Dataset DOI: 10.5061/dryad.ncjsxkt9w

Description of the data and file structure

This dataset was developed to support a trait-based assessment of fire regime risk to threatened fauna in New South Wales, Australia. The data were generated to characterise how key components of fire regimes (time since fire, fire frequency, and fire severity) interact to influence species persistence, particularly in contexts where species-specific empirical data are limited.

Data were compiled through a combination of structured expert elicitation and targeted literature synthesis. Expert elicitation workshops were used to group species with similar ecological characteristics and fire response patterns, and to develop qualitative fire response curves describing relative abundance trajectories across fire regime gradients. These outputs were subsequently refined and standardised using available trait databases and published literature.

Species were classified into 18 fire response functional groups (FRGs), representing shared patterns of survival, recovery, and abundance following fire. These groups were further synthesised into five overarching fire response strategies (FRSs), capturing consistent patterns of resilience and vulnerability across fire regime components.

This dataset contains the R Markdown script and its input data that together reproduce the fire response curve construction and the k-means clustering of FRGs into the five overarching FRS reported in the associated manuscript.

Reference material describing the species-to-group assignment rationale, fire response curve rationale, FRG definitions and trait framework, and the FRS summary/management guidance (previously circulated as working spreadsheets numbered S1–S5) is published as Supplementary Information with the manuscript on the journal's website rather than duplicated here, since that content is not required to run or reproduce the analysis in this script.

Manuscript status: accepted for publication, 21 August 2026.

Files and variables

File: CODE_FireResponseGroups_curves_clean.Rmd

R Markdown script that reproduces the fire response curve construction (from workshop-elicited data) and the k-means clustering of fire response functional groups (FRGs) into the five overarching fire response strategies (FRS) reported in the manuscript. The script's prose labels each figure it discusses with its corresponding manuscript figure number (Figure 1 through Figure 7) for reference, and points the reader to the manuscript's Supplementary Information tables for further detail (species assignment rationale, curve rationale, FRG definitions, FRS summary). No figure image files are bundled with this dataset, and several code chunks reference image filenames via markdown image embeds; without those image files present, the document will knit with broken image links at those points, though the underlying analysis chunks (curve construction, clustering) run and produce their own output independently of those references.

The script's file-reading chunks read DATA_all_species_F.csv, DATA_FaunaGroupings.csv, DATA_TSFopt.min.csv, DATA_Severityopt.min.csv, and DATA_Frequencyopt.min.csv — all five are bundled with this dataset (see below), so the script's active analysis chunks run without any external inputs.

One chunk (methods plot, building the Figure 1 study-area map) is not evaluated on knit (eval = FALSE) and is provided for provenance/reference only — it is not required to reproduce any result or figure in the manuscript, since Figure 1 is a finished/composited graphic (see its caption in the script). It draws on two external, publicly available shapefiles not bundled with this dataset:

• IBRA7 subregions (NSW/ACT): https://fed.dcceew.gov.au/datasets/5b69db6a36f1429a84ea07dd4616eb30/about
• World Heritage Areas (used to extract the Greater Blue Mountains Area):
  http://www.environment.gov.au/fed/catalog/search/resource/downloadData.page?uuid=%7B6C54FE6C-2773-47C6-8CBC-4722F29081EF%7D

Required R packages: tidyverse, dplyr, tidyr, ggplot2, ggh4x, RColorBrewer, plotly (sf, ggrepel, and ggpattern are referenced by the non-evaluated map-figure chunk but not required to run the rest of the script).

Note on package versions: the interactive plotly figures in this script were tested against ggplot2 3.5.2 and plotly 4.10.4. ggplot2 4.0.0 (a major internal rewrite from S3 to S7 classes) breaks ggplotly() with "Error in pm[[2]] : subscript out of bounds," because plotly's converter relies on ggplot2's internal structure rather than a stable public API. Users hitting this error should pin ggplot2 to version 3.5.2 (e.g., via remotes::install_version) before re-running the script.

File: DATA_all_species_F.csv

Fauna fire response curve data (relative abundance estimates), derived from expert elicitation workshops, for each of the 18 fire response functional groups. Each row gives the relative abundance estimate for one FRG at one value of one fire regime component. There are 37 rows per group (18 groups x 37 rows = 666 data rows), corresponding to the points used to draw each group's response curve.

