Biodiversity loss is a pressing global issue that threatens essential ecological services, which are fundamental to human health, nutrition, livelihoods and the environment1,2. Among the services most negatively impacted by global biodiversity loss is pollination, which is primarily carried out by insects, with honey bees (Apis mellifera) playing a pivotal role3.

It is well recognized that honey bees are essential for the pollination of nearly 75% of 115 major global food crops, boosting yields by 18–71%4,5,6,7,8. They primarily collect nectar (the main carbohydrate source) and pollen (the main protein and fat source) from a variety of flowering plants9,10. However, their health and survival are continuously compromised by climate change, pests and pathogens, pesticide exposure, and poor nutrition due to habitat loss11,12,13. Previous studies have shown that well-nourished honey bees, supported by high-quality and diverse floral resources, are more resilient to biotic and abiotic stressors, exhibit higher learning abilities and are more productive14,15,16,17. In other words, poor nutrition or nutritional stress is now recognized as a key factor undermining honey bee health globally13,18.

Human population growth and the associated demand for increased food have led to agricultural intensification, particularly through monoculture farming. The expansion of monocultural land-use reduces access to diverse forage plants, thereby threatening the dietary needs of honey bees and compromising their health and survival19. This, in turn, has led to increased honey bee colony losses, undermining global efforts to meet rising food demands20,21, which are expected to increase from 59 to 98% by 2050 as the global population reaches 9.8 billion, half of whom will reside in Africa22,23.

Pollinator-dependent food systems in Africa are increasingly vulnerable to environmental changes that threaten pollinator health and associated ecosystem services13,24,25,26. Insect-pollinated crops contribute approximately 12–50% of the intake of key micronutrients such as iron, vitamin A and B-9, depending on the crop type and region27,28. Pollinator deficits, e.g. of honey and wild bees, may lead to yield gaps of 21–59% in some countries7,8. Since 1990, Africa has lost about 4 million hectares of forest annually29,30, and between 2003 and 2019, net cropland area increased by 34%31. Urban land has also expanded by 5.9% between 2001 and 201932, a trend expected to worsen by 205033. Additionally, climate change effects, such as drought, further reduce bee forage availability34, as plants produce less nectar and pollen under such conditions35,36. The combined impacts of habitat loss and climate change underscore the urgent need for strategies that enhance forage availability within honey bee landscapes to sustain their health and ecosystem services. Although the concept of improving bee forage resources in the landscape was suggested decades ago by Shuel37 and Comba et al.38, it has gained significant recognition only recently in the USA19 and Europe39,40. In Africa, however, it remains limited, largely due to limited knowledge of local bee forage plants41. Such knowledge is also essential for characterizing honey diversity across the continent for marketing purposes since the chemical composition, organoleptic and medicinal properties of honey depend directly on the floral sources visited by honey bees42.

To address this gap, we compiled a comprehensive geospatial distribution of honey bee forage plants across Africa through an extensive literature review of reported honey and pollen foraging plant resources and retrieved the occurrence records of these plants from open-access biodiversity databases like iNaturalist and the Global Biodiversity Information Facility (GBIF). We analyzed commonalities and differences among regions to provide insights for optimizing conservation efforts. Additionally, we developed an interactive dashboard to visualize these spatial patterns, offering a decision support tool for targeted conservation of honey bee forage plant species across the continent. This interactive dashboard is expected to: (1) provide a freely-accessible geospatial database for understanding regional biodiversity and ecological interactions, particularly related to honey bee foraging resources; (2) inform conservation strategies by identifying vulnerable habitats for protection, thereby supporting pollinator health and productivity as well as ecosystem stability; and (3) serve as a foundational resource for researchers studying plant-pollinator dynamics, agricultural practices, and environmental changes affecting honey bee colony productivity, among others.

