Underrepresentation of African surveillance data in global arbovirus modeling

Introduction

Recent decades have witnessed an increasing frequency of arbovirus emergence and re-emergence across multiple regions of the world, with particularly notable expansion into Europe (1–5). Viruses such as Usutu virus and West Nile virus, once considered geographically restricted, have now established transmission cycles in temperate regions, raising concerns about the drivers of emergence and the adequacy of current surveillance systems (6, 7). These newly established transmission cycles are commonly attributed to climate change, globalization, and shifts in vector distribution, all of which can change the ecological conditions that support transmission, including vector survival, host availability, biting rates, and the duration of seasonal transmission window (8). While these explanations are well supported, they largely focus on processes occurring within regions where detection has already taken place, potentially overlooking broader structural biases in how arbovirus data are generated and interpreted.

A fundamental yet underexamined issue in arbovirus ecology is the uneven global distribution of surveillance effort. Contemporary understanding of arbovirus transmission is heavily shaped by data originating from Europe and North America, where sustained investment in vector monitoring, pathogen detection, and public health infrastructure has enabled detailed characterization of transmission dynamics (9). In contrast, large areas of Africa remain comparatively underrepresented, despite harboring high mosquito biodiversity, diverse vertebrate host communities, and environmental conditions conducive to vector-borne disease transmission (10). Here, underrepresentation refers to the limited available arbovirus surveillance data and vector competence studies, rather than the unavailability of disease burden or research activity data.

Current interpretations of arbovirus emergence are strongly shaped by unequal surveillance intensity across regions. Observed increases in arbovirus detection in Europe may reflect genuine ecological change; however, they may also be influenced by improvements in surveillance sensitivity and coverage, which increase the probability of detection in historically underrepresented regions (11, 12). In this context, an increase in arbovirus detection is not solely a biological phenomenon but also a function of observation, such that regions with limited surveillance capacity may harbor ongoing transmission that remains undetected, leading to a skewed perception of global disease distribution.

Africa remains comparatively underrepresented in several global surveillance resources, including vector–virus interaction databases, longitudinal monitoring programs, and datasets commonly used for predictive modeling. The continent supports a wide range of mosquito species, including both established arbovirus vectors and numerous understudied taxa, as well as dense human populations and extensive wildlife reservoirs (2). These characteristics create conditions that are highly favorable for arbovirus maintenance and transmission. Yet, the absence of systematic, large-scale surveillance limits the ability to detect viral circulation and to assess its contribution to regional and global transmission dynamics. As a result, Africa is often underrepresented in global datasets used to parameterize models of arbovirus risk, potentially leading to biased or incomplete predictions (13).

The consequences of this bias extend beyond regional knowledge gaps. Predictive models of disease emergence rely on representative data to infer relationships between vectors, hosts, pathogens, and environmental drivers (14), and although these biases are reported as limitations, they may not always be explicitly incorporated into models. When substantial portions of the global system are underrepresented, these models may fail to capture critical transmission pathways, including those involving less-studied vector taxa or complex ecological interactions. This limitation is particularly relevant in the context of increasing global connectivity, where movement of hosts, vectors, and pathogens can link distant geographic regions into a single epidemiological network.

Despite growing recognition of arbovirus emergence, quantitative evidence highlights substantial geographic bias in surveillance effort. Global datasets are disproportionately populated by records from a relatively small number of well-surveilled regions, including Brazil, Thailand, Singapore, Europe and North America, reflecting long-standing investment in public health infrastructure and research capacity (12, 15). In contrast, much of Africa remains underrepresented in these datasets, with surveillance often limited to outbreak-driven investigations or geographically restricted studies (14). This imbalance is particularly striking when contrasted with malaria research, where extensive entomological surveillance networks exist across the continent (10). Malaria surveillance infrastructure across Africa has been supported by decades of sustained investment through programmes such as the Global Fund and the President’s Malaria Initiative. In contrast, arbovirus surveillance has often relied on outbreak-driven funding and short-term research initiatives. Consequently, differences in surveillance capacity reflect not only scientific priorities but also divergent funding architectures.

In this perspective, we argue that Africa remains unevenly represented in global arbovirus surveillance and that this bias is not only a limitation of data availability but a structural issue that systematically distorts ecological inference and predictive modeling of arbovirus emergence. By recognizing and addressing the structural biases seen in current surveillance approaches, it may be possible to develop a more accurate and comprehensive understanding of arbovirus emergence at a global scale.

Surveillance bias and the construction of global arbovirus knowledge

Current models of arbovirus emergence and transmission are heavily influenced by data availability, yet this dependence is rarely acknowledged explicitly in model interpretation. Regions with established surveillance infrastructure contribute disproportionately to global datasets, shaping both empirical understanding and predictive modeling efforts (12). In practice, this means that a relatively small number of well-studied vector–virus systems primarily involving Aedes aegypti and Culex pipiens form the backbone of global arbovirus knowledge (9). These systems have been intensively investigated through longitudinal surveillance, laboratory competence studies, and genomic sequencing, particularly in Europe and North America.

