DOI: 10.3897/biss.10.183602

Authors: Kenzo Milleville, Dieter De Witte, Maarten Trekels, Quentin Groom, Steven Verstockt

Abstract: The digitization and online availability of herbarium specimens have increased the need for automated tools to extract specimen traits at scale. Quantitative trait data, such as leaf area or petal length, are critical for ecology, evolution, and climate change research. In recent years, automated methods such as LeafMachine2 (Weaver and Smith 2023) and Mothra (Wilson et al. 2022) have been developed to measure these traits from images, supporting large-scale studies. However, such image measurements need to be converted from pixels to standardized units (e.g., centimeters) to be useful. Existing methods estimate this conversion factor (CF) by detecting the ruler type and the tick mark locations directly. From these image coordinates, the average pixel distance between them can be calculated to estimate the CF. These methods typically require many annotated samples for each ruler type and tend to perform poorly when applied to unseen or visually distinct rulers.To overcome these limitations, we present a generic pipeline that accurately predicts the CF with few labeled rulers. The approach leverages the fact that many institutions reuse identical rulers, often differing only in institutional logos. Instead of directly classifying or predicting the tick mark locations, the pipeline uses a similarity-based approach. First, rulers on the page are detected using a YOLOv11 object detection model. Second, these rulers are cropped and normalized, ensuring the longest side is horizontal. After detection, we use Contrastive Language-Image Pre-training (CLIP) to embed each ruler and to retrieve the three most similar ruler candidates from our ground-truth dataset. Next, we use the LightGlue keypoint-matching model to detect corresponding keypoints between the query ruler and each retrieved candidate. This model detects and scores similar keypoint pairs (pixel coordinates) between two input image pairs. The candidate ruler with the highest average score across the 100 best keypoint pairs is selected as the best match. If none of the candidates had at least 10 matching keypoint pairs, we consider the query ruler a “no-match”. Finally, these keypoint pairs are used to estimate a homography, which describes how one image can be transformed to align with another taken from a different viewpoint. This transformation is then applied to the ground truth tick mark locations to calculate the CF for the query ruler.To evaluate the proposed method, we constructed a new dataset by randomly selecting 400 herbarium sheets from the Global Biodiversity Information Facility (GBIF), PlantCLEF2020, and Dillen et al. 2019. Each ruler was annotated with its bounding polygon and a type. Ruler tick marks were annotated with their coordinate locations and units (e.g., ticks_1cm). The dataset contains a diverse set of 476 rulers from over 40 institutions. We estimated our measurement error at 0.794%, which is in line with the reported 0.8% from LeafMachine2, by calculating the average standard deviation of pixel distances between each tick mark. We evaluated our approach using a leave-one-out cross-validation, treating each ruler crop as a query and the remaining dataset as candidate matches. We also evaluated our method on the 1131 rulers from LeafMachine2, which were annotated with a bounding box around the tick marks (see Fig. 1).Rulers without a matching type were excluded for evaluation. The method achieved an average CF error of 0.689% using the best-matching ruler. 94.8% of rulers had a CF error less than 2%. On the LeafMachine2 dataset, performance improved with an average CF error of 0.324%. There, 99.0% of rulers had a CF error less than 2%. We plan to investigate how different annotation strategies affect performance. Table 1 details these evaluation results. The proposed method offers several advantages over existing approaches. First, it allows an accurate conversion with fewer annotated rulers of the same type, essentially needing only one. Second, the keypoint matching process is robust to partial detections, occlusions, and low-resolution rulers, as visualized in Fig. 2. Third, only the detection model needs minimal fine-tuning when new ruler types are added to the dataset, as the matching models remain unchanged. Additionally, the prediction can still be accurate when the best-matching ruler is from a different institution (see Fig. 3).The main limitation of our method is that it requires at least one annotated ruler of a similar type to result in a good match. It also assumes the tick marks are on a straight line, resulting in poorer performance for curved rulers. Finally, we found that simple rulers, such as straight lines, tend to perform worse (see Fig. 3). This work represents part of our ongoing research to assist trait extraction at scale. We plan to expand the dataset with additional ruler types, including examples from other natural history collections. We will further refine the method to improve its robustness and generalization across institutions and to explore fallback methods when no match is found. The code and dataset are available on GitHub*1.

HTML

XML

PDF

DOI: 10.3897/biss.10.183432

Authors: Naveen Dhar

Abstract: IntroductionPassive acoustic monitoring (PAM), thanks to the technological advancement of autonomous recording units (ARUs) (Prince et al. 2019), has enabled long-term, wide-scale studies on natural and/or anthropogenic soundscapes, making it an invaluable tool for conservation (Sugai et al. 2018). Crucially, PAM can aid in detecting threats to biodiversity, such as poaching, by listening for target acoustic signatures. Array-based PAM offers benefits over other poaching monitoring methods, such as unmanned aerial vehicle or camera trap monitoring: lower costs, wider detection fields, and longer-term deployments (Lynam et al. 2025). Real-time detection can enable immediate intervention, making it a goal of the Wildlife Crime Technology Project since 2015 (World Wildlife Fund 2022). However, most acoustic detection of poaching, including existing literature using deep learning for this task, consistently mentions challenges with false positives and employs large fine-tuned models with computational costs better suited for retrospective studies (Katsis et al. 2022, Vu et al. 2024). Our research developed a lightweight deep neural network suitable for on-board, real-time processing and a sensor integration function aiming to reduce false positive rates (FPRs).MethodsA pre-partitioned open-source dataset of 35,980 four-second audio recordings from Belizean rainforests (Katsis et al. 2022) was used for training and validation. A test set from Vietnamese rainforests (Vu et al. 2024), comprising 129 background recordings and 19 gunshots, was used for further testing. Waveforms were converted to log-mel spectrograms, an image-like representation of audio displaying the range of frequencies of a signal as it varies with time. Extensive waveform augmentations were applied to the dataset during training to improve the detector's generalizability to new environments. These augmentations included shifting the signal's position in time, injecting background noise, and more.The lightweight neural network, or model, was designed with convolutional layers and recurrent units for efficient gunshot feature extraction and detection. The final model contained approximately 935k parameters and was trained for 40 epochs.Many false positives occur because common rainforest sounds resemble gunshots in their frequency-time features (Katsis et al. 2022, Vu et al. 2024). However, gunshot signals propagate farther. Cross-device communication, or the leveraging of information from multiple acoustic sensors, is proposed and investigated as a novel way to reduce the aforementioned false positives in acoustic poaching detection. The proposed Sensor Analysis and Integration Layer, or SAIL function, described in Fig. 1, was developed to integrate predictions from three or more sensors by penalizing isolated confident predictions while boosting sensor agreement.This was validated through acoustic propagation simulations on the Vietnamese dataset, where each waveform had three copies with random augmentations mimicking distance-dependent attenuation, emulating a three-sensor array. The model generated confidence scores for each copy, which were then fed into SAIL to create a final probability. Across the 1000 simulation runs performed on the dataset, Wilcoxon signed-rank tests evaluated whether median differences in mean FPR per-file exceeded zero.ResultsModel performance was evaluated on the validation partition of the Belizean dataset. A precision-recall curve (Fig. 2) showed strong generalization and 0.96 average precision. Table 1 compares the lightweight model against the leading open-source rainforest gunshot detector, or the fine-tuned ResNet18 from Katsis et al. (2022). Our lightweight model is 11MB and compresses to 964KB via TensorflowLite serialization. Table 2 summarizes SAIL simulation statistics, reporting mean FPR differences and Wilcoxon p-values alongside performance metrics. SAIL produced significant FPR reduction with minimal (1%) recall reduction.ConclusionWhile not directly extendable to efficacy in the field, statistical evaluation demonstrates that SAIL consistently reduces FPR compared to simple inference across a wide variety of simulated sensor arrays. The lightweight gunshot detector generalizes to the validation dataset at 935k parameters, retaining 94% of performance (validation F1 at 95% recall) at a fraction of the leading work's model size. Overall, efficient deep learning and SAIL present a promising approach towards cost-effective real-time poaching detection.

HTML

XML

PDF

DOI: 10.3897/biss.10.182954

Authors: Carrie Seltzer, John Wieczorek

Abstract: iNaturalist hosts over 250 million biodiversity occurrence records representing more than 500,000 unique taxa—making it one of the largest platforms for community-contributed biodiversity data. Since 2012, iNaturalist has used the Darwin Core Archive (DwC-A, GBIF 2021) format to share its research-grade observations (occurrence records) with the Global Biodiversity Information Facility (GBIF), as well as more specialized information such as interactions with other partners such as Global Biotic Interactions (GloBI). However, there are some limitations to this format. For example, only one person is credited for the identification (there are always at least 2 people involved); interactions are captured in resourceRelationship, which is not specific to interactions; and many additional iNaturalist annotations are mapped to dynamicProperies (e.g., leaf phenology).The presentation from which this extended abstract derives (Suppl. material 1) demonstrates how the Darwin Core Data Package (DwC-DP) offers an opportunity for more specialized and extensive data sharing than the DwC-A. This flexible and extensible format allows iNaturalist to represent not just Occurrences and their associated Media (images, sounds), but also relationships between those Occurrences, e.g., OrganismInteractions such as predation, pollination, parasitism, and other vital ecological connections.As a demonstration*1, a subset of iNaturalist data focused on plant-pollinator interactions was mapped to the DwC-DP. This example incorporates interactions via community-generated Observation Fields*2 even though these are not currently controlled fields in the core iNaturalist data model. In Fig. 1, the relevant tables (e.g., occurrence, agent, event, media) of the DwC-DP are connected by the schemas (prepositions, e.g., conducted by, about, based on).While Occurrence remains a foundational element, DwC-DP brings structure to a richer ecosystem of data that supports both incremental improvements and novel capabilities for data providers and consumers alike. When it comes to implementing this in production for iNaturalist, many decisions will need to be made to balance the tradeoffs of use cases unmet by DwC-A, complexity, processing time, file sizes, and more.

HTML

XML

PDF

DOI: 10.3897/biss.10.183710

Authors: Erin Satterthwaite, Ben Best, Rasmus Swalethorp, Brice Semmens, Mark Gold

Abstract: The ocean is rapidly changing through warming, sea-level rise, ocean acidification, declining oxygen concentrations, and altered circulation patterns, which are reshaping marine ecosystems worldwide. These changes necessitate a complete and holistic view of affected marine ecosystems. Long-term ocean monitoring programs are crucial for understanding these changes and guiding sustainable management. The California Cooperative Oceanic Fisheries Investigations (CalCOFI)*1, initiated in 1949, is one of the world's longest-running and most comprehensive marine ecosystem observing programs. CalCOFI collects extensive ocean data across all Essential Ocean Variables in the Southern California Current four times a year (Gallo et al. 2022). This encompasses over 36 physical and chemical parameters, 2,500 biological parameters, and 50,000 environmental DNA sequences, measured simultaneously (Satterthwaite et al. 2025). These data have illuminated long-term ecosystem patterns, supported marine management, and enhanced our understanding of how climate variability and change impact the California Current.Given its long history and collaborative nature, CalCOFI's data are fragmented across more than 30 disparate datasets, managed by different entities, served in various locations, with inconsistent naming and limited integration (Fig. 1), which hinders a holistic ecosystem understanding. Thus, CalCOFI exemplifies similar challenges and opportunities of mobilizing and integrating ocean data on a global scale.Our vision has been to modernize and build an integrated data system for CalCOFI that automates dataset ingestion, integration, and serving for timely use in research and management (Fig. 1). We are developing workflows to publish these datasets into portals promoting Findable, Accessible, Interoperable, and Reusable (FAIR; Wilkinson et al. 2016) data principles. Our initial efforts focused on two core CalCOFI datasets: Conductivity, Temperature, and Depth (CTD) bottle data and larval fish data to develop a workflow for easier access, integration, and use (Fig. 2). Data providers upload processed datasets into a shared Google Drive in simple formats (e.g., .csv), with globally unique identifiers and standardized naming conventions (Suppl. material 1).Data are automatically detected and integrated into a PostgreSQL database (Fig. 2). Newly added data trigger a system rebuild and validation, checking primary keys, spatial geometry, taxonomy, and metadata. Managers are notified of errors; otherwise, the system updates production databases, records a new version, and publishes a Github release with source files and a DOI via Zenodo. Original source data files are retained for past version reconstruction. While PostgreSQL maintains relationships between data, we also produce an analytical database in DuckDB for a single, portable database with fast querying, even on remote servers (Fig. 2). CalCOFI data and observations are accessible through multiple repositories and platforms, tools, interfaces, and portals based on user needs and funder requirements (Fig. 2; Suppl. material 1). For example, some tools support flexible analysis and visualization (e.g., NOAA Environmental Research Division's Data Access Program), others provide long-term archiving with standardized metadata (e.g., Environmental Data Initiative). We are also publishing biological observations in DarwinCore (Wieczorek et al. 2012) format with Extended Measurement or Fact and uploading them to the Ocean Biodiversity Information System (OBIS) so that CalCOFI data can be integrated with global biodiversity datasets worldwide. CalCOFI program information is on the Global Ocean Observing System Biology & Ecosystem Portal, and dataset records are structured for indexing by portals like Ocean Data Information System (ODIS) and Google Dataset Search.To facilitate data access and visualization, we also developed the CalCOFI Integrated Data Viewer*2 (Fig. 3). This web app allows users to explore decades of CalCOFI oceanographic and larval fish data across space, time, and depth.We plan to integrate additional CalCOFI datasets and collaborate with other data providers to build a more complete ecosystem-level view of the California Current. Our ultimate goal is to enhance CalCOFI data findability, integration, and usability for scientific and management communities, enabling a better understanding of ocean changes and ensuring a sustainable future for marine ecosystems and human communities.

HTML

XML

PDF

DOI: 10.3897/biss.10.184100

Authors: Liliana Ballesteros Mejia, Nirmala Seon-Massin, Gargominy Olivier, Aurélie Lacoeuilhe, Rodolphe Rougerie

Abstract: Biological diversity is being strongly affected by the rapid environmental changes of the Anthropocene. Our ability to identify species presents a major challenge for monitoring the impacts of these changes and for designing sound public policies in response. Molecular methods, such as DNA barcoding, metabarcoding, and environmental DNA (eDNA) offer promising solutions; however, the effectiveness of these approaches relies heavily on the completeness and reliability of DNA reference libraries.To address this challenge, and as part of France’s national biodiversity strategy for 2030, the National Museum of Natural History in Paris (MNHN) and the French Biodiversity Agency (OFB) are developing the national reference of genetic sequences, called GenRef. This public and open-access tool will provide high-quality DNA sequences linked to reliable taxonomic identifications of species, and where available, to voucher specimens in biological collections, for taxa present in mainland France and its overseas territories (Lacoeuilhe et al. 2025).GenRef will centralise, structure, and validate genetic reference sequences and their associated metadata using semi-automated workflows and expert review. Closely integrated with the France's national nomenclatural and taxonomic registry (TAXREF, Gargominy 2025), it will be interoperable with major international databases housing sequences such as European Nucleotide Archive (ENA), Barcode of Life Data Systems (BOLD), National Center for Biotechnology Information (NCBI GenBank), Unite (for fungal ITS sequences). In addition it will collaborate with French research institutions and natural history collections to incorporate their data and expertise, while promoting the acquisition of new data.The GenRef workflow is organised into seven phases throughout which we can see the different stages of a GenRef Dataset.Data acquisition: selection of taxon names in TAXREF (Gargominy 2025) and desired genetic marker(s), followed by automatic querying and retrieval of genetic sequences and associated metadata from major international genetic databases (e.g., ENA, BOLD, GenBank, Unite) via APIs. This step results in an initial dataset for GenRef (a working dataset for subsequenting processing, not a formal entity).Data filtering: verification of mandatory fields and optional removal of contaminant sequences, such as those originating from non-target organisms (e.g., human, cow, mouse) or symbionts like Wolbachia. The output is a filtered version of the working GenRef dataset.Data extension: addition and population of fields required for score calculations using previously compiled information. Examples of such fields may include monophyly status, average genetic distance to sequences sharing the same species name (derived from Neighbor-Joining (NJ) tree and distance matrix calculations), sequence length and percentage of ambigous base pairs. Additional metadata fields may include biogeographic status, taxonomic validity, identification by a recognized expert and specimen collection status. This step results in an extended working dataset used for downstream evaluation.Scoring and classification: calculation of two scores for each record (i) a sequence identification reliability score based on sequence quality and identification accuracy, and (ii) completeness and relevance score based on the metadata completeness and provenance. The scoring framework is designed to be transparent, internally consistent, and adaptable as GenRef evolves, allowing records to be ranked into quality tiers for subsequent validation.Expert validation: domain experts review the compiled molecular, taxonomic and metadata evidence for each record and use tools provided in the interface to confirm or adjust the automatic classifications. Their input generates an expert score and ranking, ensuring that expert judgement is integrated into the process rather than serving as a mere approval step.Moderation and arbitration: resolves potential conflicts between automatic and expert classifications. This phase aggregates the expert ratings and applies moderation rules to produce a single consensus score for each sequence. The implementation may involve a combination of algorithm checks and human oversight (final design is under development).User Interface: a dedicated web interface provides access to sequences, with flexible options for data selection, filtering and visualisation, enabling researchers and stakeholders to explore and download curated data.Ultimately, GenRef will provide access to sequences through a web-based user interface generating on-demand reference sequence libraries that can be used for different applications. These will range from the simple identification of individual specimens to surveillance of biological communities through eDNA analysis to support efforts in biodiversity knowledge, monitoring, research, conservation and policy making in France.

HTML

XML

PDF

DOI: 10.3897/biss.10.182057

Authors: Amy Frazier, Nickolas McManus, Lei Song, Patrick Roehrdanz, Wenxin Yang, Dan Willett, Susana Rodríguez Buriticá, Elkin Noguera Urbano, Joana Krieger

Abstract: Global models indicate there is not enough land to simultaneously achieve conservation targets (i.e., conserving 30% of land by 2030, or '30x30') and meet future clean energy and agriculture demands (Brock et al. 2024). Therefore, planning efforts are needed at national scales to find optimal solutions. National-scale mapping can also foster better governance and interagency collaboration to ensure plans meet both national and global commitments (Chaplin-Kramer et al. 2021, Frazier et al. 2024). We integrate diverse, multi-scale living data into dynamic forecasting models to inform actionable and sustainable land use plans to prevent biodiversity and ecosystem service loss while meeting future development needs. We use a minimum set optimization approach to identify priority areas for conservation, energy and agriculture in megadiverse Colombia out to 2050, balancing 30x30 conservation targets with energy and food demands of a growing population. We highlight potential conflict areas between competing land uses under climate change and discuss efforts to better support sustainable land use decisions-making on the ground.Methods: We extend a global, multi-sectoral optimization framework (Brock et al. 2024) to prioritize country-level conservation (biodiversity, carbon) and development (renewable energy, crops) targets under four planning scenarios: nature-first; development-first; concurrent; and concurrent 30x30. The nature and development-first scenarios prioritize meeting one sector’s targets before considering the other; the two concurrent scenarios evaluate all sector targets simultaneously, with the concurrent 30x30 ensuring at least 30% of land is conserved. Each scenario is modeled with a ‘Business as Usual’ climate (RCP 6.0) for future targets out to 2050. Scenario outputs are compared to identify areas of potential conflict between conservation and development and compared to a global analysis (Brock et al. 2024). Additional details are in Suppl. material 1.Results: A comparison of the national- and global-scale models shows considerable disagreement for conservation, less for development Fig. 1, establishing the potential shortcomings of global models to meet national conservation needs and support local land use decisions under climate change.At the national scale, under all scenarios, energy and agriculture require about 40x less land than conservation to meet future targets Suppl. material 1. The majority of land prioritized for conservation in both the nature- and development-first scenarios is in the Amazonia and Pacifica regions, with smaller fragments in the Andes Fig. 2. For energy, the northern Caribbean region La Guajira is a hotspot for meeting wind and solar targets in both scenarios and a hotspot for conflicts with conservation Suppl. material 1.National scale models built with country-specific projections can minimize potential land conflicts and better support on-the-ground decisions, but data can be difficult to acquire for model parameterization. The models benefitted from species ranges from BioModelos (Velásquez-Tibatá et al. 2019) to quantify biodiversity, but not all areas have expert-driven data. More broadly, discussions are needed on how best to set targets for biodiversity and carbon. While development targets are generally informed by future population projections and human needs, the science around conservation targets is less established. Additionally, there is a need to incorporate other social and economic factors, such as land tenure, to accurately model potential and explore co-siting of multiple land uses together.

HTML

XML

PDF

DOI: 10.3897/biss.10.186013

Authors: Ana Carolina Peralta Brichtova, Elizabeth Lawrence, Jonathan Pye, Laura Brenskelle, Mathew Biddle, Stephen Formel, Tim van der Stap, Tylar Murray

Abstract: The Standardizing Marine Biological Data Working Group (SMBD WG) is a global network of biodiversity data managers who meet regularly and collaborate year-round. It was conceptualized at the Integrated Ocean Observing System Code Sprint 2019 (IOOS Code Sprint) in Ann Arbor, Michigan to build a community of practice for aligning marine biological data to Darwin Core standards (Wieczorek et al. 2012) and sharing with the Ocean Biodiversity Information System (OBIS). The SMBD WG integrates and adopts practices from the Earth Science Information Partners - Biological Data Standards Cluster, the Biodiversity Information Standards (TDWG) and the Marine Biodiversity Observation Network Data Management and Cyberinfrastructure Working Group (MBON - DMAC). The SMBD WG aims to help data providers and data holders overcome barriers and address data-related questions, empowering our community members with the tools and knowledge to mobilize marine biological data. To achieve this, the SMBD WG coordinates specialized guidance on data standardization, quality control, and publishing workflows. Since 2022, SMBD WG has organized annual Marine Biological Data Mobilization Workshops (Biddle et al. 2024), bringing together participants from around the world for hands-on training in Findable Accessible Interoperable and Reusable data principles (Wilkinson et al. 2016), Darwin Core standards, taxonomic tools, data cleaning techniques, and publication through OBIS and the Global Biodiversity Information Facility (GBIF). The workshop curriculum follows The Carpentries evidence-based best-practices of teaching. The materials and website are openly available under a permissive GitHub MIT license. Key features include hands-on sessions; concurrent instructor-led breakout rooms with specialist volunteers; a fully open and free virtual format using Slack, Zoom, and GitHub; and dual Python/R instruction.The workshop's implementation is based on community-reported blockers, updates, and data questions shared via the SMBD WG GitHub repository*1and Slack, which remain active beyond the workshop to provide ongoing support as participants refine and standardize datasets to final publication. Although long-term impacts are still emerging, the workshops go beyond technical training, building a lasting support network through a data support hub fostering international collaboration and empowering participants to confidently navigate complex data workflows. Some participants have published multiple datasets to OBIS and GBIF, and supported similar efforts in their home institutions.Participant feedback shows that structured training initiatives strengthen technical capacity and promotes inclusive participation, enabling diverse voices to contribute to the collective effort of mobilizing marine biodiversity data. Curriculum updates are guided by lessons learned and participant feedback, ensuring that it remains relevant, impactful, and responsive to global needs. Since 2022, the number of participants has grown significantly, including the integration of Spanish-language breakout rooms in the 2024, to reduce language barriers and broaden regional engagement.

HTML

XML

PDF

DOI: 10.3897/biss.10.184686

Authors: Thore Engel, Martin Friedrichs-Manthey, Jörg Holetschek, Robert Köpke, Pascal Scherreiks, Anton Güntsch, Birgitta Koenig-Ries, Aletta Bonn

Abstract: Biodiversity data can be messy, scattered, and hard to obtain (Engel et al. 2023). National biodiversity data infrastructures fulfill essential functions for science, policy, and practice by structuring national data landscapes and bridging connections between local and global initiatives and infrastructures (Güntsch et al. 2024). As the German national research data infrastructure for biodiversity (NFDI4Biodiversity), we set out to make German biodiversity data FAIR (findable, accessible, interoperable, re-usable) and open (Luther et al. 2022). One of our products is the Lebendiger Atlas der Natur Deutschlands (LAND),*1 a national biodiversity portal that provides species occurrence data in Germany to researchers, conservation authorities, and the general public. We joined forces with the Global Biodiversity Information Facility (GBIF) as well as the national node GBIF Germany and created LAND using the GBIF-hosted portal service (Fig. 1). LAND integrates species occurrence data from different sources, including monitoring initiatives, citizen science, natural history societies, museums, and universities. Initially, the portal was mainly populated with datasets already available through GBIF. Now, our focus has shifted to mobilizing and publishing new datasets from the NFDI4Biodiveristy consortium and other national partners (e.g., Engel et al. 2025, Rehbein et al. 2025). To further enhance the service and increase the mobilization of new datasets to GBIF and LAND, we organize annual community workshops with 50–80 stakeholders from various institutions. These events are designed to strengthen capacity, foster networking, and support the development of new strategic collaborations with local and interdisciplinary partners.In this contribution, we present the current state of the portal and its key functionalities, and highlight opportunities for data holders and researchers to become involved. Furthermore, we are eager to engage with similar initiatives in other countries.

HTML

XML

PDF

DOI: 10.3897/biss.10.185867

Authors: Miwa Takahashi, Joana Paupério, Peter Woollard, Tobias Frøslev, Luke Thompson, Katherine Silliman, Stephen Formel, Christopher Hunter, Camila Babo, Kristy Deiner, Lynn Schriml, Oliver Berry

Abstract: Environmental DNA (eDNA)-derived data, particularly DNA sequences and species occurrence records, are powerful resources for various applications, including biodiversity monitoring, biosecurity, and agricultural bio-surveillance. An important part of eDNA data’s value lies in its reusability, which can be enhanced through the use of metadata that adheres to FAIR (Findable, Accessible, Interoperable, Reusable) data principles (Wilkinson et al. 2016, Abarenkov et al. 2023). While FAIR practices are gaining recognition within the eDNA community, adoption remains limited. A key barrier is that existing data standards, including Darwin Core (DwC, Darwin Core Task Group 2009) and the Minimum Information about any (x) Sequence (MIxS), lack sufficient terms to fully describe eDNA workflows, from sampling and processing to bioinformatic outputs (Takahashi et al. 2025).To address this challenge, we have initiated a collaborative eDNA Task Group supported by the Biodiversity Information Standards (TDWG) and the Genomic Standards Consortium (GSC) organisations. We build upon current standardisation efforts to develop a comprehensive and interoperable eDNA metadata checklist for formal integration into the MIxS and DwC frameworks. See the task group descriptions at TDWG*2 and GSC.*1 Our goal is to ensure compatibility with, and seamless adoption by, major biodiversity and sequence data infrastructures, such as Global Biodiversity Information Facility (GBIF), Ocean Biodiversity Information System (OBIS), and International Nucleotide Sequence Database Collaboration (INSDC). This talk will present progress of the TDWG/GSC joint eDNA initiative, including our collaborative efforts with related working groups and initiatives, such as eDNAqua-plan, the Darwin Core Data Package, DNA-derived data extension to DwC, Minimum Information about any Ancient Sequence (MInAS), and the FAIR eDNA (FAIRe) initiatives (Takahashi et al. 2025). We invite contributions and collaboration to help shape a shared metadata framework that supports the growth, reuse, and long-term utility of eDNA data worldwide.

HTML

XML

PDF

DOI: 10.3897/biss.10.182661

Authors: Régine Vignes Lebbe, Sylvain Bouquin, Noël Conruyt

Abstract: In the 1970s–1980s, artificial intelligence (AI) generated research and debate, especially around expert systems (Fig. 1) to transfer expertise in fields such as medical diagnosis (e.g., MYCIN, Buchanan and Shortliff 1984), industrial troubleshooting, and biological identification. These approaches, focused on knowledge representation and logical reasoning, forming what is now called symbolic AI (S-AI).The knowledge base (KB) management system Xper was conceived in that context (Lebbe and Vignes 1988). Its current version, Xper3 developed with modern web technologies, builds a KB including a matrix and rules, used by deductive-abductive reasoning (Fig. 2). Like other tools (e.g., Lucid, Delta, Clavis), it allows taxonomists to publish online, interactive identification keys. IKBS (Iterative Knowledge base system, Conruyt and Grosser 2007) uses a structured descriptive model and specimen descriptions as examples of the taxa, then induction allows building of a dynamic decision tree.Over the last decade, identification has been profoundly renewed by citizen science, the proliferation of automatic visual or acoustic sensors, and the rise of connectionist AI (C-AI), derived from neural networks. This data-driven AI learns from examples rather than from explicit logical rules, operating in vector spaces rather than symbolic structures. It has revolutionized taxonomic identification—as illustrated by tools like Pl@ntNet or BirdNET. However, C-AI suffers from two main limitations: its need for large annotated training datasets and its lack of explainability (e.g., ability to cite diagnostic characteristics that make up identificaton decisions).Hybrid approaches combining S-AI and C-AI are now emerging to combine their strengths and mitigate their weaknesses. Building on our long experience with S-AI, we propose three hybrid identification scenarios implemented or envisioned within the Xper3 and IKBS environment.Hybrid identification scenarios Our three scenarios assume that users provide one or several images of a biological specimen to be identified (Fig. 3).Scenario 1 In Xper3 and IKBS, users freely navigate through a graph or interactive key. In this first hybrid approach, a Convolutional Neural Network (CNN) pre-filters the search space before symbolic reasoning starts and so it targets the decision tree. Depending on the case, it can reduce either the number of taxa considered or the list of relevant characters, by classifying images into higher-level categories such as genera, morphotypes, or morphological traits. With Xper3, the CNN confidence scores are translated into weights within the KB, lowering irrelevant taxa or descriptors.Scenario 2 Here, the AI acts as an assistant verifying the user’s inputs. If the CNN detects an inconsistency between the user’s path in the key and the descriptions stored in the KB, the system issues a message such as “Please verify your answer” and provides complementary explanations or illustrations. The CNNs, trained on broader taxonomic groups than the final species level, make training easier while still providing useful feedback. Scenario 3 When several identification tools coexist (CNNs, Xper3 keys, IKBS keys, etc.), we propose a third scenario with a dynamic interaction between AIs. The hybrid system selects the most appropriate tool at each step, according to metadata describing the domain and performance of each tool. This scenario is currently explored in a project on coral reef monitoring from video transects (Ouassine et al. 2025) with a CNN trained to recognize morphogroups, Xper3 KB at genus rank, and an IKBS KB at species rank.Discussion and conclusion In identification, quality and interpretability are more important than speed, but experts are scarce. C-AI is therefore valuable to complement expert work, despite the heavy effort required to prepare training datasets and validate models. Hybridization aims to:reduce expert involvement when possible,decrease user errors during key navigation,and preserve explainable results through symbolic reasoning.C-AI can further complement S-AI upstream by facilitating the construction of KBs through automatic extraction from fauna and flora, for example.Hybridisation also allows a frugal use of AI. Moreover, hybrid systems keep humans in the loop—users remain active participants who can learn from the explicit reasoning provided by S-AI.Further hybrid developments are foreseen. Heatmaps (Selvaraju et al. 2017) could highlight informative image areas to guide users during observation. Large Language Models (LLMs) are already used to extract knowledge from floras or faunas (Nainia et al. 2024) and could soon improve interaction by rephrasing key questions and explanations in natural, multilingual language. Hybrid AI research is still in its infancy but already opens a wide range of perspectives for biodiversity studies and biological identification.

HTML

XML

PDF

DOI: 10.3897/biss.10.183035

Authors: Ayoub Nainia, Régine Vignes Lebbe, Jihad Zahir

Abstract: As artificial intelligence (AI) becomes central to biodiversity research, transforming morphological descriptions of species into structured data remains a challenge, particularly in botany, where species descriptions are rich in detail but lack formal structure. Despite the progress brought by transformer architectures (Vaswani et al. 2017) and their multilingual variants such as CamemBERT (Martin et al. 2020), most large language models (LLMs) are trained on general-domain English corpora, limiting their effectiveness for non-English biodiversity applications.We present a two-stage information extraction pipeline that fine-tunes LLaMA (Large Language Model Meta AI)- based LLMs (Touvron et al. 2023) to extract botanical traits from French floristic texts and enable species-level comparison at scale. Trained on a curated corpus of legacy botanical descriptions and expert-validated French question-answer pairs, the system demonstrates how domain-specific fine-tuning can bridge the gap between narrative botanical knowledge and structured trait data (Fig. 1). At the user-interface layer, we add a third stage that automatically converts the trait-value pairs generated by the question-answering module into a structured CSV table for downstream use.Our approach unifies question generation and trait extraction within a single information extraction workflow. First, predefined botanical traits (e.g., leaf shape, flower color) are reformulated as natural-language questions tailored to each species description. Then, using these questions as prompts, the model identifies and extracts the corresponding trait values: for example, extracting "lanceolate" as the standardized leaf shape from "lanceolate leaves." This converts unstructured text into consistent character-state pairs aligned with botanical ontologies. A proof-of-concept showing the input species descriptions processed by the pipeline is shown in Fig. 2.The LLaMA model was fine-tuned via Low-Rank Adaptation (LoRA) (Hu et al. 2021) on ~17,000 manually verified French question-answer pairs derived from floristic descriptions across multiple French-language flora, including: Flore de Nouvelle-Calédonie and Flore du Cameroun. Building on FloraNER (Named Entity Recognition) (Nainia et al. 2024a) and earlier CamemBERT-based pipelines (Nainia et al. 2024b), the current system, F-LoRA-QA (Nainia et al. 2025b), substantially improves structured trait generation from domain-specific descriptions.In evaluation, F-LoRA-QA outperformed an untuned LLaMA baseline with ~4× higher BLEU score (Bilingual Evaluation Understudy), a 16% gain in semantic similarity, and exact-match accuracy rising from 2% to 24%. Expert botanists further validated trait accuracy, completeness, fluency, and terminology, supporting critiques of over-reliance on automatic metrics for domain question answering (Nainia et al. 2025a).A key application is the enrichment of descriptor-based systems like Xper³ (Kerner et al. 2025), which rely on structured trait matrices to describe and differentiate taxa. Fig. 3 and Fig. 4 illustrate trait discovery, value extraction, and a comparison matrix with evidence links, supporting descriptor systems and interoperable biodiversity databases.

HTML

XML

PDF

DOI: 10.3897/biss.10.185022

Authors: Eduardo Klein, Jie Yun, Leo Ghignone, Michaela Miller, Xiaohu Li, Yuxuan Hu, Denisse Fierro Arcos

Abstract: IntroductionThe Great Barrier Reef (GBR) is one of the world’s most biodiverse and ecologically significant marine ecosystems. Its management has increasingly relied on integrating long-term monitoring and modelling to address threats such as ocean warming, land-based pollution, and human activities. Since 2015, comparative studies have demonstrated that data-driven approaches have achieved tangible outcomes, including reductions in pollutant levels in targeted catchments and localised coral recovery through a targeted Crown-of-Thorns Starfish (COTS) control program. However, these efforts are often constrained by fragmented data access, attribution challenges, and the urgency of monitoring climate drivers.Built on open-source technologies and aligned with FAIR (Findable, Accessible, Interoperable and Reusable) data principles, the Great Barrier Reef Data Management System (DMS) serves both the Great Barrier Reef Marine Park Authority (GBRMPA) and the broader research community. It enables users to discover, access, and analyse data through robust APIs, reducing data wrangling. As reef managers increasingly rely on scenario modelling, spatial risk assessments, and participatory monitoring frameworks, the DMS provides support for these approaches at scale.System OverviewThe DMS was developed as a key infrastructure to support the Reef 2050 Integrated Monitoring and Reporting Program (RIMReP). Led by the Integrated Marine Observing System (IMOS), the DMS is designed to serve as a scalable, cloud-native platform that aggregates, harmonises, and delivers more than 170 datasets critical to the management of the GBR. See Suppl. material 1 for a detailed system description.Impact and Use CasesEnabling Evidence-Based Decision-Making. The DMS directly supports RIMReP by providing timely, standardised data to inform management decisions. For example, the system underpins the CoTS Control Program, which relies on integrated datasets from the Australian Institute of Marine Science (AIMS), eReefs, and GBRMPA to monitor outbreaks.Accelerating Research and Innovation. A significant impact of the DMS is its support for open, reproducible science. The project maintains two public repositories: rimrep-examples and rimrep-training. These resources demonstrate how to discover and filter datasets; and extract and analyse spatial and temporal subsets of data, manually or automatically.Reducing Technical Overhead and “Data Wrangling.” One of the most immediate impacts of the DMS is the reduction in time and effort required to access and prepare data for analysis. For example, accessing satellite-derived sea surface temperature data across 18,000 layers can be done in under 12 seconds, directly connecting to the DMS storage.Lessons Learned and Future DirectionsData Harmonisation is Essential but Complex. The adoption of cloud-optimised formats (Zarr and GeoParquet) improved performance and usability across diverse datasets. However, harmonising formats from >60 providers required significant effort, especially for legacy datasets. The use of the Frictionless framework proved critical in maintaining metadata quality and consistency.Automation Enables Sustainability. Automated ingestion workflows, including “watchers” that detect updates from data providers, have reduced manual overhead and improved data freshness, crucial for long-term sustainability, particularly as the number of datasets continues to increase.Performance Matters for Usability. Improvements in file structure and indexing have significantly reduced query times, enabling real-time analyses and dashboard integration, particularly with terabyte-sized datasets such as eReefs and satellite-derived products.Engagement Drives Adoption. Training workshops and public code repositories were instrumental in building trust and capacity among users. User feedback informed improvements to documentation and use cases.ConclusionThe GBR DMS aligns with the Ocean Decade Data and Information Strategy and embodies its core principles of FAIR data practices, open-source infrastructure, and equitable access. By reducing technical barriers and enabling rapid access to high-quality environmental data, the DMS serves as a strategic enabler for transformative ocean science and management. The DMS is currently managed by AIMS, which is strategically positioned to foster collaboration among data providers, managers, researchers, and Traditional Owners of the Sea Country.

HTML

XML

PDF

DOI: 10.3897/biss.10.182096

Authors: Sanson Poon, Arianna Salili-James, Chris Addis, Avellina Leong, Hester van Schalkwyk, Naifeng Zhang, Ben Hartley

Abstract: From November 2024 to March 2025, the Natural History Museum, London (NHM), supported by the UK Department for Culture, Media and Sport (DCMS), led a national programme to explore a new model for deploying artificial intelligence (AI) across the UK cultural and heritage sector. Inspired by the Museum’s in-house AI Lab Programme (Poon 2024), which has successfully applied AI to research, collections management, and public engagement within the NHM (e.g., Poon and Leong 2024, Salili-James 2024, He et al. 2024, Zhang et al. 2025c, Gallois et al. 2025), this national initiative extended that model to a broader institutional landscape. The programme examined how AI can support biodiversity research, collections management, accessibility, and public engagement, while addressing the practical realities of AI adoption, scalability, and long-term sustainability.In partnership with the Zoological Society of London, Lyme Regis Museum, National Museums Liverpool, the Royal Agricultural University, and the Royal Botanic Garden Edinburgh, the programme delivered five AI pilot projects. Each project developed an AI-driven solution to a critical challenge: automated species classification of illegally traded snake goods (Leong et al. 2025a); automatic data extraction from index cards (Leong et al. 2025a); AI-generated alt-text for digital collections (van Schalkwyk et al. 2025); a conversational chatbot trained on herbarium data (Zhang et al. 2025b); and a herbarium digitisation quality-control tool (Zhang et al. 2025a), respectively.These pilots demonstrated how AI can deliver both operational efficiencies and public benefit. Automated transcription and text extraction have streamlined digitisation efforts, reducing the time and resources required for archival documentation. AI-driven image processing has enhanced the quality control mechanisms for digitisation, ensuring higher standards in digital preservation. Computer Vision models have provided new capabilities in species classification and object identification, offering valuable insights for conservation and research. Additionally, AI-powered accessibility tools, such as automated alt-text generation and chatbot interaction, have improved inclusivity by making collections more accessible to diverse audiences. By engaging partners of different scales, locations, disciplines, and specialisations, the programme provided a rare opportunity to test AI in diverse institutional environments. This diversity enabled robust evaluation of prototypes and fostered knowledge exchange, skills development, and capacity building across the sector. Crucially, the collaborative model demonstrated the importance of co-design: tools developed in close dialogue with institutional priorities were more readily embedded in existing practices and had clearer pathways to sustainable integration. Beyond technical outcomes, the programme highlighted sector-wide considerations for responsible AI adoption. These included: the need for consistent data standards to support model training and interoperability; designing systems that augment, rather than replace, professional expertise; ensuring infrastructure and training are in place to support long-term use; and transitioning from short-term pilots to sustainable, maintainable solutions. Collectively, the five projects suggest how the sector might progress from isolated experimentation to a coordinated ecosystem of digital innovation. Although the pilots are at proof-of-concept stage, the projects have already demonstrated measurable improvements in workflows, data quality, and accessibility. They provide a framework that other organisations can adapt, offering practical entry points into AI adoption. At the same time, the work underscores the importance of continued investment in digital transformation to ensure that heritage institutions can keep pace with rapid technological change.In summary, the NHM-DCMS AI Pilot Programme has shown that AI can be responsibly deployed in culture and heritage when solutions are co-created with institutional partners, tailored to local contexts, and designed with flexibility for future use. While still early in development, the programme has established a strong foundation for future work, charting a path towards AI applications that improve efficiency, inclusivity, and engagement across museums, galleries, and research collections in the UK and beyond.

