Participatory monitoring drives biodiversity knowledge in global protected areas

The study addresses how participatory monitoring (citizen science and community-based monitoring) contributes to and shapes the global biodiversity evidence base in terrestrial protected areas. It situates the work within the 30x30 conservation target of the Global Biodiversity Framework, ongoing debates about protected area effectiveness and indicators, and the pervasive shortfalls, geographic, and taxonomic biases in biodiversity data—especially acute in the Global South. Given limited monitoring resources and the need for locally informed management, the authors hypothesize that participatory monitoring has become a dominant source of open biodiversity data in protected areas and that its contributions differ from non-participatory monitoring in geographic, taxonomic, and threatened species coverage. They aim to quantify these contributions over time and across contexts to inform strategies that maximize conservation impact.

The paper synthesizes literature on: (1) data shortfalls and biases in biodiversity research and monitoring, noting uneven coverage and limited data for many species and regions; (2) calls for localization, community engagement, and integration across scales to improve conservation outcomes; (3) the promise of participatory monitoring to expand data volume and coverage, aided by digital platforms and improved analytical tools for unstructured data, alongside risks of replicating or introducing new spatial and taxonomic biases; (4) policy and governance contexts that enable participatory monitoring and open data sharing; and (5) the complementarity of participatory and structured, long-term monitoring in delivering fit-for-purpose conservation evidence.

Protected areas: Polygons for global protected areas were obtained from the World Database on Protected Areas (WDPA). From 254,526 areas, 236,845 terrestrial areas were retained. Attributes used included area size, IUCN protection category, and governance structure. Biodiversity data: All GBIF occurrence records were downloaded and loaded into a spatially enabled PostgreSQL/PostGIS database with WDPA polygons. Occurrences within protected area boundaries were extracted, and analyses were restricted to 2000–2021 to capture recent trends. Of the WDPA areas, 143,510 had GBIF data; 93,335 did not. The dataset comprised 486 million species occurrence records, summarized by unique combinations of taxon (n=778,949), dataset (n=11,242), protected area (n=143,510), and year; taxa were linked to IUCN Red List categories. Classification of participatory monitoring: Datasets were labeled as participatory if they involved voluntary biodiversity data collection by the public outside professional/academic roles (structured or unstructured). First, all datasets with GBIF machineTag "citizenScience" (n=494) were classified as participatory. Next, remaining datasets’ registry metadata were searched for multi-language keywords derived from citizen science terminology literature and ECSA Ten Principles translations; datasets matching terms (n=4,806) were manually screened. Datasets were classified as participatory if ≥50% of data appeared to originate from participatory monitoring (following Chandler et al.). In total, 970 datasets were classified as participatory; 10,272 were not. The approach likely yields a conservative estimate because some participatory datasets may lack indicative metadata. Analyses: The authors computed the amount (counts) and ratio (participatory/total) of participatory observations at national and protected area levels, and summarized by global region and broad taxonomic groups. They described distributions of observations per species and per IUCN Red List category for participatory vs non-participatory data, and temporal trends by year. Relationships between participatory data amount/ratio and protected area characteristics (size, IUCN category, governance) were tested using Kruskal–Wallis tests (low correlations among predictors; all Cramer’s V<0.24). Four monitoring program characteristics (program size, taxonomic focus, geographic focus among areas, geographic focus among nations) were derived (low inter-correlation; all Cramer’s V<0.30) and related via chi-squared tests to area size, IUCN category of areas where programs are active, and IUCN categories of species monitored. Analyses were conducted in R 4.1.2. Data and code are openly available on OSF.