By contrast, arbovirus surveillance in many African settings remains fragmented and is frequently driven by outbreak response activities rather than routine longitudinal monitoring. Evidence of arbovirus circulation is often generated through reactive investigations following outbreaks of diseases such as yellow fever, Rift Valley fever, dengue, or chikungunya, or through geographically restricted cross-sectional studies. Even though regional surveillance bulletins published by Africa CDC and WHO AFRO demonstrate ongoing reporting of arboviral outbreaks across the continent, comprehensive, standardized, and openly accessible surveillance systems remain limited compared with dedicated regional platforms such as the Pan American Health Organization (PAHO) Arbovirus Portal in the Americas. Consequently, arbovirus occurrence data from Africa are often less continuous and poorly integrated into global surveillance resources, complicating efforts to characterize transmission dynamics and assess emergence risk at continental scales (13–15).

This imbalance has several consequences. First, it creates a perception that arbovirus emergence is geographically concentrated in data-rich regions, reinforcing existing research priorities and funding patterns. For example, global arbovirus datasets and vector competence studies are heavily skewed toward Europe and North America, with comparatively limited representation from African countries despite their ecological suitability for transmission (16). This imbalance is further reflected in the distribution of longitudinal surveillance programmes, which are concentrated in high-income regions.

Second, it limits the identification of transmission pathways that fall outside well-characterized systems, including those involving understudied vector species or alternative host networks. Third, it constrains the development of predictive models, which are often trained on incomplete datasets and may therefore fail to capture the full ecological complexity of arbovirus transmission (17).

Importantly, these biases are self-reinforcing. Regions that generate more data are more likely to be included in future studies, while lesser surveilled regions remain marginalized, further entrenching gaps in global understanding. Addressing this issue requires not only increased data collection but also explicit recognition of how surveillance bias shapes the construction of scientific knowledge.

Implications for emergence and risk prediction

The underrepresentation of African data has direct implications for how arbovirus emergence is interpreted and predicted. Emergence is typically framed as a process driven by ecological change, including climate variability, land use transformation, and shifts in vector distribution (8). However, these interpretations depend fundamentally on the availability of baseline data against which change can be measured. In regions where historical data are sparse or absent, it becomes difficult to distinguish between true emergence and increased detection.

The recent expansion of arboviruses in Europe provides a clear example of this challenge. Enhanced surveillance systems, including coordinated vector monitoring and molecular diagnostics, have increased the likelihood of detecting viral circulation. While environmental change has contributed to transmission dynamics, the role of improved detection should not be underestimated (6). Without stronger representation of African transmission dynamics, global surveillance systems may struggle to distinguish localized outbreaks from wider shifts in arbovirus circulation.

This limitation has important consequences for predictive modeling. Many models of arbovirus risk rely on correlations between environmental variables and observed transmission events. When these models are trained predominantly on data from Europe and North America, they may implicitly assume that these regions are representative of global dynamics. However, this assumption is unlikely to hold in ecologically diverse regions such as Africa, where vector communities, host availability, and environmental conditions differ substantially.

Furthermore, this biased collection of surveillance data may obscure key components of transmission systems, including potential reservoir hosts and alternative vector pathways. For example, migratory birds are known to play a role in the spread of viruses such as West Nile virus, linking African and European transmission systems. However, limited data on viral circulation within African bird populations restricts our ability to quantify this connection. Similarly, human mobility and trade networks may facilitate pathogen movement, yet these pathways are difficult to model without comprehensive surveillance data. As illustrated in Figure 1, this uneven distribution of surveillance effort leads to biased representations of global arbovirus risk. The dots in panel A represent available surveillance evidence rather than a complete inventory of surveillance sites, drawing on public outbreak reporting platforms, arbovirus catalogues and published literature.

A largely unexplored dimension of this overlap is the potential for co-exposure and interaction between arboviruses and Plasmodium within shared vector and host systems. In regions where Anopheles mosquitoes coexist with established arbovirus vectors, the possibility of concurrent transmission or co-infection remains poorly understood. While Anopheles are primarily studied in the context of malaria, their ecological overlap with arbovirus transmission systems raises important questions. These interactions could influence vector competence, pathogen dynamics, or transmission efficiency. However, these interactions remain largely uncharacterized, particularly in African settings where surveillance is limited. Addressing this gap represents an important direction for future research in integrated vector-pathogen systems.

Taken together, these gaps suggest that current models may underestimate both the complexity and the scale of arbovirus transmission. Addressing this issue requires the incorporation of underrepresented regions into both empirical studies and modeling frameworks, as well as the development of approaches that explicitly account for data uncertainty.

Integrating overlooked vector systems

A further limitation of current surveillance frameworks is the narrow focus on a small subset of vector species. While Aedes aegypti and Culex pipiens have been extensively studied due to their established roles in arbovirus transmission, other mosquito taxa remain comparatively understudied. This focus reflects historical research priorities rather than the full diversity of ecological interactions that may influence transmission.