HTML

XML

PDF

DOI: 10.3897/biss.10.182583

Authors: Qianqian Gu, Ben Scott, Geoff Martin, Arianna Salili-James, Laurence Livermore, Peter Wing, Vincent Smith

Abstract: The digitisation of natural history collections at scale raises a range of logistical and curatorial challenges. One concern is the physical expansion of storage infrastructure. As part of the Distributed System of Scientific Collections UK (Smith et al. 2022), the Natural History Museum London (NHM) will rehouse a significant proportion of its 36 million pinned entomological specimens (~130,000 drawers). Digitisation involves imaging every specimen and attaching a physical, unique, human- and machine-readable identifier (a small card label with a Data Matrix barcode). These barcodes must be readable from above, and their addition during digitisation can increase drawer occupancy if they exceed the existing footprint. Manual estimation of the footprint increase is impractical at this scale, given the range of circumstances associated with different specimens and drawers.To address this, the NHM has developed an AI-driven approach to estimate specimen drawer expansion and support resource planning. The deep learning pipeline can automatically detect drawer objects, referred to here as classes, including specimens, labels, barcodes, notes, unit trays, and drawers, from high-resolution images. Our dataset includes 11,090 digitised pinned Coleoptera drawers from the Index Lot collection (Natural History Museum 2014), representative of typical entomological collections. The AI pipeline supports both "pre-" and "post-" digitisation analysis by calculating the specimen bounding areas, handling overlaps, and estimating the net footprint change.Our object detection models achieved high performance under real-world conditions, with a mean average precision (mAP*1) of 85.45% across all classes. Barcode detection reached 86.63% mAP, while the standard unit tray detection and unit tray type classification model achieved 99.5% accuracy. (An example of the detection outputs is shown in Fig. 1)Three different calculation methods were tested and evaluated to estimate the drawer expansion area, corresponding to three specimen expansion rates for specimens with Data Matrix barcodes attached.Rate 1 (Per-Drawer Area via Polygon Masks*2): calculates expansion rates per drawer by measuring the total area occupied by specimens and their labels, ensuring overlaps are counted only once using polygon masks, as shown in Fig. 2.Rate 2 (Total-Based Expansion Rate): calculates expansion rates by averaging the occupancy areas of barcode-attached specimens across the entire dataset.Rate 3 (Per-Specimen Expansion Rate): averages the expansion rate per specimen using only bounding box*3 areas. Every specimen contributes equally, regardless of its size, to the final value.Table 1 compares different expansion rates. Both per-drawer Rate 1 and per-specimen Rate 3 have confidence intervals. Fig. 3 and Fig. 4 show their distributions: Rate 1 is right-skewed, indicating modest area increases from overlap, while Rate 3 is bimodal*4, with a small peak near 0 (large specimens unchanged) and another around 0.3 (barcode additions on smaller specimens). The global total-based Rate 2 is the mean of the ratios without confidence intervals, providing a macro-level view.In addition to estimating drawer expansion and supporting budget planning, this tool has been designed in a modular fashion, allowing individual components to be used independently at different stages of the digitisation and curation workflow. For example, specific models can be used to flag missing labels or barcodes during digitisation, assist in tracking specimen relocation, and support downstream re-curation decisions. Importantly, the pipeline is designed to be reusable across other entomological collections, with all code to be made openly available alongside a forthcoming publication, promoting scalability within the community.Our findings demonstrate that automated spatial analysis not only improves accuracy and speed in collection management but also lays the groundwork for predictive infrastructure modelling across large-scale digitisation efforts.

HTML

XML

PDF

DOI: 10.3897/biss.10.184058

Authors: Naifeng Zhang, Arianna Salili-James, Sanson Poon, Ben Hartley, Robyn Drinkwater, Milo Phillips, Rob Cubey, Elspeth Haston

Abstract: It is estimated that there are nearly 400 million herbarium specimens held across approximately 3,500 herbaria worldwide (Davis 2023). Over the past decade, many institutions have embarked on large-scale digitisation initiatives to robustly and efficiently digitise herbarium sheet collections. While these processes can be highly productive, reaching daily rates of 600–1,000 herbarium sheets in some programmes, there remains a risk and cost associated with re-digitisation, often caused by complications during the imaging stage. To address these challenges, the Natural History Museum, London (NHM) partnered with the Royal Botanic Garden Edinburgh (RBGE) to develop an Artificial Intelligence (AI)-based workflow aimed at providing enhanced quality control of herbarium sheet digitisation. This initiative was supported by the UK Government’s Department for Culture, Media and Sport (DCMS) through the NHM–DCMS AI Pilot Programme (Poon et al. 2026).The aim of this project was to develop an AI-assisted quality control framework (Zhang et al. 2025). Trained on collections at RBGE, the tool enables automated quality checks of digitised herbarium sheets. This is done by detecting missing objects (such as colour reference cards or barcodes), identifying cropping errors, and highlighting focus issues (Fig. 1). The pipeline integrates several computational techniques, including object detection, decision trees, classical statistics, and deep learning methods. In addition to post-digitisation analysis, a user-friendly interface was also developed to support in situ quality checks during the imaging stages.Model evaluation on an independent test set demonstrated the high accuracy and robustness of the AI-assisted framework (Zhang et al. 2025). The object detection model achieved an overall precision of 0.98, recall of 0.98, and mean Average Precision scores of 0.99 (mAP50) and 0.94 (mAP50–95) across all object categories. Key reference elements, including colour reference cards, barcodes, and rulers, were detected with near-perfect accuracy, ensuring precise object localisation. We note that as model development was done on RBGE collections alone, this meant that reference elements were similar across many herbarium sheets in the sample. Lastly, the issue detection models also performed strongly, achieving a precision of approximately 1.00 for out-of-focus detection, 0.98 for over-cropping, and 0.93 for under-cropping.The proposed workflow underwent further evaluation with herbarium digitisers at RBGE, which will inform refinements to the model, software interface, and the overall pipeline. The next step is to expand on this pilot project by testing the model with collections across different institutions, leading to more refinements for generalisability. This can also help us evaluate and estimate the total proportions of digitised herbarium sheets with issues. Finally, our global aim is to provide an AI-based quality control service that can be integrated into digitisation workflows not only at RBGE but across institutions worldwide.

HTML

XML

PDF

DOI: 10.3897/biss.10.182719

Authors: Avellina Leong, Arianna Salili-James, Sanson Poon, Louise Gibson

Abstract: Introduction The wildlife trade significantly contributes to the ongoing biodiversity crisis. While charismatic taxa receive more attention (Colléony et al. 2017), reptiles, which are the most traded fauna, are often overlooked, despite the fact that approximately 90% of traded reptile species are captured from the wild (Marshall et al. 2020). Curtailing this trade begins with species identification, yet enforcement officers at border control facilities often lack the access to herpetological expertise, hindering law enforcement efforts. However, with recent advances in artificial intelligence (AI), particularly in computer vision, various automated species identification applications have emerged (Zhang et al. 2024). Thus far, computer vision models have successfully classified different reptile species. For example, Patel et al. (2020) used Fast R-CNN (Region-based Convolutional Neural Networks) and ResNet models to classify nine endemic snake species of the Galapagos Islands from live snake images to achieve an accuracy of 75%. However, existing research focuses on classifying images of live reptiles and has not ventured into classifying processed reptillian goods, where common distinguishing features such as colour, body shape and positioning are not present, posing significant challenges. Unlike previous studies, our aim is to classify species solely through the dorsal textural pattern of their skin, allowing classification without taxonomically diagnostic features.In this study, we focus on snakes and introduce a multi-model pipeline to identify illegal snake goods to the species level. Our approach involves training segmentation (identification of snake bodies within images) and classification (identification of distinct snake species) models to classify species based on skin texture patterns captured from live snake imagery. With this pattern-based classification approach, our trained model is generalised to recognise these same patterns in processed goods. This ensures that snake goods that have been processed artificially can still be classified despite losing features used to identify live reptile specimens. MethodsTo reliably build a model to classify snake goods, a large dataset of snake skins is required. However, we utilise live snake images, rather than illegal snake goods, as they are more easily accessible. Our pipeline includes four key stages to ensure the model learns skin textural patterns rather than factors like body shape or colour:isolating snakes and snake goods from their background through two segmentation models;production of the box image dataset (splitting segmented images into boxes) to ensure only skin pattern is used for model training (Fig. 1);training an EfficientnetB0 model (Tan and Le 2019) to classify snakes from the skin pattern of live snake data andapplication of the EfficientnetB0 model on snake goods. The pipeline evaluates the model's performance across eight snake classes in two stages, when:fed live snake image data, andfed illegal snake goods image dataEvaluation across these two points in the pipeline allows us to understand whether the model maintains its classification abilities from live snake images to illegal snake goods. Results and DiscussionWhen testing the model’s classification ability when fed live snake image data, an overall accuracy of 0.77 and AUC-ROC (Area Under the Curve-Receiver Operating statistic) scores ranging from 0.86 to 0.95 were obtained. These scores indicate that although final class predictions suffered some misclassification, the underlying probability estimates demonstrated strong discriminative ability across classes. When testing the model’s classification ability on illegal snake goods data, the overall accuracy is maintained at 0.77, demonstrating successful generalisation to snake goods and that classification via skin textual patterns is a viable method.In future, we plan to expand the training dataset with underrepresented species, and include more species from other CITES (Convention on International Trade in Endangered Species of Wild Fauna and Flora) taxa. Once trained optimally, pipeline deployment as a mobile app could assist law enforcement, and border and custom agencies on the ground. This would deliver a cost-effective method to combat wildlife trafficking at UK and international borders, increasing the likelihood of successful reptile trade prosecutions. Overall, we highlight the potential of interdisciplinary AI applications to wildlife trade challenges, offering promising scalability and adaptability for a time and cost-efficient tool.

HTML

XML

PDF

DOI: 10.3897/biss.10.183712

Authors: Vijay Barve, Nandita Barman, Shantanu Dey, Sohail Madan, Arjan Basu Roy

Abstract: The Big Butterfly Month (BBM) is an annual citizen science initiative launched in September 2020, dedicated to monitoring butterfly populations across diverse landscapes (Big Butterfly Month Team 2025). BBM is promoted and supported by a consortium of organizations and communities like DiversityIndia (Barve et al. 2023). This program leverages the power of community participation to gather crucial ecological data, contributing significantly to our understanding of local biodiversity and conservation needs (Kelling et al. 2019, Sekhsaria and Tayyil 2022). Two primary data collection methodologies are employed: spot occurrences, which record the presence of specific butterfly species, typically shared as a record on iNaturalist or the India Biodiversity Portal, and standardized 30-minute counts. The latter, a highly replicable method, proves invaluable for assessing species diversity, population trends (Dennis et al. 2017), and relative abundances over time and across different habitats (Pollard 1977).A core objective of the BBM initiative is the mobilization of this invaluable citizen-generated data to open-access platforms, most notably the Global Biodiversity Information Facility (GBIF Secretariat 2023). This ensures that the collected information is not only accessible to researchers and conservationists worldwide but also contributes to broader global biodiversity datasets, facilitating large-scale ecological analyses and informed policy decisions (Pocock et al. 2017).The engagement strategy focused on broad reach and support, starting with promotion across BBM channels. Participation was facilitated through virtual training sessions, with video recordings shared for later review, complemented by follow-up virtual question-and-answer sessions. To maximize accessibility, a simple, spreadsheet-based data format was adopted and shared. Finally, the initiative fostered motivation and retention through recognition for top contributing individuals and regions. Success of reaching out to various regions of India is illustrated in Fig. 1.The data workflow begins when data files are extracted from a series of spreadsheet attachments received via email. These data files are then read into R (R Core Team 2021), where they are merged and thoroughly checked for quality in terms of dates, scientific names and location details. To ensure proper attribution and tracking, the Open Researcher and Contributor ID (ORCID) of the contributing users are collected. The final, validated dataset is then converted to Darwin Core (DwC) format (Darwin Core Task Group 2009), making it compatible for its ultimate destination: publication on GBIF (Barve et al. 2024).

HTML

XML

PDF

DOI: 10.3897/biss.10.183408

Authors: Julika Voß, Caroline Höschle, Recep Can, Amel Mahjoub, Vladislav Kosarev, Kelly Macleod

Abstract: There is an ambitious global goal to protect 30% of the ocean by 2030. However, currently, only 8.1% of the ocean is protected and over 80% of the ocean is still unexplored. We thus urgently need to collect data for evidence-based decision-making to help tackle the global climate and biodiversity crises and to support coastal communities. But mapping these areas and key indicator species like whales using ship-based and aerial surveys tend to be limited in scale/scope, time-consuming, expensive and pose safety risks to surveyors, while acoustic monitoring is difficult to convert into counts of individuals for most species.To overcome these challenges, BioConsult SH and HiDef Aerial Surveying Ltd developed the service, SPACEWHALE. Bridging technological innovation with marine conservation, SPACEWHALE monitors whales and other wildlife species from space using satellite imagery and AI. We determine where whales are, provide a minimum count and investigate when they are in certain areas—quickly and over wide cross-border areas—yet at comparable costs to traditional survey methods.The “SPACEWHALE” service uses very high-resolution satellite imagery (e.g., resolution of 0.3 m per pixel). At this resolution, we can distinguish between several large whale species as well as between juveniles and adults. In addition, we can detect a diverse range of large cetaceans down to a body length of 9 meters and smaller. By taking a snapshot of the ocean area of interest, we can capture very large and remote areas almost instantaneously. Due to licensing restrictions, we are not permitted to share satellite imagery itself, but we summarise our findings in a report that contains easily processable data, such as the locations of the whales, for further analysis.We developed a semi-automatic process to evaluate satellite images combining state-of-the-art artificial intelligence with marine mammal expertise. Artificial intelligence as applied in SPACEWHALE means that data are collected and processed in a standardized way: satellite imagery is first analysed by a convolutional neural network-based object detection algorithm. Our algorithm is highly effective: a study in collaboration with Stony Brook University in New York and HiDef Aerial Surveying Ltd from the UK showed that the algorithm correctly classified 100% of images containing whales and 94% of images containing only water (Borowicz et al. 2019). Once the automated process is complete, a large team of expert reviewers validates the outputs to ensure that misclassifications such as boats, rocks, waves or sea foam are minimized, implementing a human-in-the-loop framework for comparable and quality-checked data.Applying this semi-automatic approach in diverse ecosystems, we have already successfully monitored fin whales and other species in the Bay of Biscay and in the Pelagos Sanctuary, blue whales and other marine megafauna off Mexico, humpback whales in Hawaiian waters, and southern right whales off Argentina and New Zealand. The methodology has been validated through comparison with conventional aerial and ship-based surveys demonstrating a strong agreement (Höschle et al. 2022).Satellite snapshots can be taken over previously unexplored areas, providing baseline data for environmental impact assessments for offshore wind farm operators and contributing to the identification of Important Marine Mammal Areas and Marine Protected Areas. The service can be applied at different spatial and temporal scales, from local and regional to ocean basins and over time with repeated surveys, enabling long-term monitoring, generating robust, standardized data for marine spatial planning and complementing international efforts (e. g. Marine Strategy Framework Directive, Convention on Biological Diversity targets, United Nations Ocean Decade). We are at a step where satellite-based data informs applied conservation by providing useful guidance to find solutions that balance species conservation with human use of the seas. It has the potential to elevate conservation efforts to appropriate spatial scales for wide ranging species.

HTML

XML

PDF

DOI: 10.3897/biss.10.184435

Authors: Patricia Mergen, Frederik Leliaert, Sofie Meeus, Nicole Hanquart

Abstract: There is an ever growing pressure on global platforms in the current difficult geopolitical context, and this is threatening international collaborations across continents and the tackling of acute challenges that need to be addressed at a planetary level (Lenton 2023) such as climate change or biodiversity loss (U.N.E.S.C.O. 2023). If the integrity of such platforms is not guaranteed anymore; if information becomes fragmented by being based only on data from a continent, region or country; if access from outside the place where the platform is hosted becomes restricted, then scientific research as we know it—including international collaborations—will be seriously jeopardized (Gates et al. 2025). In addition to the arguments of strategic autonomy or competitiveness, the countries or for example, European Union (EU) authorities, are panicking (unfortunately for good reason) because of the increase in number and higher impact of recent cyberattacks. Hence, governments impose ever stricter cybersecurity rules that could hamper or at least seriously slow down the pace of international collaborative scientific research. Authorities are trying to secure virtual working environments with ever increasing rules and associated administration (EU AI Act or The Digital Services Act). Notably when it comes to information-technology-related activities, such as testing new technologies not yet fully secured or working with cross-border developed software, it becomes more complex to get the authorisations to proceed and deploy in multiple countries (EU risk based approach). There are measures discussed, for example in Europe, to support global platforms that are considering hosting and mirroring or at least backing up the content. Shockingly, some authorities want to address only the European content hosted outside Europe and potentially at risk, but do not want to invest into keeping the integrity of the global systems, diminshing their usefullness for the users needing a global context for their studies. By not backing up the non-EU content as well, they overlook that the users need access to data worldwide in scope. One additional issue is that personnel data and data protection breaches are treated differently in the United States (US) and in the EU as rules differ in the matter (i.e., US Cloud Act versus EU GDPR [General Data Protection Regulation]).This presentation highlights the risky situation of the Biodiversity Heritage Library (BHL), which has to find a new host. In April 2025, BHL (BHL 2025) announced that the Smithsonian Institution would no longer host the administrative functions of the BHL consortium as of 1 January 2026. BHL represents more than 63 million freely accessible pages to the user community and is widely used with approximately 200–300k searches per day, notably on scientific names of species. We would like to analyse a model where subcriptions to lesser-used closed-source journals or articles are decreased to change to a per article on-demand payment model. The saved budget could be reallocated to support platforms such as the above mentioned BHL or as additional funding for GBIF (Global Biodiversity Information Facility) much needed by our researchers instead. Additionally, we show examples of how the EU AI Act and the Flemish Research Foundation have found workable solutions to address the risks, while impacting international collaboration as minimally as possible. For example, AI systems developed outside the EU and considered risky, would not be forbidden but undergo stricter control, with official representatives based in the EU supervising their usage, implementation, and deployment following the EU legislation. The Flemish goverment, via the FWO (Fonds voor Wetenschappelijk Onderzoek – Vlaandern), confirmed that international collaborations remain essential for excellent research, while acknowledging the risks by implementing research security procedures as a workable compromise.

HTML

XML

PDF

DOI: 10.3897/biss.10.183771

Authors: David Stella, Simona Zvěřinová, Julia Leventon

Abstract: The urgent call to reverse biodiversity loss has led to unprecedented efforts to improve biodiversity monitoring, databases, harmonize indicator frameworks, and promote open access to data (Lindenmayer et al. 2011). Global initiatives such as the Kunming-Montreal Global Biodiversity Framework (GBF) (Stephens 2023) emphasize the need for robust, interoperable biodiversity data to support evidence-based decision-making across scales. In this context, more data and better technical standards are often seen as key enablers of effective biodiversity governance. However, the assumption that increasing data volume and standardization will automatically lead to better decisions does not always hold true (e.g., Porto de Albuquerque et al. 2021)—particularly at the local level, where conservation action actually takes place. Local decision-makers often face practical, institutional, personal and epistemic barriers that limit their ability to access, interpret, or apply available data in ways that support their mandates and contexts. Meanwhile, local knowledge and plural values are rarely integrated into higher-level biodiversity data infrastructures. ENABLElocal*1 is a transdisciplinary mixed-method research project working in Czechia, Germany, and Sweden, which explores how biodiversity data is understood, accessed, and used by stakeholders involved in land use and conservation management. It highlights the importance of governance, context, and stakeholder engagement in determining whether biodiversity data becomes usable—and ultimately useful—for transformative action on the ground. The ENABLElocal project aims to bridge the persistent gap between biodiversity data—including monitoring datasets, species records, and environmental variables—and local-scale decision-making in conservation and land use. While large volumes of ecological data are increasingly accessible through national and international infrastructures, their uptake in local governance remains limited (Güntsch et al. 2024). ENABLElocal addresses this across three case studies—Germany, Sweden, and Czechia—each situated in distinct governance systems and biodiversity contexts. At CzechGlobe, the focus is on identifying institutional, social, and technical barriers that hinder effective data use in local decisions. More broadly, the project examines how biodiversity data supply-and-demand align (or not), and how governance conditions shape perceived usability.The project applies a living lab approach to foster iterative knowledge co-production with stakeholders in local biodiversity governance. Such integration of qualitative social science methods into biodiversity monitoring remains rare but essential for making data meaningful and actionable. The first round of workshops (autumn 2024) mapped stakeholder objectives, decision processes, and data types used or needed—from formal datasets to local knowledge. To examine actual data use in context, 39 semi-structured interviews were conducted (summer 2025) with local actors, exploring barriers to access, use, and relevance. A complementary survey, run across all case studies, identified commonly used datasets, unmet needs, and preferences. All activities followed harmonized protocols; interviews were coded using MaxQDA,*2 and user stories helped ground data needs in stakeholder perspectives.Preliminary results point to consistent challenges in using biodiversity data locally. Workshops revealed diverse stakeholder objectives, even within regions—complicating one-size-fits-all tools and reinforcing the need for context-sensitive approaches. Interviews highlight poor communication between data providers and users as a key institutional barrier. Many stakeholders lack spaces to express data needs or feedback on usability. Although analysis is ongoing, findings suggest governance reform may be as critical as technical improvements. Overall, the results underline ENABLElocal’s core insight: more data and better standards do not automatically lead to better decisions. Trust, communication, and participatory processes are essential to make biodiversity data actionable.ENABLElocal shows that the barriers to effective biodiversity data use are often not technical—but social, institutional, and relational. Even as global frameworks promote standardization and open data, local actors struggle to translate datasets into decisions. Addressing this requires more than better tools; it demands better relationships. Our work suggests that communication, co-production, and context-awareness should be treated as core components of biodiversity data infrastructure. Without them, even FAIR (Findability, Accessibility, Interoperability, and Reusability) data may fail to support fair outcomes (Wilkinson et al. 2016). As ENABLElocal continues, upcoming workshops (scheduled for January 2026) and interview analysis will deepen our understanding of how local governance systems can become active co-creators—not just passive users—of biodiversity knowledge.

HTML

XML

PDF

DOI: 10.3897/biss.10.182512

Authors: Ahmed Zine Eddine Marhoum, Anis Oukid, Hanane Ariouat, Nicolas Turenne, Eric Chenin, Youcef Sklab

Abstract: Herbarium collections serve as irreplaceable archives for taxonomic, ecological, and evolutionary research. Recent advances in deep learning now make it possible to design AI models capable of automatically annotating herbarium specimens (Ariouat et al. 2025, Sklab et al. 2025a, Sklab et al. 2025b), to develop intelligent systems for querying, visualizing, and exploring them. In previous work Sklab et al. 2025a, state-of-the-art Computer Vision (CV) models failed to generalize to unseen specimens, relying on superficial cues like labels and mounting paper rather than learning true plant morphology. Meanwhile, rich botanical texts remain disconnected from specimen images: CV models lack semantic depth, and LLMs extract traits but can’t ground them visually. Vision–Language Models (VLMs) emerge as a natural solution, combining the strengths of both to close this gap. By jointly reasoning across images and text, they have the potential to align visual perception with botanical semantics. However, most existing VLMs are trained on general-purpose datasets and fail to capture the specialized features inherent to herbarium specimens. To address this limitation, we investigate how adapting a VLM to the botanical domain enables it to answer expert-level questions about individual herbarium specimens.We adapted the PaliGemma-2 model (Andreas 2024), leveraging its Visual Question Answering (VQA) capabilities to align large language model knowledge with the observable content of herbarium images. In this setup, the digitized specimen is provided as visual context, and the user interacts with the model through targeted questions. The adapted model grounds its answers in visible features of the specimen—morphology, phenology, and context—rather than relying on generic species-level knowledge. By doing so, the system aims to exploit the semantic richness encoded in modern large language models while constraining it to the biological reality captured in a single herbarium sheet — a snapshot of the plant at the moment it was collected, rather than an idealized description of its entire life history. Our approach comprises three stages (Fig. 1): i) domain-specific dataset construction; ii) parameter-efficient fine-tuning; and iii) evaluation. We curated a dataset from herbarium specimens, combining digitized images with expert-annotations (28 traits). To translate structured trait annotations into a linguistically and semantically rich training corpus, we employed the LLaMA 4 (Meta AI 2025) language model to automatically generate over 50,000 question–answer pairs grounded in botanical semantics. To achieve domain adaptation under limited computational resources, we employed Quantization-aware Low-Rank Adaptation (QLoRA), which enables efficient fine-tuning by introducing trainable low-rank adapter layers into a quantized pre-trained backbone. We evaluated the fine-tuned model against two baselines, its un-fine-tuned version and the larger PaliGemma-28B model, using five metrics (Fig. 2) to quantify both semantic similarity and lexical fidelity with respect to expert annotations. Fine-tuning yielded substantial improvements across all evaluation metrics. The adapted PaliGemma-3B achieved a 40% increase in SBERT similarity (Fig. 2) relative to its base counterpart, reflecting enhanced semantic understanding of botanical traits. Likewise, gains in ROUGE-L and METEOR demonstrate the model’s capacity to employ precise botanical terminology rather than generic descriptive language. This highlights the model’s potential to ground linguistic outputs in visual evidence, enabling more accurate interpretation of herbarium specimens.

HTML

XML

PDF

DOI: 10.3897/biss.10.183020

Authors: Hervé Goëau, Vincent Espitalier, Jean-Christophe Lombardo, Christophe Botella, Toke Thomas Høye, Mads Dyrmann, Pierre Bonnet, Alexis Joly

Abstract: Invasive alien plants drive biodiversity loss and ecosystem disruption, costing billions globally. Roadsides faciliate spread, yet large-scale monitoring is difficult. Traditional methods like object detection and instance segmentation require costly, time-consuming annotations, impractical for thousands of species. Citizen science platforms like Pl@ntNet use efficient classification models for global identification, yet they struggle with high-resolution multi-species imagery. In this work we adapted the Pl@ntNet global model, trained on millions of images, to detect multiple invasive species in roadside imagery (Espitalier et al. 2025).Pl@ntNet uses a Vision Transformer trained on 6.6 million images covering 43,683 species. Optimized for single-species classification from fixed-resolution (384×384) close-ups of plant organs, it fails on high-resolution complex, multi-scale roadside scenes where plant size varies. We explored two strategies to bridge this domain gap. VaMIS (Variable Model Input Size), forces the model to process the whole roadside scene at once at a larger but non optimal resolution (1024×768). Multi-scale tiling acts like a scanner, decomposing the high-resolution image into hundreds of overlapping patches at multiple zoom levels. Each patch is analyzed independently at the model's native optimal resolution (384×384). If a species is identified with high confidence in any patch, it is considered present in the full image. Both were tested on the Danish Road Dataset (Dyrmann et al. 2021), containing 14,808 high-resolution (4024×3036) vehicle-mounted images for evaluating the detection of six invasive taxa: Cytisus scoparius, Lupinus polyphyllus, Pastinaca sativa, Reynoutria spp., Rosa rugosa, and Solidago spp. 44% of images contain no invasive species, and few include multiple taxa. Performance was measured using balanced accuracy (average recall of presence/absence classes, mitigating imbalance) and AUC (Area Under the ROC Curve, assessing global discrimination of True vs. False Positive rates). F1 and Jaccard index were used on a multi-label test set of 206 images.The tiling approach demonstrated strong generalization, achieving 84.9% balanced accuracy and 91.6% AUC, despite requiring no additional annotations, whereas VaMIS underperformed. To reduce domain shift and re-optimize internal parameters, essential for VaMIS to handle full scenes, we fine-tuned the whole transformer (VaMIS) or the classification head (tiling). Fine-tuning improved both: VaMIS rose from 66.8 to 91.8% accuracy, and tiling from 84.9 to 92.1%. On the multi-species test set, fine-tuned tiling achieved the best results (F1 = 0.85, Jaccard = 0.76), outperforming fine-tuned VaMIS and baseline CNN or YOLO, both models representing former classification and object detection baselines reproduced from previous work (Dyrmann et al. 2021) for comparison.Performance varied by species, but detection of Rosa rugosa consistently exceeded 99% AUC even without fine-tuning. Tiling was about 34x slower but far less resource-intensive to fine-tune (20 min vs 25 h).These results highlight complementary strategies. Tiling offers a generic, training-free pipeline, while VaMIS is faster but needs fine-tuning and memory. Reusing Pl@ntNet models cuts annotation costs, enabling scalable monitoring. Together, these strategies extend biodiversity monitoring to realistic roadside imagery, aligning with FAIR principles.

HTML

XML

PDF

DOI: 10.3897/biss.10.184359

Authors: Lydia Jennings

Abstract: As Indigenous Peoples continue to steward the world’s most biodiverse ecosystems, they are also asserting their rights within rapidly expanding digital environments. This talk explores how Indigenous Data Sovereignty serves as a critical tool for advancing rights-based approaches to biodiversity governance—from the tangible realms of land and water to the domains of code and cloud storage. Drawing from community-based examples, policy frameworks, and Indigenous-led data infrastructures, the presentation will highlight how Indigenous Peoples are protecting knowledge systems, asserting governance over ecological and digital territories, and challenging extractive models of conservation and data use. In an era when environmental decision-making is increasingly driven by algorithms, remote sensing, and big data, ensuring Indigenous Peoples’ control over their data is inseparable from their rights to land, life, and future-making. This keynote invites us to reimagine digital ecosystems not as deeply connected to sovereignty, relationality, and the ethical care of all forms of life.

HTML

XML

PDF

DOI: 10.3897/biss.10.183404

Authors: David Williamson, Emily Hartop, James Speed

Abstract: The world's museums collectively preserve billions of natural history specimens: identified by experts, curated, and verified. These specimens provide unique and irreplaceable historical data that can help to answer our most pressing questions relating to climate change, changes in phenology, habitat and species range, and biodiversity loss. Large-scale studies incorporating museum collections are vital to understanding these topics, but unlocking the information in collections typically requires physical access to the specimens and careful examination by experts. For the large-scale studies needed to understand global changes, involving millions of specimens across many nations, this is impractical. It is therefore increasingly important for museums to provide digital representations of specimens, comprising images and associated metadata, which can be accessed by researchers anywhere on the planet through searchable online databases. Digital specimens and modern computational analysis allow studies to be conducted at geographic and temporal scales that were previously impossible. We present two case studies demonstrating the value and limitations of such an approach.Firstly, we investigated global flowering phenology using millions of digital specimens from museums and herbaria, applying machine learning (ML) to rapidly estimate phenology from images and reveal long-term, worldwide trends (Williamson et al. 2025). This global study aggregated data from museum collections spanning three centuries, leveraging collection and digitisation efforts by hundreds of herbaria, and using modern computational methods to examine phenological trends at an unprecedented scale. We identified local changes in flowering seasonality, opening the door to fine-grained analysis of the effects of climate change on the world's flora (Fig. 1). However, this study represents only ~2% of the estimated 400 million specimens held in the world's herbaria (Davis 2023). Lack of funding for digitisation, inadequate standardisation of digitisation methods, and inconsistent metadata and data standards remain a challenge for automated studies of this type, with more than 40% of records matching our search criteria failing data quality checks (see Williamson et al. (2025) for details).In our second case study we digitised a portion of the R.H.L. Disney collection of scuttle flies (Diptera: Phoridae) at the Cambridge University Museum of Zoology. This fly family represents a "dark taxon" that may contain more than 100,000 species, of which <10% are described (Hartop et al. 2022). Dark taxa are both critical to measures of biodiversity and, by their hyperdiverse and cryptic nature, difficult to assess using traditional methods. The collection contains thousands of specimens, many of them type material, in some cases the only documented exemplar of a species. These specimens were mounted on glass slides making molecular methods such as DNA barcoding impossible. We used ML to extract morphometric landmarks to describe specimen characters and cluster specimens into species, allowing rapid identification of new specimens and letting the existing collection be tied to molecular analyses (Fig. 2). Since the collection was gathered over decades, with some specimens more than 50 years old, digitisation and ML were not a consideration in their mounting or documentation, leading to difficulties including unstandardised metadata and degradation of specimens between collection and digitisation. While it is not possible to address such problems in historical collections, this highlights the need to ensure that future collection and digitisation efforts produce data that are directly machine actionable. We welcome work such as East et al. (2025), which aims to provide guidelines for standardised, effective digitisation.By presenting these studies, and the challenges they involved, we aim to stimulate conversation on best practices around digitisation, data collection and dissemination, and the use of software tools for large-scale analysis of museum specimens to address pressing ecological questions.

HTML

XML

PDF

DOI: 10.3897/biss.10.184401

Authors: Tsiky Rabetrano, Laban Musinguzi, Melianie Raymond, Maheva Bagard Laursen

Abstract: The Global Biodiversity Information Facility (GBIF) Africa Regional Support Team offers a compelling example of how regionally embedded cooperation can empower local scientific communities and expand participation in global biodiversity data initiatives. Acting in strong coordination with the GBIF Secretariat, the Africa Support Team plays a critical role in bridging gaps between institutions, national GBIF nodes, and global infrastructure through tailored multilingual support.This extended abstract presents the team’s work supporting more than 20 countries across the continent through distributed helpdesk services, targeted virtual training, mentorship, and peer-learning initiatives. Over the past year, the Africa team has led the delivery of thematic webinars, co-facilitated multi-country training courses, and supported institutions with practical guidance for data cleaning, metadata preparation, and dataset publishing. In particular, the team organized training programs on topics including the Darwin Core (DwC) standard, data digitization and curation, use of the Integrated Publishing Toolkit (IPT), publishing using Darwin Core Archives (DwC-A) and the Data paper.Another innovation was the integration of regional feedback into the GBIF Data Use Club and pilot initiatives such as the Data camp, allowing African voices to contribute to shaping global biodiversity data communities.A key strength of this regional approach lies in its responsiveness to local needs. For example, the team’s bilingual support (English and French) ensures wider accessibility, while one-on-one sessions help institutions navigate challenges like georeferencing, IPT use, and licensing. The team also facilitates regional exchanges through interactive events.By embedding support close to the users in biodiversity informatics, the Africa Regional Support Team model enhances both technical capacity and long-term engagement, helping to respond more directly to the needs of the region. It demonstrates how locally anchored teams can increase biodiversity data publication and use while making sure African institutions not only contribute to but also benefit from global data infrastructure.

HTML

XML

PDF

DOI: 10.3897/biss.10.183654

Authors: Donat Agosti, Laurence Benichou, Tim Hirsch

Abstract: Published biodiversity research—from species descriptions and traits to distributions, interactions, and drivers of change—remains largely locked in inaccessible literature and electronic resources (Uetz and Agosti 2024, Uetz and Agosti 2025). This constitutes a major barrier to scientific progress and evidence-based policy (Fuller et al. 2014). It also limits the benefits arising from the dramatic explosion of available data through recent advances in digitisation and sharing of data from natural history collections, citizen science networks, monitoring programmes including environmental DNA (eDNA) sampling, and environmental impact assessments, among other sources. Making biodiversity knowledge openly available through connected, curated, and digitally accessible cross-domain knowledge bases is both a pressing challenge and a critical opportunity for the global research and policy communities. The Disentis Roadmap, a decadal strategy for liberating global biodiversity knowledge from scientific literature, emerged from a symposium held in Disentis, Switzerland, in August 2024, which assessed the impact of the 2014 Bouchout Declaration and identified strategic priorities for the coming decade. As of November 13, 2025, the Roadmap has been signed by 107 entities, comprising 74 individuals and 33 institutions. Discussions were grounded in existing workflows for data liberation (Fig. 1), involving a collaborative effort among key stakeholders: publishers (EJT, Muséum national d’Histoire naturelle Paris, Pensoft), Plazi, the Biodiversity Literature Repository at Zenodo, the Global Biodiversity Information Facility (GBIF), ChecklistBank, and the Biodiversity PubMedCentral hosted by the Swiss Institute of Bioinformatics Library Services (SiBILS). These efforts include the automated conversion and reuse of (as of August 2024) 198 biodiversity journals, resulting in substantial open Findable, Accessible, Interoperable and Reusable (FAIR) outputs from 95,000 publications, including 760,000 taxonomic treatments, 630,000 figures, and 1.7 million material citations available as open FAIR data in the Biodiversity Literature Repository at Zenodo and TreatmentBank. This insight to the data included in publications enables novel views of articles such as dashboards (Fig. 2, Bénichou et al. 2021) and also opens the door to create alternative research assessments (e.g. Declaration on Research Assessment (DORA). Additional input has been provided through the extensive digitization and annotation of historic texts through the Biodiversity Heritage Library (BHL), as well as the efforts of a large number of individual scientists and research groups such as Taxodros. Use cases have informed the process, including demonstration of literature reuse for assessments of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES), annotation with reference vocabularies and mining of biotic interactions at Biodiversity PMC, extraction of relevant evidence from literature for the Global Mountain Biodiversity Assessment, and the expansion of ChecklistBank as a joint initiative of GBIF and the Catalogue of Life (CoL) with highly automated, real-time input of data from new scientific publications.The signatories of the roadmap have given their support to an ambitious set of outcomes to be delivered over the next decade:100% of major public biodiversity research funders and academics publishers to enforce and enable FAIR data publicationBiodiversity publications to be accessible in machine-actionable formats, with all non-copyrightable parts of articles flowing into publicly-accessible repositories, thereby streamlining the re-use of this knowledge for further research, for science policy platforms such as IPBES, and ultimately for the wider public.Published research on biodiversity will be 'fully AI ready,' i.e., openly available for AI training and properly labelled for ingestion by machine-learning models, based on well-curated and semantically structured data, within appropriate ethical and legal frameworks.A dedicated portion of biodiversity-related research and infrastructure funding will be allocated towards ensuring that the above goals regarding access to biodiversity data and knowledge are met.The roadmap proposes creating a ‘Libroscope,’*1 through large-scale deployment of an established workflow by participating organisations, which will overcome existing barriers to the discovery and re-use of biodiversity data and knowledge currently ‘trapped’ in publications. Among the short to medium term actions proposed for advancing the Libroscope concept are:Developing a set of empirical use cases to demonstrate the value of biodiversity data liberation based on actual needs of researchers and policy makers.Developing minimum standards for machine-actionable publications to benchmark the extent of knowledge ‘liberation’ contained in different biodiversity publications.Building efficient workflows through existing research infrastructures to ensure sustainable access to open biodiversity knowledge, including establishment of appropriate governance, funding and legal frameworks.Establishing links with collection data infrastructures such as Distributed System of Scientific Collections (DiSSCo) and Integrated Digitized Biocollections (iDigBio), recognising that the role of specimens is crucial to permanently link research data and enable faster biodiversity description.Developing a global training programme on how to maximise the re-usability of biodiversity research publications, focusing on the value of liberating biodiversity data.Becoming a signatory to the Disentis Roadmap is highly encouraged.

