Biodiversity loss has been among the central topics in public, policy, and scientific debate for several decades. While its accelerating pace and impacts are well recognised, translating the vast amounts of biodiversity data into reliable indicators that can inform policy remains a challenge, requiring considerable time and technical effort.

That’s where B-Cubed comes in. Through our information campaign, “How to Be Cubed,” we will explore how our automated data pipelines transform raw biodiversity data into meaningful indicators for decision-makers. We will follow the pathway from building species occurrence cubes, to modelled cube approaches, and biodiversity indicators, exemplifying how our data pipelines support biodiversity monitoring. You can learn all about our project results from our legacy booklet.

Filling the gaps with modelled cubes

B-Cubed developed a set of modelled cube approaches that extend biodiversity data from aggregation and reporting to prediction, interpretation, and decision support. Rather than only summarising where and when biodiversity records occur, they help explain ecological patterns, compare communities, and anticipate emerging risks.

What is a modelled cube?

The modelled cubes developed by B-Cubed address complementary ecological questions at different levels of organisation. There are four types of modelled cubes:

• Suitability Cube: provides a reproducible and transparent framework for exploring how species occupy environmental space, identifying areas of high uncertainty or extrapolation, and evaluating the robustness of model predictions through space and time.

• Dissimilarity Cube: turns biodiversity records into clear, mappable signals of community change, thereby helping identify stable regions, shifting assemblages, and areas at risk of ecological reorganisation. It also formalises multi-site dissimilarities in a consistent data-cube structure, making change comparable across regions, taxa, and time.

• Network Invasibility Cube: converts complex invasion ecology into decision-ready indicators by identifying which invaders are most likely to establish, where invasions are most likely, and why (which traits and conditions drive risk). This enables more targeted surveillance, pathway control, and site-level mitigation, while remaining comparable across taxa, regions, and scenarios.

• DeepMaxent: a deep learning framework for species distribution modelling that integrates the maximum entropy principle with neural networks and enables the incorporation of complex and heterogeneous predictors, such as remote sensing imagery, climate projections, ecological context, and multimodal biodiversity data.

Each modelled cube produces indicators that describe not only a predicted pattern (such as species suitability, community dissimilarity, or invasion risk) but also the confidence with which that pattern can be interpreted. This dual focus on prediction and reliability distinguishes the B-Cubed approach from conventional modelling workflows, where uncertainty is often reported separately or not at all.

To learn how modelled cubes work, you can check out the tutorial for using a Suitability Cube, a Dissimilarity Cube, and an Invasibility Cube.

How can modelled cubes benefit biodiversity policy?

The cube framework provides scalable, reproducible, and decision-relevant outputs that align with emerging biodiversity monitoring frameworks such as the Post-2020 Global Biodiversity Framework.

• The Suitability Cubes help identify priority areas for conservation and restoration.

• The Dissimilarity Cubes support bioregional classification and tracking of community-level change.

• The Invasibility Cubes enable early warning and targeted management of invasion risks.

Together, these modelled cube approaches show how occurrence cubes can underpin a new generation of transparent, scalable, and reproducible biodiversity modelling workflows.

We would love to hear your feedback on B-Cubed’s modelled cube approaches, so feel free to email us at info@b-cubed.eu or check out the GitHub pages for Suitability Cubes, Dissimilarity Cubes, and Invasibility Cubes.

In case you have any other questions, check out our FAQ page.