FrogID is a citizen science initiative allowing Australians to record and submit frog calls for expert identification. The 2023 dataset represents the sixth annual data release, contributing to over 30 scientific papers on frog ecology, taxonomy, and conservation since 2017. Australia hosts 257 native frog species found nowhere else on Earth — yet almost one in five species are threatened with extinction due to climate change, urbanisation, disease, and invasive species.
Suggested research questions:
Which frog species are endemic to specific Australian regions?
Do different species exhibit distinct seasonal calling patterns?
Which species has the broadest geographic distribution versus the most limited range?
Loading necessary packages
My handy booster pack that allows me to install (if needed) and load my usual and favorite packages, as well as some helpful functions.
# A tibble: 9 × 2
stateProvince n
<chr> <int>
1 New South Wales 58749
2 Victoria 32383
3 Queensland 23334
4 Western Australia 10844
5 South Australia 4158
6 Tasmania 2562
7 Northern Territory 2380
8 Australian Capital Territory 2082
9 Other Territories 129
# Top observed speciescat("\n=== Top 15 Species ===\n")
# A tibble: 11 × 2
month n
<ord> <int>
1 Jan 16715
2 Feb 8934
3 Mar 6569
4 Apr 7248
5 May 4190
6 Jun 6664
7 Jul 9658
8 Aug 15852
9 Sep 20509
10 Oct 17288
11 Nov 22994
The frogID_data table contains 136,621 records spanning January–November 2023, covering 186 unique species across all Australian states and territories. A few immediate structural features stand out:
New South Wales dominates the record count (58,749 — 43% of all observations), likely reflecting both its large population of citizen scientists and the density of the FrogID app’s user base along the eastern seaboard.
Queensland leads in species richness (96 unique species), consistent with its tropical biodiversity, even though NSW contributes far more raw observations.
The seasonal signal is unmistakable. November alone accounts for nearly 23,000 observations — over 16% of the annual total — while April–June are quietest. This spring peak reflects the southern hemisphere breeding season for most temperate species.
Crinia signifera (Common Eastern Froglet) accounts for 33,630 records — nearly 25% of all observations — making it by far the most-recorded species in this dataset.
The frog_names table is a 294-row taxonomic dictionary. Most entries belong to Myobatrachidae (ground frogs, endemic to Australia and New Guinea) and Hylidae (tree frogs). A handful of rows are genus-level entries without a species epithet — these need to be filtered out before joining to avoid many-to-many matches.
Seasonal Calling Phenology: When Do Australia’s Frogs Sing?
Frog calling is tightly coupled to breeding: males call to attract females, so call frequency spikes during the breeding season. In Australia, this creates a stark biogeographic divide:
Note
The north-south seasonal split. Tropical frogs in Queensland and the Northern Territory breed during the wet season (November–April), triggered by monsoon rains. Temperate frogs in NSW, Victoria, and South Australia breed in spring and early summer (August–November), responding to warming temperatures. The FrogID data, collected across 2023, captures both signals simultaneously.
Species richness vs. observation effort by state
state_richness <- frog_data %>%group_by(stateProvince) %>%summarise(n_species =n_distinct(scientificName),n_obs =n(),.groups ="drop" ) %>%arrange(desc(n_species))cat("Species richness and observation effort by state:\n")
Species richness and observation effort by state:
print(state_richness)
# A tibble: 9 × 3
stateProvince n_species n_obs
<chr> <int> <int>
1 Queensland 96 23334
2 New South Wales 70 58749
3 Western Australia 52 10844
4 Northern Territory 33 2380
5 Victoria 28 32383
6 South Australia 20 4158
7 Australian Capital Territory 12 2082
8 Tasmania 11 2562
9 Other Territories 8 129
Queensland has 96 unique species — more than triple Victoria’s 28 — yet NSW generates 2.5× more observations. This gap between where frogs are and where people are looking is the classic citizen-science sampling bias. The FrogID team explicitly accounts for this in their conservation analyses.
Constructing the seasonal calling calendar
# Filter frog_names to binomial species only to avoid genus-level many-to-many joinfrog_names_sp <- frog_names %>%filter(str_detect(scientificName, " ")) %>%select(scientificName, commonName) %>%distinct(scientificName, .keep_all =TRUE)# Identify top 15 species by observation counttop15_species <- frog_data %>%count(scientificName, sort =TRUE) %>%head(15) %>%pull(scientificName)cat("Top 15 species selected for heatmap:\n")
# Determine peak calling month for each species (for ordering)peak_month <- monthly_matrix %>%group_by(scientificName) %>%slice_max(pct, n =1, with_ties =FALSE) %>%select(scientificName, peak_month = month_num)# Join common namesmonthly_matrix <- monthly_matrix %>%left_join(frog_names_sp, by ="scientificName") %>%left_join(peak_month, by ="scientificName") %>%mutate(# Use common name where available; fall back to italicised scientific namedisplay_name =case_when(!is.na(commonName) & commonName !="—"~paste0(commonName, "\n*", scientificName, "*"),TRUE~paste0("*", scientificName, "*") ),# Order species by peak calling monthdisplay_name =fct_reorder(display_name, peak_month),month_abbr =factor( month.abb[month_num],levels = month.abb ) )cat("\nmonthly_matrix after join and mutate: ", nrow(monthly_matrix), "rows\n")
1. Spring is frog season — for most species. The dashed box covering August–November captures the peak calling period for the majority of the top-15 species. This aligns with the southern hemisphere spring, when temperatures warm and rains arrive across eastern Australia. Crinia signifera (Common Eastern Froglet), which alone accounts for 25% of all observations, is remarkably evenly spread across the year but shows a sharp November peak.
2. The tropical outliers stand out.Uperoleia species and Queensland-heavy frogs (appearing lower in the ordered list) show relatively flat or early-year distributions, hinting at the wet-season breeding calendar of tropical Australia — though the citizen-science sampling bias toward NSW blunts this signal in the aggregate data.
3. Citizen science has a seasonality of its own. The November surge isn’t only frogs calling more — it’s also people going outside more. Disentangling phenological signal from observer effort is a persistent challenge in FrogID analyses. The research team publishes annual corrections for this bias, which is partly why FrogID requires expert acoustic verification rather than relying solely on observer labels.
4. 186 species in a single year is remarkable — and not enough. The 2023 dataset captures less than three-quarters of Australia’s 257 native frog species. Many threatened species are either in remote areas with few citizen scientists or have such small remaining populations that detection is rare. The FrogID app is actively working to close this gap, but it also underlines why the 33,000+ Crinia signifera records coexist with species recorded only once or twice in the entire dataset.
Tip
For your own exploration: The latitude and longitude fields in frogID_data are rich but deliberately coarsened (10,000m uncertainty buffers) for sensitive species. Even so, mapping species range limits — particularly the northern boundary of temperate species or the southern edge of tropical ones — against the calling calendar would be a compelling follow-up.