HTML

XML

PDF

DOI: 10.3897/biss.10.180953

Authors: Lide Izeta-Zalduendo, Rafael Miranda, Amaia Rodeles, Maria Moran-Luis, Alexia González-Ferreras, Francisco Peñas, José Barquín, David Galicia

Abstract: Over one quarter of freshwater fish species are threatened with extinction, and the presence of dams and water management affects almost half of these species (Sayer et al. 2025). Man-made barriers like dams or weirs fragment the river network and alter habitat characteristics such as flow, sediment, nutrients, and temperature regimes (Maavara et al. 2020, He et al. 2024). As a consequence, there can be declines in habitat suitability for certain species, reflected in changes in species richness and composition (He et al. 2024). Moreover, the movement of fish along the river is hindered, making the access to spawning and feeding grounds difficult and reducing the genetic flow between populations (He et al. 2024). Species Distribution Models (SDMs) are a widely used conservation tool that allow us to explain and predict the geographical distribution of species based on their relationships to environmental covariates. They can be used to make inferences on environmental relationships (Guisan et al. 2017). The aim of this study was to model the distribution of the brown trout (Salmo trutta Linnaeus 1758), a migratory freshwater fish, in mainland Spain, taking into account the impact of river fragmentation.A detailed river network, divided into 432,099 river segments, each with associated hydrologic, climatic, topographic and geological variables (hereafter, environmental variables), was used to build the models. Barrier data was extracted from European, national and regional databases and used to calculate fragmentation variables. The pre-selected variables were then filtered based on their correlations and their variance inflation factors to avoid multicollinearity. The final variables used are described in Suppl. material 1.Brown trout occurrences, a total of 19,098 records from 48,076 samplings, were obtained from the Spanish Fish Chart*1. However, only samplings from the year 2000 onward and from the study area were retained (i.e., 27,391 samplings with 13,174 brown trout occurrences). Samplings performed through electrofishing where no brown trout were recovered were marked as true absences. Samplings from the same river segment were aggregated into a single presence-absence record, considering a single occurrence as evidence of presence. Presences and absences were balanced to avoid prevalence issues. After all this processing 4,344 presences and 4,344 absences were retained to model brown trout distribution. Eighty percent of the data was used to train the SDMs and the remaining 20% was further divided into two test datasets, each containing 10% of the data.SDMs were produced using the R package Biomod2 (Thuiller et al. 2025). Eleven sets of SDMs were built depending on the variables included. One of the sets included only the environmental variables and the 10 remaining sets of models included the environmental variables and one of the 10 fragmentation variables. For every set, individual models were built first and then used to create ensemble models. Their performance was evaluated using the first and second test datasets respectively, and measured through the Area Under the Curve (AUC) of a receiver operating characteristic curve and the True Skill Statistic (TSS).The base model (i.e., the ensemble model that only included environmental variables) had an AUC of 0.911, and a TSS of 0.673. The ensemble models that included fragmentation variables had a significantly better performance on average than the base model (Fig. 1). The most important variables to explain the species' distribution were consistently climatic related to rainfall (LCP) and temperature (BFMEDT) (Suppl. material 1).The highest metrics (i.e., AUC = 0.915 and TSS = 0.689) were obtained by the ensemble model incorporating the variable DistUpDwnN, which can be interpreted as the minimal barrier-free distance. According to the model's results, the brown trout is mainly distributed in the north of mainland Spain (Fig. 2), a distribution that is favorable to its environmental requirements. However, this model predicted a decreasing probability of occurrence as the minimal barrier-free distance increased, which is probably an artifact of the river segments in the north tending to be shorter and more fragmented.

HTML

XML

PDF

DOI: 10.3897/biss.10.183630

Authors: Lyubomir Penev, Teodor Georgiev, Teodor Metodiev, Nikolay Mehandzhiyski, Iva Boyadzhieva

Abstract: ARPHA Writing Tool (AWT) is an XML-first authoring tool that allows researchers to collaborate with co-authors, editors, reviewers, proofreaders, and others to prepare a ready-to-submit manuscript in a shared online environment before publishing it in any scientific journal or platform.Being the very first authoring tool to provide ontology-linked and domain-specific semantic tagging, AWT 2.0 also supports import of structured data from external resources (e.g., Darwin Core data from GBIF, iDigBio, BOLD) and semantic enhancements made during the writing process (e.g., citation intent based on the Citation Typing Ontology (CiTO) or automated links for taxon names to trusted online resources) (Agosti et al. 2024, Benichou et al. 2023, Shotton 2010).Amongst the most notable benefits designed for the domain of biodiversity science are various biodiversity-specific data import and export features and bi-directional links with leading biodiversity data aggregators (e.g., GBIF, BOLD, Catalogue of Life, ChecklistBank, TreatmentBank, Biodiversity Literature Repository at Zenodo, European Nucleotide Archive (ENA), OpenBiodiv) (Page 2023).Originally designed for the Biodiversity Data Journal and later integrated into other journals published on the ARPHA platform, the 2.0 version of the authoring environment is now also available as a fully collaborative editing tool. As such, it allows users to export manuscripts in either PDF or JATS-XML format, before submitting them to any scientific journal or platform. Thus, the AWT 2.0 prompts wider use of XML-born and FAIR scientific publications.AWT 2.0 can also be used for collaborative authoring, community review, editing and publication of documents for institutions or international organisations. The publications are automatically available in semantically enriched HTML, well-designed PDFs and machine-readable JATS-XML. This is how, for example, the collaborative SOLO platform allows the Soils for Europe project stakeholders to contribute to the development of strategic documents—called roadmaps—in a shared environment. There, the documents are subject to at least two rounds of transparent and open community peer review before they are published on the platform (Fig. 1). External reviewers can also leave comments as long as they are logged in with their Open Researcher and Contributor ID (ORCID). Those publications serve to address knowledge gaps relevant to the EU Soil Mission and guide funding priorities (CORDIS 2025).A key feature of publishing with SOLO is the annual roadmaps update, which ensures they remain current and in tune with the latest research and emerging trends. Once their authors update them, documents retain links to their earlier versions, so that their ‘evolution’ can be easily tracked. This iterative process ensures that the platform remains a ‘living’ resource for stakeholders, providing up-to-date guidance on soil health priorities and challenges.

HTML

XML

PDF

DOI: 10.3897/biss.9.183520

Authors: Maheva Bagard Laursen, Melianie Raymond, Laura Russell, Sanja Novakovikj, April Suen

Abstract: Open science and data accessibility are widely recognized as essential to addressing today’s global challenges. Regarding biodiversity loss, greater data accessibility not only improves understanding of threats and their underlying causes, it also helps inform actionable mechanisms to address them. The Global Biodiversity Information Facility (GBIF) has at the core of its mission, the mobilization of the data, skills, and technologies needed to enable access to comprehensive biodiversity information, demonstrated by the sharing of more than three billion occurrence records through its infrastructure.However, the map of occurrence data accessible through GBIF continues to show notable imbalances in the distribution of data, which often reflect gaps in capacity for data sharing. This disparity underscores the need to strengthen the ability of all to participate in and benefit from biodiversity data sharing, as emphasized in Targets 20 and 21 of the Kunming-Montreal Global Biodiversity Framework.GBIF’s approach to capacity development responds directly to this challenge by focusing on strengthening both the capacity to mobilize biodiversity data through GBIF and the capacity to use GBIF-mediated data, areas where GBIF is best positioned to have an impact. This approach is exemplified by GBIF’s European Union-funded Biodiversity Information for Development (BID) programme, which focuses on developing capacities from individual to regional levels, within sub-Saharan Africa, Latin America and the Caribbean, and the Pacific. The BID programme aims to empower communities of practice to actively engage in data mobilization activities aligned with user needs identified regionally or nationally. GBIF’s approach to capacity development is an ongoing learning journey, as illustrated by the continuous refinement of the BID programme since its launch in 2015. From addressing the steep learning curve in the mobilization of standardized data to exploring mechanisms for cross-regional knowledge-sharing, GBIF has tailored and implemented a broad range of solutions. To date, the BID programme has contributed to increased data accessibility, addressing knowledge gaps, and building the capacity of data holders and users. Many institutions began publishing data for the first time through the programme, and several countries, such as Suriname, Zimbabwe, and Tonga, recognized the value of maintaining national biodiversity information facilities. Moreover, outputs of the BID programme, including training material and published biodiversity data, have been reused countless times, both within and outside of the original BID target regions. With renewed support in 2024, the BID programme continues to offer opportunities for engagement in calls for proposals, the development of training material and workshops reflecting the latest practices in data mobilization and use. GBIF also seeks to align and expand these activities through collaboration with other related initiatives and by actively seeking feedback from the community. By fostering collaboration across multiple scales and promoting knowledge sharing mechanisms, GBIF’s capacity development ultimately seeks to empower the global community to address biodiversity challenges collectively.

HTML

XML

PDF

DOI: 10.3897/biss.9.183444

Authors: Alice Hughes

Abstract: Globally we are beginning to realise the need for environmental policies to be built on evidence. For example, the recent Kunming- Montreal Global Biodiversity Framework (GBF) is underpinned by a monitoring framework built on a selection of indicators designed to track progress towards GBF targets. Yet, any indicator is only as good as the data used to develop it, thus understanding the limitations and assumptions in environmental data that provides the basis for indicators is crucial.The exponential growth in the availability of diverse types of ecological data has seemingly changed the problem for many ecologists trying to understand the natural world from having insufficient data to approach many ecological questions, to one of how to usefully analyse ever growing volumes of data. Yet despite this huge volume of data, biases within the data require caution to ensure their sensible use, as biases may shape the outcomes of analysis and potentially misrepresent true ecological patterns. These shortcomings have fundamental implications for the use of this data to identify trends and patterns in biodiversity, and thus for our ability to provide what is needed for science-driven policy. Here I discuss data needs, and limitations, then provide recommendations to guide sensible and effective use and interpretation of data. I discuss frameworks and standard pipelines to enable more effective use of data, and better approaches for the generation of further data to aid the development of effective conservation, policy and management. I also discuss examples of science-driven indicators currently used within policy, such as ecological conservation redlines to provide case-studies of how science can directly and effectively be used within spatial prioritisation and management. Finally, I highlight priorities ahead for both research, and biodiversity targets, as well as discussing short and longterm solutions as we fill knowledge gaps and develop more accurate model approaches.

HTML

XML

PDF

DOI: 10.3897/biss.9.177491

Authors: Elie Saliba, Sophie Pamerlon, Matthew Blissett, Ben Norton

Abstract: The Darwin Core (DwC) standard (Darwin Core Maintenance Group 2023) is the most extensively adopted vocabulary maintained by the Biodiversity Information Standards (TDWG), with over 3.5 billion published records since its official ratification in 2009. TDWG strives to promote equitable participation in biodiversity informatics by participating in and facilitating broader internationalization initiatives, including making standards accessible to non-English-speaking communities. The initial comprehensive translation of the Darwin Core Terms documentation, as well as the Establishment Means Controlled Vocabulary, Degree of Establishment Controlled Vocabulary and the Pathway Controlled Vocabulary were finalized and published in the French language in May 2025. The translation was done using the Crowdin platform*1 and can be accessed on the official Darwin Core website*2. This endeavor offered a comprehensive view of the translation process, highlighting lessons learned along the way and leading to recommendations for similar efforts in the future. Some other questions are still unanswered to this day and will require further discussions, such as the treatment of deprecated terms, or finding the most effective method for integrating translations into the ongoing maintenance workflows of standards.In contrast to standards that were recently ratified such as Latimer Core (Saliba et al. (2025)), Darwin Core's extensive legacy use and maturity present specific translation challenges. During the translation process, questions emerged about adapting the language to enhance clarity and inclusivity. For French in particular, questions around linguistic purism arose, as the use of anglicisms is variable among different francophone communities. Strict translation practices might suggest avoiding borrowed terms, but privileging terminology already unanimously established in professional communities leads to pragmatic compromises to foster usability and adoption. An example is the term 'standard' itself, which should be translated as norme (e.g., une norme ISO), but that doesn't reflect the current usage of French biodiversity information scientists who use the transparent and equally possible standard (e.g., le standard Darwin Core du TDWG) instead. In other cases, if an official wording is used in regulatory or governmental contexts (such as espèces envahissantes, see for example Comité français de l'IUCN and Office Français de la Biodiversité (2023)), it was preferred over the alternative community-used anglicism (espèces invasives), as a way to reflect the normative nature of standards.French expressions are often more verbose than their English counterparts. In response, translators must balance fidelity with conciseness, occasionally rephrasing or adapting rather than rendering terms word-for-word to preserve the clarity and economy required in technical vocabularies. This is proving to be a particular challenge for the Humboldt extension, which is currently being translated. For example, the literal translation of the label 'Is Degree of Establishment Scope Fully Reported' becomes the fairly verbose label Est-ce que le Champs d’Application du Degré d’Implantation est Entièrement Répertorié. Ensuring that TDWG standards are available in French is essential to broaden participation from underrepresented communities across Africa, the Caribbean, the Pacific, and beyond. By lowering language barriers, the biodiversity informatics community can foster inclusivity, enhance data sharing, and strengthen global stewardship of biodiversity knowledge.

HTML

XML

PDF

DOI: 10.3897/biss.10.183778

Authors: Robert Lewis, Jonas Lembrechts, Peter Walker, Chunli Li, Sigrid Engen

Abstract: Background and Rationale Despite decades of progress in ecological monitoring, primary biodiversity and environmental data remain unevenly mobilised and poorly interoperable (Hampton et al. 2015, Poisot et al. 2019). Datasets, often gathered with public funds, frequently remain inaccessible or insufficiently described, limiting their reuse in global syntheses (Culina et al. 2018). Ecologists’ concerns about trust, transparency, and control of shared data persist, particularly where data production is resource-intensive or socially embedded. These concerns echo the foundational properties of distributed ledgers, where ownership and governance are distributed across peer networks rather than centralized repositories (Lewis et al. 2023).Forests exemplify both the potential and the challenge of such decentralised infrastructures. As globally significant carbon and biodiversity reservoirs, forests are also deeply fragmented across ownership and jurisdictional boundaries. In Europe alone, over half of forested land is privately owned, yet these actors often lack mechanisms to derive tangible value from stewardship. At the same time, digital twins (macroecological models) that integrate in situ and remotely sensed data, are becoming central to forest policy and monitoring frameworks (e.g., Food and Agriculture Organization of the United Nations (FAO), Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES), Global Biodiversity Framework (GBF)). ForestWeb3 (FW3) hypothesizes that a decentralised, Findable Accessible, Interoperable, Reusable (FAIR; Wilkinson et al. 2016, Nosek et al. 2022)-aligned data network can unlock the latent value of underused biodiversity data while building trust and incentives for participation..Objectives Mobilisation and harmonisation of forest biodiversity and environmental data (Objective 1): to spearhead a shift from data curation to data stewardship through a decentralised data infrastructure built on open-source blockchain frameworks.Incentivisation and uptake (Objective 2): to design transnational pathways through which private forest owners and local communities can be economically rewarded for verifiable ecological data via nature-backed digital assets and ReFi mechanisms.Together, these objectives align technical innovation (Objective 1) with behavioural and economic motivation (Objective 2), establishing the groundwork for distributed biodiversity observatories capable of sustaining long-term ecological data flows.Methodological Approach WP 1 develops a blockchain-based data ledger with smart contracts that autonomously manage data registration, access control, and reuse. Metadata and identifiers are immutably recorded on-chain, while primary datasets remain decentralised on contributor-managed nodes. This architecture enables contributors to retain data sovereignty while ensuring transparency and traceability in reuse transactions.WP 2 extends the infrastructure to real-time environmental sensing through the integration of modular Internet ofThings (IoiT)-based microclimate sensors. These devices stream environmental data at high temporal resolution directly into the distributed ledger, forming a Decentralized Physical Infrastructure Network (DePIN) for ecological data.WP 3 links these data streams to the creation of digital twins of forest ecosystems, combining in situ biodiversity observations with satellite and climate datasets to model ecosystem integrity. These models underpin the valuation of nature-backed digital assets, a form of tokenised evidence for ecological performance, providing the data foundation for voluntary biodiversity and carbon markets.Finally, WP4 investigates forest owners’ perceptions, motivations, and barriers to adopting regenerative finance (ReFi)-based conservation mechanisms. Through interviews and a pan-European survey, it explores how varying sociocultural and institutional contexts shape engagement with emerging biodiversity credit schemes, drawing parallels to established Payment for Ecosystem Services frameworks (Kaiser et al. 2021).Significance and Legacy FW3 exemplifies the convergence of data decentralisation, digital sensing, and regenerative economics, a triad capable of transforming how ecological knowledge is produced, verified, and valued. By embedding data provenance and attribution within the infrastructure itself, we addresses long-standing issues of trust and recognition in ecological data sharing. Its incentive mechanisms offer pathways to decouple conservation finance from traditional public funding, potentially scaling stewardship and democratizing data mobilisation across millions of hectares of privately owned forest land.The project’s legacy lies in demonstrating that data infrastructures can be both scientific and economic commons, capable of sustaining biodiversity monitoring through distributed participation. Beyond its immediate technical deliverables, ForestWeb3 contributes to a broader vision of dynamic, self-sustaining ecological data ecosystems that power both global biodiversity frameworks and locally grounded conservation action.

HTML

XML

PDF

DOI: 10.3897/biss.9.183197

Authors: Hanieh Saeedi

Abstract: Introduction: Oceans cover over 70% of Earth’s surface (Cael et al. 2023). An estimated 2.2 million marine species exist, yet nearly 80% remain undescribed (Mora et al. 2011). Around 370,000 species are accepted in the World Register of Marine Species (WoRMS), but open-access occurrence data exists for only about 200,000 in the Ocean Biodiversity Information System (OBIS). Documenting marine biodiversity is vital for making evidence-based policy and management decisions in order to maintain ecosystem stability and planetary health. Initiatives such as the Census of Marine Life and open-access databases like OBIS continue to transform understanding and support the UN Decade of Ocean Science. Through data sharing and global collaboration, we can better estimate and conserve marine biodiversity by first identifying data and knowledge gaps.Methods: In two underrepresented areas, the South-West Pacific (SWP) and the Indian Ocean (IO), the current biodiversity patterns of fauna were mapped to identify the knowledge gaps and distribution patterns by depth zones. All occurrence data (Animalia) were extracted from the OBIS and the Global Biodiversity Information Facility (GBIF). The occurrence records were quality-controlled in accordance with the OBIS data quality guidelines (OBIS 2025). Only accepted marine taxa were retained after cross-referencing species names with the WoRMS. In total, 5,441,962 occurrence records in the SWP and 7,768,826 occurrence records in the IO were used in this study (Suppl. material 1).Results: The number of occurrence records decreased with depth in all taxa in both oceans. More than 60% of the occurrence records with available depth information were from shallow waters (0–200 m), highlighting significant knowledge gaps in deep-sea biodiversity. Still, more than 11 million km² of the SWP and IO had fewer than 50 occurrence records after data cleaning in the shallow waters. Based on 5-degree latitudinal bands, the higher latitudes of the SWP (0–25°S) were less sampled or the data were not reported, compared to the lower latitudes. Mid-latitudes of the SWP (30–45°S, eastern and western Australia) had the greatest distribution records, mostly related to Chordata, followed by Arthropoda and Mollusca. However, 10–25°S latitudes of the SWP had the highest number of reported species, mostly associated with Chordata, followed by Mollusca and Arthropoda. The mid-latitudes of the IO (5–30°S) were less sampled, or the data were not reported, compared to the upper and lower latitudes. Chordata occurrence records were the exception, with a peak at 10–25°S latitudes showing the highest distribution records, followed by Arthropoda. Also, 10–20°S latitudes of the IO had the greatest number of species, mostly related to Chordata, followed by Arthropoda and Mollusca (Suppl. material 2) and (Suppl. material 3).Application: The generated knowledge is crucial for strengthening biodiversity monitoring and ensuring rapid, accessible information for policymakers through science-policy interfaces such as the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES), thereby supporting the development of urgent conservation strategies for underrepresented and threatened marine ecosystems, such as the SWP and the IO, before it is too late.

HTML

XML

PDF

DOI: 10.3897/biss.9.183376

Authors: Laurence Benichou, Lyubomir Penev

Abstract: For over a decade, a lot of effort has been put into making millions of pages and taxonomic treatments available online. Yet, too often, biodiversity knowledge remains locked in hardly accessible and non-machine-readable literature and electronic resources. Indeed, most prospective publications are still produced in an “old-fashioned” way (PDF), thereby posing a risk of following a pattern we have seen with legacy literature from the past centuries.Even though access barriers can be demolished using retroconversion (i.e., converting a PDF into an XML annotated file once the article is published), it is far more efficient to have author- and editor-vetted annotations and semantic enhancements to the texts and data ready ahead of publication. Doing so avoids discrepancies, and ensures earlier and rapid dissemination and re-use of data. Beyond open access, the challenge for publishers is to reach interoperability, which would ultimately shorten the too-long time to find and analyse past literature (Fontaine et al. 2012) and therefore speed up the description of taxonomy.Making data within publications findable, accessible, interoperable, and reusable (FAIR) implies text structuration, semantic annotations*1, and standardization. Furthermore, to link data, information and knowledge contained in literature and other electronic resources to uniquely identifiable components (i.e., images, tables, references, taxonomic treatments; see Fig. 1) means that those would be re-usable in research and policy covering biodiversity and other domains. Ultimately, this enables immediate re-use of the data, and the integration of publications into a comprehensive global biodiversity knowledge graph.Following the vision of the Disentis roadmap*2, this presentation will outline and demonstrate the value of advanced methods for scholarly publishing, production and FAIRisation of biodiversity data during the journal production process. We described the latest developments in the scientific publishing industry and—more specifically—how to publish linked and semantically enhanced research outcomes. The presentation clarifies the concepts that are relevant within the publication for semantic enhancement. Highly-automated, XML-first, AI-assisted existing workflows for semantic enhancement of articles are introduced, including the related standards, all dedicated to making published data FAIR by creating bi-directional linked data from and to the published article. Semantization goes beyond metadata and all relevant information for biodiversity are isolated, annotated, and linked with persistent identifiers (Chester et al. 2019, Agosti et al. 2022) including: taxonomic treatment, taxon names, material examined, and material citation, the latter including country, locality, coordinates, collection date, collectors, and specimen code.Once annotated, the XML-based publication can be disseminated through a partnership with Plazi and all components pushed to relevant databases, e.g., Ocellus, Global Biodiversity Information Facility (GBIF), Biodiversity PMC (PubMed Central), ChecklistBank, National Center for Biotechnology Information (NCBI). The potential reuse of data is multiple and will benefit the publishers as well as the research community. Indeed, this wide dissemination of data contained within their articles enables publishers to create dashboards to measure e.g., numbers of articles published, numbers for data described, the richness of the data, the number of occurrences by continent.The benefits are multiple: not only do XML-first journal production workflows save time, effort and provide FAIR-born data on the day of publication, they optimize and accelerate the dissemination of data while providing human- and machine-readable publications. Semantically structured FAIR data pave the way for training and use of AI in publishing.

HTML

XML

PDF

DOI: 10.3897/biss.9.183350

Authors: Ram Dayal Vaishnav, Vijay Barve, Lavanya Bhat, A. Ashwin, Anukriti Sharma, Chitra Shanker

Abstract: The City Nature Challenge (CNC) has rapidly expanded across India from 2023 to 2025, leading to a surge in biodiversity observations on iNaturalist. While the volume of data is impressive, its long-term research and conservation value depends heavily on data quality (Vattakaven et al. 2022)—particularly the proportion of observations reaching "Research Grade" (RG). This study investigates patterns and barriers related to RG achievement in CNC India observations over three consecutive years using iNaturalist data (iNaturalist Community 2023, iNaturalist Community 2024, iNaturalist Community 2025).We analyze RG percentages at regional, state, and city scales, comparing the total number of observations year by year (Fig. 1). By tracking these metrics from 2023 to 2025, we quantify the "identification lag" (Fig. 2)—the gap between data quantity (uploads) and data quality (verified records). Our analysis identifies whether growth is driven solely by participation or if it is supported by a parallel increase in expert identification (Fig. 3). In addition to temporal (year-wise) comparisons, we also investigate spatial (geographical) variations in data quality across India.We also identify taxa and locations with disproportionately high numbers of "Needs ID" observations (which are not yet considered as RG; Fig. 4). This analysis is conducted both horizontally—i.e., comparing taxonomic groups (such as, Plantae, Fungi, Animalia, Insecta), and vertically—i.e., comparing taxonomic rank (such as, family, genus, species). This approach helps us to evaluate whether these bottlenecks stem from inherent identification challenges, observer behavior, or a lack of specialized taxonomic expertise within the local identifier community.Our results indicate a widening gap between upload volume and RG attainment (Fig. 2), particularly in high-performing regions. This disparity is most acute for taxa requiring specialized expertise and for observations at the species level. These findings offer practical insights for enhancing community engagement, directing expert identification efforts, and designing support mechanisms for new observers.By providing a nuanced perspective on the evolution of data quality within a massive, volunteer-driven dataset, this presentation contributes to broader discussions on sustaining and improving the living data ecosystems that support contemporary biodiversity science.

HTML

XML

PDF

DOI: 10.3897/biss.9.183132

Authors: Majid Vafadar

Abstract: The Museum für Naturkunde Berlin (MfN) DataHub*1 is an open-source web service and workflow engine developed to execute automated data-integration and migration workflows in continuous and parallel scenarios. Data migration and integration remain major challenges in publishing biodiversity data that follow international standards. To overcome these, a centralized service was created to coordinate and concentrate the computational power required for large-scale data transformation. Deployed at the Museum für Naturkunde Berlin, it now serves as the core of the institution’s scientific data-management infrastructure.This service supports digitization pipelines by iterating through datasets and files to integrate them into designated target systems. Following the ETL (Extract, Transform, Load) (Moreau 2015) principle, it extracts data from internal databases and shared storages via secure protocols such as SMB*2 and SFTP*3, transforms and validates them, and loads the compliant outputs through target-system API endpoints. The overall architecture and data flow are shown in Fig. 1.Operations are controlled through a web dashboard that allows execution and monitoring of pipelines either manually, automatically, or with AI-agent assistance. Implemented using the Django Web Framework, the service runs modular Python scripts and exposes all functions through RESTful APIs. A MCP*4 server provides an AI-readable interface, enabling both human- and machine-driven operations.Connected to the museum’s storage systems, the DataHub validates, enriches, and transforms data with a dedicated validator ensuring each record meets predefined structural and semantic rules. A persistent integration pipeline imports datasets into the museum’s Specify collection management system and its digital catalog, making data accessible for research and public use. It also prepares standardized packages for external partners such as GBIF (Global Biodiversity Information Facility), following the Darwin Core (Wieczorek et al. 2012) format and specific project requirements. Integration with field-data applications like ODK (Open Data Kit) ensures mobile data collection can enter the same pipeline. All operational steps, including validation, transformation, and API transactions are fully logged for transparency and reproducibility of the operation.A key innovation is the AI-integration layer, linking through the MCP*4 server to an AI agent built with LangChain library and Qwen3 LLM*6 model, executed locally via Ollama platform. This component assists with workflow orchestration by optimizing tasks order, resource allocation, error recovery and live reports thereby reducing manual supervision.By combining ETL*7 pipelines with AI-assisted orchestration, the DataHub provides a flexible, scalable engine adaptable to different collection domains. Its modular and open-source design promotes reproducibility and extension to new systems. The centralized architecture enhances data quality and FAIR (Findability, Accessibility, Interoperability, and Reusability) Wilkinson et al. 2016 compliance, offering a practical and scalable solution that can be deployed in natural-history institutions aiming to modernize their digital-collection infrastructures and ensure the continuous availability of reliable, high-quality biodiversity data. The open-source code and technical documentation of the MfN DataHub are available on the museum's GitHub repository.*1

HTML

XML

PDF

DOI: 10.3897/biss.9.183060

Authors: Jason Best, Sam Ekberg, Mare Nazaire, Ana Niño, Krishna Shenoy

Abstract: IntroductionThe Extended Specimen Network (ESN) (Lendemer et al. 2019) concept recognizes that scientific value extends beyond physical specimens to include field notes, field photographs, publications, and other research products. The Sherwin Carlquist ESN (Anonymous 2025) unites the legacy of Dr. Sherwin Carlquist (1930–2021), a prominent American botanist whose career focused on island biogeography and plant anatomy. This ESN connects Carlquist's physical specimens at California Botanic Garden (CalBG) with his archival materials at the Botanical Research Institute of Texas (BRIT), demonstrating both the potential and challenges of creating comprehensive ESNs across multiple institutions and digital platforms.Project Infrastructure and GoalsThe project established several objectives:imaging and transcribing approximately 9,000 specimensimaging and transcribing 26 field notebooks;digitizing and describing approximately 100,000 field photographs with metadataestablishing linkages between field images and botanical specimensapplying discovered linkages across digital platformsDigital records are distributed across the Consortium of California Herbaria (CCH2)*1 , a Symbiota*2 portal for biological specimens, and the Portal to Texas History (PTH)*3 for archival materials.Cross-Platform ChallengesImplementation of the ESN on the digital platforms revealed some limitations. The PTH's restriction to Dublin Core metadata*4 forced use of generic relationship terms "references" and "isReferencedBy"—both lacking semantic specificity. PTH also offered limited external linking capability (with relationships and URLs only describable in free text) and lacked the ability to search or filter to target only archival records with linked resources. Symbiota initially lacked functionality for linking to external archival items, a limitation addressed by creating a linked resources module through collaboration with the Symbiota developers.Both platforms shared limitations: difficulty creating complex linkages for entities like people, taxa, and publications; limited integration with existing linked data networks; and insufficient semantic richness for expressing some ESN relationships.Linked Data ApproachLinked data represents structured information interlinked through semantic relationships, enabling machine-readable queries. This aligns naturally with ESNs, where specimens, images, notes, and publications form interconnected information networks. Wikidata, containing millions of structured items, represents the largest openly accessible linked data resource. Wikibase, the open-source software underlying Wikidata, offers institutions the opportunity to implement similar infrastructure while connecting to the broader linked open data ecosystem. We established a Wikibase instance*5 to experiment with and explore the capabilities and potential of the platform as a complementary extension of the Carlquist ESN created in our primary digital platforms.Implementation MethodsTo populate the Wikibase ESN, we began by aggregating data using multiple approaches: Sickle*6 (a Python module) retrieved PTH archival records; Symbiota's Darwin Core Archive (GBIF 2021) provided specimen data; and scripts using the Biodiversity Heritage Library (BHL) API*7 retrieved publication data. Data preparation involved iterative cleaning and standardization using OpenRefine*8. Wikibase properties were created using the WikibaseIntegrator*9 Python module, datasets were uploaded via WikibaseIntegrator with SPARQL*10 queries, and relationships established using the Wikibase QuickStatements tool*11.ResultsThe Carlquist ESN Wikibase instance contains 8,941 specimen records linking to California Botanic Garden/CCH2; 13,462 field image and notebook records linking to BRIT/PTH; 301 publication records (37 linking to BHL); and biographical data for Carlquist linking to Wikipedia and Wikidata. Each record maintains bidirectional links to source platforms, creating cross-institutional relationship networks.This integrated approach enables capabilities that were challenging or impossible within original platforms: specimens link to field images across institutions; publications connect to referenced specimens and images; complex queries retrieve related items across data types; and the ESN connects to broader linked data networks through Wikidata integration.ConclusionWikibase offers significant promise as an ESN enhancement layer. Rather than replacing existing infrastructure, it provides complementary capabilities: specimen and extension relationships across platforms; expansion beyond platform limitations; comprehensive complex relationships; cross-dataset queries; inclusion of entities not in collection systems; and openness for community contribution. Before Wikibase can be fully embraced for implementation of an ESN, more effort will be required to accurately model entities such as specimens, field notebooks, and field photos. Nevertheless, this experimental implementation demonstrates the feasibility and value for institutions seeking to extend digital collections beyond traditional platform constraints while participating in the linked open data ecosystem.

HTML

XML

PDF

DOI: 10.3897/biss.9.183306

Authors: Catherine Jardine, Denis Lepage, Kyle Horner

Abstract: Targets 20 and 21 of the Kunming-Montreal Global Biodiversity Framework establish that access to good data and innovative data products are crucial to halting and reversing biodiversity loss, and that biodiversity data access has implications for all targets of the framework. Birds Canada’s NatureCounts platform*1 seeks to meet these needs by supporting easy and accurate data collection, interpreting data to produce meaningful knowledge and data products, and sharing data according to the FAIR principles (Findable, Accessible, Interperable, Reusable) to support conservation action and policy.NatureCounts supports the collection of robust biodiversity data by professional and volunteer-based monitoring programs. The NatureCounts mobile app and web interface are customizable data collection solutions that integrate standardized data from the field directly into a sharing-ready repository using a standardized schema. The flexible architecture accommodates nearly any monitoring protocol, while a user-friendly interface, unique tools, and instantaneous data upload incentivize adoption, encouraging FAIR data participation by projects of all types and sizes.Data collected using these tools are uploaded to the NatureCounts database. Hosting over 250 million records, this massive repository holds endless potential for conservation applications. Tools including an online data explorer and R package facilitate easy data access by researchers and conservationists. Flexible data access permissions support the security of sensitive records and Indigenous data sovereignty. Various data products support research and conservation, and directly address the targets of the Global Biodiversity Framework. For example, a dedicated workflow underpins the process of identifying Canada’s Key Biodiversity Areas—spaces designated as vital to the conservation of biodiversity in Canada—in accordance with Target 3. Another uses the data to set and evaluate federal population goals for Canada’s birds for the federal government, integrating biodiversity into decision making as per Target 14. A third feeds data directly into the Canadian process for identifying endangered species, addressing extinction risk as specified in Target 4 and seamlessly connecting data collection to policy. NatureCounts also processes over 9000 requests for raw data yearly by the conservation community. Users query and filter the data, then access them either through a browser-based download portal or the dedicated naturecounts R package.*2 To help NatureCounts data users interpret raw data, the Birds Canada GitHub page*3 contains publicly available repositories and documents that detail workflows for processing and analyzing data accessed through NatureCounts. These well-documented, tested, and easily shared repositories ensure reproducible research practices. To date, data from NatureCounts have supported over 4200 scientific publications and an immeasurable amount of unpublished work. Data from NatureCounts are used for species assessments, land use planning, impact assessment, academic research, climate change mitigation, and much more, allowing data from NatureCounts to be used in pursuit of nearly every target in the framework. Through the ongoing development of NatureCounts, Birds Canada aims to fulfill the goals of the Kunming-Montreal Global Biodiversity Framework, and make measurable progress for biodiversity in Canada.

HTML

XML

PDF

DOI: 10.3897/biss.9.183181

Authors: David Fichtmueller, Sofie Meeus, Quentin Groom, Anton Güntsch

Abstract: Understanding the current landscape of taxonomic expertise is crucial for addressing the growing challenges in biodiversity research, conservation, and data management. There remains a lack of comprehensive insight into who the active taxonomists are, where they work and which taxon groups they specialize in. Within the European TETTRIs project (Transforming European Taxonomy through Training, Research and Innovations) we aim to fill that gap by systematically determining the taxonomic expertise in Europe, using different sources. De Nolf et al. (2023) showed how to retrieve expertise from taxonomic publications.This talk focuses on an approach to retrieve taxonomic expertise from specimen data. It will highlight how data quality issues can be a significant hurdle for effective data reuse. We consider a current European expert as anyone who has either collected or identified a specimen in the last 10 years and who is working in or coming from a country of the European Political Community (EPC).The core workflow of the specimen approach is shown in Fig. 1, however for this talk we want to focus on the first three steps.The core data consists of around 20.5 million specimen records retrieved from GBIF via the new SQL-API in the DarwinCore (DwC) format (GBIF.org 2025). From this dataset, the columns for names and identifiers of the people who collected or identified the specimen (dwc:recordedBy and dwc:recordedByID, as well as dwc:identifiedBy and dwc:identifiedByID respectively), were extracted, combined and aggregated, as is demonstrated in Fig. 2. This reduced the number of records to 570,000 unique name and identifier combinations.Only around 2% of the name/identifier combinations actually had an identifier associated with them. This limited the usability of the data for the next processing steps as any subsequent findings would be based on only this small subsample. Thus, further cleaning, disambiguation and matching of names became necessary.The biggest obstacle during the cleanup phase were huge inconsistencies with which names where written. In DwC, the fields dwc:recordedBy and dwc:identifiedBy are not repeatable, therefore it is common to have the names of multiple persons in each field. The recomended way of concatenating multiple records using the vertical bar symbol (" | ") is rarely used. More often, enumerations are made using commas, semicolons, the ampersand symbol, or the word "and" in various languages. Other issues were the inconsistent ordering and separation of given and family names as well as group attributions or non-name content, such as numbers, affiliations or roles.Further processing was done using a Python script that took the different formats, rules and exceptions into account, to split the combined names into individual names and then their given and family name parts, reducing the number of records to 430,000. At this point we would like to acknowledge that many of the processing steps rely on generalization and heuristics that will not do justice to many names, as well as our strong emphasis on European name customs. We are aware of these shortcomings and recognize that people’s names are more complex than this approach is able to handle, as shown by the insightful and entertaining article "Falsehoods Programmers Believe About Names" (McKenzie 2010). Despite this, we still consider our general conclusions as valid, as the process relies on knowledge aggregation.The cleaned-up names could then be better matched to other GBIF records, Bionomia, Wikidata and matches from the taxonomic publications approach. While the results of the subsequent analysis are beyond the scope of this talk, the necessary preprocessing steps clearly showed that data coming from different sources is messy and requires extensive cleanup in order to properly reuse it. We encourage people to leverage their roles in the biodiversity research community to improve data and foster easy data reuse. Individual researchers, software developers and journal editors and reviewers can all play a part, e.g., by using and promoting ORCIDs, improving data entry mechanisms, or warning about data quality issues and inconsistencies. Together, we can improve data quality and data consistency in our field, as they are key factors for effective data reuse across research projects.

HTML

XML

PDF

DOI: 10.3897/biss.9.183295

Authors: Christian Bölling, Théo Léger, Joachim Snellings, Margot Belot, Franziska Schuster

Abstract: Natural history collections preserve invaluable records of biodiversity across time and space. Each specimen is typically accompanied by one or more labels documenting provenance, locality, and contextual data. Mobilizing this information is crucial for research on biodiversity change, biogeography, and taxonomy. To date, much of this data remains inaccessible for computational knowledge engineering approaches because manually processing and converting the sources into structured, interoperable data formats is a labor-intensive challenge due to the volume, heterogeneity, and complexity of the documents—and curatorial resources for fulfilling these tasks are generally insufficient. Artificial Intelligence (AI)-based methodologies can significantly accelerate this process and make it economically scalable.We present KIEBIDS*1, an open-source framework for specifying and executing AI-based workflows for information extraction from specimen label images. Following a linear data-pipeline architecture, workflows comprise five sequential functional steps for information extraction:image pre-processing (to prepare input images for subsequent analysis),layout analysis (to identify image regions that are relevant for information extraction),optical character recognition (to identify text on the syntactical level),semantic parsing (to identify text that references categories of interest),entity linking (to identify entities of interest mentioned in the text with authority records).Modularity and adaptability are central design principles for the framework's architecture. Each function can be realized by one or more modules that operate independently through file-based input and output, enabling substitution or extension as new technologies emerge. This ensures flexible adaptation to various information extraction goals or new data domains.In the current release, image pre-processing is implemented using the OpenCV framework with steps for resizing, grayscale conversion, noise reduction, and binarization. Layout analysis, based on the Segment Anything Model, identifies image regions that depict labels. Character recognition is implemented using two alternative modules. Besides EasyOCR, Moondream is used to leverage locally-deployable vision-language model (VLM) technology. Semantic parsing is implemented using spaCy and regular expressions, as rule-based parsing has proven efficient for syntactically well-defined entities, such as dates or coordinates, given the sparse context of label texts. Entity linking, in the current release, is realized for geographical place names using the GeoNames application programming interface (API).The input for a given workflow run consists of document images and configuration parameters. The configuration parameters encompasssettings for the pipeline as a whole, such as location of input and output files and execution mode,the configurable settings for each functional step of the pipeline, e.g., models to be used, model parameters or the tag selection for the semantic tagging.The workflow's output are PAGE-XML files containing image annotations, including the extracted and annotated text. Optionally, intermediate data and evaluation metrics can be assessed. Integrating seamlessly with Python codebases, Prefect is used for scheduling, monitoring, and graphical user interaction.By combining existing open frameworks rather than developing new components, the project leverages recent advances in computer vision and natural language processing to mobilize biodiversity data. Future developments will focus on improving user experience, integrating better models for handwritten text, and expanding semantic analysis capabilities. KIEBIDS' source code*2 is openly available and locally deployable with moderate hardware requirements.

