Biodiversity Building Blocks for Policy

The specification was implemented by GBIF as a new service that allows users to query and aggregate species occurrence data with the Structured Query Language (SQL). This well-known language can handle diverse and complex data operations and be used for more than the generation of data cubes. The service also provides a number of functions that shield users from having to create complex aggregation queries themselves (such as grid assignment).

Any registered user of GBIF can use the service. The easiest way to do so is through the GBIF website (https://www.gbif.org/), where one can create a cube through a series of dropdown menus and checkboxes. No SQL knowledge is required for this, but advanced users can switch to a SQL editor to customise their query further.

Users can also interact with the service programmatically using the command line interface or the updated libraries in R (rgbif) and Python (pygbif). As with other GBIF occurrence download formats, all successful queries are assigned unique and permanent DOIs. This enables FAIR citation, attributes data publishers and aids reproducibility of downstream analyses.

The Dissimilarity Cube is a reproducible, multidimensional framework for analysing how biological communities change across space and time. Rather than focusing on the predicted occurrence of individual species, it quantifies variation in assemblage composition, thereby providing a community-level perspective on biodiversity change.

By turning biodiversity records into mappable signals of community change, the Dissimilarity Cube helps identify stable regions, shifting assemblages, and areas at risk of ecological reorganisation. It therefore provides a scalable and policy-relevant approach for monitoring biodiversity structure, detecting emerging novel communities, and supporting spatial conservation planning.

The Network Invasibility Cube is a unified framework for assessing invasion risk by integrating three major dimensions of biological invasions: invader traits, abiotic suitability, and biotic resistance from resident communities. This allows invasion processes to be represented within a single, transparent, and interoperable modelling structure.

The Network Invasibility Cube supports surveillance, prioritisation, and management by highlighting invasion hotspots and high-risk species x trait combinations. Because the workflow is modular and reproducible, it also enables consistent comparison across taxa, regions, and scenarios.

DeepMaxent is a deep learning framework for species distribution modelling that integrates the maximum entropy principle with neural networks. Beyond classical environmental variables, it provides a flexible foundation to incorporate complex and heterogeneous predictors such as remote sensing imagery, climate projections, ecological context, and multimodal biodiversity data. The method relies on a normalised Poisson point-process loss, enabling scalable training with stochastic gradient descent and efficient multi-species modelling while accounting for spatial sampling biases. DeepMaxent improves predictive accuracy compared to classical Maxent and opens new opportunities for large-scale, multimodal biodiversity monitoring using deep learning.

By comparing status and trend estimates from aggregated data cubes with those obtained from structured monitoring data, B-Cubed developed guidelines for producing reliable indicators and trends from occurrence cubes. These guidelines emphasise the importance of clearly defining the scope of the analysis so that data are aggregated at an appropriate level, conducting an exploratory assessment of the aggregated data to identify common gaps or errors, and applying targeted filtering. All filtering, harmonisation, and data-quality decisions should be documented in a transparent and traceable manner.

The pdindicatoR package was developed to facilitate producing phylogenetic diversity (PD) maps from user-provided phylogenetic trees and GBIF species occurrence cubes. The workflow output highlights phylogenetic diversity hotspots and quantifies the proportion of PD currently safeguarded within existing protected areas. The package can be used as a tool to locate hotspots of phylogenetic diversity, guide conservation planning by identifying priority regions for protection and assist with long-term monitoring and reporting on PD conservation status.

The impacts of alien taxa indicators present a straightforward method for estimating the harm caused by alien (non-native) species, using freely available data and a clear, step-by-step process. As non-native species spread and multiply, they can seriously affect local plants, animals, and entire ecosystems, not to mention impacts on agriculture and health. By combining information about where these alien species occur (from online databases) with the evaluations of their ecological effects (from an international standardised process that gathers information on impacts and collates them into a standard score), we developed a practical tool that gives an “impact value.” This value reflects both how widespread a species is and how serious its negative effects can be. Researchers, conservationists, and policymakers can use these values to identify which non-native species are likely to be problematic, allowing managers to address the urgent future problems. This method thus helps guide actions to protect biodiversity and maintain healthy ecosystems into the future.

The R package dubicube supports exploratory analysis, uncertainty quantification, and evidence-based interpretation of biodiversity indicators. It tackles the need to assess the reliability and interpretability of biodiversity indicators derived from species occurrence data by implementing methods to evaluate data robustness, identify limitations in spatial, temporal, and taxonomic coverage, and quantify how sensitive indicators are to individual species. Uncertainty is explicitly estimated using resampling techniques, allowing confidence intervals and effect classifications to support cautious interpretation of trends. Together, these approaches reduce the risk of overconfidence in indicator results and improve transparency in analysis. All proposed methods are implemented in the R package dubicube, which supports exploratory analysis, uncertainty quantification, and evidence-based interpretation of biodiversity indicators.

The b3verse is an integrated ecosystem comprising eleven R packages and the b3data data package. The b3verse consolidates biodiversity indicator workflows into a coherent, reproducible, and open-source framework, covering general biodiversity indicators (e.g. b3gbi), phylogenetic diversity (pdindicatoR), alien species impacts (impIndicator), uncertainty and robustness assessment (dubicube), and modelling tools such as dissmapr and invasimapr. The ecosystem follows strict development guidelines, includes automated testing and continuous integration via R-universe, and is permanently archived on Zenodo. Comprehensive documentation and tutorials ensure accessibility, and the modular design provides the technical foundation for cloud-based dashboards and policy-oriented applications.