Compilation of a comprehensive list of nectar and pollen forage plants for honey bees in Africa

Three main sources were used to compile a comprehensive list of honey bee forage plants in Africa. First, an online search for published studies was conducted using the Web of Science and Google Scholar. Searches combined relevant germane keywords (e.g. “honey bee”, “forage plants”, “nectar”, “pollen”, “Africa”) with Boolean operators (e.g. OR, AND) and modifiers (e.g. quotation marks for exact phrases) over a 20-year period. To capture country-level data, we repeated the searches for all 54 African countries by substituting “Africa” with the names of each specific country, including “Cameroon”, “Kenya”, “Ethiopia”, “South Africa”, “Ghana”, “Rwanda”, and “Benin”, among others. This process yielded over 68 published articles. From these, 21 high-quality articles were retained based on the following inclusion criteria: (a) presence of the scientific name of the forage plant, (b) plant family, (c) floral resource provided (nectar, pollen, or both), (d) flowering period, (e) plant type (perennial, annual or biennial shrubs, trees, herbs, etc.), (f) location (country) and/or (g) relevance of the forage resource to other pollinators such as insects or birds. We extracted these attributes from these resources using a structured data collection form to ensure standardization. Certain plant families were excluded if they were of limited value to honey bees. These included Solanaceae, which typically require buzz pollination and may contain compounds toxic to mammals including humans43, and Anacardiaceae, which are largely wind-pollinated with short flowering periods. Despite the known toxicity of some Euphorbiaceae species44,45, this family was included in our list due to its prevalence in farming systems and documented use by honey bees46. Second, additional plant records were obtained from books on honey bee forage plants in Ethiopia47,48, South Africa49 and Tanzania50 using the same inclusion criteria and extraction process. Lastly, species lists were obtained from selected national partners in Rwanda, Kenya, Cameroon and Ghana. These partners, who work directly with beekeeping communities, shared locally important plant species known to enhance landscape suitability for honey production. Verbal consent was obtained prior to data sharing, and all submissions were evaluated against the established inclusion criteria.

Through this systematic and rigorous process, a total of 1,248 honey bee forage plant species from 91 families across Africa were compiled into the database. The native and exotic status of each plant species was determined using Plants of the World Online (https://powo.science.kew.org/) or iNaturalist (https://www.inaturalist.org/).

Automated retrieval of georeferenced distribution data for honey bee forage plants

The global georeferenced distribution data for honey bee forage plants in our list were retrieved from GBIF51 and iNaturalist52 using automated Python-based APIs (Application Programming Interfaces). The pygbif library accessed GBIF records53, while the pyinaturalist library retrieved iNaturalist data52. All data files downloaded from GBIF and iNaturalist were initially stored in a cloud repository and subsequently merged using Python’s Geopandas library54, enabling efficient handling of 2,074,182 georeferenced records while maintaining dataset consistency.

Data cleaning and preprocessing

A multi-step cleaning process was implemented to ensure the reliability and coherence of the initial dataset of 2,074,182 georeferenced records. Duplicate entries with identical coordinates, species names and dates were removed. Taxonomic inconsistencies were resolved through cross-referencing with standardized taxonomic databases. Missing geographic data were supplemented using available country, city, and locality information. This step improved the spatial resolution and completeness of the dataset. To further ensure data accuracy, inconsistencies in geographic locations such as mismatched country, region, or continent entries were corrected by performing a global administrative boundaries shapefile sourced from an open-access geospatial database. Each coordinate was assigned to its corresponding country, region, and continent through spatial overlay techniques implemented using the Python library geopandas. This step ensured that all records were geospatially accurate and aligned with administrative boundaries. Additionally, erroneous points, such as those located in oceans or outside plausible terrestrial boundaries, were identified and removed. Following cleaning and validation, the dataset was refined to 1,572,790 high-quality occurrence records, providing a reliable foundation for analyzing honey bee forage plant distribution.