HTML

XML

PDF

DOI: 10.3897/biss.9.183171

Authors: Jutta Buschbom, James Macklin, Deborah Paul, Elizabeth R. Ellwood, Alina Freire-Fierro, Wouter Addink, Teresa J. Mayfield-Meyer, Breda Zimkus, Gil Nelson, Kerstin Lehnert, Katelin Pearson, Mariko Kageyama

Abstract: The bio- and geodiversity communities aim to build digital infrastructures capable of delivering high-quality, well-structured, and richly interlinked data at a scale that enables powerful analyses to support the conservation and management of Earth’s diversity. Informaticians, scientists, and resource managers face the challenge of globally accelerating and upscaling open data availability. Achieving this requires sustainable, resilient, and bi-directional socio-technical connections that promote interactions and feedback both ways across the data landscape. Such interactions will connect global and national repositories to local providers, data stewards, and data users in ways that are context-sensitive and adapted to partners' location, time and communities.Although communities ranging from individual local experts to the governing bodies of global platform consortia are highly motivated and productive, they struggle to manage the rapidly growing volume of data, keep pace with technical innovation, and adhere to shared standards and best practices. Persistent challenges include insufficient recognition and visibility for contributors; functional and operational limitations in infrastructure; and inadequate long-term funding. Although the importance of the life cycles of data, infrastructures, and services is well-recognized, insufficient expertise in assessing the value that they represent and generate, and integrating those benefits into global, whole-community finance strategies presents an additional challenge. However, methodologies developed by global initiatives such as the United Nations System of Environmental-Economic Accounting (SEEA; United Nations et al. 2025) and UN Biodiversity Finance (BIOFIN; Cruz-Trinidad et al. 2024) provide innovative examples for addressing these. Without mechanisms that return generated value to data providers, stewards, and infrastructure developers, the community cannot initiate the self-reinforcing cycle of social and technical capacity growth needed for global upscaling. Instead, the current resource-limited landscape threatens the robustness of community networks and the long-term sustainability of existing infrastructures.The insights from comprehensive community outreach and infrastructure reviews conducted by the U.S.-based BIOFAIR (Building an Integrated, Open, Findable, Accessible, Interoperable, and Reusable) Data Network form the foundation for a living roadmap designed to support effective solutions and sustainable models for global biodiversity data infrastructures (Kunkel et al. 2025). Informed by these findings, the International Partners for the Digital Extended Specimen (IPDES) network are discussing shared and unique strengths and gaps across partners at different levels in this ecosystem. Adding theory of change (Rice et al. 2020), as well as socioeconomic and biodiversity finance considerations to the BIOFAIR roadmap, members of IPDES developed an extended model of infrastructure sustainability and a project proposal. Key elements include:involving economists in the process of integrating data, infrastructure, and services in information-economy value chains into natural capital accounting frameworks (e.g., SEEA), and developing comprehensive finance plans for bio/geodiversity data (e.g. adapting the methodology of BIOFIN);building governance and organizational frameworks that support transparent, community-centered decision-making;engaging the global community through coordinated participation and shared infrastructures;conducting pilot implementations that apply accounting methods to key use cases; andperforming risk assessments of, and establishing safeguards for, the arising feedback loops in a socioeconomic system of returning value to data providers, stewards and users.Together, these actions outline a path toward community-wide finance strategies that can support sustainable, scalable, and globally coherent biodiversity data infrastructures.

HTML

XML

PDF

DOI: 10.3897/biss.9.183270

Authors: Sachit Rajbhandari, Eric Raes, Philipp Bayer, Matthew Smith, Andrew Bissett, Bruce Deagle, Jodie Van De Kamp, Pascal Craw, Miwa Takahashi, Katherine Tattersall

Abstract: Environmental DNA (eDNA) is increasingly used to monitor biodiversity, biosecurity and invasive species, providing insights into species presence across ecosystems. As eDNA datasets grow, interoperability and accessibility are crucial. OBIS Australia (OBIS-AU), Australia’s node of the United Nations Educational, Scientific and Cultural Organization (UNESCO) International Oceanographic Data and Information Exchange (IODE) Ocean Biodiversity Information System (OBIS), hosted by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) National Collections and Marine Infrastructure (NCMI), promotes use of the DNA Derived Data Extension in Darwin Core (DwC) to standardise publication of eDNA and metabarcoding data Wieczorek et al. (2012), Abarenkov et al. (2023). OBIS-AU has published over 26 datasets with 21 million eDNA records to the OBIS.OBIS-AU is developing a scalable and interoperable eDNA data publishing pipeline that integrates tools such as the Findable, Accessible, Interoperable, Reusable eDNA (FAIRe) suite and the Global Biodiversity Information Facility (GBIF) Metabarcoding Data Toolkit (MDT) to transform diverse eDNA source data into Darwin Core Archives (DwC-A) for publication to OBIS, GBIF, and Atlas of Living Australia (ALA). By leveraging metadata standards including DwC, DNA Derived Data Extension, Minimum Information about any (X) Sequence (MIxS) and the FAIRe metadata checklist, the pipeline enables standardised, FAIR-compliant data publishing Takahashi et al. (2025), Meyer et al. (2023). It supports multiple transformation pathways, ensuring that eDNA datasets are consistent, reusable, and aligned with global biodiversity data infrastructures.The FAIR eDNA initiative enhances the FAIRness of eDNA data by extending standards like DwC and MIxS with eDNA-tailored metadata terms Takahashi et al. (2025). The FAIRe tools (FAIRe-ator, FAIRe-fier, and FAIRe2MDT) facilitate creation, validation, and conversion of standardised metadata to improve interoperability and reusability across platforms. The Metabarcoding Data Toolkit (MDT) is an open-source web tool that streamlines publishing of eDNA metabarcoding data by converting common data structures [e.g., Operational Taxonomic Unit (OTU) tables, taxonomy, metadata, Format for All Sequences from All Species (FASTA) files] into DwC-A GBIF Secretariat (2024).This modular design allows the pipeline to accommodate diverse data types and processing workflows while ensuring compatibility with global biodiversity data standards. The pipeline as shown in Fig. 1 provides four main pathways for converting source data into DwC-A and publishing them via the Integrated Publishing Toolkit (IPT): Directly publishing data already formatted as DwC-A; Transforming source data using a simple DwC pipeline with custom DwC transformation script; Generating DwC-A file from source data via the MDT tool; and Using FAIRe tools and converting it to DwC-A via the MDT tool or a custom transformation script. The Australian Microbiome (AM) Initiative is a national collaborative research program characterising microbial diversity across Australia’s terrestrial, freshwater, coastal, and marine environments. The AM data pipeline depicted in Fig. 2 transforms data stored in the AM Data Portal using the MDT tool to generate DwC-A for publication to global repositories.The Globalising Marine Biodiversity Observations (GLOMBO) Partnership is a collaboration between CSIRO and the Minderoo Foundation aimed at improving the large-scale monitoring of Australia’s vast marine ecosystems by deploying automated eDNA sampling systems to gather samples continuously during voyages, with the first installation taking place on CSIRO’s research vessel Investigator. This approach will be expanded through a network of “ships of opportunity,” encompassing research, commercial, and tourist vessels that contribute to a nationwide eDNA monitoring effort. The scalable data pipeline is proposed to automate, integrate, and disseminate eDNA datasets, enabling comprehensive, real-time insights into marine biodiversity across Australia’s oceans as illustrated in Fig. 3.Recording eDNA-derived species occurrences presents several challenges. One example is taxonomic ambiguity, often caused by incomplete reference databases like GenBank, World Register of Marine Species (WoRMS), Barcode of Life Data (BOLD), or SILVA. Linking eDNA sequence reads to biodiversity occurrence records is complex and requires expert knowledge and infrastructure integrating sequence data, metadata, and taxonomy. Technical barriers, limited engagement, and lack of incentives to make data accessible, hinder open access to eDNA data. OBIS-AU is addressing these challenges by exploring tools like GBIF’s MDT, OBIS’s Pacific Islands Marine bioinvasions Alert Network (PacMAN) pipeline, FAIRe tools, and AI-based tools as well as expert support and developing new automation pipelines to assist data publishers. OBIS-AU has published eDNA data using DwC Occurrence Core and DNA Derived Data Extension and is now testing a publication model with the DwC Event Core to better capture sampling context and improve integration, interoperability, and reuse of complex eDNA datasets. OBIS-AU intends to align with the new DwC Data Package to support modular publishing of marine biodiversity data.

HTML

XML

PDF

DOI: 10.3897/biss.9.183023

Authors: Susanna Ioni, Kryštof Chytrý, Kathi Schleidt, Stefan Jetschny, Heimo Rainer

Abstract: European habitats are classified under a framework developed by the European Topic Centre for Biodiversity for the European Environment Agency, as part of the European Nature Information System (EUNIS) (Davies et al. 2004). All terrestrial, freshwater, and marine habitats follow a hierarchical classification based on physical features, human influence, and dominant vegetation (Moss 2008, Chytrý et al. 2020). Distribution maps are provided and modelled using occurrence data of indicator species collected from vegetation surveys (Hennekens 2017).Although the system may seem accurate, when we first plotted the distribution of the main species of our habitat study case, EUNIS Habitat S22 ‘Alpine and subalpine ericoid heath’ (European Environment Agency 2019), we observed that occurrence data, e.g., from sources like the Global Biodiversity Information Facility (GBIF), often fell outside the mapped areas of the habitat. Furthermore, important occurrence data sources, such as herbaria, were left out of the official distribution mapping, representing, in our view, a significant shortcoming of the EUNIS system. This study addresses these gaps by integrating diverse sources of in situ occurrence data (herbaria, vegetation surveys, citizen science) through a machine learning approach to complement the current EUNIS mapping.Specifically, we modelled the distributions of diagnostic species of the Habitat S22, using species distribution models (SDMs). For this purpose, we retrieved occurrence data from GBIF, identified by the accepted names as well as taxonomic synonyms, using the R package rgbif (Chamberlain et al. 2025), and utilised the Darwin Core (Wieczorek et al. 2012) standard.Data were filtered to include European occurrences with spatial coordinates and uncertainty of <500 m, and only spring and summer months of 1980–2024. For modelling itself, they were stratified into a 1-km grid. As SDM predictors, we used proxies for macroclimate and topography. Climatic predictors included CHELSA Bioclim variables of mean annual temperature, temperature seasonality, annual precipitation, precipitation seasonality, and an aridity index (Zomer et al. 2022). For topography, we used the digital terrain model, Copernicus, and calculated slope and indices for heat load (McCune and Keon 2002), topographical ruggedness (Riley et al. 1999), and topographical wetness (Beven and Kirkby 1979), using the spatialEco R package (Evans and Murphy 2021) and SAGA GIS (Conrad et al. 2015). Data were integrated into data cubes, and correlations among species occurrences and predictors were tested. We supplemented the occurrence data with pseudo-absences sampled within a buffer around presence points (Fallgatter et al. 2025). We fitted ensemble SDMs weighted by true-skill statistics scores based on independent cross-validation.We modelled two spatial resolutions in two regions:continental Europe at 1-km resolution, andthe European Alps at 100-m resolution.Predicted species distributions were aggregated into cumulative distribution maps. Those were further validated by overlapping them with the distribution of the habitat based on vegetation plots classified by an expert system as provided by the European Vegetation Archive (EVA) plots at 1-km resolution. Predictions were also compared with the official EUNIS probability map for Habitat S22.Correlation analyses confirmed the ecological features of the Habitat S22 indicated by the EUNIS classification. Our modelled ranges largely overlapped with the distribution of EVA plots and the EUNIS probability map, but also revealed mismatches at lower elevations and in the Scandinavian region. These differences decreased when fewer species were combined in cumulative predictions.Our findings show that SDMs based on occurrence data from different sources can validate and refine expert-defined habitat maps, offering a complementary and data-driven approach.

HTML

XML

PDF

DOI: 10.3897/biss.9.182910

Authors: Ishaipiriyan Karunakularatnam, Nicky Nicolson, Priscila Reis, Magda Upton

Abstract: The Kew Herbarium houses over seven million specimens and represents one of the world's most significant botanical collections. Recent large-scale digitisation has made over 6.4 million specimens (Royal Botanic Gardens, Kew 2025) globally accessible, supporting active research into biodiversity and ecology. However, this progress has primarily been focused on herbarium sheets, leaving many specialist collections under-digitised and difficult to access. These specialist collections, stored in folders of specimen sheets, are often accompanied by printed catalogues containing the key metadata. Digitisation of these collections is difficult to process using standard herbarium workflows (which are scaled efficiently across millions of specimen sheets) due to their highly varying structure and format. This diversity makes automation more challenging so transcription is often manual, which is slow, expensive, and prone to errors. As main herbarium digitisation nears completion, adaptable, cost-effective methods are needed for specialist catalogues.This project has developed a modular, semi-automated pipeline to extract structured data from scanned pages of botanical catalogues (images/PDFs) using the Lightfoot catalogue (Bowden 1989) for development and testing. This pipeline allows user configuration across different sub-collections and extraction models. It uses Optical Character Recognition (OCR) and Large Language Models (LLMs) based transcription, allowing adaptation to different catalogue formats through prompt changes, without the need for costly custom pipelines. It consists of four stages: image processing; text extraction; data structuring; and transcription.Provided with the scanned pages, image processing (with DeepDoctection (Meyer 2024)) is performed to identify and crop regions of interest (ROI) which are normalised for text extraction. Text extraction is performed on each ROI, and the extracted text is saved sequentially. Tesseract OCR (Smith 2007) is used by default due to its adaptability and strong community support, but support for LLMs with visual reasoning, such as Qwen2.5 (Yang et al. 2024), is also implemented. Before transcription, the extracted data is cleaned and chunked with respect to family names and species names with the help of TaxoNERD (Le Guillarme and Thuiller 2022). During transcription, species-level text chunks are passed to a LLM with user-defined instructions (based on catalogue) and the pre-defined structure of output. The output from this previous step is saved in both JSON and CSV formats.Evaluation on the Lightfoot catalogue yielded F1-scores of 86%, 87% and 91% for family names, species names and folder descriptions, respectively, indicating strong extraction performance. Additionally folder descriptions had Word Error Rate (WER) and Character Error Rate (CER) of 24% and 18% reflecting reasonable baseline accuracy given the variable quality of the text.Despite this, key challenges remain: OCR accuracy drops on degraded or low-resolution scans, and LLM hallucinations can misclassify species names or description fields. These issues highlight the need for a quality control layer, such as curator validation, and refining prompt strategies to reduce errors. Nevertheless, this pipeline demonstrates that significant efficiency gains can be achieved while reducing reliance on manual data entry.This project provides an operational workflow for the digitisation of herbarium catalogue data using OCR and LLMs. Future work will focus on improving OCR accuracy, refining LLM prompts and chunking strategy. Additional testing will be performed on the Hanbury and other specialist catalogues to further validate the pipeline’s performance on varying formats. Ultimately, this approach contributes to the broader goal of accelerating digitisation across herbaria, ensuring valuable historical records are preserved and made accessible for future research.

HTML

XML

PDF

DOI: 10.3897/biss.9.183162

Authors: Patricia Mergen, Elie Saliba, Thomas Pape, Richard Pyle

Abstract: Taxonomy sensu lato is lately regaining some of its previous prominence in many biodiversity information-related activities and projects, notably for the needed taxonomic backbone in the EU Research Intrastructure Distributed System of Scientific Collections (DiSSCo) (Woodburn et al. 2022), or in training programs such as Transforming European Taxonomy through Training, Research, and Innovations (Tettris). However, the increased visibility has yet to happen for α-taxonomy, which is focused on performing revisions and describing and naming new species. The so-called taxonomic impediment (Engel et al. 2021) remains acute, with fewer and fewer experts available for a wide range of taxa, and a progressive erosion of the specialized expertise needed to document biodiversity.When describing a new species or revising a taxon, respecting the naming conventions governed by the official nomenclatural Codes is essential. There are five Codes currently aknowledged, such as the ICN — International Code of Nomenclature for algae, fungi, and plant, the ICZN — International Code of Zoological Nomenclature, ICNP — International Code of Nomenclature of Prokaryotes, ICNCP — International Code of Nomenclature for Cultivated Plants and ICTV Code — International Committee on Taxonomy of Viruses. However, the complexity of the Codes and the long learning curve can act as a barrier, discouraging early career scientists from engaging in α-taxonomy.Modernising or introducing changes to the existing Codes also involves very strict and relatively lengthy procedures, making it difficult to keep up with technological changes. To overcome these issues, and find common ground, Biodiversity Information Standards (TDWG) (with the support of the TDWG Partnerships and Fundraising Sub-Committee) and the ICZN (International Commission on Zoological Nomenclature) signed a Memorandum of Understanding (MoU). The document is available on the TDWG website*1. The ultimate goal of this MoU is to incorporate correct terminology for zoological nomenclature into TDWG standards with appropriate controlled vocabularies and definitions. It also intends to explore how nomenclature, alongside taxonomy, can be incorporated into biodiversity informatics research projects to attract a new generation of α-taxonomists, and reconnect with the roots of TDWG. This will involve developing tools andapplications and channel the use of software and AI to support new species descriptions and revisions that comply with the Codes of nomenclature. It should also involve organising training sessions dedicated to nomenclatural concepts and encouraging the development of enriched taxonomic publications. Another priority is to strengthen mentoring networks so that early-career researchers can learn directly from experienced taxonomists, ensuring that knowledge of best practices is passed on. This expended to all nomenclatural codes beyond the zoological example discussed here.A coordinated effort between nomenclatural authorities and biodiversity informatics specialists is essential to safeguard and advance online taxonomy. By combining robust standards, innovative technologies, and targeted capacity-building, the community can ensure accurate and efficient species documentation. These actions will hopefully inspirenew generations to explore, describe, and protect world's diversity.

HTML

XML

PDF

DOI: 10.3897/biss.9.183246

Authors: Quentin Groom, Tim Adriaens, Amy Davis, Peter Desmet, Damiano Oldoni, Diederik Strubbe, Lien Reyserhove, Sonia Vanderhoeven

Abstract: Research exists to make an impact. It is not necessarily aimed at profit, but seeks to meet societal, environmental and/or cultural needs, while also generating and sharing new knowledge. Sometimes this knowledge has a direct application; other times, it serves curiosity. This blend gives research its vitality.Yet research is typically packaged into short-lived projects. They begin with a question or funding call and end when resources run out. The tension is clear: the specific questions may be time-bound, but the value of the knowledge, data and tools is not. Too often, outputs lose visibility and become hard to reuse. There is also a tension between internal project coherence and broader alignment with institutional strategies, collaborators and policy frameworks. Trade-offs are inevitable between speed and sustainability, innovation and standardisation, and serving current needs versus preparing for future ones. This interplay between the ephemeral and the enduring is central to the sustainability and value of research.Our reflections draw on several recent projects. The Belgian TrIAS project (2017–2021) built open data pipelines for science and policy and designed its workflows with reuse in mind. These were not one-off solutions but adaptable foundations. Later projects such as B-Cubed (2023–2026), LIFE RIPARIAS (2021–2026), GuardIAS (2025–2027) and OneSTOP (2025–2028) build on this base (Katsanevakis 2024, Groom 2025). While their objectives differ, all shared a concern for durability. They reuse methods, extend infrastructures, and align with open standards. Durability is never accidental. It requires researchers to consider sustainability from the outset. Datasets are published under CC0, software under open-source licenses, and publications under CC-BY.To endure, research outputs must be embedded in open, stable and widely used infrastructures, such as GBIF, Zenodo, GitHub, Wikidata. These platforms increase visibility and allow others to build upon prior work. Standards are equally essential. The frameworks developed by Biodiversity Information Standards (TDWG) and others enable interoperability, so data from different sources can be combined, compared and reused (Hardisty and Roberts 2013).We’ve learned to design workflows for reuse. For example, the WiSDM species distribution modelling workflow developed in TrIAS has been picked up in OneSTOP and GuardIAS (Davis et al. 2024). This reuse saves effort but also builds continuity as each project leaves a trail others can follow.Furthermore, when tools are co-developed, they are more likely to be maintained and adapted. Communities of practice allow knowledge and lessons to flow from one project into the next. Some of the most lasting outcomes of research are the relationships it creates. Durability requires deliberate choices. Openness, interoperability and relevance ensure that research lives beyond its funding cycle. Aligning with standards, designing for reuse, and investing in shared infrastructure all help secure this. Communication is part of this, including clear documentation, identifiers, and presentation or contextualization that make outputs understandable and trustworthy.The real lesson is that long-term impact doesn’t come from isolated brilliance, but from embedded, reusable, shareable work. Funders, institutions and researchers all have a role to play in making this shift, and if we succeed, even short projects can leave lasting legacies.

HTML

XML

PDF

DOI: 10.3897/biss.9.183199

Authors: Arturo Ariño, Piluca Álvarez Fidalgo, Ana Amezcua, Berta Caballero-López, Francisco Cabrero-Sañudo, Angel Chaves, Marina González Cristóbal, Sandra Grzechnik, María Imas, Irene Lobato-Vila, Sergio Montagud, Mercedes París, Adrián Sánchez Albert, Manuel Sánchez Ruiz, Celia Santos, Robert Wilson, David Galicia

Abstract: The exactness, fitness-for-purpose (FFP) and reliability of primary biodiversity data can be enhanced by additional data beyond the basic taxon-location-date triad (Hill et al. 2010). Often, the only available data are the labels in legacy specimen collections. The digitization process is most efficient if all available information can be collected at once in a single event of specimen handling, rather than in separate phases. There is, however, a compromise between producing a faster catalogue for immediate use and housekeeping, and an accurate, wider-FFP database where all data have been thoroughly checked.Recognizing potential sources of error at digitization time may help making choices. During a dataset integration procedure, the quality and reliability of the data capture was analyzed. The dataset consisted of transcribed label data of over 58K pinned insects of agriculturally-relevant groups in XXth-century collections at six institutions in Spain, that resulted in almost 6000 collector strings. But collector names could beunidentified;misread;ambiguous,duplicated under variants; ormisplaced or misattributed to/from another entity, e.g. a location.This resulted in a high entropy level where one collector could be databased in multiple ways, artificially inflating the corresponding catalogues. The entropy was much higher in collections where collectors contributed few specimens, which is the case for university-based collections.By using simple indexing and cross-referencing techniques, the roster of names was significantly reduced, but full disambiguation of collectors required mining ancillary sources and consulting with people with long-standing knowledge of the collections. Overall, 42% of collector names were in error, resulting in excess entropy. METHODSSpecimen data were collated from the TETTRIS INC-STEP Project of Spanish pollinators complemented by some agriculturally-relevant groups in six collections deposited at five academic and research institutions in Madrid, Barcelona, Valencia and Pamplona (see Suppl. material 1 for full details). MCNB, MNCN, MUVHN were chiefly historical and created by researchers over long careers, while MZNA-R, MZNA-Z, UCME came mainly from academic coursework activities.Our procedures agreed with the overall strategy of Groom et al. (2022), while devising a specific workflow. Disambiguation and attribution included a number of steps (Suppl. material 1). First, names were normalized to facilitate grouping and sorting (7 steps). Then, ambiguous names were (whenever possible) attributed to actual persons by internal checks (clustering of names, matching localities and dates), consultation with external references (e.g. student lists), or consultation with collection curators (5 steps). Names were finally given a four-level identity qualification resulting from the disambiguation exercise (see Suppl. material 1). For further analyses, “very low” and “low” levels were considered a poor attribution, while “medium” and “full” levels were deemed good attribution.RESULTSThe disambiguation exercise reduced collector names from 5939 to 3473 (a 41.5% decrease). The total entropy, measured as Shannon’s H’, decreased by 8% while Simpson’s dominance D increased by 29%, as expected (merging names created larger collections for some collectors). However, there were differences among collections. The three academic collections had both higher diversity and higher diversity reduction than the historical collections (Fig. 1). Historical collections tended to be much more concentrated (30 specimens per collector, 25% single-specimen collectors) than the academic collections (8 and 61%, respectively).These differences can be tracked to how the collections were formed. Historical collections tended to be well documented and created by researchers spanning longer careers, while academic collections were often linked to works created during coursework. Fig. 2 shows the career spans of collectors. The abrupt end of collection at the turn of the century for academic collections can be linked to the introduction of restrictive legislation in Spain (a mandatory reduction of course loads, and unaffordable permission requirements). Full disambiguation often requires manual, time-consuming verifications against a variety of external sources. Automated disambiguation procedures elevated good identification from an initial 21% to 47% in MZNA-R, but it was not possible to use manual checking against sources for this dataset. In contrast, the similar MZNA-Z collection could be checked against coursework lists, which resulted in a 80% good attribution rate (Fig. 3). Overall, 57% of the collector names across collections had a good attribution.CONCLUSIONDoing disambiguation exercises by people with contemporary knowledge of the collections appears to be critical to avoid losing much information from the collections. Curators have an invaluable knowledge of the collections, could locate relevant documents about them, and are the ones able to reduce collection entropy by at least 32%. Their early retirement would negate a significant way to disambiguate names that left little or no trace in the literature record.

HTML

XML

PDF

DOI: 10.3897/biss.9.181958

Authors: Emily Symington, James Richardson, Peter Moonlight

Abstract: Cacao (Theobroma cacao L.) is a globally important crop, forming the basis of the multi-billion-dollar chocolate industry and supporting livelihoods in tropical regions worldwide (Houston and Wyer 2012). Cacao is highly sensitive to drought stress, a significant threat to plant growth and survival, pod yield, and bean quality (Mensah et al. 2023). Periods of water deficit can cause yield losses up to 50%, likely worsening with climate change-induced increases in drought frequency and intensity (Gateau-Rey et al. 2018).Theobroma cacao is a member of Theobromateae, a tribe in the Malvaceae family with 40 species in three genera known collectively as cacao wild relatives (CWRs). It is likely that some CWRs are adapted to drought and may provide valuable genetic material suitable for “future proofing” chocolate production against climate change (Medina and Laliberte 2017). While there have been low success rates for cross-breeding Theobroma species, recent advances in plant molecular biology are making the use of CWRs in crop breeding programs much more viable (Bohra et al. 2022).This project has two aims:Develop a database of occurrence records of all CWRs and use this to produce Species Distribution Models (SDMs). The SDMs will provide data on the climatic tolerance of each CWR and indicate which of them are likely sources of drought tolerance genes/traits.Assessing the conservation status of CWRs is crucial in order to protect this natural bank of potentially useful genetic resources. We will predict the CWR range changes under various climate and land-use scenarios to determine their conservation status.Distribution data have been collated from taxonomic treatments (Cuatrecasas 1964, Schultes 1958, Colli-Silva in prep.), large and globally important herbaria (e.g., Herbaria Nacional Colombiano (COL), The New York Botanical Garden (NY)) and Latin American herbaria (e.g., Instituto Nacional de Pesquisas Amazonia (INPA), Pontificia Universidad Católica del Ecuador (QCA), abbreviations follow Thiers (2001)).Specimens will be taxonomically verified with reference to the taxonomic literature, georeferenced to within 5 km (Hackeloeer et al. 2014) and cleaned through automatic (Zizka et al. 2019) and manual plotting in geographic space to check for outliers. Spatial biases will be minimized using spatial filtering (Kramer‐Schadt et al. 2013).We will use three SDM methodologies, all fit with data from freely available global datasets (Karger et al. 2017):MaxEnt, the most commonly used SDM methodology;one of the recently developed trait-based SDMs designed to incorporate prior knowledge about trade-offs and plasticity in functional traits, minimising the risk of model over-prediction (Benito Garzón et al. 2019);a novel, modified SDM methodology that accounts for CO2 fertilisation in future predictions (Prentice et al. 2022).These SDMs will estimate Predicted Niche Occupancy profiles (Heibl and Calenge 2018) that willcharacterise drought resistance of CWRs,test for associations with species’ genes/traits.To assess the conservation status of these CWRs, SDMs will be projected into future climate and emission scenarios, represented by the Intergovernmental Panel on Climate Change's Shared Socioeconomic Pathways, mediated by models of future land-use change. Species will be characterised by the overall change in their projected range sizes and the loss of their current range sizes. The results will be used to assess the conservation status of CWRs under International Union for Conservation of Nature criteria (IUCN 2012).

HTML

XML

PDF

DOI: 10.3897/biss.9.182514

Authors: Salza Palpurina, Kate Ingenloff

Abstract: Vegetation plots (relevés)—records of all plant species with their abundance in fixed-size plots in terrestrial vegetation—might be independent, arranged in nested designs, or established as permanent plots for resurvey. Given their complexity, representing vegetation plot data in global infrastructures such as the Global Biodiversity Information Facility (GBIF) requires thoughtful use of Darwin Core (DwC) and extensions (Darwin Core Task Group 2009). In DwC, a relevé corresponds to a single dwc:Event, hence, the recommended way to publish such data in GBIF is using the Event Core with at least two additional extensions: Occurrence and Relevé (GBIF 2018). The GBIF Relevé extension provides standard measurements across vegetation layers, with explicitly defined units and precision (e.g., “tree cover (%)”). Other plot-level measurements (e.g., soil pH) can be described via the DwC Measurement or Facts extension. The recently ratified Humboldt extension (TDWG Humboldt Extension Task Group 2024) expands the DwC Event class to capture the sampling context of more complex plot designs by extending the descriptions of sampling events, making it especially relevant for nested and repeated sampling. However, recommendations are still evolving (Ingenloff et al. 2025) and it has not been extensively tested with vegetation plot data (Suppl. material 2). Out of the 374 databases listed in the Global Index of Vegetation‐Plot Databases (GIVD; accessed on 15.09.2025), only a few are currently published via GBIF (e.g., Hennekens 2018, Kuzemko et al. 2024, Swacha et al. 2025), and mostly as Occurrence class datasets.As a case study, we present the dataset by Palpurina (2025) exported from Turboveg 2 (Hennekens and Schaminée 2001), the standard software for vegetation data management in Europe. For this dataset, we used an R script*1 to directly access the database and export the data as Darwin Core Archive (DwC-A) using an Event Core with Occurrence, Relevé and Humboldt extensions. Regarding the Humboldt extension, field mapping required one-to-many and many-to-one relationships between Turboveg and DwC. Because some Turboveg 2 fields lacked sufficient detail, the script had to include hardcoded values and manual enrichment steps to populate missing DwC fields and ensure schema compliance. Our talk (Suppl. material 1) addresses challenges in applying the Humboldt extension to vegetation-plot data, focusing on common practices for capturing key data critical for vegetation data interoperability.

HTML

XML

PDF

DOI: 10.3897/biss.9.181272

Authors: Filipi Soares, Luís Ferreira Pires, Luiz Santos, Debora Drucker, Dilvan Moreira, Fernando Elias Corrêa, Kelly Braghetto, Alexandre Delbem, Antonio Saraiva

Abstract: Modeling species names in biodiversity ontologies is particularly difficult in multilingual contexts, where semantic conflation often occurs. A good example is the common name "pimenta." In Brazilian Portuguese, experts usually refer to Capsicum spp. (chili peppers), while its direct translation “pepper” in English often denotes Piper nigrum (black pepper) (Soares et al. 2025a). In Brazilian markets, however, Piper nigrum is more accurately associated with “pimenta-do-reino" (“pimenta-negra”). This issue was observed on Wikipedia, when translating the Portuguese page for “pimenta” into English, the entry switches from Capsicum spp. to black pepper (Piper nigrum), showing how easily semantic drift can appear in multilingual data modeling. The correct association between common names used in the agricultural market with species would be a way to avoid the misunderstanding of these cultural differences. However, another challenge in vocabulary management emerges, which is how to manage species names in ontologies to keep them updated as the taxonomy itself updates. Some agriculturally controlled vocabularies, such as Agrotermos (Telles et al. 2024) lack automated mechanisms for updating taxonomic classifications. For example, Prochilodus cearensis, Prochilodus scrofa, and Prochilodus margravii are all listed in Agrotermos as preferred terms, i.e., the authorized, standard term selected to represent a concept in a controlled vocabulary, while according to the Global Biodiversity Information Facility (GBIF) Backbone Taxonomy (GBIF Secretariat 2023) these names are synonyms, as shown in Table 1. When developing the Agricultural Product Types Ontology (APTO), which was designed to represent products traded in Brazilian agricultural markets based on Agrotermos and AGROVOC, we proposed two approaches using generative AI, specifically OpenAI's ChatGPT-4, as a semantic engineering assistant to automate the inclusion of scientific names in the ontology:Prompt-based queries with a plugin accessing the GBIF APIA ChatGPT-generated Python script that converted GBIF taxonomy data into Web Ontology Language (OWL) formatThese AI-supported methods automated the construction of APTO’s “Organism” module, integrating taxonomic hierarchies and managing synonyms. ChatGPT effectively identified synonymy (e.g., see Table 1) and reduced manual labor in ontology development. The first approach is no longer reproducible since OpenAI has replaced plugins by GPTs. As such, we are currently developing a GPT named Taxonomy OWLizer 2.0*1, which is an evolution of the first approach described in that paper. Concerns about scalability, reproducibility, and hallucinations (false, made-up information) remain, highlighting the need for expert oversight throughout the process. When ChatGPT was used without API access, hallucinations appeared more frequently. For instance, when asked to check a list of plant species names for typos, it incorrectly suggested that Euterpe edulis was a synonym of Euterpe oleracea, even though both are recognized as distinct species in widely used catalogues such as the GBIF Backbone Taxonomy (Soares et al. 2025a).This case study demonstrates that generative AI can support but not yet replace human-led ontology development. It also emphasizes AI’s potential contribution to biodiversity informatics, particularly for managing evolving and multilingual vocabularies. All tools and source code related to our work are archived on Zenodo (Soares et al. 2025b). Detailed protocols are provided in Soares et al. 2025a.

HTML

XML

PDF

DOI: 10.3897/biss.9.182246

Authors: Esteban Marentes Herrera, Mario Mora Cardona, Jennifer C. Girón Duque

Abstract: Colombia is a mega-biodiverse country, however the number of known species present in the country remains underestimated due to a lack of information, especially for understudied groups that are difficult to sample and identify, such as beetles (Coleoptera). It is not possible to obtain a precise number of species for this group using conventional methods, due to their high diversity and lack of resources for taxonomic research. For this reason, a modelling exercise was conducted using the Cross Industry Standard Process for Data Mining (CRISP-DM) (Chapman et al. 1999), with occurrence records published through SiB Colombia (Sistema de información sobre biodiversidad de Colombia) and GBIF (Global Biodiversity Information Facility). These records were labeled using checklists of various families and subfamilies compiled by experts of Coleoptera de Colombia, and enhanced by relevant abiotic measurements available in open repositories, which were identified as having an impact on the distribution of beetles. The best model was used to indirectly predict the number of beetle species in the Antioquia department, which has a large amount of available data and a variety of representative habitats.Occurrence data for the order Coleoptera were downloaded directly from GBIF as of March 2025 (GBIF 2025a), using a global filter for the order Coleoptera. This dataset contains 32,356,574 rows, which include information on taxonomy, locality, coordinates, date, publishing entity, collector and type of record. Occurrence records of plant species associated with beetles were obtained from GBIF by downloading all biological records of the kingdom Plantae for Colombia (GBIF 2025b) and were spatially intersected with the Coleoptera data using a buffer of 1000 m. The abiotic variables selected included wind speed, relative humidity, precipitation, solar radiation, temperature, ecosystems/habitats, elevation above sea level and their relevant information can be found at Herrera (2025).Finally these data were labeled with the number of species per family of Coleoptera in Colombia compiled from available checklists published in GBIF via SiB Colombia by the research group Coleoptera de Colombia*1 (Coleoptera de Colombia 2025). This data was used to annotate the final dataset and served as the dependent variable used in the regression models.Four commonly used regression techniques were selected (linear regression, multilayer perceptron, deep neural networks and Random Forest) after a literature review of the most common methods used for species prediction Herrera (2025). These were built in Python, fine-tuning them for optimal performance. Data was split into training and testing sets to ensure validity and the best model was selected based on standard accuracy metrics and confirmed using cross-validation.An escalated deep neural network was selected after getting the best values in the three metrics and was used to make predictions for the number of species per family in Antioquia and Colombia. The script with all the details is available in Suppl. material 1.For Antioquia, the model predicted a minimum of 2,007, a maximum of 9,381, and an average value of 4,210 species of beetles. Using the same model for Colombia, there was a minimum of 3,226, a maximum of 14,991 and an average value of 6,420 species predicted. These predictions were compared to the last published checklist, which listed 6,170 species recorded from Colombia and was consistent with the number of species described from the country since 1945. These predictions were shared with expert coleopterists, who qualitatively assessed the results.However, depending on the particular beetle family, experts thought the models may have overestimated or underestimated predicted values, highlighting the general uncertainty in species estimates and how artificial intelligence (AI) models are also unreliable in predicting a clear value, serving only as a starting point or providing a wide range to work with.

HTML

XML

PDF

DOI: 10.3897/biss.9.182085

Authors: Giulia Micai, Astrid Hille, Heimo Rainer

Abstract: Eduard Hackel (1850–1926) was an Austrian botanist best known for his pioneering work on grasses (Poaceae), becoming one of the most famous and respected agrostologists of his time.In 2024, the botanical department of the Natural History Museum of Vienna received as a donation a collection of more than 700 letters sent to Eduard Hackel between 1870 and 1932 by botanists writing from 40 different countries, in 6 different languages. In botany, supplementary materials, such as the personal correspondence of scientists, field notes or diaries, hold a great potential, since they might contain the only documentation of a scientist’s thought processes, ideas and observations, filling in gaps on missing or poorly documented specimens and ideas. They may record information that is otherwise hard to deduce from the specimens alone, which can be vital for scientific research (Rinaldo et al. 2013). In order to make the information contained in these types of materials findable and searchable by anyone and from anywhere, it needs to be digitised and transcribed. Transcription projects are time consuming, intellectually intensive, and expensive for an organisation to set up and manage. Crowdsourcing is a sustainable strategy for transcribing large collections and enhancing descriptive metadata. Despite the increasingly prominent role of artificial intelligence, crowdsourcing is still an indispensable part of the future of libraries, museums, and archives because it solves problems, strengthens the sense of community between the users and the public, adds value to the collection, and creates engagement.The value of scientists’ private correspondence as an historical source has been recognised for a long time already, but recently the concept of "correspondence networks," which connect several people into webs of exchange, became distinct objects of historical studies themselves. (Ogilvie 2016). The transcription process started with the use of the AI-powered platform Transkribus, which enables the transcription of handwritten text into machine-readable and searchable texts.The challenges we faced in the transcription are intrinsic to the nature of our project: dealing with old, obsolete hand-written scripts (e.g., old Kurrent in German); letters written by different senders; and letters including different languages.In order to overcome these challenges, we had to use different transcription models for the AI-driven transcription, and rely on various volunteers with different linguistic backgrounds and skills.After the transcription, the letters were uploaded to Goobi Workflow, which is a tool for the management and display odigitisation projects. Transcribing the letters made it possible to extract relevant data from the texts about botanical specimens and their collectors and allowed the linkage of biographical data on Wikidata and Bionomia with specimen data on JACQ and GBIF (Global Biodiversity Information Facility). With the use of Transkribus and the help of our volunteers, we were able to transcribe 864 pages and identify 140 senders. Each one of the botanists was linked to their Q-number, which allows the linkage to their biographical data on Wikidata and Bionomia. In various letters, the sender included a botanical specimen and requested help from E. Hackel for its identification. In various cases, thanks to the information contained in the text, we have been able to identify the specimen within E. Hackel’s personal herbarium, deposited at the Natural History Museum of Vienna. Among the senders, we identified 17 Italian scientists. One of the publicly accessible outcomes of the investigation of Hackel’s exchange with the 17 Italians is the online exhibition on the Europeana platform in an English*1 and Italian*2 version. A second result was the curated story of E. Hackel and his 17 Italian correspondents by the Transforming European Taxonomy through Training, Research and Innovations (TETTRI) project in the form of a curated website*3, which links their biographical data from Wikipedia, Wikidata, and Bionomia, and visuals from Wikimedia. Furthermore, the letters are publicly available for consultation together with their transcription*4 via Trankribus*5 sites. This contribution will provide an insight into the possibilities of extracting data from natural history collections and interlinking them to implement their usability and accessibility for research institutions and the general public.