• Participatory dominance: Since 2000, participatory monitoring produced 77% of GBIF biodiversity data for terrestrial protected areas. It is the sole GBIF data source for 25% of protected areas. Among the 61% of areas with any GBIF data, participatory data form the majority in 75% and >90% in 59% of areas. The participatory share accelerated around 2008 due to rising participatory growth and declining addition of non-participatory data to GBIF. - Geographic patterns: Data volumes are highest in wealthier nations, but many low-data countries derive a high proportion from participatory sources. Substantial variation exists among protected areas within countries, indicating national policies can shape open biodiversity knowledge. - Taxonomic coverage: Participatory monitoring contributes >50% of GBIF data since 2000 for birds, invertebrates, fungi, reptiles, and amphibians within protected areas. Distributions of observations per species are similarly or less skewed for participatory vs non-participatory data in birds, reptiles, and amphibians. - Threatened species: For IUCN-assessed taxa, participatory sources record threatened species less frequently than non-participatory data (Kruskal–Wallis H=279, p<0.001), yet 47% of threatened species recorded by participatory monitoring within protected areas were not recorded by other means, indicating complementarity. - Protected area context: Larger areas host the largest participatory datasets (Kruskal–Wallis H=18167, p<0.001), while smaller areas rely more heavily on participatory data proportionally (H=12824, p<0.001). Stricter IUCN categories have more participatory data in absolute terms (H=1263, p<0.001), but less strict areas have a higher participatory proportion (H=581, p<0.001). Areas managed by local communities show the greatest overall participatory contributions; those managed by non-profits have the highest participatory proportion. - Program characteristics: Smaller and narrowly focused (taxonomically or geographically) programs are slightly more likely to collect data on threatened and data deficient species and are more active in large and less strictly protected areas. Small and strictly protected areas are more often sampled by large, taxonomically and geographically diverse programs.

The findings demonstrate that participatory monitoring has transformed biodiversity knowledge in protected areas, now supplying a large and growing majority of open data on GBIF. This creates opportunities for fine-scale biodiversity trend analyses in large areas and provides crucial information in smaller or multi-use protected areas where professional monitoring is limited. Participatory contributions differ from non-participatory data across geographies and taxa and uniquely add records for many threatened species, underscoring complementarity rather than redundancy. However, the concurrent decline in non-participatory data being published on GBIF raises concerns, whether due to reduced collection or lags/barriers in data sharing. Given different strengths and weaknesses, participatory data should complement, not replace, structured long-term monitoring. Uneven global growth, biased toward wealthier countries, persists, yet strong roles for community-based monitoring in some developing nations suggest policy and infrastructure can enable broader participation and data sharing. Context matters: small areas often depend largely on participatory inputs to inform urgent conservation decisions and prioritization, while large areas can leverage massive participatory datasets for high-resolution analyses. Locally managed areas and non-profit governance show high participatory uptake, aligning with emerging emphasis on OECMs. Smaller, locally led programs can better target threatened or data-deficient species, implement tailored protocols, and enhance civic engagement, but may lack resources and visibility. National policies to support participatory monitoring and open data, capacity building, and best-practice sharing can amplify impact. Advances in methods for unstructured data further increase the fitness-for-purpose of participatory datasets for protected area management and assessment.

Participatory monitoring now underpins open biodiversity knowledge in protected areas globally, expanding geographic and taxonomic coverage and uniquely documenting many threatened species. Its strengths are complementary to structured, professional monitoring, and together they can better support conservation assessments and management. To maximize conservation impact, the authors recommend: enabling national policy environments for participatory monitoring; strengthening open data infrastructure and incentives for both participatory and non-participatory data sharing; fostering locally led, focused programs alongside large-scale platforms; and disseminating best practices and analytical advances to enhance data quality and applicability at local and network scales. Continued recognition and development of participatory monitoring are critical to improve protected area outcomes and to achieve ambitious area-based conservation targets.

The analysis includes only biodiversity datasets published to GBIF, excluding many participatory and non-participatory programs that share data locally or via other platforms. This likely underestimates contributions from small, taxonomically or geographically focused participatory programs and emerging data sources (e.g., social media, online platforms) with limited GBIF integration. The observed decline in non-participatory data on GBIF could reflect decreased collection or delays/barriers in data publication. Additional constraints include sensitive species data, privacy, Indigenous data sovereignty, and local data governance considerations that limit open sharing. As with unstructured occurrence data generally, spatial and taxonomic biases remain, though analytical methods can mitigate some biases.