Designing an interactive dashboard

An interactive dashboard was developed to visualize the distribution of the cleaned dataset. This dashboard serves as a support tool for understanding spatial patterns and optimizing conservation efforts tailored to specific ecological zones, particularly across the African continent. The cleaned dataset was uploaded to an open-source database called PostgreSQL (version 14.0) using the spatial extension PostGIS (PostgreSQL for Geographic Information Systems)55. This was achieved by integrating and reorganizing the data into a structured database model (Fig. 1). This approach ensured the datasets were harmonized, allowing for efficient storage, retrieval, and analysis of honey bee forage plant records while preventing potential duplicated records. The georeferenced data were integrated with a GeoServer56 to facilitate efficient querying and retrieval of spatial information. This process involved creating a GeoServer datastore linked to our relational database, ensuring seamless access to the geospatial data. Specifically, the georeferenced information stored in the “observations” table (see Fig. 1) was published as a tile layer. This setup allowed optimized visualization and interaction with geospatial data, enabling users to explore and analyze location-specific information via map-based interfaces. For instance, users can select a specific continent and plant family to visualize the top 20 most reported plant species within that family, the reporting trend of all species in that family over the last 20 years and view their global distribution under the “statistics” section. Alternatively, users can select a specific plant family under the “map” section to visualize plant distribution and their native (indicated by red dots) versus exotic (indicated by blue dots) status.

Structured entity relationship diagram derived from merged data.

Full size image

Dashboard architecture overview and user interface

To ensure simplicity and scalability, we adopted a four-layer data model architecture as a back end for the dashboard. This well-established software design pattern organizes applications into distinct layers, each responsible for a specific function57. This structure enhances the separation of concerns, making the dashboard system more manageable, easier to test, and highly scalable. Our design features a four-layer architecture (Fig. 2), comprising the following layers: (i) presentation layer: responsible for delivering information to the user, this layer focuses on providing anintuitive and interactive user experience. It includes features like dynamic charts, geospatial mapping, and other visualization tools to present complex data clearly; (ii) service layer: acts as the intermediary between the presentation and business logic layers. It provides well-defined application endpoints that implement core system functionalities, facilitating smooth data exchange and feature execution; (iii) business logic layer: this layer encapsulates all essential business processes, including statistical models, key functions, system entity definitions, and database access logic. It ensures that business rules are consistently applied across the system; and (iv) a data layer: Dedicated to storing and managing all system data, this layer handles information retrieval and persistence. It maintains the integrity and security of the data while supporting efficient query execution. This architectural approach ensures that each layer operates independently, enabling modular development and future scalability.

Layered architecture of the system with associated technologies stack. Base map source: OpenStreetMap, rendered using OpenLayers Version 10.5.0.

Full size image

The user interface of the dashboard was developed using advanced technologies to optimize usability and minimize latency in data retrieval and display. Since the final product is a web-based system, the technology stack prioritized JavaScript-based libraries and frameworks. Specifically, Chart.js was used for interactive chart visualization58, while React and Next.js were chosen to optimize rendering and ensure smooth interactivity59. Additionally, OpenLayers was integrated as a front-end solution to facilitate the display and exploration of georeferenced data through interactive maps.

Composition of native and exotic honey bee forage plant species occurring in Africa

Through our comprehensive literature abstraction, a total of 1,248 honey bee forage plant species, belonging to 91 families, were identified across Africa (Fig. S1). Most of these species (74%) were native to the continent (Fig. 3A). The top 20 most reported plant families supporting honey bee nutrition are shown in Fig. 3B. Most of these families had a higher proportion of native species, except for Myrtaceae and Rosaceae, which contained a considerable proportion of exotic plants.

Number of native and exotic honey bee forage plant species among the 1,248 identified plants (A), and among the top 20 most reported plant families supporting honey bee nutrition in Africa (B).

Full size image

The dashboard system

Of the 1,248 identified honey bee forage resources reported in Africa, 819 (66%) had occurrence records on GBIF and iNaturalist. A web-based dashboard developed in this study provides an interactive, user-friendly tool for visualizing the distribution of these 819 species across African regions and globally (https://beehealth.icipe.org). This interactive dashboard allows users to explore several features: under the “Stats” section (Fig. 4), users can view statistics for the top 20 most reported plant species within a selected plant family by continent or globally. They can also examine temporal trends in species occurrence over the last 20 years, as recorded on global biodiversity platforms, alongside their corresponding distribution maps. Additionally, a pie chart displays the regional distribution of observations, highlighting regions with abundant or limited occurrence records. Under the “Map” section, users can further explore the distribution of each plant family across African regions, with their native versus exotic status indicated by blue and orange dots, respectively. The map supports zoom functionality and allows direct access to location names where the plant species was recorded by clicking individual blue or orange dots.