HTML

XML

PDF

DOI: 10.3897/biss.9.182213

Authors: A. Ashwin, Ram Dayal Vaishnav, Lavanya Bhat, Farida Tampal

Abstract: Rapid urban expansion in Hyderabad continues to present significant challenges to sustaining urban biodiversity, particularly the intricate fauna-flora relationships underpinning city ecosystem health. To address these challenges, the Hyderabad Annual Tree Biodiversity Survey—a collaborative citizen science program led by WWF-India’s Hyderabad office—was conducted for the second consecutive year in 2025, building on the model and findings of the 2024 survey.The 2025 survey mobilized over 150 volunteers across 32 teams and 18 locations encompassing all six city zones. Volunteers underwent a structured two-part training series, covering survey protocols, biodiversity identification, and mobile app usage (iNaturalist, Merlin Bird ID, eBird), resulting in improved data collection quality and participant engagement. Tree biodiversity, and the faunal biodiversity found exclusively on the trees, were surveyed extensively for avenue and park trees with diameter at breast height (DBH) greater than 10 cm, employing community-led observations, photographic documentation, and digital citizen science platforms.A total of 837 unique trees were documented, representing 108 species, along with 1,885 faunal observations—primarily ants/termites, birds, and spiders. Notably, avenue trees (trees beside broad city or town roads) comprised over 82% of the sample and supported 93% of the observed animals (Fig. 1), highlighting their immense contribution to urban biodiversity. Flora-fauna records increased to 923 entries in 2025, reflecting deepened community involvement and broader spatial coverage compared to 2024, with 23 locations surveyed from 6 city zones and 753 unique trees recorded from 797 entries.The top three tree species covered in the 2024 survey were Peltophorum pterocarpum (Copperpod), Azadirachta indica (Neem), and Milletia pinnata (Pongamia). This result does not vastly differ, with the top three tree species covered in 2025 survey being the same (Fig. 2).Native species findings reaffirmed that trees native to India and Telangana hosted a higher diversity of animal life (Fig. 3). However, overall faunal observations were lower than last year, possibly due to sampling time, Air Quality Index (AQI) variation, and urban noise factors. More than half of surveyed sites exhibited “Poor” AQI, correlating with reduced animal abundance, and 42% of the samples corresponded to high vehicular activity zones—underscoring the dual pressures of pollution and traffic on urban ecological health.The trees were also compared on the basis of the faunal diversity hosted on them, in order to see the effect of the species, the nativeness, and the location (avenue vs non-avenue), and the top 5 trees were thus sorted on the basis of the absolute value or the total faunal abundance (Fig. 4), and the mean faunal diversity (Fig. 5).Key improvements in the data collection process, learned from previous year's survey, included enhanced digital workflows, publicly available forms for data entry, standardized morphometric measures (DBH), and use of GBIF to make the dataset accessible for wider research and replication. Persistent challenges involved data consistency, incomplete fields, difficulties with citizen sciene app registration, and some mismatch between faunal observations and tree records. Volunteer feedback stressed the need for offline data entry options, more comprehensive training, and expanded urban habitat coverage to further improve reporting accuracy and inclusiveness.The survey’s outcome demonstrates both the vulnerability and resilience of Hyderabad’s urban biodiversity. It advocates for prioritizing native species in urban greening, pollution mitigation efforts, and wider adoption of citizen science approaches. Future recommendations include protocol refinement for insect and individual counts, unbiased sampling across city habitats, and ongoing offline training and support for volunteers. The survey not only fosters a robust ecological inventory but also strengthens public stewardship essential for sustaining biodiversity in fast-growing urban landscapes.

HTML

XML

PDF

DOI: 10.3897/biss.9.181733

Authors: Giulio Martellucci, Hervé Goëau, Pierre Bonnet, Fabrice Vinatier, Alexis Joly

Abstract: Recent developments in European environmental legislation, such as Directive (EU) 2024/1203 on environmental protection through criminal law and Regulation (EU) 2023/1115 on imported deforestation, highlight the need for robust and scalable methods to inventory and monitor plant communities. In this context, vegetation quadrats, rectangular ground plots widely used in ecological surveys, are central to biodiversity monitoring strategies, including forensic applications where reliable species-level identification can provide evidence for law enforcement and compliance monitoring.Traditional quadrat-based inventories are essential for tracking habitat quality, ecosystem change, and anthropogenic impacts. However, they are labor-intensive, costly, and rely on taxonomic expertise that is not always available at scale. This creates a gap between policy demands and operational monitoring capacity. PlantCLEF 2025 (Martellucci et al. 2025) directly addresses this challenge by introducing a new benchmark task: the automatic identification of all species present in high-resolution quadrat images, framing the problem as a multi-label classification challenge.The dataset integrates several complementary components designed to advance automated plant identification in ecological contexts. It includes approximately 1.4 million close-up single-species images covering about 7,800 taxa from Southwestern Europe, providing a strong foundation for species-level recognition. In addition, around 2,100 annotated high-resolution quadrat images capture realistic vegetation plots where multiple species co-occur (Fig. 1), thus supporting the development and evaluation of multi-label models.Participants were tasked with predicting all species visible in quadrat images, with evaluation based on a sample-averaged F1 score at the quadrat level. PlantCLEF 2025 goes beyond traditional image classification in biodiversity research by introducing a task that requires models to address multiple challenges simultaneously. The limited availability of labeled quadrat images to ensure model generalization across ecosystems further imposes training the model on citizen-science images, which are typically close-up and organ-centered on single plants, thereby introducing a strong domain shift. The task is further compounded by the presence of numerous, overlapping taxa within the same image, as well as high species diversity, strong visual ambiguities, occlusions, phenological variation, and the presence of small or non-vascular plants. The 2025 challenge attracted over 500 participants and 38 research teams worldwide, submitting 659 distinct methods. Despite the difficulty, results show significant progress: the best approaches without metadata achieved a sample-averaged F1 of 0.35, compared to 0.29 in the previous edition. Most top-performing methods relied on the DINOv2 (Goëau et al. 2024) pretrained vision transformer (Goëau et al. 2024), often combined with multi-scale tiling strategies to process high-resolution inputs.Additional insights emerged from the challenge. Preprocessing test images proved critical, with performance substantially improved when compression settings and resizing algorithms, such as Lanczos interpolation, were carefully aligned across training and evaluation data.Several approaches were tested to explicitly suppress non-plant regions; while effective in stabilizing results, none were both lightweight and generalizable. Finally, leaderboard analyses revealed partcipants often faced overfitting when fine-tuning an excessive number of hyperparameters.By framing quadrat identification as a benchmarked AI challenge, PlantCLEF 2025 provides a controlled testbed for methods directly relevant to forensic biodiversity. It promotes the development of robust, scalable, and explainable models that can support digital evidence pipelines in environmental law, including the EU’s efforts against habitat destruction and imported deforestation.The open and collaborative nature of PlantCLEF, hosted on Kaggle and integrated into the CVPR-FGVC workshop and the CLEF conference, ensures reproducibility and fosters cross-disciplinary dialogue between computer scientists, ecologists, and biodiversity monitoring authorities.

HTML

XML

PDF

DOI: 10.3897/biss.9.181898

Authors: Mathias Dillen, Ben Norton, Cat Chapman, Elspeth Haston, Marie Briggs

Abstract: The MIDS standard (Minimum Information about a Digital Specimen) aims to indicate the digitisation level of natural history specimens on a simple scale. Much progress has been made over the past year to standardise this process for optimal reusability and scalability. The Simple Standard for Sharing Ontological Mappings (SSSOM) framework (Matentzoglu et al. 2022) has been adopted by the MIDS task group for this purpose and mappings are now available for Darwin Core archives and ABCD. The use of SSSOM mapping sets should facilitate the assessment of digitisation status over time, for different kinds of specimens and for different data standards and their serialisation. In this presentation, we will show the results of calculating MIDS levels for those different use cases when applied to several datasets acquired from the Global Biodiversity Information Facility (GBIF), using the MIDSCalculator and the GBIF MIDS Checker tools. We will determine the reasons why MIDS levels were (not) met, highlighting the impact of MIDS information element definitions and, in particular, their mappings to different standards and the stability of their implementation in different kinds of software. These results should show the feasibility of enabling MIDS across different infrastructures and promote implementation of the MIDS standard.To acquire an overview of the current digitisation status of specimen data, we downloaded an export of specimens from GBIF (GBIF.org 2025f) using its new SQL API service to minimize data storage requirements for unrequired data. To calculate the MIDS scores for this large dataset - 265 million records - we used an adapted script re-using the MIDSCalculator calculation scripts in batches of 10 million (omitting the RShiny interface). We subsequently summarised the results per MIDS information element and level and performed follow-up analyses, again using the GBIF SQL API, to look into the causes of some MIDS information elements often being absent in the GBIF data. Fig. 1 shows the overall results per information element. 4% failed to meet MIDS level 0 and would be considered undigitised. 69% met MIDS level 0, 25% level 1 and 2% level 2. No specimen met level 3, but this was because Darwin Core terms mapped to MediaID are only available in multimedia extensions to Darwin Core and were not available through the SQL API, so achieving this level was impossible through this method.Significant impacts were observed for several information elements: An unmapped Organization caused ca. 10 million specimens to remain at the level of "undigitized" (technically MIDS level -1) (GBIF.org 2025a). More than half of the datasets with specimens seemed to omit a last Modified date, preventing the specimens in them from reaching MIDS level 1 (GBIF.org 2025b). Object Type was even more impactful, missing from 173 million specimens or almost 2 out of every 3 specimens (GBIF.org 2025e). Collecting Number was missing for more than half of the specimens, but this one is difficult to assess properly in the absence of a reliable vocabulary for unknown, blank and non-applicable values (GBIF.org 2025g, GBIF.org 2025d, GBIF.org 2025c). This number is very important for a significant amount of specimens, as it is often used to refer to them in literature, but many specimens do not use these numbers at all.These results show that calculating MIDS even at larger scales is possible and insightful. The technical workflow used could still be optimized in several ways, including parallel computing, more efficient storage methods and more powerful software frameworks than R, and potentially be leveraged to show the evolution of MIDS values, and hence digitization, through time using past GBIF snapshots.

HTML

XML

PDF

DOI: 10.3897/biss.9.181968

Authors: Michael Elliott, Manuel Luciano, Jose Fortes

Abstract: The ability of large language models (LLMs) to interact with their environment through function-calling mechanisms has enabled the development of agentic systems - artificial intelligence (AI) systems that can:understand natural language requests,create an action plan to address the response (e.g., retrieve data from an online source),execute the action plan through the use of tools, and theninterpret the outcomes of their actions to generate natural language response.As research tools, agentic systems have the potential to transform how researchers interact with scientific data. They allow users to interact with the system through requests rather than instructions (though users may still issue instructions if desired), letting agents autonomously plan and execute the labor-intensive tasks of finding, retrieving, processing, and compiling information across one or more information sources.General-purpose agentic systems struggle to make effective use of scientific data and retrieval tools due to lack of domain expertise, and also produce information that is difficult to verify. To address these issues, we have created iChatBio, an agentic system specifically designed to a) incorporate expert biodiversity knowledge into its decision-making processes and b) provide traceability of information back to sources. This is achieved through a multi-agent design in which one general-purpose chat agent breaks down complex user requests into specific information needs which are fulfilled by expert agents, which are specialized to find or process certain types of information. iChatBio currently connects with expert agents that access data served by iDigBio*1, GBIF*2, iNaturalist, BHL*3, OBIS*4, and ALA*5, with more expert agents being added regularly. iChatBio also connects with several expert agents that provide generic data processing and visualization functions. This synergistic combination of data retrieval experts and data processing experts enables new, emergent data use cases beyond the scope of any individual data provider. This presentation discusses the following use cases:Exploration: includes finding, retrieving, summarizing, and comparing biodiversity data from diverse sourcesCuration: includes compiling existing records into new thematic datasets, fixing spelling and formatting errors, combining duplicate records, and inferring missing field valuesExtension: includes linking specimen records to related records, genetic data, literature, and more, to form extended specimen recordsiChatBio makes integrated data services more easily accessible through its LLM-powered natural language interface. For example, the use of multilingual LLMs like OpenAI’s GPT-4.1 allows users to interact with biodiversity portals and the data they serve in the user’s native language (so long as the LLM is proficient in the language). Additionally, although human-friendly data portals are often designed to support specific use cases, data providers also often host machine-friendly Application Programming Interfaces (APIs) which support many more use cases through highly configurable functions. Such APIs typically require significant effort and expertise on the user’s part to use effectively; iChatBio’s expert agents, however, allow users to make use of such APIs without any experience. Expert agents can even educate users about APIs by demonstrating how to construct API calls for particular use cases.iChatBio is an evolution of the agentic system presented in Elliott et al. 2024, improving upon the original system in both system design and breadth of integrations with biodiversity data services. Whereas the original system design was monolithic (all agents existed as part of a single software application), iChatBio is a distributed system (agents run as independent web servers) in which expert agents communicate over the Internet using the Agent2Agent (A2A) protocol (Surapaneni et al. 2025). This distributed design enables community-driven agent development, where expert agents can be independently managed and improved by community members. Moreover, A2A compliance allows expert agents developed for iChatBio to potentially participate in other agentic systems, and likewise existing A2A-supporting agents can potentially connect to iChatBio. To facilitate community contributions of expert agents, we have assembled an iChatBio Software Development Kit (iChatBio SDK), available in the Python Package Index.

HTML

XML

PDF

DOI: 10.3897/biss.9.181877

Authors: Elspeth Haston, Cat Chapman, Ben Norton, Stefan Seifert, Sam Leeflang, Wouter Addink, Mathias Dillen

Abstract: The digitisation of the world’s natural science collections is expanding massively and providing a unique global resource for answering some of the most fundamental bio- and geodiversity questions. However, digitisation at this scale can only be done in stages, increasing the variation in the level of digitisation both between and within collections. The ability to measure and monitor the level of digitisation of each individual specimen and, by extension, each collection on a national or global scale has never been more important. The Minimum Information about a Digital Specimen (MIDS) standard is being developed to provide an international digitisation standard within the Biodiversity Information Standards (TDWG) organisation. The standard, along with implementations which can calculate the MIDS level of specimens and, by extension, datasets, provide users with tools to help develop a digitisation strategy as well as plan, manage and monitor a digitisation programme, including prioritisation and data enhancement. For researchers, whilst the MIDS level of published specimens does not provide information about the quality of the data present, it does indicate the expected amount of associated data for intended analyses, including geographic coordinates and identifiers.The MIDS website provides access to the current draft of the standard. The four MIDS levels (MIDS0 to MIDS3) are described, and for each level the purpose is included. The purpose has been key to defining the information that is required to be present.The information recorded for each specimen is categorised into information elements. A detailed schema for the information elements includes the label, definition, usage note, purpose and examples as well as the disciplines for which each element is required (Biology, Geology, Palaeontology). The information elements required for each MIDS level are cumulative, with each level adding additional information relevant for the purpose of the level. Specimen data recorded in collection databases and submitted to international aggregators such as the Global Biodiversity Information Facility (GBIF) need to be mapped to the information elements to enable the calculation of the MIDS level.The website explains how the Simple Standard for Sharing Ontology Mappings (SSSOM) (Matentzoglu et al. 2022) is being used to map Darwin Core (DwC) (Wieczorek et al. 2012)and Access to Biological Collection Data (ABCD) (Access to Biological Collection Data task group 2007) terms to MIDS. It provides a tabulated quick reference of MIDS mappings with a filter option to enable users to quickly review the mapping by MIDS level or by information element.Several tools have been developed, implementing MIDS to calculate the MIDS levels of datasets, and the website provides links to these. As the MIDS standard is not yet ratified, there have been updates which are not reflected in all the tools currently available. There is a reference implementation as part of the open Digital Specimen model with the Distributed System of Scientific Collections (DiSSCo) where the MIDS level is calculated for each digital specimen. The MIDS Calculator can be used to calculate the MIDS score for DwC archive and ABCD biological datasets based on the current version of MIDS. Additional functionality for geological and palaeontological datasets is being developed. As the MIDS standard approaches public review we encourage curators and collection managers to test out the functionality on their collection data and provide feedback using the MIDS GitHub repository.

HTML

XML

PDF

DOI: 10.3897/biss.9.181619

Authors: Pierre Bonnet, Antoine Affouard, Mathias Chouet, Jean-Christophe Lombardo, Hugo Gresse, Thomas Paillot, Vanessa Hequet, Hervé Goëau, Alexis Joly

Abstract: IntroductionCitizen science platforms such as Pl@ntNet (Lefort et al. 2025), iNaturalist, and Flora Incognita have enabled the collection of tens of millions of plant occurrences over the last decade, many shared via the Global Biodiversity Information Facility (GBIF). Thanks to rapid advances in AI-assisted image identification and geospatial analysis, these data streams are increasingly being used for investigative purposes: from monitoring trade flows of Convention on International Trade in Endangered Species (CITES)-listed plants to detecting invasive species, documenting illegal cultivation in protected areas, or time-stamping habitat disturbances.Results and operational workAI-assisted observations can support biodiversity-related investigations in several distinct ways, depending on who produces the observations and how they are used. First, automated identification can help customs officers prioritize inspections at border entry points by flagging potentially protected plants—such as rare succulents or orchids—visible in luggage or shipments. In addition, AI-assisted citizen-science observations provide localized confirmation of species’ presence in landscapes. Spatiotemporal clustering of such records can indicate the time and location of activities of interest, such as the sudden appearance of a critical invasive species at a sensitive site. Finally, AI-supported citizen-science data can contribute to early-warning systems by revealing anomalies in species-community patterns, prompting environmental agencies responsible for monitoring specific areas to conduct targeted field verification before ecological impacts intensify.To transform this potential into probative value suitable for management actions, several points are essential. The authenticity of observations requires a strong link between the digital asset and its source, exploiting the metadata of the device used for data collection and the identity of the citizen scientist. Integrity relies on secure storage, requiring the implementation of protection systems and monitoring of any possible transformation. The reliability of the observations themselves is linked to the implementation of calibrated AI scores, as well as the traceability of the models and versions used to generate the predictions. The reproducibility of the employed methods requires transparent metadata that complies with selected standards, in particular through the use of standardized fields, e.g., based on terms of the Darwin Core standard (Darwin Core Maintenance Group 2023).Scientific and ethical constraints must be taken into account from the outset. Rigorous quantification of uncertainty (calibrated confidence of AI models, identification of consensus predictions involving expert arbitration) are necessary to confirm the presence/absence (or even the abundance) of certain species or groups of species. Ethical safeguards should prevent harm by paying particular attention to sensitive localities, ensuring General Data Protection Regulation (GDPR)-compliant processing of personal data.Current European projects are making concrete progress in meeting these requirements. Within "safeGUARDing biodiversity and critical ecosystem services across sectors and scales" (GUARDEN) Horizon Europe EU project, we are co-designing workflows with agencies and non-governmental organizations (NGOs) to explore AI-assisted identification and classification services with transparency.DiscussionStrengthening the reliability of citizen-science infrastructures for potential use in formal investigations requires careful attention to sustainability and interoperability. This includes developing a sustainable management model, backward-compatible APIs, and institutional partnerships to ensure preservation and legal interoperability. Maintaining community trust remains paramount: contributors must see how their observations translate into preservation results while retaining control over visibility and attribution.Future improvements will consolidate this trajectory. On a large scale, content authenticity should be strengthened by adopting a regularly updated taxonomic repository, drawing on the latest available resources, resulting from work carried out through the Plants of the World Online (POWO) and World Flora Online (WFO) platforms. These taxonomic standardization efforts should be complemented by the use of vocabularies from the Darwin Core standard, as well as other standards from the Biodiversity Information Standards (TDWG) community (such as the World Geographical Scheme for Recording Plant Distributions WGSRPD for example), in order to better guarantee the provenance of the produced data and to clearly document the roles of the various reviewers involved, whether automated AI-based classifiers, volunteer contributors validating observations, or expert botanists providing authoritative identifications.The robustness and security of implemented services should be reinforced by additional tests to limit poor data quality (drawing in particular on the work of a dedicated group*1 within the TDWG community), the detection of out-of-distribution sources and uncertainties, and the verification of duplicates. Strategies to improve the quality of decisions will need to involve humans, e.g., via active learning workflows (mechanisms that automatically request expert review when uncertainty or legal risk is highest), and expert-based consensus protocols preserving the transparency of divergent opinions. Multimodal validation, combining ground-based photos with drone and satellite images, as well as acoustic sensors, can strengthen chains of evidence.ConclusionAI‑assisted, citizen‑generated observations already support screening, intelligence, and early warning in biodiversity enforcement. To elevate their evidential value, we must institutionalize provenance, uncertainty, and ethics by design, linking community platforms, international standards, and operational partners. The GUARDEN and Modern Approaches to the Monitoring of BiOdiversity (MAMBO) pilots illustrate a pathway toward forensic‑ready biodiversity data systems that complement traditional monitoring and help protect living systems in real time.

HTML

XML

PDF

DOI: 10.3897/biss.9.181523

Authors: Giulia Micai, Wouter Addink, Anton Güntsch, Heimo Rainer

Abstract: The digitisation of natural history collections is nowadays among the most critical challenges and strategic efforts faced by museums and scientific institutions around the world.In the frame of the EU-funded project TETTRIs (Transforming European Taxonomy through Training, Research and Innovations), we built upon existing platforms for specimen annotation, such as Les Herbonautes (Muséum national d'Histoire naturelle, Paris), Die Herbonauten (Botanischer Garten und Botanisches Museum Berlin) and DoeDat (Plantentuin Meise), to extract new data from botanical specimens physically deposited in various European herbaria and increase the engagement of the general public with collections.The objectives of this work have been to expand the use of these pre-existing tools to a broader community of citizen scientists; to reach different linguistic communities; and to get the platforms to be used in more countries. We also investigated the connection of these tools with the new DiSSCo Digital Specimen infrastructure (DiSSCOver). DiSSCOver enables annotations to be made directly on the specimen data connected through this infrastructure.The digitisation of herbarium specimens significantly enhances access to the collections for researchers and institutions located all over the world and facilitates large-scale studies. Furthermore, digitisation helps to preserve the physical specimens by reducing the need for handling and transportation. It also allows novel uses of metadata in combination with digital material: for example, herbarium label data have been used to study shifts in plant phenology and also to generate species distribution models (since accurate models rely on large numbers of data points that can efficiently be achieved via the repositories of digitized herbarium records).Through the joint use of the three aforementioned crowdsourcing annotation platforms, we have been able to reach various communities of volunteers who helped overcome linguistic barriers that might make data extraction difficult on a local scale. They have made the information on the labels machine-readable, searchable and safely stored. Relying on a broad, varied and knowledgeable volunteer community allowed us to overcome the difficulties that are typical of label transcriptions. One of the main challenges has been the transcription of difficult handwriting or old, obsolete scripts. Also, we successfully overcame linguistic barriers by delivering specimens otherwise hard to annotate on a local scale to the corresponding linguistic community. Thanks to the knowledge of amateur specialists it has been possible to correctly attribute locations in presence of, for example, historical names of locations and to fill up incomplete data. By involving a large number of volunteers and processing high numbers of specimens in a limited time, the process of transcription sped up drastically.Since the beginning of the project, we have been able to launch and complete nine digitisation campaigns: three on Les Herbonautes, three on Die Herbonauten and three on DoeDat. By doing so, we have been able to transcribe the label data of 5,603 specimens written in 3 different languages, involving approximately 115 volunteers. In particular, a special effort was put into delivering specimens with labels written in German, French and Dutch languages to the corresponding linguistic communities: for example, labels in French but specimens conserved at Berlin have been transcribed online by a French-speaking community of volunteers, and so on, overcoming the linguistic barrier that kept this material from being transcribed. This action contributed to creating an extended community of volunteers involved in more than one annotation platform at the same time, promoting a knowledge exchange across national borders.

HTML

XML

PDF

DOI: 10.3897/biss.9.181567

Authors: Sabine von Mering

Abstract: The Museum für Naturkunde Berlin (MfN) is an integrated research museum of the Leibniz Association. As part of the MfN Future Plan, the vast natural history collection is opened up and digitised; a digital catalogue, currently under development, will provide access to this collection comprising approximately 30 million objects from zoology, palaeontology, and mineralogy. Within this transformation process, an open knowledge infrastructure for all will be created, providing access for researchers as well as the wider public.The Museum Knowledge Graph project aims to improve the transparency, accessibility and discoverability of the collection. In addition, it facilitates monitoring and reporting, e.g., in the context of institutional evaluations. For some years, the open knowledge base Wikidata has been used to create identifiers for different entities related to the museum and its collection. After starting with a project focussing on collection agents (von Mering et al. 2024), Wikidata is now used to provide more structured, human- and machine-readable data connected to the institution. Currently, Wikidata identifiers for people and events are integrated in our collection management system, Specify.The evolving Museum Knowledge Graph provides information about the museum in time and space. It contains data on entities such as taxa, specimens, people, places, publications, and events, as well as their relationships, supporting better contextualisation of collection and research data. The collaborative and multilingual tool, Wikidata, links to external databases and their identifiers, thus serving as a central hub for Linked Open Data (LOD).The current focus is on the following entities (Fig. 1), including:People (1000+): Collection agents with different roles working at or affiliated with the museum or contributing to the collection; well-known collectors as well as hidden figures; technical staff; and underrepresented or marginalised groups, e.g., local guides, indigenous contributors, women in natural history.Collection objects (close to 1 million): Persistent identifiers, so-called NURIs (Naturkundemuseum URIs), are generated for MfN specimens (based on Consortium of European Taxonomic Facilities (CETAF) Stable Identifiers; Groom et al. 2017).Publications (200+): Books or scholarly articles dealing with the collection or institutional history, e.g., accession catalogues, published catalogues of type specimens (von Mering in press), about research expeditions commissioned or organised by the museum, also publications in the MfN journals published by Pensoft.Events (20+): Important historical events such as research expeditions (von Mering et al. 2024b), milestones in the history of the museum (e.g., new building, acquisition or moving of a collection) or conferences held at the museum.Organisational units and subcollections (20+): Centres, departments and subcollections (also predecessor institutions) of the MfN.Agents and expeditions are linked to estates and records in the MfN Archive, with inventory numbers and links to available finding aids that can improve access to relevant sources such as field notebooks, correspondence, diaries, etc., helping us also to uncover hidden contributors and historical contextualisation.The dynamically growing Museum Knowledge Graph is part of work done in the Research & Innovation field, Data for Nature, which is part of the science programme, Collection Future. Currently, we are scaling up, importing data from several subcollections, including identifiers for people following disambiguation, and for research expeditions. The newly created WikiProject, MfN Berlin Names, will support keeping track of named entities related to the museum. In addition to tracking MfN-related citations of publications and specimens, we aim to measure the number of connections, the reuse of data and linked media, and thus have an additional indicator for impact. Edits and additions to newly created Wikidata items are monitored, different tools developed within the Wiki community allow the community of users to receive such metrics.The example of the Museum für Naturkunde Berlin (Fig. 2) demonstrates the value of Wikidata for unlocking, connecting and contextualising information about natural history collections in a Museum Knowledge Graph. Contributing to the larger LOD ecosystem will help to answer interdisciplinary questions and discover unexpected connections.

HTML

XML

PDF

DOI: 10.3897/biss.9.181465

Authors: Elspeth Haston, Hannah Atkins, Marie Briggs, Helena Brown, Robyn Drinkwater, Hui Ke Christopher Ng, Heleen Plaisier, Michael Sundue

Abstract: As the digitisation of natural science collections progresses, the need for tools and skills for the staff who curate those collections becomes increasingly important and urgent. The specimens held in these collections are fundamental to research and are a critical source of data and knowledge urgently needed for the current biodiversity crisis. Without taxonomic curation, the names attached to these specimens may be synonyms, misapplied or incomplete. One study examined specimens from 40 herbaria in 21 countries and found that half of specimens from two tropical plant families had the wrong names applied (Goodwin et al. 2015). These issues may have a significant impact on research undertaken based on these collections.Curation of natural science collections covers a huge range of activities and responsibilities. Taxonomic curation is the management and updating of the taxon names applied to the specimens in collections. Taxonomic curation can be carried out at different levels. A key aspect of taxonomic curation is providing access to accepted classifications across multiple taxonomic levels and enabling comparisons with alternative systems, including those already in use within a collection. At the specimen level, tools to access specimen citations, identifications of duplicates as well as managing online determinations are becoming more important for curation staff and volunteers. These processes require digital skills for curators. For curation staff and volunteers to develop similar skills and be trained in the use of technical tools, adequate and accessible training documentation is required. Before digitisation, collections were frequently curated using regional taxonomic accounts and it was quite common to have specimens from different regions arranged differently, resulting in multiple taxonomies through the collection. With digitisation, there is usually a single taxonomy in the database which pushes us more towards aligning all specimens to a single classification. The digitised specimen record will normally include the name under which the specimen is physically filed, and it is critical that this is always aligned to ensure findability of the specimen. This is a huge task and the combination of the growth of the collections and the lack of growth in staffing levels means that we must look at more efficient options for doing this work. Case studies in the Herbarium of the Royal Botanic Garden Edinburgh were undertaken to review the process of curation of several digitised taxa and one undigitised taxon. This allowed us to explore the impact of digitisation on curation practices. We also investigated the additional training requirements for staff to be able to use the increasing number of online tools and resources available. The case studies included the genera Plectranthus and Coleus in the Lamiaceae (Fig. 2, Streptocarpus in the Gesneriaceae (Fig. 1) and the family Lycopodiaceae. We identified 4 levels of curation:High level to delimit and organise the collection at Kingdom, Phylum and Class level;Family level to delimit and organise families and the genera within them;Genera and species level to delimit and organise genera and species; andSpecimen level to identify individual specimens in the collection. The appropriate level of taxonomic curation will depend on the scope of the taxa being curated and the time and resources available. Using the case studies at the Royal Botanic Garden Edinburgh, we are developing more effective and efficient taxonomic curation workflows for digitised collections. For each workflow, we include the tools and resources available as well as consider the skills, knowledge, and training required. The initial workflow for the genera and species level comprises ten steps:Curation requirements and prioritisation;Classification review;Taxonomic scope;Review and comparison of taxonomy in literature;Review and comparison of taxonomy in the collection management system (CMS);Recuration planning and communication across the organisation;Updating of names and taxonomy in the CMS;Reorganisation and updating of specimens and folders (Fig. 2) and specimen records in the CMS;Updating of indexes; andUpdating of cabinet labels. The development of workflows has identified requirements for tools and resources, as well as requirements for skills. This approach can also standardise taxonomic curation practices and make it easier to train curation staff, giving them guidance for their work.

HTML

XML

PDF

DOI: 10.3897/biss.9.181439

Authors: Claus Weiland, Daniel Bauer, Desalegn Chala, Dag Endresen, Jonas Grieb, Zakaria Kehel, Marcella Orwick-Rydmark, Danaële Puechmaille, Gabriela Zuquim

Abstract: The European Green Deal addresses a wide array of critical topics regarding the interaction and exchange between society and ecosystems involving human-induced climate change, biodiversity loss, and deforestation. To support the development of appropriate and sustainable mitigation strategies, both political frameworks as well as large-scale scenario forecast and simulation infrastructures such as Destination Earth Data Lake (DEDL, Puechmaille et al. 2025) and the Green Deal Data Space (Penninga et al. 2021) are currently implemented. Setting up these data spaces requires the consolidation and harmonization of vast amounts of data, among them data from long-term biodiversity monitoring (e.g. eDNA, Abarenkov et al. 2023), near real-time data captured by IoT devices (e.g. camera traps, Grieb et al. (2025)) and large-scale Earth observation data (e.g. ERA5 global reanalysis, Hersbach et al. (2020)). Key tools and research methodologies for integration of such data have been developed in the Destination Earth-linked Biodiversity Digital Twin project (BioDT). Biodiversity digital twins provide digital replicas of biodiversity patterns and processes (Khan et al. 2025), for FAIR-compliant reuse and cross-platform portability, they are packaged in BioDT using the RO-Crate specification. Resulting datasets as well as computational workflows are bundled as digital objects and annotated with rich machine-interpretable metadata using schema.org and its extensions Bioschemas and Croissant (high-level metadata specification supporting AI-Readiness, Bolleman et al. 2025).The workflow.earth pilot was developed to migrate and functionally integrate BioDT's digital twin for crop wild relatives (BioDT CWR) into DEDL. BioDT CWR is designed to identify genetic resources in crop wild relatives (CWR) and aid in creating strategies to mitigate food-related crises linked to climate change (Chala et al. 2024). CWR are crucial sources of untapped genetic diversity, offering greater adaptability to extreme weather events such as droughts, cold waves, and heavy precipitation. They are therefore of particular importance to ensure food security with regard to Sustainable Development Goals (notably Goal 2 “End hunger, achieve food security and improved nutrition and promote sustainable agriculture”, United Nations 2023).Major objectives for workflow.earth areto enhance visibility and long-term availability of the BioDT CWR,to improve its prediction capabilities using DEDL’s near data processing services, andto facilitate (in perspective) the orchestrated interplay with other digital twins such as the Climate Change Adaptation DT (Weiland et al. 2024).On top of RO-Crate, workflow.earth employs FAIR Signposting to add machine-interpretable links to human-readable digital objects using link relation types (RFC8288, Fig. 1) to present the web topology of a resource (Soiland-Reyes et al. 2024).In this presentation, we will demonstrate the deployment of agrobiodiversity workflows in DEDL using workflow.earth and the publication of model and results as webby FDOs.

HTML

XML

PDF

DOI: 10.3897/biss.9.180327

Authors: Diane Srivastava, David Hunt, Sandra Binning, Sandra Emry, Ellen Bledsoe, Jessica Reemeyer, Jason Pither

Abstract: The need for data rescue. Data are dying all around us: when they are stored in inaccessible or fragile repositories, in proprietary formats, on defunct media or without complete metadata. In environmental science, such data loss has high societal and scientific costs. For example, failure to archive ecological data focused on the Exxon Valdez oil spill represents an estimated loss of >100 million USD (Bledsoe et al. 2022). Loss of environmental data can also represent erasure of historic baselines that can never be recollected.The Living Data Project (LDP). This award-winning program helps scientists and organizations archive valuable “legacy” datasets, making them permanently open and accessible (Bledsoe et al. 2022). We pair data custodians with graduate students who we train in reproducible data management and provide mentoring by postdoctoral researchers and faculty and assistance by undergraduates. By training Canada’s next generation of scientists, we aim to ensure that data are never lost again. Over the last five years, we have trained 305 students, across 93 internships, rescuing > 2000 years of data. LDP student working groups then analyze archived data to investigate pressing biodiversity questions and develop open access teaching resources.Benefits of data rescue. LDP-rescued datasets, many stretching back into the last century, provide missing historical baselines that enable the detection of environmental change – such as change in lake ice formation and melt over the last 150 years. Rescue of historical datasets, like surveys of urban birds or plant communities collected five decades ago, sets the stage for contemporary re-surveys. Combining rescued data with contemporary data can also yield discoveries such as legacy effects of pesticides on stream health (Sugden et al. 2025). LDP rescue of classic studies in ecological and evolutionary theory (Darwin’s Finches, Serengeti ungulates, and freshwater sticklebacks) enables new analyses that were computationally impossible at the time. LDP projects help environmental organizations archive data over broad spatial and temporal scales, critical for metrics such as the Watershed Reports or Living Planet Index. LDP also assists non-professionals to archive data; e.g., helping communities archive water quality data on DataStream enables easy comparison to national standards. Data rescue can knit together disparate datasets in relational databases, such as LDP compilation of fish, invertebrate, phytoplankton, and limnology data from the Canadian government’s Turkey Lakes Watershed research. The benefits of data rescue will pay incalculable dividends for decades to come. However, to ensure successful data rescue, we recommend the following:Rules for success:Prevent scope creep. When non-essential data is included in the initial scope of the project or the scope broadens during the rescue process, concluding the project is challenging. It is better to successfully archive a subset of the data than fail to archive any.Involve data collectors and custodians throughout. They provide important contextual knowledge, help decipher obscure data codes and can suggest constraints to values, facilitating validation. Formalizing their knowledge in the metadata ensures future usability.Continually champion projects until the dataset DOI is minted. It is easy for projects to stall in the final stages. Delaying archiving to add more years or variables or to publish one more paper increases the risk of no data being archived. Encourage the use of versioning or embargo periods, available for many repositories, to deposit promptly the core data while enabling later additions and public release.Manage adaptively. Problems like contradictory data, inconsistent use of terms, and heterogeneous collection methods may only be apparent after working with the data. At the LDP, we use high-level checks of progress throughout each rescue project to adaptively manage the scope and personnel needed, informed by our experience in a wide variety of projects.Summary. Environmental data rescue has many benefits, including establishing baselines and change in the environment, enabling new analyses and composite indices, connecting disparate data, facilitating community science, and training the next generation of researchers in open science. However, data rescue projects requires continued and careful management to ensure success, which we defined as the deposition of data in open, accessible and permanent repositories.

HTML

XML

PDF

DOI: 10.3897/biss.9.181374

Authors: Rémi Palard, César Leblanc, Lukas Picek, Benjamin Deneu, Thomas Paillot, Antoine Affouard, Mathias Chouet, Maximilien Servajean, Pierre Bonnet, Alexis Joly

Abstract: The increasing frequency of environmental crimes (Rippingille 2023), from illegal deforestation and habitat degradation to biopiracy, requires new digital tools for collecting, verifying, and communicating biodiversity evidence. GeoPl@ntNet addresses this need by transforming citizen-science and expert plant observations into auditable, high-resolution biodiversity intelligence suitable for research, conservation, and forensic applications.Built upon the globally used Pl@ntNet infrastructure, which hosts more than 20 million users contributing georeferenced plant observations from 200 countries, GeoPl@ntNet Leblanc et al. (2025a) integrates this vast dataset with environmental predictors to generate predictive biodiversity maps at 50×50m spatial resolution. At its core, GeoPl@ntNet employs a multi-species deep learning model (DeepSDM Botella et al. 2018) and a Habitat Distribution Model based on a transformer model.The DeepSDM is first trained on approximately 30 million Global Biodiversity Information Facility (GBIF) species occurrence records and 5 million European Vegetation Archive (EVA) surveys to infer species presence, habitat types, and biodiversity indicators across Europe. The model combines environmental variables, Sentinel-2 RGB and near-infrared imagery, elevation, land cover, and long-term climatic time series. Species distribution predictions are then computed over a European grid of 50×50 m cells within 25×25 km meta-tiles. The species distribution predictions are subsequently fed into a transformer-based model (Pl@ntBERT Leblanc et al. 2025b) to predict habitats under the EUNIS 2020 standard. Finally, derived biodiversity indicators, such as counts of protected, local, or invasive species, are aggregated into reproducible spatial layers that users can explore, download, or integrate via WMS or STAC services.Beyond ecological research, GeoPl@ntNet demonstrates strong potential for environmental forensics. Investigators can detect non-native or invasive taxa near ports and industrial areas, monitor biodiversity decline before and after disturbance events, and generate standardized biodiversity reports that support legal or regulatory proceedings.GeoPl@ntNet implements a provenance-based auditing system that tracks the full lifecycle of each biodiversity layer. Every map is linked to machine-readable metadata describing the source datasets, model versions, and environmental predictors used to generate it. These metadata are exposed through STAC and WMS services, enabling external verification, reproducibility of outputs, and legally traceable biodiversity reporting.The platform's design: Nuxt and Leaflet for the frontend, Python backend with MapProxy and TiTiler services, also ensures transparency and reproducibility, enabling alignment with TDWG data standards including World Geographic Scheme for Recording Plant species Distributions (WGSRPD) and EUNIS habitat classifications. By coupling scalable AI models trained on both citizen-science and expert data with diverse maps and reports, GeoPl@ntNet connects biodiversity informatics with environmental governance, enabling auditable, site-level verification of biodiversity integrity.Accessible via https://geo.plantnet.org, the platform will be extended to global coverage and integrated within Pl@ntNet to support scalable and legally robust biodiversity assessments worldwide.

