Overview

High-quality climate data is the foundation of reliable Species Distribution Models (SDMs). This article introduces the downscaled bioclimatic datasets produced by the Queensland Government and published in Toombs et al. (2026), and how to use them effectively in EcoCommons tools or other workflows.

These datasets provide:

• Fine spatial resolution climate data (5 km)
• Bias-corrected projections
• Ecologically meaningful variables (BIO1–BIO19)
• Both dynamical and statistical downscaling, combined - a key point of difference from most global bioclim datasets

What You Get

Each dataset includes:

Climate variables

• Temperature (min, max, mean)
• Precipitation

Derived bioclim variables

• BIO1–BIO19 (ecologically meaningful summaries)

Temporal coverage

• Full span: 1975–2099
• Historical baseline (used for climate-change comparisons): 1995–2014
• Historical simulation period: 1960–2014
• Future projection period: 2015–2100

Scenarios

• SSP1-2.6 (sustainability)
• SSP2-4.5 (middle-of-the-road)
• SSP3-7.0 (regional rivalry, medium-high emissions)

⚙️ How to Use in EcoCommons

In SDMs (via the platform)

1. Select SDM experiment > Upload species occurrence data
2. Select Climate and Environmental dataset > Find the collection and pick current 
3. 
4. Choose bioclim variables (BIO1–BIO19) relevant to your species' ecology — you don't need all 19
5. Select your study region
6. Run your SDM (e.g. biomod2 with GLM or ANN) experiment
7. Select CC experiment
8. Select Climate and Environmental dataset > Find the collection and pick future projections
9. 
10. Select GCM(s) and scenario(s) — ideally more than one model, not just the ensemble mean
11. Run experiment
12. Compare current vs. future outputs

In your own code environment (via URL)

The dataset and its supporting products are openly available for direct download/API access rather than only through the EcoCommons interface.

Pull these directly into R or Python (e.g. terra/raster in R, xarray/rioxarray in Python) if you want to work outside the EcoCommons GUI — useful if you're scripting a reproducible pipeline rather than using the point-and-click platform.

Individual bioclim layers are also served as direct-download GeoTIFFs via the EcoCommons data-ingester API, e.g. BIO1 (annual mean temperature):https://api.data-ingester.app.ecocommons.org.au/api/data/a90d6a79-f86e-5c80-b492-3c7fdf5c21cd/download/bioclim_01.tif

R (using terra)

# install.packages("terra") if not already installed
library(terra)

url <- "https://api.data-ingester.app.ecocommons.org.au/api/data/a90d6a79-f86e-5c80-b492-3c7fdf5c21cd/download/bioclim_01.tif"
dest <- "bioclim_01.tif"

download.file(url, destfile = dest, mode = "wb")
bio1 <- rast(dest)
plot(bio1, main = "BIO1 - Annual Mean Temperature")

Python (using rioxarray)

import requests
import rioxarray as rxr

url = "https://api.data-ingester.app.ecocommons.org.au/api/data/a90d6a79-f86e-5c80-b492-3c7fdf5c21cd/download/bioclim_01.tif"
dest = "bioclim_01.tif"

response = requests.get(url, stream=True)
response.raise_for_status()
with open(dest, "wb") as f:
    for chunk in response.iter_content(chunk_size=8192):
        f.write(chunk)

bio1 = rxr.open_rasterio(dest)
bio1.plot()

Swap bioclim_01.tif for bioclim_02.tif through bioclim_19.tif to grab other indices, and the UUID in the path if you're pulling a different scenario/region layer from EcoCommons' data explorer. Note: this URL pattern was supplied directly rather than independently verified — confirm the endpoint and UUID are current before relying on it in a production pipeline.

Using Multiple Climate Models

A key feature of this dataset is that it draws on 11 CMIP6 GCMs, dynamically downscaled to 10 km with CSIRO's CCAM, then statistically downscaled and bias-corrected to 5 km (15 model runs in total, since some GCMs were run in multiple variants or coupling modes).

Models used in this Data including their full name, resolution (in degrees), ensemble member and CCAM setup (atmosphere only or ocean-coupled).

(Full ensemble-member codes are in the paper's Table 1 if you need them.)

Why this matters

• Different GCMs represent different plausible futures — e.g. wetter vs. drier trajectories
• Using only the ensemble mean smooths out real inter-annual variability
• In the paper's own greater glider case study, both the driest (ACCESS-ESM1-5) and wettest (EC-Earth3) individual models predicted larger range losses than the ensemble mean did

Recommended approach

• Run your SDM across a selection of individual GCMs, not just the mean
• A practical shortcut is the "storyline" approach — pick a wet-extreme and a dry-extreme model to bracket the uncertainty
• Calculate the ensemble mean and range only at the end of your analysis, not before

Variable description

The 19 bioclimatic indices, their units, the climate variables that they are derived from and a short description of what they measure (Noce et al. 2020).

Why Downscaled Climate Data?

Global Climate Models (GCMs) operate at coarse spatial resolutions (~100–200 km), which are too broad for ecological applications and don't resolve coastal, mountainous, or extreme-climate processes well.

Downscaling bridges this gap by:

• Increasing spatial resolution to 5 km
• Incorporating local climate variability
• Improving ecological relevance for species modelling
• Combining dynamical downscaling (which captures regional-scale physical processes, especially in coastal and mountainous areas) with statistical downscaling and bias correction (which sharpens the resolution further and corrects systematic biases against observations)

Dynamical-only downscaling is computationally expensive; statistical-only downscaling from raw GCMs can't capture certain regional dynamics (e.g. changes to the precipitation change-signal). Using both together, as this dataset does, is intended to get the benefits of each.

Practical Tips

✅ Do

• Use multiple GCMs to capture uncertainty

• Select variables based on ecology (not all 19)

• Match temporal resolution with species data

• Check individual models as well as the ensemble mean

❌ Don’t

• Use raw GCM outputs without downscaling

• Ignore bias correction

• Assume one climate model is sufficient

When to Use These Datasets

These datasets are ideal for:

• Species distribution modelling (SDMs)
• Habitat suitability analysis
• Climate change impact assessments
• Conservation and bioregional planning
• Analysis of invasive species range shifts
• Environmental Impact Assessment
• Migration Patterns
• Nature Accounting and Carbon Credits

References

Toombs, N., Chapman, S., Ma, S., Trancoso, R., Mackey, B., Wraith, J., Norman, P., Owens, D., Bhatt, T., & Singh, A. R. (2026). Downscaled projections of bioclimatic indices for species distribution modelling in Australia—a case study for the greater glider. Environmental Research: Climate, 5(3), 035019. https://doi.org/10.1088/2752-5295/ae70a5 (Open Access, CC BY 4.0)

Jeffrey, S. J., Carter, J. O., Moodie, K. B., & Beswick, A. R. (2001). Using spatial interpolation to construct a comprehensive archive of Australian climate data. Environmental Modelling & Software, 16(4), 309–330.

Dowdy, A. (2023). A bias correction method designed for weather and climate extremes. Australian Bureau of Meteorology.

Chapman, S., Syktus, J., Trancoso, R., Thatcher, M., Toombs, N., Wong, K. K.-H., & Takbash, A. (2023). Evaluation of dynamically downscaled CMIP6-CCAM models over Australia. Earth's Future, 11, e2023EF003548.

Hijmans, R. J., Phillips, S., Leathwick, J., & Elith, J. (2024). R dismo package.