In Africa, this limitation is particularly evident in Anopheles mosquitoes. These species are central to malaria transmission and are therefore among the most intensively studied vectors worldwide. Extensive infrastructure exists for their monitoring, including entomological surveillance, insecticide resistance tracking, and intervention programs (10). However, despite this detailed knowledge base, Anopheles mosquitoes are rarely included in arbovirus surveillance or predictive models.

This exclusion is difficult to justify from an ecological perspective. Although Anopheles mosquitoes differ ecologically from many established arbovirus vectors, several overwintering studies have reported their presence alongside Culex and Aedes mosquitoes in shared hibernation sites, including bunkers, cellars, and other sheltered environments used during diapause. These studies have occasionally detected arboviruses in overwintering mosquitoes, yet surveillance efforts have largely focused on established vector species. Consequently, the potential role of Anopheles within overwintering transmission systems and arbovirus persistence remains poorly understood (4, 18, 19). Additionally, experimental studies have demonstrated that some Anopheles species can support arbovirus infection under laboratory conditions, although their vector competence may vary (18, 19). The systematic exclusion of Anopheles from arbovirus frameworks represents not only a taxonomic gap but a conceptual limitation in how vector–pathogen systems are defined.

The separation of malaria and arbovirus research systems further reinforces this gap. Surveillance programs are typically organized around specific diseases, with limited integration across pathogen types. As a result, opportunities to detect co-exposure or shared transmission pathways are often missed (20, 21). This is particularly relevant in high-transmission settings, where multiple pathogens may circulate within the same ecological system.

Recognizing and addressing these overlaps does not imply that Anopheles mosquitoes are major arbovirus vectors in all contexts (21). Rather, it highlights the need to move beyond rigid classifications of “primary” and “secondary” vectors and to consider a broader range of ecological interactions. Even limited contributions from neglected vectors could influence transmission dynamics, particularly in regions with high vector density and environmental variability.

Toward an integrated surveillance framework

Addressing uneven African arbovirus surveillance requires a shift from fragmented, pathogen-specific approaches to integrated, scalable surveillance frameworks. We propose a conceptual framework based on three interconnected components.

First, surveillance must expand from reactive, outbreak-driven sampling to systematic and longitudinal monitoring across diverse ecological settings (20, 21). This includes routine mosquito sampling across environmental gradients, coupled with pathogen screening using both targeted diagnostics and using untargeted genomic approaches as a tool to understand transmission. Such strategies would enable the detection of both known and previously uncharacterized viruses.

Second, surveillance systems must move beyond a narrow focus on established vector species to incorporate broader vector communities. This requires standardized approaches for assessing vector competence and the inclusion of understudied taxa, such as Anopheles, Mansonia, and other regionally abundant genera (19, 22, 23). Expanding vector coverage would improve detection of alternative transmission pathways and reduce taxonomic bias in surveillance data. There is an urgent need to understand the role of Anopheles in arbovirus epidemiology, as such, further study is needed to define Anopheles vectorial capacity for arboviruses.

Third, integration of heterogeneous data sources is essential for improving predictive capacity. Combining field surveillance data with laboratory-derived vector competence data, environmental variables, Plasmodium interactions, and host distribution information would enable the development of more robust predictive models. Machine learning and network-based approaches offer promising tools for identifying potential vector–virus associations and prioritizing surveillance efforts under conditions of data scarcity (17, 19, 20).

Importantly, existing malaria surveillance infrastructure provides a practical foundation for implementing such integrated systems (21). Although integration offers potential efficiencies, it is not universally cost-free. Existing malaria surveillance systems were designed around specific epidemiological objectives and may require additional diagnostic capacity, personnel training, sampling modifications, and financial investment to support arbovirus monitoring (21, 24, 25). The feasibility of integration is therefore likely to depend on the target pathogen, vector ecology, available laboratory infrastructure, and surveillance objectives.

Conclusion

Global understanding of arbovirus emergence is shaped as much by where we look as by what we find. The apparent expansion of arboviruses into Europe has prompted important advances in surveillance and modeling, but these efforts remain constrained by a fundamental imbalance in data availability. Africa, despite its ecological importance, does not have sufficient and consistent arbovirus surveillance. Addressing this gap is essential for improving both empirical understanding and predictive capacity. By expanding surveillance in underrepresented regions, incorporating diverse vector systems, and adopting integrated data-driven approaches, it will be possible to develop a more accurate and comprehensive picture of global arbovirus dynamics. Such efforts are not only necessary for understanding current transmission patterns but are critical for anticipating and mitigating future emergence in an increasingly interconnected world. In conclusion, we find that until surveillance inequalities are addressed in Africa and other underrepresented regions facing similar surveillance constraints (including fragile health systems, isolated settings and areas with logistical barriers to routine monitoring) global arbovirus models will continue to provide only a partial representation of transmission risk in an increasingly interconnected world.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fitd.2026.1894025/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Thaneeya Duangchinda, National Science and Technology Development Agency Yothee Office, Thailand

Reviewed by: Lara Ferrero Gómez, Universidade Jean Piaget de Cabo, Cabo Verde

Kathryn Mary Edenborough, Monash University, Australia