HTML

XML

PDF

DOI: 10.3897/biss.9.180774

Authors: Zakaria Doubabi, Hanane Ariouat, Eric Chenin, Jihad Zahir, Christophe Denis, Youcef Sklab

Abstract: Bone classification is fundamental in paleontology, evolutionary biology, and museum collection management, where natural history museums house billions of specimens documenting biodiversity. Traditional bone identification has relied on specialist expertise and manual methods, but with rapid digitization of collections and advances in 3D scanning, computer-assisted approaches are increasingly adopted. Among digital formats, 3D point clouds (Qi et al. 2017a) produced using scanning techniques such as surface scanning, photogrammetry, and computed tomography (CT) have become preferred representations as they preserve fine structural details while remaining computationally efficient. However, the absence of inherent ordering in 3D point space poses fundamental challenges for conventional deep learning models, and scaling manual classification to vast digital collections remains problematic.This study evaluates and compares state-of-the-art deep learning architectures, specifically designed for 3D point cloud processing, on mammalian bone classification. We assembled a dataset of 1,641 3D bone specimens representing 14 bone-type classes, 302 species, 203 genera, 40 families, and 28 orders, obtained through digitization as part of the e-COL+ project*1, which aims to digitize a wide range of zoological specimens in 3D and is led by the MNHN, Paris. This taxonomic breadth increases morphological variability and provides a valuable benchmark for evaluating generalization capacity. All 3D meshes were standardized by converting to point clouds, downsampled to 1,024 points using Furthest Point Sampling (FPS) algorithm, and normalized to unit spheres (Fig. 1). Training data underwent augmentation including random rotations, scaling, and Gaussian jittering, with an 80–20 train-validation split. We evaluated seven architectures: PointNet (Qi et al. 2017a), PointNet++ (Qi et al. 2017b), Dynamic Graph CNN (DGCNN) (Wang et al. 2019), Point Transformer (Zhao et al. 2021), Point Cloud Transformer (Guo et al. 2021), PointMLP (Ma et al. 2022) and PointNeXt (Qian et al. 2022). Models were implemented in PyTorch, trained for 150 epochs using Adam optimizer with step-based learning rate scheduling, and employed cross-entropy loss with label smoothing and class weighting to address imbalance.Results in Table 1 demonstrate clear performance differences in overall accuracy, mean accuracy and f1-score metrics across model families. PointMLP achieved the highest performance with 98.29% ± 0.2% overall accuracy, showing excellent discriminative capacity and stable generalization. PointNeXt and Point Transformer followed closely with 98.19% ± 0.5% and 97.89% ± 2.4% overall accuracy respectively, highlighting effective feature learning capabilities. Earlier architectures like PointNet performed substantially worse at 82.53% ± 2.4% overall accuracy, reflecting limited capacity for capturing local geometric dependencies. The progression from PointNet to PointNet++ (87.95% ± 0.5%) demonstrates the importance of hierarchical feature learning, while DGCNN (93.27% ± 1.7%) and PCT (95.88% ± 0.8%) show intermediate performance levels. Notably, PointMLP exhibited the lowest variance across all metrics, indicating superior training stability and generalization consistency.The findings reveal that architectures emphasizing sophisticated feature aggregation mechanisms (PointMLP, PointNeXt, Point Transformer) currently provide the most accurate and generalizable solutions for bone point cloud classification. Modern MLP-based approaches like PointMLP demonstrate particular effectiveness, combining computational efficiency with superior performance (98.01% ± 0.3% mean accuracy, 97.89% ± 0.3% mean F1-score). Advanced architectures including Point Transformer and PointNeXt achieve comparable accuracy levels (>97% overall accuracy) while offering different trade-offs between model complexity and geometric reasoning capabilities. Earlier generation models (PointNet, PointNet++) remain viable for baseline comparisons but exhibit significant performance gaps relative to state-of-the-art architectures.This research demonstrates the potential of point cloud deep learning for advancing automated morphological analysis in biodiversity research. Despite limited training data, these models accurately classify complex biological structures, with top-performing architectures exceeding 98% classification accuracy, underscoring deep learning effectiveness for such applications. The success of modern deep learning architectures opens new possibilities for large-scale automated classification of museum collections, supporting digitization efforts of natural history institutions worldwide. Future work should focus on investigating the scalability of high-performing architectures to larger taxonomic datasets and exploring multi-modal approaches combining geometric and taxonomic information.

HTML

XML

PDF

DOI: 10.3897/biss.9.181431

Authors: Yi-Ming Gan, John Wieczorek, Anton Van de Putte, Cecilie Svenningsen, Pablo Deschepper, Charlie Plasman, Tim Robertson

Abstract: The BROKE-West fish dataset (Australian Antarctic Data Centre 2022) originates from the BROKE-West voyage, a survey conducted in a data-poor area of the Southern Ocean in 2006. This dataset forms part of the baseline biodiversity assessment of the Cosmonaut and Cooperation Seas.The dataset was initially published in 2008 as a single flat Darwin Core Archive (DwC-A) occurrence table without extensions. Although the original dataset contained valuable occurrence data, it omitted much of the sampling context, including survey design, target scope, and material relationships. In 2016, the dataset was reused to test the extended MeasurementOrFact (eMoF) extension developed by Ocean Biodiversity Information System (OBIS) (De Pooter et al. 2017), which enabled the inclusion of environmental and biological measurements (Van de Putte 2016). In 2023, the Humboldt Ecological Inventory extension (TDWG Humboldt Extension Task Group and Biodiversity Information Standards (TDWG) 2024) was applied to describe the survey design and scope, linking each trawl to its survey targets and sampling context (Van de Putte et al. 2025).The presentation from which this extended abstract derives (Gan et al. 2025) describes the limitations of DwC-A for this dataset and how DwC-DP overcomes these.While DwC-A remains useful, several concepts cannot be well represented in this format.Marine datasets often contain both community-level measurements (e.g., abundance or total weight of an occurrence) and individual-level measurements (e.g., the standard length of a single fish from that same occurrence). The subsetting relationship is not explicit in DwC-A.Survey context and survey target are flattened into a single Humboldt Extension table. If each survey target combination is recorded as a separate row, survey context information, such as eco:isTaxonomicScopeFullyReported of a dwc:Event, must be repeated for each target.The use of identifiers can be confusing: because the Occurrence extension requires a unique dwc:occurrenceID per row, the sample identifier representing a dwc:MaterialEntity must be used as dwc:occurrenceID. This leads to properties of the material (e.g., dry weight) being linked via eMoF to an Occurrence instead of to the material itself.Although photographs can be linked using dwc:associatedMedia, this approach omits key media metadata, such as the license.Prey items found in the stomach contents of fish from an occurrence are difficult to model in DwC-A.Today, the dataset has been fully restructured (Gan and Van de Putte 2025) using the proposed Darwin Core Data Package (DwC-DP) (Darwin Core Maintenance Group 2025). DwC-DP replaces the single-core architecture with a linked-table model that overcomes the limitations of the star schema. This new structure allows an explicit representation of Occurrence and Material, preserving the meaning of dwc:occurrenceID and dwc:materialEntityID. It also supports explicit relationships such as isPartOf and derivedFrom. For example, a prey item with dwc:materialEntityID “AAV3FF_00337_stomach_002_Euphausiid” is dwc:derivedFromMaterialEntityID “AAV3FF_00337” with dwc:derivationType “stomach content of”.This restructuring has revealed the full analytical potential of legacy datasets. The BROKE-West case demonstrates how DwC-DP can represent data that were previously unpublished or too complex to format in DwC-A, such as material-to-material derivations and media linked to materials.The following three examples highlight DwC-DP’s analytical flexibility.Non-detections for an event can be inferred by comparing occurrences to survey targets. If all occurrences of a survey target are fully reported, an event without that survey target indicates a non-detection.Fish length-frequency distributions can be computed directly by linking Material Assertion to Material and then to Event, aligning with the Essential Ocean Variable (EOV) sub-variable Fish Abundance and Distribution (Global Ocean Observing System 2025).Diet composition can be represented through material relationships between predator stomachs and prey taxa — an analysis previously impossible in DwC-A because stomach-content data were unpublished. An OrganismInteraction table was not used since many prey items were unidentifiable or digested and their ingestion time unknown, making the creation of corresponding prey Occurrences difficult. These examples underline how DwC-DP enhances both data representation and the interpretability and reusability of historical datasets.Adopting DwC-DP brings major benefits but also challenges. It better preserves data meaning, supports richer links between entities, and suits complex marine survey datasets. However, it requires more identifiers, more relational workflows, and careful curation to keep identifiers stable over time.DwC-DP is a major shift toward a relational biodiversity data model that accommodates complex ecological surveys. The BROKE-West case study shows how DwC-DP enriches flat datasets and more accurately represents survey complexity.

HTML

XML

PDF

DOI: 10.3897/biss.9.181234

Authors: Wouter Addink, Sam Leeflang

Abstract: The accelerating decline of biodiversity demands urgent action to mobilise and connect the knowledge held in natural science collections. These collections represent centuries of accumulated specimen-based evidence, offering traceable and historical information essential for scientific discovery, conservation, and policy. Yet despite decades of progress in digitisation, the majority of specimen information remains inaccessible, through lack of digitisation or fragmentation across institutions and data silos. Coordinated collaboration, shared infrastructure and industrialisation of digitisation with help from machines, are needed to unlock the full potential of these resources at the scale, required to address global biodiversity challenges.The Distributed System of Scientific Collections (DiSSCo) is designed to meet this challenge. By bringing together institutions, expertise, and services across Europe, DiSSCo aims to provide a unified platform where both humans and machines can collectively work on enhancing, linking, and curating specimen data (Koureas et al. 2018). Central to this vision is the need for a common, comprehensive and extensible data model that can handle the diversity and complexity of specimen-related data and metadata, support interoperability across systems, ensure long-term accessibility of data, and compliance with FAIR (Findable, Accessible, Interoperable, Reusable) Digital Object (FDO*1) principles to support machine actionability.For this, DiSSCo has developed the open Digital Specimen (openDS)*2 specification. It is an all-encompassing framework for describing, linking, and preserving digital representations of physical specimens (Hardisty et al. 2019). Unlike isolated schemas developed for specific domains, openDS functions as a shared language across the diverse service providers and communities, not only contributing to DiSSCo, but globally and across domain boundaries. It provides the flexibility needed to represent any type of specimen-related (meta-)data while embedding mechanisms for persistent identification, versioning, annotation, and provenance.The design of openDS builds on a strong foundation of existing community standards and persistent identifiers. It extends and integrates TDWG (Biodiversity Information Standards) standards, including Darwin Core, the Chronometric Age extension, and Audiovisual Core, while also incorporating broader semantic web standards such as Schema.org, World Wide Web Consortium (W3C) Web Annotation, and the PROV Ontology (PROV-O). In this way, openDS can be regarded as an implementation*3 of the Darwin Core Conceptual Model*4, ensuring lossless data transfer to and from Darwin Core Data Packages (DwC-DP). This interoperability guarantees that institutions adopting openDS can continue to align with widely used biodiversity data standards while benefiting from new capabilities provided by FAIR Digital Objects and DwC-DP.In 2024, openDS underwent a public community review *5 coordinated by DiSSCo, to collect feedback from biodiversity information experts worldwide. Since December 2024, openDS has been the basis of the DiSSCo Digital Specimen data infrastructure, underpinning key services such as Digital Specimen Digital Object Identifier (DOI) assignment, metadata integration, and interoperability with collection management systems and data infrastructures. By using FDOs, openDS ensures that each digital specimen is not merely a static record but a dynamic, citable, versioned, and machine-actionable object that can be enriched and reused across research workflows.The structure of openDS reflects the complexity and richness of specimen data. The model consists of eight top-level classes*2 , complemented by a set of generic nested classes and hundreds of terms that together cover a wide range of specimen attributes, contextual information, and derived data. While this richness may initially appear daunting, it is precisely what enables openDS to accommodate the diversity of natural science collections, from herbarium sheets, fossil specimens and rock samples to genomic extractions and environmental samples. The minimum information for what constitutes a “Digital Specimen” is described by the (draft) MIDS (Minimimum Information about a Digital Specimen) standard (Haston and Chapman 2022). Through careful alignment with FAIR principles and added semantics, openDS also ensures that digital specimens can be linked to a wide variety of related digital resources, including digital media, DNA sequences, publications, ecological observations, and nanopublications.By adopting openDS, DiSSCo is not only enabling a pan-European integration of natural science collections but also contributing to the global conversation on how best to structure, govern, and sustain FAIR Digital Objects for science. The model exemplifies how community standards can be extended and harmonised into a framework that supports both human curation and machine-actionable, AI-supported workflows. For biodiversity researchers, this means improved access to specimen information including provenance about data changes, persistence, and citability of specimen data. For institutions, it provides a sustainable pathway to align local digitisation efforts with international infrastructures and standards to improve data quality and data access.In the conference we present the open Digital Specimen model, describe its design principles and foundations in community standards and use cases, and illustrate how it implements FAIR Digital Objects in practice. By guiding participants through the structure of openDS and demonstrating its role in linking collections, data infrastructures, and research, we aim to show how this new model contributes to a future where digital specimens play a key role in enhanced access to specimen data by researchers.

HTML

XML

PDF

DOI: 10.3897/biss.9.181329

Authors: Peter Bozakov, Andrea Battisti, Andree Cappellari, Davide Nardi, Hervé Jactel

Abstract: Forests play a key role in achieving the objectives of the European Green Deal due to their multifunctional role. They provide multiple ecosystem services, such as soil formation, nutrient cycling, provisioning of timber, food, and water, climate regulation, and flood control, while also being of central importance for biodiversity conservation (Brockerhoff et al. 2017). However, insect and pathogen pests increasingly threaten forests, alone or after abiotic disturbances such as drought, fire, and storms (Romagnoli et al. 2023). Among pests, invasive alien species are of great concern because they can have a strong impact on forest biodiversity and economy, and many efforts are being made to regulate them in the EU (Ghelardini et al. 2017). In addition, several native pests are expanding their range because of climate change and are called emerging pest species (Jactel et al. 2019; Simler-Williamson et al. 2019). Research is therefore essential to support early detection of emerging and invasive pests, monitor their spread, and inform phytosanitary measures that can prevent or mitigate their impacts.In this context, the Horizon 2020 project HOMED (HOlistic Management of Emerging forest pests and Diseases), coordinated by Hervé Jactel from the National Research Institute for Agriculture, Food and the Environment (France), developed a holistic and multi-actor approach to develop tools and methods for the management of emerging native and non-native pests threatening European forests, focusing on all phases of invasion, from prevention to surveillance to eradication and control. The project also broadened the definition of “forest health”, including trees not only in forests, but also in nurseries, urban and rural areas, owing to the key epidemiological role of these components. The major outputs of the project are synthesised in a special issue of the journal NeoBiota, and include a participatory tool for reporting forest damage data (Silvalert), improved multi-funnel traps with generic pheromone attractants (Santoiemma et al. 2024), pioneering studies on pest detection using remote sensing (Nardi et al. 2023), and sustained stakeholder engagement through six targeted policy briefs. Importantly, HOMED underscored the need for interdisciplinary collaboration across forest entomology and pathology, economics, and social sciences, and highlighted the priority of developing innovative tools for the early detection and eradication of pests.Building on this foundation, the Horizon Europe project FORSAID (FORest Surveillance with Artificial Intelligence and Digital technologies), coordinated by Andrea Battisti from the University of Padova (Italy), continues and expands HOMED's legacy. FORSAID focuses on designing a comprehensive and technologically advanced framework for the early detection and surveillance of EU-regulated forest pests, integrating a wide range of innovative tools, including satellite- and drone-based remote sensing, smart traps with automatic monitoring capacity, robotic devices for rapid insect identification, environmental DNA methods for early pest detection, and citizen science approaches supported by mobile applications. Stakeholder engagement is central to this process, ensuring that research outputs are aligned with end-user needs and can be effectively implemented in practice. The project aims to integrate these methods into a coherent and operational monitoring and assessment system for regulated pests, thereby supporting the development of a novel pan-European strategy for forest pest management. FORSAID also ensures continuity and knowledge transfer from HOMED. For example, data on pest distribution collected in HOMED through Silvalert now serves as a baseline for FORSAID remote sensing activities. Generic attractants validated globally in HOMED are being incorporated into next-generation smart traps. Lastly, stakeholder surveys from HOMED provide direct input into FORSAID's research agenda. By maintaining and extending these outputs, FORSAID exemplifies how scientific legacies can be consolidated and advanced across projects.Together, HOMED and FORSAID demonstrate the value of sustained research, innovation, and cross-disciplinary collaboration in forest health management, and their combined contributions highlight the importance of open data, participatory science, and stakeholder-driven approaches in building robust strategies against emerging forest pests.

HTML

XML

PDF

DOI: 10.3897/biss.9.181043

Authors: John Wieczorek, Tim Robertson, Paula Zermoglio

Abstract: After more than fifteen years of use, the Darwin Core Archive (DwC-A) (GBIF 2021) has become the most popular means of sharing species occurrence and sampling-event data. But it has a limitation: in its “star-schema” structure, all information must connect directly to a single “core” table. As biodiversity data publishing expands to include molecular, ecological, and behavioral information, publishers increasingly encounter cases where valuable relationships among entities—events, agents, organisms, materials, media—cannot be expressed, resulting in the loss of information.The Darwin Core Data Package (DwC-DP) (Darwin Core Maintenance Group 2025a) arose from this tension. It represents a practical evolution of Darwin Core standard (Wieczorek et al. 2012), retaining its familiar vocabulary while adopting a more flexible, relational structure. Rather than replacing DwC-A, DwC-DP complements it, offering an alternative for datasets that challenge DwC-A’s constraints. DwC-DP supports nested hierarchies of MaterialEntities, Media, and Occurrences in addition to the hierarchical Events already possible in DwC-A.DwC-DP was developed in parallel with and implements a proposed Darwin Core Conceptual Model, (Darwin Core Maintenance Group 2025b) which provides the semantic foundation for the relationships it supports. The proposed Darwin Core Data Package Guide, under review starting 2025-09-17, embodies specifications for using the open Frictionless Data Package framework (Fowler et al. 2017). The Global Biodiversity Information Facility (GBIF) led iterative prototyping and testing beginning in July 2021. The two documents went under public review as potential additions to the Darwin Core standard in late 2025. One might ask, “Why not simply modify the Darwin Core Text Guide specification (Darwin Core Maintenance Group 2023), on which DwC-A is based, to encompass DwC-DP’s capabilities?” If the DwC-A specification were changed to rely on Data Package technology alone, existing Darwin Core Archives would become obsolete. That would violate the principle that users satisfied with DwC-As should be able to continue to use them unimpeded. Conversely, a dual specification preserving backward compatibility would not be interpretable by standard tools. By supporting both, we gain the advantage of broader compatibility and integration across disciplines that data packages offer as products of a wider data-science community.The presentation (Suppl. material 1) from which this extended abstract derives introduces DwC-DP as a story about expressiveness. Using two contrasting narratives—a “DwC-A detective short story” and a “DwC-DP epic novel”—it illustrates how richer relational connections can transform how biodiversity data are documented and the stories they can tell. The “detective short story” is a metaphor for DwC-A’s thematic constraints: it follows a single bee, observed, photographed, collected, and identified—an entirely solvable case using DwC-A. The “epic novel” elaborates on a pollination-survey network, in which the survey target consists of bees visiting flowers (organism interactions), material samples yield DNA sequences to aid identification, and observations, media, identifications, and measurements interlink across multiple events and materials. DwC-DP can fully capture this intricate web of relationships. DwC-A cannot.The use cases that inspired DwC-DP depend on one or more kinds of Events (Occurrence Events, Survey Events, Collecting Events, OrganismInteraction Events), but DwC-DP has no concept of a “core” table as DwC-A does. Instead, Darwin Core Events, Occurrences, and MaterialEntities function as “foundation classes,” to which additional “modules” may be attached for richer information: agent roles, assertions, bibliographic resources, chronometric ages, geological contexts, identifiers, identifications, media, nucleotide analyses, organisms, organism interactions, protocols, provenance, surveys, and usage policies. Each module contributes an additional layer of narrative complexity to the “epic novel.”While offering great potential for richness and depth, DwC-DP’s modular schemas and simplification mechanisms (e.g., embedded agent identifiers for common roles) let publishers include only what they need—an abridged version of the “epic novel.” DwC-DP resembles DwC-A, both use tabular data files and metadata in Ecological Markup Language (EML; Fegraus et al. 2005). The key difference is that DwC-DP describes its structure in a JSON file (datapackage.json), whereas DwC-A uses an XML file (meta.xml). The similarity provides a bridge for those familiar with DwC-A who wish to explore DwC-DP’s greater expressiveness.Who should be interested in DwC-DP? Data publishers who struggle to fit complex data into DwC-A; researchers seeking complete, reproducible, citable datasets with their full context preserved; policy and decision-makers relying on diverse primary biodiversity data; and data aggregators who need to support any of the above.

HTML

XML

PDF

DOI: 10.3897/biss.9.181155

Authors: Tiago Lubiana, Richard Littauer, Siobhan Leachman, Jan Ainali, Manoj Karingamadathil, Andra Waagmeester, Heidi Meudt, Dario Taraborelli

Abstract: With over 50 million observations per year, iNaturalist is one of the world's most successful citizen science projects, uniting millions of people worldwide in observing, sharing, and identifying nature (Mason et al. 2025). iNaturalist and Wikipedia have much in common: they are both collaborative, large-scale, open infrastructures made by volunteer communities with long-reaching impact on human knowledge. Wikipedia and its sister projects, Wikimedia Commons and Wikidata, also influence biodiversity research worldwide via their numerous, widely-accessed entries on nature topics (Lubiana 2025).While contributions to both projects are primarily made by individual volunteers, a community of iNaturalist contributors that first convened at Wikimania 2018, a Wikimedia conference in Cape Town, South Africa, started WikiProject Biodiversity. The coordinated contributors facilitate content flow in three main ways: by directly adding content to iNaturalist and sharing it on Wikimedia projects, by supporting the development of tools, and by promoting campaigns and outreach activities (for example, by participating at Living Data 2025, in Bogotá, Colombia). To enable the seamless upload of iNaturalist images to Wikimedia Commons (which in turn enables their reuse on Wikipedia and other Wikimedia projects), this volunteer community has developed a diverse set of open source tools, detailed in Table 1. These volunteer-driven tools are released under open source licenses, making it possible for anyone to join, contribute and expand on the work. While there is no formal content partnership between iNaturalist and the Wikimedia Foundation, the volunteer community that connects the two platforms follows a similar pattern as GLAM (Galleries, Libraries, Archives and Museums) partnerships in Wikimedia. This ensures that iNaturalist content on Wikimedia enjoys a wealth of resources for tracking and reporting impact, including statistics for uploads and page views. Using Wikimedia's open Commons Impact Metrics API, we can identify over 220,000 images on Wikimedia Commons sourced from iNaturalist. A fourth of those images (around 53,000) are reused to illustrate content on Wikimedia projects (such as English Wikipedia). On average, iNaturalist images receive around 27 million monthly views across Wikimedia platforms (as of August 2025). When these images make it to the English Wikipedia main page, views may go as high as 196 million in one month, which was the count for May 2025 (Fig. 1).iNaturalist content also finds its way into the Wikimedia ecosystem via Wikidata, a collaboratively edited multilingual knowledge graph, which also has members engaged with biodiversity informatics (Groom and Paul 2020, von Mering et al. 2025). Wikidata currently holds five kinds of iNaturalist identifiers: place, taxon and user IDs enrich information on Wikidata, while photo and observation IDs contribute to structured metadata on Commons. This infrastructure supports and enriches iNaturalist itself, as the platform relies on Wikidata links to select Wikipedia content as the default description of taxon pages.To sum up, we invite the reader to join the community of iNaturalist contributors and Wikimedians: there are a countless ways that biodiversity lovers can join and support this work (von Mering et al. 2025, Meudt and Leachman 2025). By uploading iNaturalist photos, joining WikiProject Biodiversity, and licensing your photos as CC0, CC-BY or CC-BY-SA, you will be joining a welcoming community, and helping to expand the free (as in freedom) biodiversity knowledge on the internet.

HTML

XML

PDF

DOI: 10.3897/biss.9.180280

Authors: Debora Drucker, Filipi Soares, Jorrit Poelen, José Augusto Salim

Abstract: Plant-pollinator interactions play a pivotal role in ecosystem functioning and sustainable agriculture. However, plant-polinator datasets are scattered across various networks, in country-specific initiatives, and stored in isolated silos, making them difficult to access by scientists and decision-makers. By promoting the adoption of Findable, Acessible, Interoperable, and Reusable (FAIR) data standards (Wilkinson et al. 2016) across multiple initiatives worldwide, we are working to transform the fragmented nature of these datasets and make data on plant-pollinator interactions widely available. As the biodiversity community advances towards FAIR data, machine-actionable metadata has emerged as a critical enabler for scalable data assessment, discovery, and reuse. However, while FAIR principles emphasize machine-readability, many datasets are still evaluated manually or lack structured metadata entirely, limiting their integration into global platforms. This study shares practical insights from the WorldFAIR Agricultural Biodiversity Case Study, in which we operationalized machine-actionable FAIR metadata for the review of plant-pollinator interaction datasets (Drucker et al. 2024).We developed a semi-automated workflow to assist in evaluating datasets against the FAIR principles using tools from the Global Biotic Interactions initiative (GloBI, Poelen et al. 2014). The GloBI bots "Nomer" and "Elton" can read structured metadata from standard vocabularies such as Darwin Core (DwC), Ecological Metadata Language (EML), and the Plant-Pollinator Interactions (PPI) vocabulary. Nomer focuses on taxonomic alignment with several taxonomic catalogues, such as GBIF Backbone and Catalogue of Life. Elton extracts species interactions from datasets of various structures and formats, including DwC-Archives.By relying on machine-readable metadata, the bots were able to flag inconsistencies, suggest improvements, and generate repeatable reports across pilot projects in Argentina, Brazil, the African continent, Kenya, Colombia, East Africa, Central Asia, and the USA (for example, Elton et al. 2025). This helped researchers assess dataset interoperability without needing full access to the raw data, a crucial feature given legal and institutional access constraints. To make the data review report readable for researchers, GloBI's bots produce a document resembling a data publication, complete with a title, authors, publication date, abstract, introduction, and other relevant sections. Our results underscore the transformative role of machine-actionable metadata in biodiversity data governance. Automating FAIR assessments not only increases transparency and repeatability but also accelerates the integration of datasets into platforms like GloBI. While human expertise remains essential, tools like Nomer and Elton demonstrate that FAIR assessment can evolve beyond checklists to become dynamic, scalable, and integrated into the data lifecycle. Resources and code are openly available in the online repositories Zenodo*3 and GitHub*2, and are summarized in the GloBI platform*1.To help alleviate the burden of manually reviewing data as part of scientific publication review, we propose deploying domain-specific, automated data review processes that enable researchers to better understand how to make their data easier to review and reuse. Recognizing that publishing reusable, integrated data remains mostly a manual process, we recommend that plant-pollinator and species interaction datasets be registered with one or more infrastructures (e.g., GloBI, GBIF) to benefit from the domain-specific data review services they offer. We suggest that data publishers continue to collaborate on building, maintaining, and improving similar infrastructures to assess and increase the quality and FAIRness of published scientific data.Through the adoption of standards such as Ecological Metadata Language, Darwin Core, Plant-Pollinator Interactions Vocabulary, and Relation Ontology, we aim to enhance the understanding of how plant-pollinator interactions contribute to sustaining life on Earth while ensuring that data is easily findable, accessible, interoperable, and reusable for further research and analysis (FAIR).

HTML

XML

PDF

DOI: 10.3897/biss.9.180398

Authors: Jonas Grieb, James Macklin, David Shorthouse, Christian Bölling, Volker Lohrmann, Michaela Grein, Etta Grotrian, Satpal Bilkhu, Claus Weiland

Abstract: Provenance plays an important role in natural history collections, but capturing this information accurately, proved challenging for legacy digital collection management systems. DINA*1 is being developed to address these limitations through a process-oriented data model that more effectively captures the complexity and context of provenance information (Bölling et al. 2022). DINA is an open-source, robust sample- and specimen-based collection management system in production and developed by an unincorporated international consortium of technologists and practitioners of the natural sciences (Glöckler et al. 2020). Its governance and data models help to foster the adoption of FAIR principles (Findable, Accessible, Interopable, Reusable) and to integrate objects across science domains. DINA's innovative "samplistic" data model records metadata on stepwise, hierarchical processes that generate physical and digital derivatives from parent material samples, accommodating complex real-life sample trajectories, for example: A fossil specimen is acquired by a museum, which contains the remains of several organisms in hardened resin. Later, it is sawed into smaller pieces; some pieces are stored under new catalog numbers, a tiny piece is sent for a destructive C14 (radioactive carbon) analysis (Fig. 1 a).A naturally deceased individual of a mammal species is collected under a material transfer agreement (MTA). Later, subsamples like teeth, bones, and tissues undergo different preparation and preservation processes and are finally stored in specialized collections, each with its own identifier (scheme); a subsample might even be retrieved for sequencing. Strong provenance tracking is required to ensure that regulatory constraints like the MTA are consistently passed down to any derivatives.A soil core is collected from a sampling site and registered together with observational metadata according to the MIxS (Minimum Information about Any Sequence) soil extension standard. Several subsamples are extracted before the remaining core is preserved. The subsamples are subjected to DNA extraction and sequencing to analyze environmental DNA (eDNA). The resulting sequences are then processed computationally to identify the species present (Fig. 1 b).In all examples, the processing history of each subsample can be modeled in DINA so provenance can always be traced back to the original sample. Knowledge from any derivative analysis can be linked back to the parent sample and other derivatives. This ensures that regulatory restrictions applying to a parent material sample are passed down consistently. Besides parent-child relationships, the model can represent other associations, e.g., host, parasite, and vector relationships linking samples in different collections. Each sample in the provenance chain, as well as first-class related objects like projects, collections, events, people, protocols, and storage, has a system-defined globally unique identifier. DINA integrates with external systems where possible via persistent identifiers, including reuse of scientific names from community-curated biodiversity sources through the Global Names Architecture.DINA’s application programming interface simplifies data import and migration and can be used by scientific programming languages like Python or R to access data objects or their relationships. Flexible user-defined “managed attributes” enhance object and derivative metadata when standards do not yet exist. When standards like MIxS or MIDS (Minimum Information about a Digital Specimen) exist, these are incorporated as field extensions.Looking ahead, we aim to further strengthen linkages and provenance tracking in DINA, extending support to model complex bio-geo relationships and enhancing linkage of material samples to external resources such as publications.

HTML

XML

PDF

DOI: 10.3897/biss.9.178659

Authors: Robert Brylka, Jonas Grieb, Ronny Gey, Claudia Müller, Stephan Frickenhaus, Claus Weiland

Abstract: Data repositories play a central role in enabling sustainable research data management and long-term accessibility of scientific information. Within the Earth System Science (ESS) domain, a large number of repositories exist, often operated by different institutions and communities. However, their heterogeneous levels of standardization and varying technical maturity present a challenge for cross-repository data integration. Moreover, researchers seeking reliable and sustainable repositories for data deposition require transparent indicators of trustworthiness and interoperability to make informed decisions. Within NFDI4Earth - the German National Research Data Infrastructure for the Earth System Sciences (Bernard et al. 2021) - the NFDI4Earth Label is being developed to address these challenges. The NFDI4Earth Label provides a framework for assessing the FAIRness of research data infrastructures, referring to the principles of Findability, Accessibility, Interoperability, and Reusability (Wilkinson et al. 2016). The assessment focuses particularly on technical interoperability and organizational trustworthiness (Grieb et al. 2024).Its overarching goal is to support the harmonization of data infrastructures by: Defining concrete metrics that enable a systematic assessment of interoperability and trustworthiness within an evolving ecosystem of digital ESS services, and Providing recommendations to close identified gaps and improve the FAIR alignment of participating repositories.The framework integrates three complementary assessment components:An automated assessment, which evaluates metadata about repositories already registered in the international re3data registry of data repositories. This fosters collaboration with an established community resource and ensures up-to-date information on repository standards and interfaces. The extracted information is evaluated against a set of predefined criteria, with particular focus on metadata standards, availability of application programming interfaces (APIs), and the use of persistent identifiers.A self-assessment, in which repository representatives provide information on governance, sustainability, and institutional policies that cannot be automatically retrieved. Repositories already certified with the CoreTrustSeal (CTS) are recognized as fulfilling this component, reflecting the shared objectives and overlapping standards between CTS and the NFDI4Earth Label. An optional F-UJI-based evaluation which analyses a subset of datasets at the metadata level, providing additional insights into the FAIRness of data holdings (Huber and Devaraju 2021).Each component contributes to a defined set of evaluation metrics that address the FAIR principles. To obtain the Label, repositories must meet a minimum threshold; partial fulfillment results in actionable recommendations for improvement. The entire assessment workflow is implemented within the established OneStop4All service of NFDI4Earth (see Fig. 1), which provides the technical infrastructure for performing and managing evaluations. Repositories that meet the requirements are awarded the NFDI4Earth Label, which is visualized as a digital badge confirming the successful assessment.In summary, the NFDI4Earth Label provides a practical and community-driven approach to harmonize research data infrastructures, fostering transparency, interoperability, and trust across repositories. While it is currently being developed within the Earth System Sciences domain, the underlying concept and assessment framework are designed to be transferable to other disciplines, supporting broader alignment and integration into international data ecosystems such as the European Open Science Cloud (EOSC).

HTML

XML

PDF

DOI: 10.3897/biss.9.180578

Authors: Stijn Cooleman, André Heughebaert, Dimitri Brosens

Abstract: The effective reuse of biodiversity data is essential for informing evidence-based policy. As the Global Biodiversity Information Facility (GBIF) increasingly aggregates data from diverse biological surveys, machine-readable metadata becomes progressively crucial. Structuring survey metadata from textual descriptions into standardized format improves their comparability and alignment with the FAIR (Findable, Accessible, Interoperable, Reusable) principles (Wilkinson et al. 2016). The Humboldt Extension for Ecological Inventories (TDWG Humboldt Extension Task Group 2024) provides a controlled vocabulary to capture key methodological details of surveys and monitoring, enabling more robust analyses of biodiversity indices across spatial and temporal scales.At the Belgian Biodiversity Platform (BBPF), we apply the Humboldt Extension guidelines to enhance two bird monitoring datasets on GBIF. Although both represent counts of observed birds in Flemish areas, they differ in spatial scope and methodology: the Common Breeding bird monitoring in Flanders (Piesschaert et al. 2025) dataset, published by the Research Institute for Nature and Forest (INBO), uses point counts across 1x1 km squares throughout the entire Flemish region, while the Bird census counts at the Zwin Nature Park (Faveyts and Cooleman 2025) dataset, published by the BBPF, combines point and transect counts within one location.In this presentation, we review our practices for selecting and mapping Humboldt Extension terms in both sampling-event datasets. We thereby highlight our findings on how to structure their events’ metadata hierarchically to reflect spatial nesting, sampling effort, and taxonomic scope. For instance, both protocols focus on the same target taxonomic scope at the class' rank (Aves). A critical distinction, however, is that only the Common Breeding bird monitoring protocol excludes certain behavior-based observations of species groups breeding outside the counting area, such as foraging gulls (Larinae) and herons (Ardeinae) or corvids (Corvidae) in flocks flying over. Consequently, the excludedTaxonomicScope term is technically appropriate for mapping that methodological difference. Generically, we propose a schema that facilitates the identification of protocol-specific nuances and supports the more accurate derivation of species densities, abundances, and absences.Implementing the Humboldt Extension to Darwin Core in these real-world cases demonstrates the practical value of structured metadata for enhancing biodiversity data interoperability and reusability. This standardization generally contributes to meeting targets of the Global Biodiversity Framework.

HTML

XML

PDF

DOI: 10.3897/biss.9.178685

Authors: Ana Carolina Mendes dos Santos, Sónia Maria Carvalho Ribeiro

Abstract: Brazil harbors the richest biodiversity on Earth, accounting for 15–20% of all known species worldwide (Lewinsohn 2006). Its territory also encompasses significant cultural diversity, represented by 355 officially recognized Indigenous lands, 253 Afro-descendent territories, and 28 groups of traditional communities such as rubber tappers, coconut breakers and veredeiros (de Mendonça 2009). Despite this potential, the economy remains anchored in an agro-export model based on exotic commodity crops – most notably soybeans, strongly associated with deforestation and rising inequalities (Lobão and Staduto 2020). In contrast, Brazil’s sociobiodiversity, the combination of socio-cultural and biological richness, underpins the potential for a sociobiodiversity-based bioeconomy (Garrett et al. 2024), an approach that integrates traditional knowledge, sustainable use of native species, and innovation into broader economic agendas. Yet, the potential for developing this bioeconomy remains constrained. A central barrier is the invisibility of sociobiodiversity in official statistics.To examine how sociobiodiversity products and local markets are represented in official datasets, we conducted a local-scale study in Northern Minas Gerais, within the Cerrado biome – a global biodiversity hotspot increasingly threatened by agribusiness-driven deforestation. We compared production data from the Brazilian Institute of Geography and Statistics (IBGE) and subsidy data from the National Supply Company (Conab) with figures provided by the Cooperative Grande Sertão, a leading local cooperative that works directly with 400 families across nearly 40 municipalities. IBGE is the official agency responsible for compiling national statistics on the production of non-timber forest products such as açaí (Euterpe oleracea), Brazil nut (Bertholletia excelsa) and natural rubber (Hevea brasiliensis), among others, and Conab is the governmental company that monitors products eligible for public procurement. Our results revealed two main key messages. First, value chains of native species that are socially and economically central to traditional communities are largely absent from systematic monitoring regarding their use, production, or market flows, corroborating previous studies (Porro 2019, Carvalho Ribeiro et al. 2024), and excluded from public policies. Of the 18 native species collected and commercialized by agroextractivists locally, only four (umbu – Spondias tuberosa, mangaba – Hancornia speciosa, pequi – Caryocar brasiliensis, and buriti – Mauritia flexuosa) are tracked in official datasets. Only five species, including macaúba (Acrocomia aculeata), are recognized under public procurement programs, despite their socio-economic relevance. Even for pequi, the most consolidated product, official data is incomplete (Fig. 1). Among the 1,434 municipalities in the Cerrado, its natural distribution area, only 292 (20%) appear in IBGE statistics, as the institute only records volumes above one ton. This implies that in 42% of municipalities, pequi production exists but remains unregistered, while in 38% there is no data at all.Our second finding was the marked discrepancy between official statistics and actual production reported by cooperatives. In four leading pequi-producing municipalities, official records diverged substantially from local data. In Cônego Marinho, for instance, just 30 agroextractivists collected six times more pequi than IBGE registered for the entire municipality in 2023, and in Miravânia, two collectors harvested more than five times the official figure (Fig. 1). These discrepancies may arise from multiple factors, including the informal or subsistence character of many extractive activities. As a result, the national statistics may fail to capture the full economic, ecological, and cultural importance of native species to rural and traditional communities. Understanding these data mismatches is essential for improving the accuracy and inclusiveness of biodiversity and bioeconomy indicators.Our study underscores the need to strengthen local monitoring capacities and create integrated data systems that combine official records with community-based knowledge to unlock the transformative potential of Brazil’s sociobiodiversity. Investments in territorialized data governance are essential to ensure more accurate representation of sociobiodiversity value chains, enabling place-based policies that align biodiversity conservation with inclusive economic development. Strengthening the visibility of native species in data systems is needed not only to improve accuracy but also to ensure informed decision-making, equitable benefit-sharing, and resilience in the face of environmental and social challenges. Making sociobiodiversity visible in data is a prerequisite for building a sustainable and just bioeconomy.

HTML

XML

PDF

DOI: 10.3897/biss.9.180359

Authors: Mamta Sharma, P Abhilash

Abstract: Insects, the most diverse group of organisms, are a global concern due to climate change. They are extensively distributed, playing vital ecological functions. Alterations in their spatial distribution can threaten human health and food security. India hosts over 67,000 insect species (Banerjee et al. 2025), and given the increased average temperatures and projected climate change for the region, understanding the vulnerabilities of these species is imperative. By utilising the widely adopted Species Distribution Modelling (SDM) techniques, insect distributions are mapped worldwide. Analysis of regional modelling studies can identify gaps, and alignment with global efforts to inform effective management strategies.In this study, we conducted a systematic review using Web of Science and Scopus databases for SDM studies utilizing occurrence records from India for known insect orders (Wang et al. 2016). The research maps trend of insect SDMs and delineates key insect groups, models, predictors, occurrence data sources, and applied disciplines. A total of 118 papers were selected for final analysis. The selection method and document list is provided in Suppl. material 1. This review covers global studies and shows a greater application of SDM to insects than previously documented for India (Roy et al. 2022, Sarkar et al. 2024).The SDM studies demonstrate a cumulative rise starting from the year 2008, with a notable increase from 2017 onwards. These studies were classified as global, regional, national, and sub-national with respect to India (Fig. 1). Spatial scope analysis showed that studies at the country level and below are fewest in number (17.80%), whereas majority of studies were conducted at global level (61.86%). Further analysis of the corresponding author’s country revealed that, beyond national level, China is a major contributor to these studies. The literature (80.51%) was identified as most utilized data source for species data. Additionally, while global studies leveraged databases like Global Biodiversity Information Facility (GBIF) and Centre for Agriculture and Bioscience International (CABI), national studies primarily depend on field data. Most of the research were focused on five orders: Coleoptera, Diptera, Hemiptera, Hymenoptera, and Lepidoptera, commonly known as beetles, true flies, true bugs, ants/bees, and butterflies/moths, respectively. These studies modelled distributions of 171 insect species from 10 orders and 90 genera. The majority of research centered on modelling the current distribution (58.25%) of insect invasions in agricultural and forested areas. Among the five orders, Hemiptera had the highest number of studies, in addition to the highest ratio of 1.21 genera per study. Two genera with the highest frequency of studies (7.63%) were of agricultural pests, fruit flies (Bactrocera, Diptera) and armyworms (Spodoptera, Lepidoptera). In the human health area, disease vector mosquitoes (Diptera) were the most studied, whereas in conservation context, studies on bees (Hymenoptera) were most numerous for their role as pollinators. A large proportion of studies (79.66%) used a single model, whereas 12.71% utilized ensemble methods. The most frequently used modelling algorithm was Maxent (88.98%), while use of other models remained below 12%. All studies used climatic data, followed by topographic variables in 27.12% studies.This analysis reveals that there is vast potential for further studying insect diversity through SDMs. Coverage across insect functional roles may improve socioeconomic relevance of SDMs, particularly to address challenges of emerging vector diseases and decline of beneficial insects due to climate change. Further, utilising a diverse array of models for comparative assessment, may increase result reliability. Increased use of diverse data sources at national level can identify data inaccuracies and promote better data publishing practices in the region. The increased data application in SDMs may also encourage higher data generation for various species of local importance through collaboration of scientific experts and database managers. This study underscores the criticality of prioritizing national biodiversity data management as a measure for climate change adaptation and emphasizes the need for advanced biogeographic research to inform evidence-based actions.

