
foot_net <- convert(foot_net, to_spatial_subdivision)

foot_simple <- convert(foot_net, to_spatial_smooth)
plot(foot_simple)

foot_net <- foot_net %>% 
  activate("edges") %>% 
  mutate(weight = edge_length())

library(ggplot2)
ggplot() +
  geom_sf(data = st_as_sf(foot_net, "edges"),
          mapping = aes(colour = as.numeric(weight))) +
  labs(colour = "Edge length (m)")

Interactive map

At this point, it might be interesting to create an interactive visualisation to zoom into, which might help spot issues with the data.

library(tmap)
tmap_mode("view") # set to interactive mode
tm_basemap("CartoDB.Positron") +
tm_shape(st_as_sf(foot_net, "edges")) +
  tm_lines(col = "footway", palette = "Accent", colorNA = "red") +
tm_shape(st_as_sf(foot_net, "nodes")) +
  tm_dots()

foot_net <- foot_net %>% 
  activate(nodes) %>% 
  mutate(neighbourhood = local_size(order = 6)) %>% 
  filter(neighbourhood > 5)

Isodistances

Let’s now draw an isodistance around the Kurilpa Library in West End. First, download the library’s feature from OSM, and calculate its centroid (because it is mapped as a building):

kurilpa_lib <- opq_osm_id(id = 523925261, type = "way") %>% 
  osmdata_sf() %>% 
  .$osm_polygons %>% 
  st_centroid() %>% 
  st_set_crs(4326) %>% # (if the following step generates a GDAL error)
  st_transform(crs = 7856)

Then, calculate the isodistance for a distance smaller or equal to 1 km (more or less a 15-minute walk), using the node_distance_from() function from tidygraph:

foot_net <- activate(foot_net, "nodes")
iso <- foot_net %>%
  mutate(dist_to_lib = node_distance_from(
    st_nearest_feature(kurilpa_lib, foot_net),
    weights = as.numeric(weight))
    ) %>% 
  dplyr::filter(dist_to_lib <= 1000)

Finally, draw a polygon around the isodistance, and plot everything:

iso_poly <- iso %>%
  st_geometry() %>%
  st_combine() %>%
  st_convex_hull()
plot(foot_net, col = "grey")
plot(iso_poly, col = NA, border = "black", lwd = 3, add = TRUE)
plot(iso, col = "lightgreen", add = TRUE)
plot(kurilpa_lib, col = "red", pch = 8, cex = 2, lwd = 2, add = TRUE)

Shortest path

Let’s now calculate the shortest distance from the South Brisbane Sailing Club to the library.

# get the location of the Sailing Club's entrance
sailing_club <- opq_osm_id(id = 7622867925, type = "node") %>% 
  osmdata_sf() %>% 
  .$osm_points %>%
  st_set_crs(4326) %>% # (if the following step generates a GDAL error)
  st_transform(crs = 7856)

# calculate the shortest path
shortest <- foot_net %>% 
  activate(edges) %>% 
  st_network_paths(from = sailing_club, to = kurilpa_lib)

# extract the node IDs
node_path <- shortest %>%
  slice(1) %>%
  pull(node_paths) %>%
  unlist()

# only keep the network for these nodeIDs
path_sf <- foot_net %>% 
  activate(nodes) %>% 
  slice(node_path) %>% 
  st_as_sf("edges")

# visualise
tm_basemap("CartoDB.Positron") +
tm_shape(path_sf) +
  tm_lines(col = "red") +
tm_shape(sailing_club) +
  tm_dots(col = "blue", size = 0.1) +
tm_shape(kurilpa_lib) +
  tm_dots(col = "green", size = 0.1)

The granularity of the OSM data available for this area allows us to create a precise path for pedestrians, switching sides of roads only when a crossing is reached. However, every new analysis of the network might reveal more interesting information about the data we have used. For example, the shortest path calculated might go through private residential paths (usually blocked by gates), and avoiding those would mean extra steps in pre-processing (for example keeping track of existing OSM tags like barrier=* on nodes and access=private on ways).

Further resources

• sfnetworks’ documentation, including its 5 articles, which inspired much of this tutorial
• tidygraph’s gigantic collection of graph manipulation functions
• Transportation chapter in the free Geocomputation with R book (Lovelace, Nowosad, Muenchow)
• Spatial network analysis with the {sfnetworks} package, video on analysing OSM data with sfnetworks, by Renate Thiede, for R-Ladies Johannesburg (2021)
• Spatial networks in R with sf and tidygraph, article by van der Meer, Abad and Lovelace (2019)
• Other related packages:
  • ggraph for graph visualisation
  • stplanr for sustainable transport planning and modeling
  • cppRouting for calculating distances, shortest paths and isochrones/isodistances
  • dodgr for distance and time calculations on directed graphs (and its vignette on street network routing based on OSM data)
  • osrm for using a routing API based on OSM data

R session details

System and packages used to run the code when this page was last rendered.

sessionInfo()

Legal

Apart for illustrations in which the copyright is mentioned, this article is released under a CC-BY 4.0 licence. All data used for the data visualisations use OpenStreetMap data, which is © OpenStreetMap contributors but released under an ODBL licence.