Using biodiversity features to promote an Ecosystem-Based Conservation framework in terrestrial ecosystems

PLOS One | https://doi.org/10.1371/journal.pone.0336705 November 21, 2025 5 / 18 reclassified into ecosystem types following the UNEP [51] classes. All spatial operations, including clipping, reclassifi- cation, and visualisation, were conducted in QGIS version 3.38.0. The resulting map depicts eight types; grasslands, miombo woodland, acacia savanna, coastal forest, mangrove, moorland, montane forest, and flooded savanna (Fig 1). Delineating PAs and their proportions in different ecosystems PAs are clearly defined geographical spaces, recognised, dedicated, and managed, through legal or other effective means, to achieve the long-term conservation of nature and associated ecosystem services and cultural values as defined by the IUCN [52]. For our study, we used three categories of land use within ecosystems: 1) Priority protected areas (PPAs) which included National parks, Ngorongoro conservation area, Nature reserves, Game reserves, and For- est reserves; 2) other protected areas (OPAs) which included Wildlife Management Areas, Game Controlled Areas and Game Open Areas; and 3) unprotected areas (UAs) which included included settlements, farmlands and pastures. PPAs differ from OPAs based on the level of human intervention allowed. In PPAs, human activities are either prohibited (e.g., nature reserves, national parks) or very limited (e.g., only sport hunting in game reserves). OPAs, on the other hand, allow human activities such as settlements, cultivation, and livestock keeping, while focusing on the sustainable use of natural

resources. We analysed PA coverage across different ecosystem types using a spatial dataset from the Tanzania Wild- life Research Institute (TAWIRI). We used the QGIS intersect tool to determine the area of each ecosystem type within each category (Fig 3). We also determined the number of PAs within each ecosystem type, and for any PA that extended beyond one ecosystem type, its count was assigned to the ecosystem where the largest portion of its area was located. Establishing distribution estimates of biodiversity features As a consequence of a general lack of biodiversity information at ecosystem level, we derived a pragmatic approach for obtaining distribution estimates of three biodiversity features across different ecosystems, to assess progress towards achieving the Kunming-Montreal Global Biodiversity Framework goals. We assessed biodiversity distribution by conduct- ing a literature search of biodiversity occurrence within Tanzania’s major terrestrial ecosystems, focused on three biodiver- sity features (vertebrates, invertebrates, and plants), which are the most studied in terrestrial ecosystems [34]. Ecosystem boundaries guided this as our primary unit of analysis. We conducted a comprehensive literature search in Scopus and Google Scholar databases from May 2023 to August 2023. These databases were chosen for their ease of access and extensive coverage. In both databases, results were restricted to peer-reviewed articles, sorted by relevance, and without time limitation. We utilised tailored search strings

that were designed to be ecosystem- and taxon-specific, combining habitat descriptors with biodiversity and taxonomic keywords using Boolean operators (AND, OR). The complete set of search strings is included in the supplementary materials (S1 Table). Each search string was applied in each database separately for vertebrates, invertebrates, and plants within a given ecosystem, and the number of articles retrieved was recorded per database, per ecosystem and per taxon. In Google Scholar, we screened abstracts of up to 1,000 articles per search string, due to Google Scholar platform constraints, whereas in Scopus, all retrieved articles were screened. The search date and the number of articles retrieved were documented for each ecosystem-taxon combination. The first-level screening assessed titles and abstracts when necessary to determine whether each study was biodiversity-related and conducted in Tanzania. In Scopus, the geographic filter automatically excluded non-Tanzanian studies, while biodiversity relevance was assessed manually. In Google Scholar, both biodiversity relevance and geo- graphic scope were determined manually from titles and abstracts. Studies were excluded at this stage only if they were clearly irrelevant to biodiversity or explicitly indicated a study area outside Tanzania. For each database, the number of studies identified as relevant after screening was recorded by taxon and ecosystem. All Studies retained after the first- level screening were merged across databases by taxon and ecosystem, and duplicates were eliminated so that each

study was only counted once in the aggregation (Fig 2).