HTML

XML

PDF

DOI: 10.3897/biss.9.180293

Authors: James Macklin, Tony Kess, Satpal Bilkhu, Ian Bradbury

Abstract: Genomic technologies provide the highest resolution molecular information on species biology, and can help us understand risks and potential for adaptation among species in a changing environment. The Genomic Adaptation and Resilience to Climate Change (GenARCC) project uses molecular tools to identify ecosystem composition, pathogen and pest prevalence, and adaptive capacity within species. This Government of Canada project purposefully takes a multi-department/agency approach, drawing on complementary expertise and centralized infrastructure to address the complexity of climate change spanning multiple levels of trophic and taxonomic diversity. The GenARCC project’s goal is to develop capacity to use genomics to assess, predict, and adapt to climate change for the protection of Canada's biodiversity, ecosystem resilience, food security, and health. Together, the data and expertise generated in this project represent the largest single national effort and combined dataset to address climate change impacts across species and ecosystems with molecular data. GenARCC comprises research across a diversity of environments and species, covering forest and tundra, arctic, marine, and agricultural ecosystems. Genomic, climate and phenomic datasets have been generated and centralized to better understand impacts to biodiversity across biological levels of organisation. To facilitate storage and analysis of this data, we have deployed a common high performance compute environment allowing participants across multiple Canadian departments and agencies to leverage shared infrastructure, datasets and workflows. Project outputs and inputs are managed through a dedicated sample data research management platform, DINA (Fig. 1). DINA utilizes a highly flexible data model with a core set of fields that can be extended using domain-based standards as field-extensions and/or managed attributes that are user defined, with both able to leverage controlled vocabularies. Strong process-based provenance is maintained for samples and their derivatives. Samples managed, for example, range from individual specimens of bacteria and viruses, plants, insects, fish, and mammals to environmental samples of soil, water, and external and internal microbiomes. A robust API facilitates data migration and customized export supporting data analysis and publication (GitHub Repo). The project has produced several standardized pipelines using Snakemake and Nextflow for efficiency and ease of comparison. An effort was also made to standardize Canadian climatic data for modelling allowing for cross-species comparisons (Marquis 2024).Additionally, project participants benefit from training on scientific computing, as well as genomic methods in discussion forums for data analysis and integration best practices. This access to computing resources and training has supported publication of more than 30 studies that can be found on the GenARCC publication website. Results from this project will inform evidence-based policy to support conservation of biodiversity, as well as management of natural resources and key species across ecological realms.

HTML

XML

PDF

DOI: 10.3897/biss.9.179930

Authors: Christophe Van Neste, Pierre Bonnet

Abstract: With climate change and accelerating biodiversity decline, lawmakers have been implementing legislation to curtail the negative impact of human actors on further decline. A few noteworthy legal acts are the European Union Deforestation Regulation (EUDR, Reg. EU 2023/1115) in 2023 and the updated European Environmental Crime Directive (ECD, Dir. EU 2024/1203) in 2024. Enforcing these legal instruments requires forensic scientific services able to provide traceable, auditable evidence for provenance, habitat change, and chain-of-custody.IntroductionBiodiversity loss represents a systemic risk to human well-being and planetary function; this has been synthesised in the IPBES Global Assessment (IPBES 2019) and earlier conservation science (Soulé 1985. Economic thinking incorporated ecological risk through natural-capital accounting and valuation of ecosystem services (Costanza et al. 1997), which helped motivate legal instruments that translate ecological risk into enforceable rules. The EUDR obliges companies placing forest-risk commodities (FRCs: timber, coffee, cocoa, soy, palm oil, rubber) on the EU market to exercise due diligence to exclude links to deforestation and degradation, while Member States must verify compliance using scientific evidence.Results and operational workWithin the Horizon Europe GUARDEN project (https://cordis.europa.eu/project/id/101060693) we developed a decision-support application (DSA) to inform policy choices with biodiversity data. Use of DSAs in regulatory contexts raises forensic questions: who bears responsibility for decisions, how is uncertainty documented, and can DSAs be adapted to produce court-usable artefacts as evidence? Converting digital policy tools into forensic-grade outputs requires provenance capture, immutable audit trails, model-versioning, and explicit uncertainty propagation.Dr. P. Bonnet has contributed significantly to the monitoring of plants through the Pl@ntNet initiative (Lefort et al. 2024). Together with Dr. A. Joly they cary out research to access the uncertainty in biodiversity modelling and emphasise quantifying identification error, observer bias; these metrics define whether species-level inferences reach forensic viability. Model-sensitivity analyses can be inverted to identify minimal sampling density and taxonomic resolution needed to meet legal evidentiary thresholds in priority areas.Dr. Ch. Van Neste established an atomic-elemental laboratory at Meise Botanic Garden to characterise elemental fingerprints of FRCs and build reference baselines in collaboration with Dr. V. Declerck from World Forest ID (Deklerck et al. 2024). The lab was designed with forensic principles: documented Standard Operating Procedures (SOPs), instrument calibration records, and an open-source traceability platform deployed via Infrastructure-as-Code (IaC) to enable reproducible, auditable instances for other labs or corporate due-diligence systems (https://github.com/AgentschapPlantentuinMeise/MeiseCSI). DiscussionA recurring concern at Living Data 2025 was incomplete metadata and sampling bias in historical and contemporary biodiversity databases; these omissions produce a “data shadow” of unseen assumptions and missing observations that undermines forensic credibility (Leonelli et al. 2017). Forensic utility demands (1) standardised metadata schemas capturing geolocation with uncertainty, timestamp, collector identity, sampling protocol identifiers, chain-of-custody tags, and Certified Reference Material (CRM) usage; (2) reference baselines (elemental, isotopic, genetic) with documented spatial coverage and measurement uncertainty; (3) accredited lab networks and inter-lab proficiency testing; (4) auditable software (immutable logs, versioned models, IaC); and (5) training for expert witnesses to communicate quantified uncertainty against legal standards of proof.ConclusionThis session acted as a call to action. Decision-support tools are necessary but insufficient without forensic design: SOPs, metadata-first capture, uncertainty quantification, accredited analyses, and capacity building for experts and communities. Registering high-quality and comprehensive biodiversity metadata is a minimal requirement to enable enforcement of legal instruments such as the EUDR and the strengthened ECD; the clock is ticking.

HTML

XML

PDF

DOI: 10.3897/biss.9.179381

Authors: Juan Cancino, Paola González-Kother, Jaime Villafaña, Natalia Rebolledo, Leonardo Mondaca

Abstract: iNaturalist*1 is a digital citizen science platform that allows the presence of species to be identified and geolocated. In Chile, it was adopted by the Ministry of the Environment*2 and is widely used by academics, students and the public. This platform has become a valuable educational tool at the Catholic University of the Most Holy Conception (USCS)*3, allowing students to experience being protagonists in the generation of knowledge, which is made instantly available to a community of millions of naturalists around the world. In the process, students learn to recognize the various species that exist in a given study site, a fundamental step to value and promote the conservation of biodiversity. The objective of this work is to share the progress made with 14 projects created since 2021 to the present, brought together under the umbrella project "Naturalists in Action"*4 (Fig. 1), with its concrete application to the registration of biodiversity on the main university campus, and moving forward into projects that involve community organizations. One hundred and eighty-three students from 23 undergraduate and graduate programs at UCSC (Fig. 2) have participated in these projects, with more than 6400 observations, 1190 species registered and identified, with the online collaboration of more than 800 specialists and naturalists in general (Fig. 1). Nearly 70% of the students who learned to use the application in curricular activities at UCSC have remained active on iNaturalist (Fig. 3), showing the interest generated by the knowledge of local biodiversity. In agreement with Di Cecco et al. (2021), a few of them stayed very active (Fig. 3). The most recent actions involve the creation of collection projects for areas delimited by a .kml file, such as the one created for the San Andrés Campus*5. In addition, students and academics have collaborated with citizen organizations that wish to promote the conservation of nature, in places of interest, by virtue of the biodiversity they host, or for the cultural value they represent. Emblematic iNaturalist projects are the ones generated to accompany the conference cycles created in alliance with the Museo de Historia Natural de Concepción (2024), Museo de Historia Natural de Concepción (2025a), Museo de Historia Natural de Concepción (2025b), entitled “Luciérnagas en Concepción”*6 and “Puya chilensis Santuario de la Naturaleza Península de Hualpén”*7. Both include lectures at the Museum and registration of people interested in sighting the organisms of interest in the field. Fireflies and P. chilensis are two types of charismatic organisms that help to promote the conservation of habitats and biodiversity on a local scale.

HTML

XML

PDF

DOI: 10.3897/biss.9.179292

Authors: Laura Salazar Jaramillo, Yiehitson Correa, Daniel van der Post, Aida Vasco

Abstract: The demand for natural products that promote sustainable production chains is growing worldwide. Honey, while not a staple for food security due to its limited nutritional contribution, represents a valuable resource to foster dietary diversity, responsible consumption, and forest conservation. Its production depends directly on access to diverse flora with low pesticide exposure and requires minimal soil transformation or external inputs, making it a low-impact practice. Beekeeping also contributes to the maintenance of pollinator populations, which are key for biodiversity and sustainable agriculture. Consumer interest in the origin and authenticity of honey has grown in recent years, especially for Apis mellifera products (Escuredo and Seijo 2024, Mascarello et al. 2024).Less known is the diversity and economic potential of stingless bees (Meliponini). Across tropical and neotropical regions, nearly 500 species have been described (Grüter 2020), many of which produce honey and act as key pollinators. In Colombia, about 34 species are reportedly used in meliponiculture (Nates-Parra and Rosso-Londoño 2013). This practice is expanding in rural areas, where it complements agriculture and is compatible with family-centered management. Traditional knowledge attributes medicinal properties to stingless bee products (Sgariglia et al. 2010, Rosales 2012). Although clinical validation remains limited, biochemical studies of their honeys have identified distinctive compositions rich in phenolic compounds and other bioactive molecules, providing preliminary evidence of antioxidant and anti-inflammatory properties that warrant further research (Gomes et al. 2022).These properties are linked to microbial communities naturally present in stingless bee honeys. Fermentative yeasts and lactic acid bacteria shape these products, creating unique ecological niches with potential probiotic and bioactive value (da Silva et al. 2024). Microbial assessment is also essential to detect possible pathogens, as not all stingless bee species are recommended for meliponiculture: Trigona species, for example, may incorporate feces or dead animal tissue into nests, raising sanitary concerns (Gómez et al. 2023)Developing markets for stingless bee products requires robust methodologies for authentication and safety. We propose environmental DNA (eDNA) as a tool to address three dimensions of traceability and diversity assessment:Entomological origin of honey. Despite vast bee diversity, taxonomic expertise has declined, limiting our ability to monitor wild bees. DNA-based approaches, particularly metabarcoding, enable verification of the bee taxa involved in honey production—critical in regions with limited expertise. Challenges include primer selection for non-Apis bees, expanding reference databases, and training in molecular methods. Incorporating eDNA and metabarcoding into product traceability could democratize biodiversity monitoring and inform public policy.Botanical foraging spectrum. DNA from bee pollen and honey reveals floral visitation patterns, offering a proxy for ecosystem health. Progress is constrained by incomplete plant reference libraries. Diversity indices—borrowed from microbiome studies—can help characterize floral communities, though interpretation must account for variable foraging behavior across bee species.Microbial diversity and bioprospecting. Stingless bee honeys act as natural “filters,” concentrating microbial communities with industrial, probiotic, or antibiotic potential. Although generally safe, rigorous biosafety evaluation is needed to balance opportunities with risks. Reliable isolation and cultivation methods will be key to characterize diversity, establish reference libraries, and develop pipelines for future bioprospecting.Our pilot project, in the Magdalena Medio region of Colombia, applies eDNA analyses to bees, honey, pollen, and associated microorganisms. The study focuses on three genera of stingless bees maintained in boxes derived from colonies collected in situ in the El Silencio nature reserve of Fundación Biodiversa Colombia, ensuring foraging in a pesticide-free environment with high floral diversity. As one of the first efforts to apply DNA-based tools to stingless bee biodiversity surveys and product authentication, this initiative highlights the need for stronger collaboration, standardized data frameworks, and investment to unlock the potential of eDNA for conservation and bioeconomic development.

HTML

XML

PDF

DOI: 10.3897/biss.9.175153

Authors: Henrik Lantz, Daniel Brink, Rory Crean, Quentin Ågren, Hanna Kultima, Bengt Persson, Johan Rung, Angela Fuentes-Pardo

Abstract: Non-human eukaryotic genome assemblies and annotations are frequently dispersed across repositories, supplementary materials, or remain unpublished. Their access often requires technical expertise and large downloads, hindering usability. The Swedish Reference Genome Portal*1 (SRGP) offers a web platform to easily find, explore, and share genomic data from non-human eukaryotic species studied in Sweden.The SRGP gathers assemblies and annotations into species-specific pages with an interactive JBrowse2 genome browser, metadata, and download links (Fig. 1). This web interface provides user-friendly visualisations aimed at broad user communities, particularly in biodiversity and evolution.Unlike global resources such as Ensembl or the UCSC Genome Browser, which focus on model organisms, the SRGP showcases species studied in Swedish institutions. It fosters community-driven data sharing and broadens access to genomic annotations that are rarely published (e.g., from population genomics), while adhering to Findable, Accessible, Interoperable and Reusable (FAIR) principles. In doing so, it bridges global infrastructures with local research needs, enabling both specialists and non-specialists to make better use of genomic data.The technical architecture of the SRGP (Fig. 2) is designed to make the service easy to use and reuse by others:Simple and secure: Built with Hugo, a static site generator that requires no databases or heavy servers, enabling lightweight maintenance.Lightweight and fast: Composed of prebuilt pages and files for quick loading and easy scalability.Open and reusable: Fully open-source, allowing others to adapt and extend the code.Easy to update: New species or datasets can be added by editing configuration files and rebuilding the site.Portable: Packaged with Docker for easy deployment anywhere, including Kubernetes clusters.User-friendly: Integrates JBrowse2 for interactive exploration of genomic data.The SRGP is designed for simple deployment and maintenance. Genome data and annotations are processed with automated scripts in Docker containers to ensure consistency and reproducibility. The website is built with Hugo, which generates fast, lightweight pages requiring no database. Both data and pages are bundled into Docker images and deployed on a Kubernetes cluster, with ArgoCD handling updates automatically. This setup provides reliable operation on standard infrastructure, easy scaling for new datasets, and secure, low-maintenance access to genomic resources (Fig. 2).In summary, the SRGP makes previously hidden genomic data, especially from local or specialised communities, easy to find and explore through a single, user-friendly browser. It lowers access barriers while remaining simple, fast, and scalable. By unifying scattered resources, it provides a reusable model for other countries, institutions, or research groups. This presentation is available in (presentación disponible en español en) Fuentes‐Pardo et al. (2025).

HTML

XML

PDF

DOI: 10.3897/biss.9.179190

Authors: Nicole Kearney, Kelli Trei, Bianca Crowley, JJ Dearborn, Joseph deVeer, Colleen Funkhouser, Elisa Herrmann, Richard Wragg, David Iggulden

Abstract: In April 2025, the Biodiversity Heritage Library (BHL) released a public statement announcing that the Smithsonian Institution, BHL’s administrative host since 2006, would cease to host BHL on 1 January 2026. This decision placed one of the world’s largest and longest-running open biodiversity data infrastructures at a critical inflection point.BHL is a global consortium of libraries, museums, and herbaria digitizing and openly sharing biodiversity literature. It provides over 63 million freely accessible pages spanning five centuries of publishing; connecting taxonomic data, historical context, and research infrastructures. The sudden withdrawal of institutional hosting threatened BHL’s continuity and exposed the fragility of community-led initiatives reliant on a single institutional anchor.The transition from the Smithsonian required immediate action, careful planning, and international collaboration. Within weeks, BHL convened its partners and launched a Call for Support to identify new administrative, technical, and financial hosts. Open through mid-2025 and soft-closed on 31 August, the call yielded promising offers. To guide decisions, BHL developed Guiding Principles for the Transition, reaffirming its commitment to open access, community-led governance, inclusive collections, modernized infrastructure, and FAIR Principles (Findable, Accessible, Interoperable, Reusable). At the time of writing (September 2025), BHL’s future remains uncertain. Hosting and financial commitments received thus far will provide interim stability for 1–3 years, securing much-needed breathing space to implement a durable governance and funding model, and to address critical questions: how to distribute long-term hosting across multiple organizations and countries; how to secure sustainable in-kind, philanthropic, and grant-based support; and how to adapt BHL’s technical architecture for scalability, interoperability, and continuity beyond the transition.The Call for Support also raised awareness of BHL’s tenuous future among its global user base, sparking widespread recognition of BHL’s value to science and conservation, and surfacing the depth of global commitment to BHL. Support has poured in from partners and the wider community of biodiversity researchers and enthusiasts, ranging from firm offers of technical hosting to pledges of cookies. This solidarity demonstrates that BHL is more than a repository of literature: it is a shared scientific resource, deeply embedded in the global biodiversity informatics ecosystem.This set the stage for a fundraising campaign inviting researchers and the public to support BHL’s sustainability. Concurrently, BHL is pursuing grants to strengthen its technical foundation, upgrade core systems, expand persistent identifier coverage, and enhance interoperability with related infrastructures. While these efforts cannot replace institutional partners, they will diversify revenue streams and strengthen the collective stewardship of BHL.In August 2025, the transition took a major step forward with the establishment of fiscal sponsorship with the Council on Library and Information Resources (CLIR). This partnership provided BHL with an immediate financial and legal home, enabling the consortium to accept donations, pursue grants, and manage staff contracts. It also marked an important shift towards independence. No longer constrained by the requirements and shifting priorities of a large U.S.-based, federally affiliated institution, BHL now has the flexibility to operate as a neutral, international, community-governed infrastructure.The BHL transition is an unfolding case study in navigating infrastructural precarity, illustrating the fragility and resilience of community-led infrastructures, and the importance of distributed governance, safeguarded data, transparent values, sustainable funding, and diversified support. Most notably, the shift to a more autonomous structure has intensified the need to modernize BHL’s platform to ensure continued relevance as bioinformatics advances. The foundation is already being laid for a more agile, distributed, and globally resilient BHL, capable of continuing its mission to connect the world to freely accessible, discoverable, and interoperable biodiversity knowledge.Securing this future requires new partners, sustained investment, and modernized technical infrastructure. Long-term commitments are needed to share responsibility for hosting and funding this global resource. Institutions, funders, and individuals are called upon to safeguard the world’s collective biodiversity knowledge. To follow progress and explore ways to contribute, see ongoing #BHLTransition updates on the BHL blog.

HTML

XML

PDF

DOI: 10.3897/biss.9.178694

Authors: Julien Homo, Kévin Darty, Shadia Kilouchi

Abstract: Context and motivationEcoacoustics is an emerging discipline that studies environmental sounds across spatial and temporal scales to understand biodiversity and ecosystem dynamics, from species behaviour to landscape-level change (Sueur and Farina 2015).Soundscapes encompass biological, abiotic, and anthropogenic acoustic sources, providing a comprehensive and dynamic view of ecosystems. Despite the increasing abundance of recordings collected worldwide, environmental acoustic data remain scattered, inconsistently described, and difficult to reuse due to the absence of shared international standards and interoperable workflows (Sugai et al. 2019).To unlock this latent wealth into actionable biodiversity knowledge, structuring and interoperability are essential.The Sounds of Life initiativeLaunched in 2023, Sounds of Life (SoL) unites over a dozen French institutions bridging ecology, ecoacoustics, geography, and information sciences. Coordinated by CNRS-UMR Passages and labelised by Huma-Num, the French national infrastructure for digital humanities, SoL aims to build a shared FAIR framework (Findable, Accessible, Interoperable, Reusable) for environmental acoustic data. The initiative brings together research groups across France such as CEFE, ISYEB, CESCO, INRAE, Lab-STICC, and others, combining biodiversity expertise with data science and infrastructure design.At the heart of the project is a FAIR-by-design platform that operationalizes the full pipeline: ingest, enrich, annotate, and publish. Through a web-based user interface (Homo et al. 2025), users can import audio files from local or cloud sources. A CSV mapping step lets users attach a first layer of structured metadata on upload, while built-in validators check required and recommended fields for GBIF-ready publication. These imported records are shared in a library, enabling collaboration in workspaces. Vocabulary services power autocomplete and normalization. A customizable spectrogram viewer Fig. 1 enables users to annotate precise time-frequency regions, add comments and provenance. Sets of records, called datasets, can be defined with mappings to Darwin Core terms and packaged as Darwin Core Archives (DwCA) with GBIF compliance indicators. Support for persistent identifiers in publication workflows is planned for the near future. Lessons learned from the pilotBuilding SoL around the FAIR principles revealed key lessons in representing acoustic data within biodiversity standards such as Darwin Core. A first challenge concerned multi-species soundscapes, which are difficult to represent within models that assume a single occurrence per recording. While Darwin Core handles single-taxon recordings well, real soundscapes often include several species and non-biological sounds. In our GBIF/TDWG-aligned profile, each recording is modeled as an Event and each detection as an Occurrence linked to that Event, with the media held via the Simple Multimedia extension. Time-frequency annotations are stored alongside detections, and exports reference the same media record. This design remains GBIF-compatible but reveals friction when publishing dense, multi-species scenes, highlighting the need for a community pattern or lightweight extension supporting multi-label audio.Another key finding concerned vocabulary load. Beyond the about 190 Darwin Core terms, more than 60 additional fields from 10 domain vocabularies were required for taxonomy, geography, devices, and traits. To keep forms usable, we applied progressive field display strategy (required or optional), use-case presets, and GBIF-ready validation. A minimal acoustic profile (about 25 fields) complemented by an extended layer could provide a practical balance between expressiveness and usability.We also faced structural challenges with the lack of a central schema enforcing consistency between Event, Occurrence, and MeasurementOrFact cores. We added consistency rules and export checks to reduce hidden errors. Documenting such patterns in a concise TDWG guidance note would help others align acoustic workflows with existing biodiversity standards.Overall, these lessons underline that interoperability is as much about design and usability as it is about technical standards. Shared profiles, cross-field validation rules, and clear community guidance will make FAIR acoustics both achievable and sustainable at scale.Collaboration, AI-readiness, and next stepsBeyond technical design, SoL enables collaborative curation between ecologists, data specialists, and sound engineers, aligning vocabularies and improving reproducibility across studies, supporting the growing use of environmental sound as a measurable ecological signal.While the current pilot focuses on metadata management and annotation, it is intentionally AI-ready. Consistent, well-structured, and traceable data are prerequisites for automated detection, soundscape classification, and large-scale monitoring. Recent syntheses (e.g., Hoefer et al. (2023)) in passive acoustic monitoring emphasize the need for standardized, comparable datasets to enable robust cross-site analyses.Looking ahead, SoL aims to scale carefully beyond its pilot phase and contribute back to the TDWG/GBIF community through shared profiles and reference datasets. In the short term, through mid-2026, we are seeking input from TDWG and GBIF experts, ecoacoustics researchers, and data publishers toreview our Darwin Core profile for acoustic data (Event/Occurrence + Multimedia) and controlled vocabularies for sound types and scenes;test end-to-end exports (DwC-A to IPT/GBIF) using real multi-species soundscapes;contribute pilot datasets across biomes to evaluate FAIRness, traceability, and reusability; andco-design AI-ready benchmark corpora with clear licensing, balanced labels, and robust metadata.Our goal is to align these efforts with European and global infrastructures so that acoustic evidence becomes durable and comparable at scale.

HTML

XML

PDF

DOI: 10.3897/biss.9.179182

Authors: Cristina Di Muri, Ernesto Azzurro, Pierluigi Strafella, Ilaria Rosati

Abstract: Notwithstanding the global interest and ongoing efforts to manage Invasive Alien Species (IAS), limit their spread and/or their impact on native communities and ecosystems, the path to this goal remains lengthy and fraught with significant challenges.A variety of approaches to support an efficient management of IAS have been outlined, including effective data sharing, iterative collaboration between stakeholders, as well as scientific communication and engagement, among others (Onley et al. 2025). Specifically for data sharing, several guidelines have emerged to ensure the accurate management and communication of scientific information, including the FAIR (Findable, Accessible, Interoperable, Reusable) principles and open science practices. FAIR adoptions are vital to enable the interoperablity and reusability of digital research products (Wilkinson et al. 2016), and open science enhances their transparency and reliability (Bertram et al. 2023). In addition to such broadly applicable guidelines, specific recommendations and standards have emerged for IAS research and data management (Groom et al. 2017; Groom et al. 2019). Implementing effective data management practices alongside enhanced collaboration and communication between stakeholders can support the evidence-based management of IAS.The opportunity to test and validate this integrated approach was provided by USEit - Use of operational synergies for the study and integrated management of invasive alien species in Italy - a multidisciplinary, multi-ecosystem, and multi-taxonomic Italian project focusing on IAS. USEit combines expertise from scientists studying IAS in different ecosystems (aquatic and terrestrial) to find synergies across different data collection and data management methods, ultimately outlining operational and technical recommendations suitable for adoption within and beyond the project. The recommendations were drafted based on an evaluation of the responses to a national survey, subsequently reviewed by national experts of biological invasions, and validated through SWOT analysis, i.e. semi-quantitative analysis of Strengths, Weaknesses, Opportunities, and Threats identified for each recommendation (Azzurro et al. 2025). Such recommendations were adopted and further expanded within USEit. Through this process, detailed technical guidelines for IAS data management were collaboratively defined and incorporated into the project's Data Management Plan (DMP). The DMP was published as a machine-actionable, open access document describing standards, schemas, and repositories used for a diverse array of (meta)data, including occurrences, stable isotopes, elemental composition, remote sensing, DNA metabarcoding, citizen science, and acoustic telemetry (Rosati and Di Muri 2025). Overall, USEit data, services and related metadata are published into open-access repositories and available through the LifeWatch Italy Data Portal and Metadata Catalogue (LifeWatch Italy Data Portal 2025; Tarallo et al. 2025). All digital research products are assigned with persistent identifiers, described by clear metadata and enriched with controlled vocabularies to enable their interpretation, reuse and interoperation. In addition, all digital products generated and published within the project are shared within a unique landing page representing the main access point to USEit research outcomes as well as to communication material and events for public engagements (LifeWatch Italy 2025). These results could be readily translated into practical solutions for IAS management, showing that the implementation of community-shared FAIR practices across multidisciplinary contexts enhances our ability to manage natural resources effectively while fostering new opportunities for scientific and societal growth.

HTML

XML

PDF

DOI: 10.3897/biss.9.179170

Authors: Andrea Tarallo, Alberto Basset, Francesco Cozzoli, Francesco De Leo, Davide Raho, Ilaria Rosati

Abstract: Citizen science (CS) refers to the active participation of non-professional scientists in scientific research, where they contribute to data collection, analysis, and problem-solving (e.g. Jansen et al. (2024)). It has gained traction in recent years as a means to democratise science and enhance large-scale environmental monitoring, particularly in biodiversity and ecosystem research (Danielsen et al. 2024). The widespread use of mobile phones, together with built-in geolocation features, has facilitated the development and use of mobile applications for CS (Hognogi et al. 2023). However, their development and use require substantial effort, including the need for IT expertise to develop and maintain the app, the infrastructural costs associated with data collection and management, private information security, and the increasing demand for generating Findable, Accessible, Interoperable and Reusable data (Wilkinson et al. 2016). In this context, LifeWatch Italy (Tarallo et al. 2025) has developed a CS Platform*1, a comprehensive online resource designed to support every stage of a CS project, from project development to data collection and management. The CS Platform could help overcome the current fragmentation of CS initiatives in Italy by providing a unified framework for managing and sharing information, enhancing understanding of key environmental issues such as biodiversity loss and the spread of alien species. It offers various functionalities (Fig. 1): A knowledge hub dedicated to fundamental knowledge about CS, including its history, methodologies, and key principles. Users can access explanatory texts with bibliographic references and a curated collection of scientific publications searchable via filters. Best practices and tutorials for developing and managing CS projects. This section leads users through the essential steps of project formulation, community engagement, data management, and communication. Each step includes case studies, external resources, and video tutorials. A searchable repository of training materials to help users develop skills in CS methodologies and project management. A catalogue of CS projects, visualised on an interactive map. Users can filter projects by topic, age group, language, country, organisation, and project status. Registered users can contribute by adding new or existing projects. A toolkit allowing users to create mini-websites to showcase and disseminate information and results from their participatory projects, as well as web applications for data collection by citizens in a customisable manner for free. The user who creates a new project on the platform becomes the project administrator of their own website and web application. The project administrator can set up a customised observation-based data collection system tailored to their specific research needs, such as the collection of audio files, pictures, numerical data, etc., along with a customisable list of metadata. Project administrators can access and validate all the user-submitted data and generate analytical reports directly from the web app. The data are downloadable in CSV format compatible with the LifeWatch Italy Data Portal. Fields are described using controlled terminologies such as Darwin Core so that the data are also compatible with international initiatives (e.g., the Global Biodiversity Information Facility).Although the CS Platform has been released relatively recently and is fully operational, it has not yet been widely adopted. Dissemination activities are planned to encourage broader adoption. The objective is not to compete with well-established platforms, but rather to promote the use of the CS Platform in order to support a more uniform and coordinated approach to citizen-generated biodiversity data in the Italian context. This vision is supported by the platform’s integration within the wider LifeWatch Italy ecosystem (Tarallo et al. 2025). Its practical impact has been demonstrated through its use in national projects, such as the Hic Sunt Lupi, which monitors the wolf populations in the Apulia region (Italy). Under this framework, a dedicated web app is currently supporting a CS initiative to record wolf observations. Users can upload a photo to confirm the sighting. The system automatically records geographic coordinates, prompting the observer to confirm or modify them. Optional additional metadata can also be provided, such as whether human interactions occurred, the number of wolves observed, and the presumed life stage.Future developments will focus on building a direct integration of (meta)data flows to the LifeWatch Italy Data Portal and Metadata Catalogue.

HTML

XML

PDF

DOI: 10.3897/biss.9.178568

Authors: Sandra Emry, Ellen Bledsoe, David Hunt, Jason Pither, Jessica Reemeyer, Diane Srivastava

Abstract: Understanding biodiversity trends through time is complex (Johnson et al. 2024), and there is an urgent need to synthesize data across time and space to understand changes. Technological advances have expanded our ability to collect contemporary environmental data, yet understanding biodiversity trends also requires looking to the past. Previously collected data are often not only scattered across formats, especially analog, but also disciplines and institutions. Thus, valuable biodiversity records are difficult to access and underused in research and decision-making (Moritz et al. 2011). The Living Data Project (www.ciee-icee.ca/ldp.html), an initiative of the Canadian Society of Ecology and Evolution, brings together historical and contemporary sources to create curated datasets that inform regional and national biodiversity assessments. Central to data rescue and synthesis is collaboration: datasets are often historical, as are the expertise and networks of the data holders. Building biodiversity knowledge systems, therefore, depends on partnerships spanning institutions, disciplines, and skill sets. Our collaborative workflow moves through four key stages:identifying and collecting legacy datasets, often in partnership with archivists, librarians, and data stewards;digitizing and standardizing data into interoperable formats with the help of graduate students, postdoctoral researchers, and faculty;analyzing biodiversity trends through interdisciplinary collaboration among ecologists, statisticians, and spatial analysts;delivering outputs that reach both science and society, from peer-reviewed publications to policy tools. We illustrate this process through two case studies.A: Historical pesticide spraying in New Brunswick, CanadaStep 1: Forestry pesticide application records were historically managed at the provincial level, with files scattered across four provinces. Some records were digitized, but many remained only in hard copy. Government researchers recognized the value of these datasets for calculating cumulative pesticide loads but lacked the capacity to rescue them. Step 2: A team of graduate students, mentored by a postdoctoral researcher and faculty, compiled, digitized, and standardized pesticide application records. The digitized dataset is archived on the Government of Canada’s open data portal and published as a data paper in Ecology (Heartz et al. 2023). Step 3: A working group (WG) convened ecologists, toxicologists, statisticians, and GIS specialists to integrate the pesticide dataset with aquatic community data from the Canadian Aquatic Biomonitoring Network. This WG collaboratively defined criteria for recovery, evaluated legacy effects of multiple pesticides, and assessed drivers of aquatic community resilience. Step 4: This project produced new insights into ecosystem recovery from pesticide disturbances, culminating in an article in Environmental Pollution (Sugden et al. 2025), presentations at national conferences, and StoryMaps to engage the public.B: The Living Planet Index as a biodiversity indicator.Step 1: The Living Planet Index (LPI) database, managed by the Zoological Society of London (ZSL) and World Wildlife Fund (WWF), aggregates more than 38,000 vertebrate time series worldwide, including over 8,000 from Canada (928 species from 1950–2018). Source data include peer-reviewed literature, government reports, and citizen science. Many Canadian records lacked georeferencing, however, and concerns remained about potential biases due to these gaps. Step 2: Graduate students, postdocs, and faculty mentors worked with WWF and ZSL staff to georeference Canadian LPI time-series data, harmonizing variables such as location, taxonomic resolution, and monitoring methods. The resulting database provided the foundation for two subsequent international WGs on LPI data sensitivity and representativeness. Step 3: Subgroups of these WGs asked: How sensitive is the LPI to different treatments of zeros, short or sparse time series, or outliers?; What is the minimum sample size needed to reproduce the overall trend?; Can indicator species be identified for poorly monitored areas? Step 4: These collaborations generated multiple outputs. A FACETS article (Currie et al. 2022) explores the identification of indicator species; a manuscript currently in review tests the sensitivity and robustness of the LPI (Currie et al. 2025). A toolkit for applying sensitivity analyses to national indices is also in development for conservation practitioners. These outputs ensure that biodiversity indicators remain both scientifically rigorous and policy-relevant, supporting Canada’s international reporting obligations.ConclusionBy combining legacy datasets and interdisciplinary expertise, we can unlock biodiversity data that would otherwise remain inaccessible. A collaborative workflow—identifying, digitizing, analyzing, and delivering—can transform disparate datasets into knowledge that informs both science and policy in an era of rapid ecological change.

HTML

XML

PDF

DOI: 10.3897/biss.9.178687

Authors: Ashleigh Whittaker, Jack Plummer, Nicky Nicolson

Abstract: As we progress towards a globally accessible natural history collection, the ways in which we digitally curate, share and use our data will inevitably change. Digital specimen records enable access by in-country experts, increase the opportunity for specimen enhancement by the scientific community, and improve the breadth of research to which the specimens contribute.Digitisation workflows capture an image often with minimal transcription; they do not include the further enhancement of specimen records, such as georeferencing - the addition of coordinates to text based locality information. Typically used to georeference herbarium specimens, the point radius method assigns a coordinate and a measurement of maximum uncertainty (Wieczorek and Chapman 2020). Some herbarium specimens that lack coordinates still cannot be georeferenced using this method due to poor label data or the unavailability of contextual information.Data derived from herbarium specimens form the basis of species distribution modelling, taxonomic research and IUCN extinction risk assessments. If a specimen record does not have coordinates, then it is often discarded early in the data mining process, highlighting the importance of data enhancement through georeferencing, which confirms the species occurrence in space and time.The Kunming-Montreal Global Biodiversity Framework targets help to guide the most important applications of collection data for biodiversity conservation; Target 4, for example, aims to halt species extinction and protect genetic diversity (Convention on Biological Diversity 2023). Extinction risk assessments of plant species are underpinned by distribution maps derived from preserved specimens and observation records with coordinate information either recorded at the time of collection or through subsequent georeferencing of the specimen. However, over half of all herbarium specimens on Global Biodiversity Information Facility (GBIF) do not have coordinates and so robust extinction risk assessments often require a georeferencing step before applying the IUCN’s criteria. Other data types contributing to extinction risk assessment, such as population size and trend are often lacking and, where available, subjective and restricted to an expert’s firsthand knowledge of the species, making the data untraceable for the wider community (Nic Lughadha et al. 2019, Willis et al. 2003).As highlighted in Bloom et al. (2018), the estimated distribution of a species differs depending on the origin of the coordinate data. For example, non-georeferenced records tend to inflate the estimated distribution of a species. Working with collaborators with regional geographical expertise, and using datasets following standardised protocols, will more likely result in accurate and precise georeferenced records.As herbarium specimen labels can contain qualitative context on collection localities, the process of georeferencing is subjective and so an indicator of confidence and the georeferencer's method is important for end user trust and usability of the record. As more herbarium specimens are digitised, there is a growing need to produce georeferenced locality data at scale.. Although there are robust tools that can be used to supplement manual georeferencing, georeferencing of all specimens in natural history collections currently lacking coordinates is not feasible to resource, as georeferencing is time and resource heavy.The Royal Botanic Gardens, Kew now contributes 5.8 million herbarium specimen records to GBIF, many of which will have between three and six duplicate specimens located in herbaria around the world. Enhancing these records with existing georeferencing efforts accumulated through the completion of several thousand extinction risk assessments will reduce duplicated effort and uncover georeferenced localities that would otherwise go undocumented. This dataset will also help in establishing protocols to apply when georeferencing plant collections in the future, particularly in data-poor tropical regions.

HTML

XML

PDF

DOI: 10.3897/biss.9.178671

Authors: K. Samanta Orellana, Zabdi López, Maura Quezada, Jiichiro Yoshimoto, Lucia Prado, Ana Ambrocio, Nico Franz, Edward Gilbert

Abstract: Symbiota is an open source software for the management and mobilization of biodiversity information (Gries et al. 2014). Symbiota generated portals work as online Collection Management Systems (CMS) and medium-scale aggregators, facilitating digitization and mobilization taks simultaneously. CMS features for live-managed collections include a form to directly transcribe specimen data and upload images, as well as more advanced label editing, loan management, and georeferencing tools. Aggregator features allow administrators to integrate snapshot collections that are not directly managed in the portals. Currently, more than 60 portals are maintained by the Symbiota Support Hub, now based at the University of Kansas. Each portal is managed by a community of biodiversity researchers, often focused on specific taxa or regions. The Guatemala Biodiversity Portal (GBP) is a Symbiota-based platform, dedicated to digitization and mobilization of natural history collections in the country (Orellana et al. 2024), serving academic, non-governmental and governmental institutions. In five years of activity (September 2020 to September 2025), the GBP enabled the digitization and mobilization of more than 58,000 records of Guatemalan collections and biodiversity surveys (Fig. 1). Additionally, the GBP integrates specimen and observation records shared by foreign collections and iNaturalist, with taxonomic resources available in IUCN, Catalog of Life, World Register of Marine Species, and the Biodiversity Heritage Library, among others, increasing accessibility of specialized information for local communities.At least 86% of these records (more than 50,000) have also been mobilized to the Global Biodiversity Information Facility (GBIF) to increase their accessibility (GBIF 2025,Fig. 2). From GBIF, data can be cited in scientific publications and harvested by external platforms like Bionomia (Shorthouse 2020), aiding local researchers to claim credit for their collected and identified specimens. Moreover, the direct data mobilization to GBIF, has facilitated the aggregation of 30 collections and over 50,000 records to this global platform. The digital presence of Guatemalan collections has allowed the use of the data in at least 500 scientific publications, including studies developed by local scientists. Furthermore, Guatemalan collections and their digitized records are now available in the Global Registry of Scientific Collections (GRSciColl), increasing their international visibility. The GBP has also facilitated development of Symbiota documentation in Spanish, as well as online and in-person training in Guatemala. Regular workshops, meetings, and social media communications keep the community engaged and updated about the GBP. Online activities have also reached hundreds of users in natural history collections across Latin America, increasing capacity for the mobilization of biodiversity data in the region. The poster in Spanish presented during the conference is available as Supplementary Material (Suppl. material 1).