A screenshot of the dashboard interface showing statistics on the 20 most reported Acanthaceae plants in Africa, illustrating the occurrence trend of all species in this family over the last 20 years on open-access biodiversity platforms. It also shows their global distribution on maps, with dark blue regions indicating areas with the highest occurrence records. Their distribution across African regions is also presented, with red and blue dots representing native and exotic plants, respectively.

Full size image

Global trends in Temporal reporting of honey bee forage plants in open-access biodiversity platforms

Of the 1,248 identified honey bee forage species, Africa had the highest documentation rate (775 species, 62%), followed by the Americas (550 species, 44%), Europe (446 species, 36%), Asia (445 species, 36%), and Oceania (415 species, 33%). Species occurrence records have increased significantly across all continents, particularly in the 21 st century (Fig. 5). Records prior to 2000 were relatively sparse, but a pronounced surge occurred from 2011 to 2020, continuing into 2021–2024, although reporting rates varied by continent. For example, despite a substantial increase in reporting from Africa since the early 2000 s, overall reporting volume in 2024 remained approximately 22–34% lower than in Europe and the Americas, respectively (Fig. S2). In fact, the Americas and Europe showed the most pronounced increase in 2024, whereas Asia and Oceania exhibited slower growth in reporting compared to these two continents.

Temporal trends in honey bee forage plant occurrence records across continents. ArcGIS desktop 10.8 (ArcMap) (https://www.esri.com/en-us/home) running under an advanced licence was used to generate the figure.

Full size image

Global and regional disparities in honey bee forage plant documentation

The global distribution of the 819 bee forage plant species recorded on open-access biodiversity platforms revealed distinct patterns at both global and regional levels (Fig. S4). Globally, Africa accounted for 32.7% of total occurrence records, followed by the Americas (29.9%), Europe (20.9%), Oceania (8.5%) and Asia (7.9%). Within each continent, this documentation was uneven. In Africa, Southern Africa recorded the highest number of documented species, followed by Eastern Africa, while Western, Central and Northern Africa were underrepresented. Similar disparities were evident in the Americas, where Northern and Central America had more records than South America and the Caribbean. In Europe, species records were concentrated in the Southern and Eastern regions, while in Asia, the Eastern and Southern regions had the most occurrence records.

Bee forage plant families also varied regionally. Asteraceae and/or Fabaceae were the most reported plant families in Southern and Eastern Africa, Northern and Central America, Southern and Eastern Europe, and Eastern and Southern Asia (Fig. S3). In Oceania, Myrtaceae and Fabaceae dominated the records (Fig. S3). Interestingly, 308 plant species from 61 families were reported in the Sahara (spanning Western, Central, Northern, and Eastern Africa) and Kalahari (restricted to Southern Africa) deserts (Fig. 6; Fig. S4), with 84% of these plants being native to Africa. Of these, 157 species (51%) were reported in the Kalahari, 62 species (20%) in the Sahara, and 89 species (29%) in both deserts.

Top 10 most reported plant species in the Kalahari and Sahara deserts.

Full size image

This study compiled a comprehensive geospatial database of 1,248 honey bee forage plant species reported in Africa, representing 91 plant families. Occurrence data for 819 of these species (66%) were retrieved from GBIF and iNaturalist, totaling over 1.57 million records globally. In recent years, reports of African honey bee forage plants on these open-access biodiversity databases have increased considerably (Fig. 5 & Fig. S2). Visualization of their distribution patterns through our interactive web-based dashboard revealed considerable disparities in species reporting across continents and regions. At the continental level, Africa still lags behind Europe and the Americas in the overall reporting volume (Fig. 5 & Fig. S2). This gap reflects differences in investments, with substantially greater funding and research capacity supporting pollinator research60,61,62,63,64 and citizen science initiatives65,66 in Northern America and Europe, where plant-pollinator interaction data are more systematically collected and reported, compared to developing regions like Africa where such frameworks are still emerging. Additionally, structured plant-pollinator databases67,68(https://saveplants.org/plant-pollinator-interaction-explorer/) and implementation of pollinator-friendly policies61,64 further underpin coordinated biodiversity documentation in developed regions. These reasons also explain South Africa’s leadership in species records across Africa62,65,69,70,71,72. Other African countries in Eastern and Western regions have also made progress in biodiversity research and citizen science initiatives69,71 and attracted funding for environmental projects62, but at a smaller scale relative to South Africa. Addressing these disparities will require scaling up biodiversity research efforts, promoting citizen science in biodiversity documentation, increasing both national and international funding, and strengthening conservation policies to advance pollinators and their landscape conservation strategies throughout the continent.