Variables:

• Group: Fauna fire response functional group, Group 1 to Group 18. See the manuscript's Supplementary Information for group definitions and species-to-group assignments.
• Type: The fire regime component the row belongs to: TSF (time since fire), Severity, or Frequency.
• Xaxis: The value of the fire regime component on the x-axis of the response curve. Units/scale depend on Type. TSF is continuous, in years since last fire (observed values: 0, 1, 2, 3, 4, 5, 6, 8, 10, 12, 16, 20, 24, 28, 32, 36, 40). Severity is categorical, linked to Fire Extent and Severity Mapping (FESM) categories, coded 0-8 (even values are the named FESM categories: 0 = Unburnt, 2 = Low, 4 = Moderate, 6 = High, 8 = Extreme; odd values 1, 3, 5, 7 are intermediate transition points added for curve resolution). Frequency is an integer count of fires occurring in the preceding 50-year period, 0-10.
• RelativeAbundAve: Expert-elicited mean/mode relative abundance estimate at that point on the curve, where 0 = complete loss of the group and 1 = theoretical carrying capacity.
• RelativeAbundLow: Lower bound of the elicited relative abundance estimate.
• RelativeAbundUp: Upper bound of the elicited relative abundance estimate.

File: DATA_FaunaGroupings.csv

FRG-to-trait-axis lookup table, read by the script to facet the fire response curve plots by each FRG's defining trait combination.

Variables:

• Group: Fauna fire response functional group, Group 1 to Group 18 (matches Group in DATA_all_species_F.csv).
• Flee: Whether the group can flee/seek fire-resilient shelter during the initial fire event (Yes/No).
• Ecosystem: Ecosystem specialisation and recovery time — ES-s = ecosystem specialist, short-term recovery (<2 years); ES-l = ecosystem specialist, long-term recovery (>2 years); EG = ecosystem generalist.
• Resource: Resource/shelter specialisation and recovery time — RS-s = resource specialist, short-term recovery (<2 years); RS-l = resource specialist, long-term recovery (>2 years); RG = resource generalist.
• Group_summary: A plain-English restatement of the Flee/Ecosystem/Resource codes above for that group, so each row is self-explanatory without needing to cross-reference the code legend.

For the broader synthesis of these 18 FRGs into five overarching fire response strategies (FRS) — including a qualitative description of each strategy's typical response pattern and which FRGs belong to it — see the "Figure 7: fire response strategies" section of CODE_FireResponseGroups_curves_clean.Rmd. That synthesis is part of this script's own narrative text (not a separate table), so it is fully included in this dataset.

Files: DATA_TSFopt.min.csv, DATA_Severityopt.min.csv, DATA_Frequencyopt.min.csv

Per-FRG optimal and minimal relative abundance coordinates for each fire regime component, used as the input to the k-means clustering step (ordinations chunk) that derives the five overarching fire response strategies (FRS) from the 18 FRGs.

Variables (each file):

• Group: Fauna fire response functional group ("Fauna Group 1"–"Fauna Group 18").
• TSF / Severity / Frequency: The value of the corresponding fire regime component at which the group's relative abundance is at its optimal (maximum) or minimal (minimum), taken from the elicited response curve in DATA_all_species_F.csv.
• RelativeAbundAve: The optimal or minimal relative abundance value itself.
• State: Whether this row is the group's "Optimal" or "Minimal" coordinate.

Reproducing the analysis

Open CODE_FireResponseGroups_curves_clean.Rmd in RStudio and knit, or run chunks interactively, with this folder as the working directory. See the file description above for required R packages and the ggplot2 version-compatibility note.

Code/software

An R Markdown script (CODE_FireResponseGroups_curves_clean.Rmd) and its input data (DATA_all_species_F.csv, DATA_FaunaGroupings.csv, DATA_TSFopt.min.csv,
DATA_Severityopt.min.csv, DATA_Frequencyopt.min.csv) are included to reproduce the fire response curve construction and FRG clustering analyses reported in the manuscript.

Access information

These data are provided for reuse in ecological modelling, fire regime scenario analysis, and conservation planning. The dataset is particularly suited to applications requiring generalised or trait-based representations of species responses to fire in data-limited contexts.

Users should note that:

• Fire response curves and group assignments are based on expert elicitation supported by available literature and should be interpreted as informed approximations rather than precise empirical measurements.
• Management guidance derived from this dataset is indicative and context-dependent; see the manuscript's Supplementary Information for the full fire response strategy summaries and management directives.
• No figure image files are bundled with this dataset; figures referenced by number in the Rmd script are available in the published manuscript.
• Reference material on species-to-group assignments, curve construction rationale, FRG definitions, the trait framework, and FRS management guidance is published as Supplementary Information with the manuscript rather than included here, as it is not required to run or reproduce this script.
• No sensitive or restricted data are included in this dataset.

Please cite the associated publication when using these data.

Funding

NSW Department of Climate Change, Energy, the Environment and Water

Contact

Victoria A. Reynolds — victoria.reynolds@unimelb.edu.au