HTML

XML

PDF

DOI: 10.3897/biss.9.176413

Authors: Filipi Soares, Antonio Saraiva, Luís Ferreira Pires, Luiz Santos, Debora Drucker

Abstract: The growing demand for high-quality, interoperable biodiversity data exposes the current challenges of standardizing taxonomic information across platforms. While the Global Biodiversity Information Facility Backbone Taxonomy (GBIF Secretariat 2023) offers a valuable reference for species names, translating this information into formal ontology structures for semantic integration remains a complex task. A key challenge lies in the dynamic nature of taxonomic knowledge: as GBIF-BT is continuously updated, ontologies built manually from its data risk becoming outdated or misaligned. Without automated mechanisms for synchronizing with these updates, such ontologies may fail to reflect current taxonomic consensus, limiting their interoperability and long-term usefulness in biodiversity informatics.To address this problem, we developed the Taxonomy OWLizer (TOWLizer), a lightweight web application that allows users to convert species names into Web Ontology Language (OWL) ontologies based on taxonomic data retrieved directly from the GBIF Species API, as detailed in Fig. 1. Users input one or multiple scientific names, and the application fetches, organizes, and outputs a structured OWL file, facilitating taxonomic integration into semantic web projects. OWLizer automates synonym handling (see Fig. 2) and taxonomic hierarchy generation, and reuses GBIF URIs in the code, following linked data principles (Berners-Lee 2006). It aims to lower the technical barriers for researchers working in biodiversity informatics who need machine-actionable taxonomies but may lack programming expertise.The application was implemented using HTML, JavaScript, Bootstrap, and a Flask backend, with ChatGPT-4 providing support for code drafting and debugging, which accelerated prototyping and streamlined development. The tool is freely available through a GitHub-hosted interface with the backend deployed on Render, and a backup archived in Zenodo (Soares 2025).TOWLizer has some limitations though. Its performance depends on free-tier hosting services: the backend may enter a sleep state after periods of inactivity, causing occasional delays, while caching is limited to local storage on a single device and browser. The algorithm is also sensitive to typographical errors in scientific names; a recommended workflow is to combine ChatGPT for spelling verification with TOWLizer for synonym management. Finally, GBIF URLs used in the application do not yet support content negotiation, which constrains their reusability in some semantic web contexts. We tested TOWLizer in the development of a real-world ontology, namely the Agricultural Product Types Ontology (APTO), which was designed to represent agricultural commodities in Brazil. In this context, we addressed the content negotiation issue by replacing the GBIF URLs with URIs from the APTO namespace. These URIs are served via AgroPortal (Jonquet et al. 2018), enabling proper content negotiation.Despite these constraints, the Taxonomy OWLizer has proven to be a practical and accessible way to generate OWL representations of taxonomic data. It provides an initial step toward automating taxonomic ontology construction, highlights the potential of AI-assisted development, and contributes to ongoing discussions in biodiversity informatics about sustainable, interoperable, and machine-actionable taxonomies. More details on the tool development are provided in Soares et al. 2025.

HTML

XML

PDF

DOI: 10.3897/biss.9.179162

Authors: Stanley Blum, David Bloom

Abstract: The International Working Group on Taxonomic Databases (TDWG)*1 was established in 1985 to promote biodiversity data sharing and collaboration among organizations, projects, and people who manage biodiversity information. To achieve these goals TDWG develops standards for publishing biodiversity data and provides a forum for advancing biodiversity informatics.TDWG provides critical elements for developing biodiversity information standards, including: an organizational structure, processes for ratifying and maintaining standards, and the collaboration tools available to participants. TDWG's role as an open standards body creates a complementary partnership with the Global Biodiversity Information Facility (GBIF). TDWG’s organizational structure is established in its Constitution and includes an Executive Committee, a suite of standing committees for on-going operational concerns, a Technical Architecture Group (TAG), and an array of interest, task and maintenance groups that develop and maintain standards. These groups and committees provide TDWG’s administration, maintain existing standards and best practices, and support the efforts of the community to create, expand and implement data standards using current methodologies of computer and information science. TDWG’s Process prescribes how interest and task groups are established to develop standards. Interest groups typically form organically when people with a shared interest in a particular aspect of biodiversity science determine that the science can be better supported by integrating information they create and manage. An interest group is established by writing a charter that describes its purpose and scope, the way of work, and how to become involved. The group designates one or more conveners, who are then responsible for managing the group and communicating with the Executive. The conveners submit the charter to the Executive for approval, and the approved charter is posted on the TDWG website and announced via email and social media. In addition, collaboration tools, such as a GitHub repository, mailing list, and slack channel can be created to support the group’s work. Interest groups explore problems and plan the development of potential solutions. When a plan has taken shape, a task group is established within the interest group by drafting its charter using a similar template and having it approved by the Executive. The essential difference between an interest and task group is that an interest group continues as long as the community’s interest sustains it, whereas a task group works to a schedule and dissolves when the task is finished. Throughout the exploration and planning phases, an interest group should seek advice from the TAG on methods for developing the standard and the framework for documenting it.The TDWG Process also specifies how a standard gets ratified. When a draft is complete the conveners submit it to the Executive, which then appoints a review manager to shepherd the draft through review. Ratification entails both expert and public review, combining the familiar practice of scientific peer review and the internet practice of open source improvement. The review manager summarizes the review outcomes and makes a recommendation to the Executive. When a consensus indicates that the standard is usable and contains no serious problems, the Executive ratifies the standard. The standard is registered and a landing page with a standard suite of metadata is posted on the TDWG website. If the standard contains terms and definitions, URL targets are created in https://rs.tdwg.org. In many cases, a website specific to the standard is established in the TDWG domain, and guides people in how to use the standard. Finally, a maintenance group is established to receive feedback and modification proposals from the community, and ensure that the standard continues to improve. Maintenance groups use a similar, but less cumbersome process, the Vocabulary Maintenance Standard (VMS), to add, change, or deprecate elements of ratified standards.When TDWG was founded, members were focussed primarily on taxonomic compilations; basic species information, such as scientific names and synonyms, geographic distributions, and bibliographies of essential literature. During the 1990s, the number and size of primary biodiversity databases (e.g., natural history collection databases) grew exponentially, and soon it was recognized that a global data infrastructure would be needed to integrate all these data resources and support biodiversity science (Robbins 1996). GBIF was established in 2001 to meet that need, and since then TDWG and GBIF have been engaged in a complementary informal partnership. TDWG’s role in this partnership has been to establish and maintain standards for the key concepts in primary biodiversity data. GBIF's role has been to develop and maintain the technology that enables access to the world's biodiversity data as a well integrated resource. By necessity, GBIF was created through international agreement, as an organization responsive to governments, ensuring that this global data infrastructure is effective and sustained. TDWG, on the other hand, is an open standards organization that enables any member of the global community to participate in developing the data concepts required for biodiversity science.While TDWG's basic standards have helped to integrate immense quantities of primary biodiversity data, they have also imposed limitations. For example, presence-only occurrence data supports only a limited class of methods for species distribution modeling. Our more recent efforts, however, are broadening the kinds of primary data that can be brought to bear on documenting biodiversity and its change over time. These include the Humboldt extension for surveys and monitoring, and the environmental DNA standard (eDNA) for documenting occurrences through the traces of DNA left in the environment. Finally, the Darwin Core Data Package, recently developed by GBIF and currently in review by TDWG, should enable us to mobilize more kinds of data, broaden our community, and support new advances in biodiversity science.

HTML

XML

PDF

DOI: 10.3897/biss.9.177820

Authors: Paula Castellões, Luciana Frazão, Cristine Moraes, Leonardo Marques da Cruz

Abstract: To conserve and restore marine biodiversity, it is essential to critically assess the data generated so far. Recent findings showing healthy phytoplankton cells in deeper, aphotic layers challenge the traditional view that they occur only within the photic zone. This highlights the need to rethink outdated concepts and reassess current sampling strategies. Open, standardized databases offer a valuable opportunity to identify knowledge gaps and inconsistencies, enabling a paradigm shift in our understanding of marine ecosystem functioning.In this study, we used the data from the Environmental Characterization Database (ECD) (Castellões et al. 2021) which compiles results from Environmental Baseline Studies (EBS) conducted within the scope of oil and gas (O&G) Brazilian licensing processes, to analyze the microphytoplankton biodiversity in the water column, both within and below the photic layer. The ECD is a platform for storing marine environmental data generated by projects addressing water and sediment quality, as well as aquatic biota.Among the 17 EBS currently available in the ECD, only four included phytoplankton sampled throughout the entire water column. Although the implementation of EBS is defined by specific guidelines issued by the Brazilian environmental agency (IBAMA/MMA 2019), phytoplankton sampling strategies are not standardized. These studies, carried out in the Southwest Atlantic, reveal the presence of healthy microphytoplankton cells from 200 m to nearly 3,000 m depth. Although cell density decreases with depth, taxonomic richness remained remarkable, evidencing unexpected biodiversity in the aphotic zone. Diatoms, dinoflagellates, and cyanobacteria are the most representative groups in the deep ocean, considering both taxonomic richness and density (cells L-1). Among diatoms Chaetoceros spp., Cylindrotheca closterium, Pseudo-nitzschia cf. delicatissima, and Thalassionema nitzschioides are the most frequent taxa in the aphotic layer. Dinoflagellates are dominated by Heterocapsa spp., Oxytoxum gladiolus, Prorocentrum spp. and the order Gymnodiniales, while cyanobacteria are represented mainly by the orders Nostocales and Oscillatoriales. Coccolithophorids and silicoflagellates (Dictyocha fibula) are also present.The occurrence of microphytoplankton in aphotic layers was previously reported in mesopelagic (300–500 m) and bathypelagic waters (1,000–3,000 m) of the eastern Indian Ocean (Liu et al. 2022), and between 2,000 and 4,000 m depth across subtropical and tropical areas of the Atlantic, Indian, and Pacific Oceans (Agustí et al. 2015). More recently, the presence of healthy microphytoplankton cells down to 1,000 m depth was also recorded in the Santos Basin, Brazil (Frazão et al. 2025). These results suggest that opal ballasting plays an important role in transport of the particles containing these cells to the deep-sea (Cael et al. 2024), also providing evidence of vertical connectivity between surface and the deep ocean, and reinforcing the role of microphytoplankton to the biological carbon pump. They further suggest an important structural carbon flux that has not been fully recognized due to sampling being restricted to the euphotic layer, highlighting the importance of expanding phytoplankton sampling strategies beyond the photic layers. Finally, our study reinforces the value of open-access environmental databases for achieving a more comprehensive understanding of deep-ocean biodiversity, its role in biogeochemical cycles, and its ecosystem services.

HTML

XML

PDF

DOI: 10.3897/biss.9.177490

Authors: Elie Saliba, Jutta Buschbom, Sharon Grant, Janeen Jones, Ben Norton, Maarten Trekels, Kate Webbink

Abstract: Internationalization of standards documentation is essential for pursuing global interoperability through the adoption of data standards that can be understood and competently applied throughout the world and across sociocultural contexts. Ratified by Biodiversity Information Standards (TDWG) in 2024, the Latimer Core (LtC) data standard focuses on the representation and discovery of natural science collections (Woodburn et al. 2022). The first complete translation of the LtC standard documentation*1 was published in June 2025 into French facilitating access to the standard for francophone communities.Translating biodiversity data standards such as Latimer Core into French presents a series of intertwined linguistic and technical challenges. The rigor of the translation effort depends on consistent terminology inside a given standard, and is achieved through careful reuse of formulas, such as 'recommended best practice', and the support of translation management tools such as Crowdin*2 to ensure uniformity. In addition to intra-standard consistency, the reuse of Darwin Core terms, which were translated prior to the ratification of Latimer Core (see Saliba et al. (2025)), requires caution when retranslating definitions, to ensure homogeneity across standards.Aside from linguistic elements discussed in part in Saliba et al. (2025), challenges like documenting translation work remain. The latter is largely informal, relying on collaborative platforms and personal notes, which underscore the potential need for more structured, reproducible workflows, especially in the context of multiple translators working together on a given language, notably those with strong regional variants. Similarly, no universal threshold has been defined for the minimum content needed to achieve a “functional” translation. An incremental approach, beginning with labels and definitions and progressively expanding to webpage elements such as headers and footers, non-normative complementary information and supplementary documentation seems to emerge as good practice. To address these issues and others, a recommendation document aimed at defining good practices and workflows for translating standards is being prepared.Finally, the Latimer Core maintenance group is experimenting with having a point of contact for translation to act as a bridge between translators, the standard maintenance group, and users. The point of contact can answer domain-specific questions, gather feedback from users and report errors to the relevant translator.Ensuring that TDWG standards are available in French is a good way to broaden participation among underrepresented scientific communities across Africa, the Caribbean, the Pacific and other francophone regions. Beyond opening doors for these audiences, the translation process itself offers a unique opportunity for contributors to deepen their understanding of a standard while making it, and subsequently connected standards, accessible to others. Far from being a mere technical task, translation is an intellectually rewarding and collaborative endeavor that amplifies the global relevance of TDWG’s work—ultimately enriching both the standards and the communities they serve.

HTML

XML

PDF

DOI: 10.3897/biss.9.178513

Authors: Caitlin Thorn, Jessica Zimmermann

Abstract: At the Museum für Naturkunde Berlin (MfN), the Digital Catalog Infrastructure (DCI) team is experimenting with project management tools that extend beyond their traditional use in the technology sector. The team, comprising developers, data and media managers, and digitization managers, works collaboratively to build and maintain the systems that support the museum’s complex digital data landscape. To manage this multifaceted work, the team introduced scrum practices within a workplace not typically structured for agile methodologies. Guided by certified scrum masters, agile principles were adapted to fit the museum context through the use of short iterations, structured meetings, and sprint planning boards designed to organize priorities and improve communication across diverse groups. This work presents the lessons learned from implementing scrum in a museum setting. It highlights what succeeded, challenges encountered, and how the balance between flexibility and structure was maintained to keep the team aligned.The five pillars of a museum: collecting, preserving, researching, exhibiting, and education (Deutscher Museumsbund e.V. et al. 2023) form the basis of the need for project management at the MfN. Curators, developers, conservators, researchers and many others work together, communicating across different professional languages. Museums are not just about collections, they are a complex ecosystem of people, data, and projects, and project management helps connect these moving parts. Museums are also becoming increasingly digital, and museum work is long-term, but often resources are only secured in the short term.The MfN faced several challenges. Multiple teams of developers carried out complex projects independently with little communication between them. Insight into what projects others were doing was limited, resulting in repeated efforts and delays. Then in 2023, a cyberattack hit MfN’s systems and exposed the need for better visibility and coordination among the groups and teams. Following the emergence of a new project requiring collaboration among previously distinct teams for the creation of a digital data catalog, it became clear that a solution was needed to improve collaboration.The DCI team began by exploring several project management frameworks. It was quickly decided that agile project management might be suitable for the team. Agile project management is an approach that originated in software development and is often used in software teams. Its focus is on flexibility and teamwork to deliver working products (Beck et al. 2001).Specifically, the team decided to adopt scrum as a structure for its work following several workshops. Scrum is a type of agile project management. It uses short work cycles, called sprints, and involves regular team meetings called stand-ups. It also includes reviews and retrospectives to reflect on completed work and how the team functions. There are a few core positions within the team: product owners (POs), who maintain the backlog of tasks following communication with stakeholders, and scrum masters (SMs), who facilitate meetings and remove barriers to work (Karlesky and Vander Voord 2008). Within the DCI team, scrum masters were responsible for hosting meetings, enabling clear communication, and removing barriers to work. For the DCI team, working with agile and scrum meant working in short, clear cycles following three week sprints, rather than following long, rigid plans and providing the ability to adjust as needed.Within the process, the DCI team faced the challenge of adapting scrum for a museum context. Key differences between a typical scrum and that in the DCI team are highlighted in Table 1 below.The DCI team did not copy scrum exactly but translated and adapted it to make sense for the museum context. Since using scrum, the team’s development successes became clearer and there was a notable increase in communication and collaboration among developers. The team gained a shared awareness of projects, both in and outside the team, as information was passed on more freely during daily stand-up meetings. There was quicker coordination when problems arose, and more teamwork and accountability. Structured retrospectives provided opportunities for reflection on team processes and progress.There were, of course, hurdles along the way: the sudden increase in meetings was overwhelming at first. It took a lot of work to break down the tasks for the backlog. Understanding the timing of these tasks remained an ongoing challenge. There was agile jargon to become familiar with and a need to establish in-house definitions. As the only team at MfN working agile, the process was not always straightforward. Training sessions for the whole team, led by external trainers, established a common foundation for the project, followed by additional training for POs and SMs. Reflecting on the process, it made sense not to apply the entire scrum framework at once. Methods could be adapted to each institution and tailored to its context, focusing on what the team needed. The DCI team needed greater cohesiveness. Alternatively, if a team needed to quickly produce products, those aspects of agile and scrum could be highlighted.Agile can work anywhere, including a museum. The DCI project at MfN bridges a traditional research-focused workplace with modern development methodologies. The initiative demonstrates that, with thoughtful adaptation, agile methods can bring clarity, focus, and momentum to traditional scientific institutions.

HTML

XML

PDF

DOI: 10.3897/biss.9.177966

Authors: Jutta Buschbom, Elizabeth R. Ellwood, Sharif Islam, Eva Häffner, Alexandra Pavlova

Abstract: Effective decision-making on biodiversity restoration would greatly benefit from baseline data on intraspecific genetic diversity, the ability to integrate it across species for each location, and efficient systems for monitoring changes of genetic diversity in response to management interventions and environmental dynamics. This requires large sets of well-curated information-rich FAIR (Findable, Accessible, Interoperable, Reusable) Digital Objects (FDOs; Schultes and Wittenburg 2019), implemented as, for instance, Digital Extended Specimens (DES), which are representing digitized field samples, their derived genomic data and associated information (Hardisty et al. 2022). These use-case-driven sets of structured (meta)data from many providers need to be merged, modified and further extended on demand. Existing workflows and work environments have to be redesigned to accelerate this process and achieve seamless integration to be able to scale to the needs of efficient and effective worldwide monitoring under the United Nations Kunming-Montreal Global Biodiversity Framework.Contributing to data and infrastructure development processes in support of global monitoring is also the global effort to generate high-quality reference genomes. The Earth BioGenome communities, including the European Reference Genome Atlas (ERGA) and Biodiversity Genomics Europe (BGE) communities, drive the development of standardization and harmonization for well-designed sampling and comprehensive FAIR and CARE (Collective Benefit, Authority to Control, Responsibility, Ethics) (meta)data (Buzan et al. 2025). Their goal is reproducible analytical pipelines that can be assembled on the fly for implementing sophisticated statistical approaches and algorithms.Such research-focused pipelines prepare the development and application of globally adopted workflow templates for the calculation of Essential Biodiversity Variables (EBVs). These templates are under development within the Group on Earth Observations Biodiversity Observation Network (GEO BON) (Lumbierres et al. 2025). Their output can be used for globally aligned and interpretable planning, monitoring, reporting and reviewing (see CBD/COP/16/L.33).Achieving global agreement on a (small) set of jointly used interoperable vocabularies and ontologies for (meta)data and machine-actionable operations is a communication, community-capacity development, and negotiation process that requires significant resources, engagement across sociocultural groups and geographies, patience and time. Ongoing processes towards these goals continue to be organized and promoted by, e.g., Biodiversity Information Standards (TDWG), the Global Biodiversity Information Facility (GBIF), GEO BON, and the Ocean Biodiversity Information System (OBIS), as well as large continental networks.We propose to bridge the gap before standardization and harmonization are in place and thereby facilitate and accelerate such efforts. Taking a pragmatic approach, our objective is to contribute a lightweight "pocket" dataspace that provides interoperability and data governance in connection with a digital platform for global genetic monitoring (Fig. 1).The pocket dataspace would allow existing platforms and tools to be connected easily through community-provided mappings between workflow element-specific formats, terms, data and operations stored in an open repository. This approach would enable users to take advantage of the core strengths of existing software products and the expertise of their associated communities, while quickly sharing data and their work between specialized solutions. At the same time, data would be FAIRified and CAREd-for, promoting attribution, transparency and responsibility. The functions of the pocket dataspace can be prerequisites for a transition to machine-actionable operations usable to agentic AI. As a general-purpose interlinking and translation component, the pocket dataspace aims to be the missing link between distributed, federated, non-standard-compliant and undocumented data, governance regimes, provenance logs and software output, and the need for transparent, well-governed and versatile conservation applications.One of these conservation applications will be the proposed platform for monitoring global genetic diversity. The platform will aggregate and visualize externally-linked population-genetic data and summary metrics that are the results of analysis pipelines enabled by, e.g., the pocket dataspace. Its aim is to provide visualization and support dataset and analysis management for local to global conservation efforts. It would store uploaded or linked genetic diversity metrics, perform selected automated analyses for continuously updated genetic diversity measures, as well as provide a starting point for user-designed analyses. The objective of our initial use case is to analyze three basic measures of population-genetic diversity based on genome-wide sequencing data as a first step towards operationalizing genetic monitoring at scale. Together, the pocket dataspace and monitoring platform for genetic diversity data have the goal to support a digital ecosystem that is foremost flexible, requiring low investments by users, and be able to quickly integrate both inter- and transdisciplinary data as well as existing powerful platforms and well-tested analytical pipelines and functionality.

HTML

XML

PDF

DOI: 10.3897/biss.9.177440

Authors: Cristiana Serejo, Mariane Targino, Fernanda Amaral, Aluf Elias, Marcus Vinícius De Almeida

Abstract: Museums and their natural history collections (NHCs) have become increasingly interconnected through the digitization of information about specimens (Bakker et al. 2020). In the last two decades, some South American countries have joined biodiversity data digitization programs and projects, including Brazil (Nelson and Ellis 2018). In this sense, initiatives such as Brazilian Biodiversity Information System (SiBBr), a country-level branch of the Global Biodiversity Information Facility (GBIF), aim to encourage and establish ways for the digitization of biological collections. Established in 1818, the Museu Nacional of the Universidade Federal do Rio de Janeiro (MN), is the oldest scientific institution in Brazil and ranks among the most important repositories of NHCs in the Global South (Serejo et al. in press). Since its foundation, the MN has played an important role in promoting and disseminating public knowledge, such as culture, education, and scientific research (Rodrigues-Carvalho et al. 2012). This role is extremely important, given that Brazilian institutions often face resource limitations and structural vulnerabilities (Marinoni et al. 2024). On September 2, 2018, a catastrophic fire engulfed the Palácio de São Cristóvão, destroying significant portions of its collections (Zamudio et al. 2018). Despite this devastating event and the loss of the vast majority of specimens deposited in the NHCs of that building, some holdings had been extensively digitized and had their information preserved through the SiBBr project, such as the entomological collection (De Almeida et al. 2021). Collections located in other MN's buildings and not affected by the fire, such as the vertebrate collections, have continued to be digitized to the present day through the COLBIO project ("Restructuring the Museu Nacional/UFRJ through its Biological Collections: Digitization, Curation, and Metadata Management for Open Access"). The objective of this work is to publicize the digitization progress of the MN’s NHCs, as well as the availability of their open data through the SiBBr. All data studied are based on the MN institution and its NHCs. In 2022, the COLBIO project was launched, alongside the creation of the Laboratory of Digital Collections (COLDIGI), a hub for developing and managing digital infrastructure. New photographic stations and digitization protocols were implemented, and long-standing databases previously managed in spreadsheets are being migrated to Specify 7. This marks the first fully institutional initiative dedicated to the digitization of MN’s NHCs, with an initial focus on rare, historical, and type material. Four major collection areas are currently involved: entomology, invertebrates, vertebrates, and botany. By December 2024, 22,081 images had been captured, representing 4,544 specimens across 3,743 species. Among these, 3,805 images correspond to nomenclatural types (783 specimens and 645 species). Thousands of entomological type images previously digitized under the SiBBr project were crucial for digitally preserving part of the collection, as nearly all physical specimens were lost in the fire (see De Almeida et al. 2021). Final image editing has been completed for zoological collections digitized under the COLBIO (Fig. 1) and SiBBr (Fig. 2) projects, as well as individual initiatives, totaling 2,768 images from seven collections. A key goal of the COLBIO project is the migration of zoological collection databases to Specify 7. Currently, 18 databases are hosted on the Specify Consortium’s cloud service, which provides automated daily backups. To date, 14 datasets have been published on the SiBBr, totaling 510,956 publicly available records, most without images. Besides SiBBr, an institutional platform for MN’s collections is under consideration. The COLBIO project and related digitization initiatives demonstrate how digital technologies now function both as safeguards and as strategic extensions of the MN’s scientific mission, ensuring the preservation, accessibility, and resilience of biological data, even under catastrophic circumstances.

HTML

XML

PDF

DOI: 10.3897/biss.9.177306

Authors: Catherine Jardine, Denis Lepage, Kyle Horner

Abstract: Robust, accessible data are critical for the conservation of biodiversity. Biodiversity monitoring data is collected by numerous professional and volunteer projects all over the world, but finding and accessing these disparate datasets for use in research and conservation is problematic. Integrating individual datasets into a metadatabase that abides the FAIR principles by being Findable, Accessible, Interoperable, and Reusable may be the best way to maximize the usefulness of these abundant data, but barriers such as cost and a lack of access to necessary tools may prohibit this. Birds Canada seeks to resolve this long-standing issue with the NatureCounts mobile app. The app’s extremely flexible and customizable data architecture accommodates a wide variety of data collection methodologies. Users define protocols tailored to their monitoring project, then the app dynamically renders user-friendly and intuitive data entry forms specific to each of the project’s protocols. In this way, it can provide mobile data entry support to almost any standardized avian survey with just a few hours of setup. The app includes many innovative tools, such as a first-of-its-kind point-count feature that uses satellite imagery to improve and digitize one of the most important methodologies for biodiversity monitoring. Data collected through the app flow into the FAIR-abiding NatureCounts database, which—at 360 million records—is the largest biodiversity database in Canada. Data in NatureCounts can be made available via browser-based tools, an R package, and several data products. NatureCounts supports over 9000 data requests annually from the conservation community. Data from NatureCounts have been used for species assessments, land use planning, impact assessment, climate change mitigation, and over 4200 scientific publications to date. Flexible data access permissions support the security of sensitive records and Indigenous data sovereignty.Where data permissions allow, datasets collected through the app can also be integrated into the Global Biodiversity Information Facility (GBIF). By improving accuracy in several key areas and removing the tedious step of data transcription, the app offers immense time and cost savings and reduces errors. Its user-friendly interface (Fig. 1) walks observers through protocols, ensuring consistency and increasing accessibility for staff and volunteers of varying skill levels. These benefits incentivise programs to adopt the platform. In doing so, their data are integrated into a FAIR repository at the outset, eliminating the need for re-formatting and integration, steps which are often overlooked or unfunded. By providing software tools across the data pipeline, from collection through application, the NatureCounts platform seeks to support projects in collecting robust data and meeting the FAIR principles of data management. This support allows projects of all sizes and capacities to make a positive contribution to the data landscape and to achieve the targets of the The Kunming-Montreal Global Biodiversity Framework (Convention on Biological Diversity 2022).

HTML

XML

PDF

DOI: 10.3897/biss.9.177303

Authors: K. Samanta Orellana, Edward Gilbert, Katelin Pearson, Greg Post, Mark Fisher, Nico Franz

Abstract: Symbiota is an open-source software for the management and mobilization of biodiversity data (Gries et al. 2014). Online Symbiota portals, often with regional or taxonomic scope, are used by more than 1,200 natural history collections to directly digitize and manage their specimen records. A simplified form with all relevant terms from the Darwin Core standard (Wieczorek et al. 2012), allow users to easily transcribe the information about the specimens (Fig. 1). At the same time, Symbiota portals facilitate the integration of images and other linked resources to each record, increasing the digital information available for each specimen (Hardisty et al. 2022). Symbiota portals can also incorporate snapshots of external datasets, expanding the accessibility of biodiversity information for the community of users. Data integration and specialized tools in Symbiota portals enable the generation of interactive resources to advance biodiversity research (e.g. checklists and taxonomic keys). Most Symbiota portals have been funded by projects in the United States, and are managed by the Symbiota Support Hub (SSH), based at the University of Kansas. However, easy access to existing portals has encouraged their use by collections worldwide. Translations to Spanish, French and Portuguese also increase the participation of international users in Symbiota communities. Moreover, by following biodiversity data standards (e.g., Darwin Core) and accessibility principles (Wilkinson et al. 2016), collections digitized in Symbiota portals can be further mobilized to regional and global aggregators (e.g., Integrated Digitized Biocollections, Global Biodiversity Integration Facility), increasing the reach of the data.In Latin America, at least 140 collections in 12 countries are actively using Symbiota portals to manage and mobilize their specimen data. Portals like Neotropical Flora, Mexican Herbaria Network, and InvertEBase are currently hosting more than 500,000 botanical, zoological and paleontological specimen records. Communities in portals like Consortium of Lichen Herbaria and Guatemala Biodiversity are also developing interactive resources for the study of local biodiversity, including checklists for specific taxa (Yánez-Ayabaca et al. 2023) or regions (Castillo-Cabrera et al. 2025). Furthermore, collections from Guatemala, the Dominican Republic, and Costa Rica are sharing more than 64,000 records to GBIF directly from Symbiota portals. Data mobilization to this global aggregator has increased data access, generating over one thousand citations in scientific literature for these natural history collections. Additional perspectives about the use of Symbiota portals in Latin America, and their potential for biodiversity data integration and use were shared during the presentation. Original slides in Spanish are available as supplementary material Suppl. material 1.

HTML

XML

PDF

DOI: 10.3897/biss.9.175400

Authors: Alinta Krauth, Jason Nelson

Abstract: 1.0 IntroductionWe present two works of digital poetry that explore the use of ecological sensor data as potential co-authors. By drawing on forest microclimate data, these works position nonhuman systems as active participants in poetic production, and we consider the ways in which this destabilises traditional notions of authorial intentionality. As ecological data infrastructures become increasingly complex, there is a need to explore their role beyond scientific domains (Åsberg 2024). In tandem to this need, scholars in the arts and humanities insist on the development of cultural and creative practices that recognize the agency of non-human ecological actors (Morton 2010, Krauth 2024, Bennett 2010). The practice of digital poetry offers an experimental mode for translating ecological data and processes into culturally resonant experiences.2.0 MethodologyWe explore this through a creative practice-led examination of how ecological data can be transformed into multisensory poetic experiences, experimenting practically with ways in which digital poetry may unveil environmental actors as co-authors of poetic texts and participants in creative production. Our work uses ecological data provided by the Norwegian Institute for Nature Research. 3.0 Background Digital poetry is poetry derived through computational processes and networked media (Rettberg 2018). Since the mid-twentieth century, poets have experimented with algorithms and software to expand their practice beyond the printed page (Funkhouser 2007, Glazier 2002). Meaning in such works arises through technotextual interplay between code, medium, and the reader (Hayles 2002). Reading digital poetry requires nontrivial engagement, sometimes closer to that of a computer game, through which it destabilizes traditional notions of authorship (Aarseth 1997). Poststructuralism—a philosophy emphasising knowledge and language structures as fluid (Barthes 1977, Foucault 1991)—further unsettled notions of the author in the twentieth century, and this shift is seen in the human/machine co-production often associated with digital poetry (Flores 2019).The use of biosensing data as a material in digital poetry allows creators to consider ecological actors as non-representational generative systems. For example, Ackerman and Richert’s Getting Plants to Write Poems (Ackerman and Richert 2014) demonstrates this by mapping plant bioelectricity into textual outputs, while Nelson’s Vholoce (Nelson 2007) and Matsuda’s Every Thing Every Time (Matsuda 2017) each use data pulled from weather and city sensors to structure verse and determine poetic outputs. Arguably, authorship in these works is not just an interaction between code, medium, and reader (as suggested by Hayles (2002)) but also of ecological conditions and nonhuman living entities that can shape digital poetry production through their actions or existence.4.0. Discussing two pilot projects Two digital poetry interfaces have so far been produced, which showcase the potential for human/nonhuman co-creation with ecological systems:4.1 (very) small stories A web-application interprets moisture and temperature readings of hyper-specific locations across Norway's Sigdal, Kvam and Grong forests. Each location holds four sensors that produce separate readings. A codeset reads and interprets these four sensor readings as combinatory fragments of text, producing fictionalized translations of the forest's thoughts, which change each time the temperature changes by 0.1 Celsius, or moisture changes by 0.1%. The resulting poetry is intentionally visually unstable, echoing the difficulty of translating non-human data into a human language (Krauth 2025). Instead of treating each data reading as a sentence-equivalent, the system parses each sensor as a quarter of a line, enabling a full line of poetry based on all four sensor readings(Fig. 1). 4.2 we are both data we are both data is an installation that draws from the same datasets as (very) small stories, but further includes a bespoke galvanic skin response-to-MIDI touchable controller, where moisture readings from the skin of human interactors is interpreted on-screen into poetic text, and juxtaposed next to poetic text derived from the moisture readings of our forest sensor data. In this setup, both human participant and climate data hold a similar level of creative authorship over the work's visual and textual outcomes; while it may be a passive writing process for both, it aims to allow for the recognition of similar agencies, under the metaphor of becoming physically in sync with the microclimate of a forest(Fig. 2). 5. Early ConclusionsDigital poetry demonstrates how creative practice can operate as a bridge between data and culture. But it also opens up questions regarding the role of nonhuman agency in poetic authorship. By allowing microclimate data to shape poetic forms and outcomes in our two pilot projects, the intimate climates of forest locations have emerged as participants in poetic construction. This adds to a growing awareness in posthuman literature studies (or indeed returns to many Indigenous ideologies) that the ecological and climactic processes and systems around us influence digital, technological and creative productions.

HTML

XML

PDF

DOI: 10.3897/biss.9.173953

Authors: Folco Soffietti, Gianni Sinni, Francesco Musco

Abstract: 1. IntroductionDigital transformation is reshaping how environmental governance documents are communicated, accessed, and used. Marine spatial planning (MSP), increasingly adopted across Europe to coordinate sea uses through ecosystem-based and integrated approaches (Zaucha et al. 2025), is framed by Directive 2014/89/EU on MSP (European Union 2014) and relies on interoperable spatial data infrastructures, as promoted by the Directive 2007/2/EC of the European Parliament and of the Council of 14 March 2007 establishing an Infrastructure for Spatial Information in the European Community (INSPIRE) (European Union 2007). In France, the evolving relationship between Marine Protected Areas (MPAs) and MSP reflects complex institutional and spatial dynamics (Trouillet 2020, Trouillet and Jay 2021). In this context, the authority of the Parc naturel marin de l’estuaire de la Gironde et de la mer des Pertuis transposed its management plan into an interactive web platform. This study examines the design, usability, and governance value of this portal, asking how digital design and spatial data management align with EU MSP standards and human-centred design principles.2. Employed MethodologyDocument analysis of the official management plan (Office Français de la Biodiversité 2018), its cartographic annexes, and the SIMAtlantic (Supporting the Implementation of Maritime Spatial Planning in the Atlantic) report describing the web tool’s development (Alloncle et al. 2021).Semi-structured expert interview with a former officer of the French Office for Biodiversity, focusing on motivations, target users, legal relationships, participatory processes, technical choices, and reflections on governance value.Systematic exploration of the portal, guided by usability and user-experience literature (Soares et al. 2022) and ISO 9241-210:2019 standards (International Organization for Standardization (ISO) 2019).3. ResultsThe portal is not a legal document but an interactive translation of the official plan. The paper plan remains the regulatory reference, while the website serves communicative and operational purposes for key stakeholders.The platform uses a dual-screen layout: a left pane with a Web-GIS viewer and a right pane with factsheets. Users explore ecological, socio-economic, and governance stakes (enjeux), objectives (finalités), actions, and monitoring protocols. Relational links connect goals, measures, and cartography. Exportable maps and shareable URLs enable decision-making uses.Developed in the framework of the SIMAtlantic project (Alloncle et al. 2021), the portal addresses challenges of lengthy, static marine plans: opacity, updateability, and difficulty in visualising spatial interactions. Its value lies in making governance rationale visible.The portal balances functionality and simplicity. It aligns with ISO 9241-210:2019 principles (International Organization for Standardization (ISO) 2019), being task-oriented and accessible. Digital features include real-time exploration, overlays of ecological and socio-economic variables, and relational navigation between management elements. Yet, fixed layout and performance issues can hinder mobile or low-bandwidth users.4. Discussion and conclusionThe Gironde portal illustrates a new genre of digital planning artefacts, which both visualises and operationalises spatial strategies. Its relational architecture reinforces the interconnectedness of stakes, objectives, and actions in a transparent way. It demonstrates how digital tools can recalibrate legal design and governance: although not binding, the portal fosters legal consciousness by making planning rationales traceable and accessible (Trouillet 2020). In fact, its design has altered how planning is accessed.This hybrid status—between normativity and informativity (Pasa and Sinni 2024)—mirrors debates in legal design and digital governance. The platform also shows how MSP and MPA governance can converge through shared digital infrastructures (Trouillet and Jay 2021), accommodating multiple user typologies and aligning with EU MSP implementation challenges (Zaucha et al. 2025). Importantly, it confirms that design is not merely aesthetic: usability and cognitive load management (Soares et al. 2022) are fundamental to making spatial planning understandable and actionable. Relational navigation and modular presentation have made complex governance strategies more accessible and transparent. The case suggests that digital infrastructures may improve policy accessibility. This hybrid nature demands further study, particularly on how trust, accountability, and participation evolve when policy artefacts become interactive infrastructures.

HTML

XML

PDF

DOI: 10.3897/biss.9.174854

Authors: Rose Barrowcliffe, Dakota Feirer, Alex Ip, Ben Foley

Abstract: Herbaria hold millions of specimens collected from Indigenous lands around the world. The specimens may represent a piece of the puzzle that allows Indigenous peoples to care for their Traditional Country ('Country' for short), build sustainable incomes, and continue cultural practices. Indigenous data sovereignty (IDSov) means people have the right to manage and control these specimens as part of their environment and resources (Carroll et al. 2021, Maiam nayri Wingara and Australian Governance Institute 2018), but Indigenous peoples' ability to exercise these rights is limited due to the poor findability of the specimens.This project, led by a Butchulla and Bundjalung-Gumgganggbir Macquarie University research team and partnering with New York Botanical Garden (NYBG), Local Contexts, the Language Data Commons of Australia (LDaCA), and Australian Academic and Research Network (AARNet), pilots processes for improving the findability of, and reconnecting botanical specimens to Aboriginal and Torres Strait Islander peoples. NYBG staff assisted with understanding the priorities and processes that have resulted in some Australian records being findable but not others. The Local Contexts staff shared learnings from metadata enrichment to similar items in other collections. LDaCA provided initial programming support to map the specimens' collection points and suggested language datasets to indicate who the relevant Traditional Owners might be. AARNet provided the geospatial analysis necessary for the process of inferring connections between large datasets of botanical specimens and Aboriginal groups.Reconnecting the NYBG specimens to their Aboriginal origins involved three steps. First, a dataset of NYBG’s Australian-sourced specimens was exported from the Global Biodiversity Information Facility (GBIF). The second step was mapping those specimens to their collection points using the specimens' geolocation metadata. The final step is inferring a relationship between the specimen and Aboriginal Traditional Owner (TO) groups. It is not the place of collecting institutions to decide whose Country is whose; that is for Indigenous people to decide, but in Australia, there is a dearth of Aboriginal-created and owned datasets that self-define Country. As an interim measure, we used open geospatial datasets that have a range of Aboriginal-derived data that describe some forms of connection to Country. These include the Native Lands dataset*1, the Asher and Moseley Glottography (The Glottography Consortium 2007), and the Native Title Tribunal’s Native Title Determinations and Indigenous Land Use Agreements datasets*2. Situating the collection points within geographic areas defined in these datasets was used to infer relationships between the NYBG specimens and TO groups.As an extra measure, we audited a selection of NYBG Australian specimens' metadata in both physical and digital form (through a dataset derived from the NYBG’s C.V. Starr Virtual Herbarium online catalogue). This highlighted that there were large amounts of “dark data” in NYBG collections that lack digitisation, processing through optical character recognition (OCR), or transcription to make them findable to Aboriginal peoples through the C.V. Starr Virtual Herbarium and GBIF. Specimens can also be considered dark data if they lack appropriate metadata to make the record findable from a user's point of view. In our pilot, the lack of geolocation metadata inhibited our ability to connect the specimen to Country. While we could not control NYBG's digitisation, OCR and transcription prioritisation, we could enrich specimen geolocation metadata to maximise the number of specimens that could be reconnected to Country. The project plan was amended to include an extra step of using natural language processing (NLP) to parse place names from the ‘locality’ metadata. The place names were then looked up in the Australian Place Name Composite Gazetteer dataset, resulting in a further 3224 specimens with geolocation metadata.The final phase of the pilot is to feed the enriched metadata back into the NYBG catalogue to assist Aboriginal peoples in finding the specimens collected from their Country. The ultimate goal of the project is for Aboriginal peoples to work directly with NYBG to exercise their rights in relation to their botanical kin stewarded by NYBG.

HTML

XML

PDF