The geospatial dashboard developed alongside the database directly supports these efforts by highlighting spatial disparities in honey bee forage plant reporting across Africa. It clearly identifies significant data gaps, notably in Western, Central and Northern Africa, where GBIF and iNaturalist records are sparse compared to Southern and Eastern Africa. By pinpointing these geographic gaps, the dashboard offers actionable insights to prioritize research and direct conservation investments where they are most needed. In data-rich regions, such as Southern and Eastern Africa, the dashboard provides practitioners, including beekeepers, researchers, land managers, conservationists and policymakers, with detailed information on forage plant distribution, flowering periods and resource types provided by these plant species (nectar, pollen, or both). This information can be leveraged to implement honey bee-friendly land management practices, such as habitat restoration, targeted planting of high-value forage species, and/or protection of critical floral resources during key foraging or lean seasons. Ultimately, the database and dashboard function together to support evidence-based decision-making, aiming to improve pollinator health and productivity while enhancing conservation of both wild and managed pollinators and their supporting habitats across Africa. To maximize this potential, future refinements are needed, including incorporating field data on wild pollinators and their interactions with visiting plant species, as well as integrating these biodiversity data with climate and land-use models, as done elsewhere67,68. Such integration could eventually support emerging carbon market mechanisms that incentivize ecosystem services, like pollination and carbon sequestration, provided by biodiverse landscapes.

Analysis of the database revealed that a large proportion of honey bee forage plant families and species in Africa are native, indicating that African honey bees predominantly rely on native flora for nutrition. These native plants are well adapted to the local environmental conditions, and can provide abundant nectar and/or pollen to honey bees and other local pollinators, thereby strengthening continental biodiversity73,74. However, this predominance of native plants in supporting honey bee diet may be geographic-specific. For example, a recent study in Kenya found that exotic forage species accounted for 67% of the nutrition of A. m. scutellata compared to native species (33%) in Taita Taveta County26. Our findings also showed that the proportion of native versus exotic plants varies across African countries (https://beehealth.icipe.org) and plant families (Fig. 3; Fig. S1). The Myrtaceae family, for instance, ranks fourth in abundant but features significantly more exotic plants, mostly Eucalyptus species, which are recognized as good nectar sources for honey bees worldwide75. Eucalyptus trees have been introduced in many African countries, with cultivation varying by region, reflecting regional differences in trade, land use and management practices for these species. South Africa and Ethiopia have the largest Eucalyptus cultivation areas, each covering over 500,000 hectares, while Kenya cultivates about 100,000 hectares76,77. Their high water consumption and occasional allelopathic effects on native plant species may contribute to these regional differences76,78,79. Taken together, these results underscore the importance of understanding local context when developing country-specific strategies to sustainably integrate both native and exotic forage plants into the landscape. Such tailored strategies can support resilient bee nutrition and productivity while maintaining ecosystem service delivery. Achieving this also requires more detailed knowledge of the nutrient profiles of nectar and pollen provided by these plant species to manage honey and wild bees.

As mentioned above, nectar and pollen supply distinct nutrients for honey bee health9,80. While all nutrients are necessary, lipid intake is crucial, particularly the essential polyunsaturated fatty acids like linolenic acid (omega-3) and linoleic acid (omega-6). These fatty acids support brain function, foraging activity, development, and overall colony performance14,81. They also serve as energy reserves during periods of nectar and pollen scarcity82,83. In this study, pollen-rewarding plant families such as Asteraceae contributed most to African honey bee nutrition (Figs. 3 and 5), though their pollen generally exhibits a low protein-to-lipid ratio and poor digestibility10,84. In contrast, pollen from the Asphodelaceae family, which promotes ovarian development in African worker honey bees, has higher protein content compared to the Asteraceae pollen85. Other key families, including Fabaceae and Rosaceae, which also significantly supported African honey bee nutrition, produce large amounts of pollen with a high protein-to-lipid ratio10. On the other hand, nectar-rewarding plants, including those from the Malvaceae, Lamiaceae, and Euphorbiaceae families, which were among the top 10 most abundant families in this study produce pollen with a low protein-to-lipid ratio. For example, Eucalyptus pollen is low in omega-3 fatty acids and has a high omega-6 to omega-3 ratio14. An unbalanced omega-6 to omega-3 ratio and deficiency in omega-3 have been linked to impaired learning and development in honey bees14,81. Other nectar-rewarding families such as Acanthaceae and Rubiaceae, also contributed considerably to bee nutrition in this study, but their nutritional profiles remain poorly understood in Africa. Overall, knowledge gaps persist regarding the nutritional ecology of honey bees and other insect pollinators in Africa, and how climate, land-use changes and seasonality affect the nutritional quality of nectar and pollen of flowering plants41. Therefore, future studies addressing these gaps and incorporating nutritional metrics into the dataset would greatly enhance its value for research and conservation planning.

A notable finding in this study is the identification of desert-adapted forage species in the Sahara and Kalahari ecosystems. These plants, predominantly native to Africa, may offer climate-resilient forage alternatives as pollinators increasingly face drought-driven resource scarcity34. Interestingly, over twice as many plant species were recorded in the Kalahari compared to the Sahara, a disparity likely reflecting differences in research efforts as well as climatic and historical biogeography factors. Similar patterns were observed in semi-arid areas of Kenya, where Ochungo et al.24 and Chege et al.26 reported how honey bee forage diversity and/or protein content vary with land-use change. These findings emphasize the importance of understanding forage availability in the context of local environmental pressures. Additionally, while the findings of desert-adapted species highlight their ecological resilience, future work should place them within the broader spectrum of African ecosystems. For instance, assessing the adaptability of identified honey bee forage plants across diverse ecosystems, including tropical, montane, savannah and coastal habitats, some of which are native to or introduced outside Africa, could reveal their potential to thrive under varied environmental conditions. Comparing resilience and ecological functions across these ecosystems can inform targeted conservation and restoration strategies under climate chance scenarios.

Overall, this database represents the most comprehensive compilation to date of honey bee forage plant species and their distribution across African regions, highlighting areas with rich or limited occurrence records to support conservation prioritization, habitat restoration, and region-specific planting strategies. By providing both raw data and an interactive web-based visualization tool, it enhances data accessibility, reproducibility, and foster interdisciplinary collaboration across pollination ecology, sustainable agriculture, and biodiversity research. Importantly, the database can be integrated with other available ecological and climate datasets, as well as nutritional metrics, to enable robust predictions for habitat restoration and identification of climate-sensitive species or habitats in the future. Enhancing its value will also require greater investment in field data collection on plant-wild pollinator interactions across diverse ecosystems, strengthened institutional collaboration in data sharing, and innovative approaches such as citizen science to expand data coverage and foster community engagement. These advances will support more detailed analyses of forage dynamics and significantly contribute to evidence-based conservation planning, climate-resilient agricultural practices, and long-term pollinator sustainability strategies across Africa and beyond.

The datasets that support the findings of this study are archived at https://doi.org/10.5281/zenodo.15181534. However, certain data, including unpublished indigenous knowledge and information provided by national partners, are subject to access restrictions due to sensitivities around data ownership and potential implications for competitiveness in the international honey trade. Consequently, these restricted datasets are available from the corresponding author upon reasonable request and subject to appropriate data sharing agreements. The codes used for data processing and analysis are freely accessible via the GitHub repository (https://github.com/Vansnoden/hBeePlantForage).