Malaria around large dams in Africa: effect of environmental and transmission endemicity factors

Rainfall variability disrupts agricultural productivity, contributes to disasters associated with floods and droughts, and hinders economic growth in Africa [1, 2]. To cope with these challenges and provide a platform for advancing water security and sustainable development, the African continent has entered a new era of dam construction [3]. A great number of large and small dams are currently under construction particularly in sub-Saharan Africa (SSA)—the region with the lowest per capita water use in the world [4]. To support Africa’s dam building plans, the World Bank has renewed its financial support over recent years [5].

Dams alter landscapes, increasing the abundance of standing water and drastically changing aquatic ecosystems and the ecological functions associated with rivers. Inevitably these changes result in a range of consequences beyond the direct envisioned benefits of infrastructure development. Dams’ impacts on the transmission of vector-borne diseases such as malaria comprise a particularly pressing public health challenge in Africa. A number of studies document that dams often increase the rate of malaria in communities living close to reservoirs (hereafter referred as reservoir communities), especially in areas of unstable malaria transmission [6‐12].

Dams’ impact on malaria typically results from the expansion of habitats suitable for Anopheles mosquito breeding. Availability of breeding habitats in the vicinity of dams and the reservoirs they form is often dramatically exacerbated by the expansion of agricultural activities that create numerous breeding habitats for mosquito vectors [12]. Nonetheless, the effect of the expansion of such habitats is context specific since climatic and biophysical factors governing malaria transmission remain important. Rainfall creates breeding habitats that may affect the abundance of breeding sites, even in reservoir communities [13]. Temperature determines the rate of growth of a mosquito’s aquatic stage and the sporogonic development of the malaria parasite inside adult mosquitoes [14, 15]. Similarly, biophysical factors such as geography and land use can influence the suitability of habitats for mosquito development.

Landscape features and ecologic variables have widely been used to explain and predict the spatial variation of malaria incidence in a certain geographical area [16‐20]. For instance, several studies indicated that distance to breeding habitats contributed to the heterogeneity in malaria risk across regions [16‐18]. In Dakar, Senegal, a study showed that malaria incidence in households within 160 m from a permanent marsh was 74% while the incidence in those located at 900 m were only 17% [18]. Similarly, evidence suggests that village distance to reservoir shoreline and wind direction are significantly correlated with malaria incidence around Ethiopian dams [21, 22]. In the specific context of water storage reservoirs, Jobin [23] suggested that topographical maps may be useful to gauge suitability for vector development along shoreline sites. In controlled laboratory experiments, Endo et al. [24] identified that the steeper the slope of the shoreline, the less suitable they are for vector development. A recent study, noted that the impact of dams on malaria transmission varies in different ecological settings due to local topographic and climatic factors [21].

To better understand the impact of dams on malaria, recent efforts have moved beyond identification of elevated malaria in the vicinity of individual reservoirs to estimate their aggregated contribution to population at risk and disease transmission in Africa [9, 25]. Keiser et al. [9] estimated that 9 million people were at risk of malaria transmission due to large dams in Africa. Kibret et al. [25] conservatively estimated that the presence of 1286 large dams across the region increased the risk for at least 15 million people and were associated with more than 1.1 million additional malaria cases annually. Another recent study further predicted that, without further control measures, transmission around reservoirs could triple by the 2080s as a consequence of climate change and population increases [26].

While efforts to improve understanding of the malaria consequences of large dams mark important progress, past efforts toward Africa-wide analysis have been limited in three ways: (i) reliance on a crude rectangular approximation for reservoir delineation; (ii) assumption that recent levels of malaria transmission are static, whereas actual conditions are dynamic and evolving; and (iii) insufficient focus on the interactions between environmental variables and reservoir presence that influence transmission levels.

In order to devise appropriate malaria control measures, it is crucial to accurately determine the effect of dams on malaria and understand the underlying factors that lead to increased or decreased malaria incidence in their vicinity. This paper reports on the findings of a study to determine the effect of large dams on malaria in sub-Saharan Africa using an approach that incorporated precise reservoir delineation and explicitly considered evolving transmission levels. The study also explored the role of environmental factors in determining the rate of malaria disease incidence in the vicinity of reservoirs.

Anopheles arabiensis, Anopheles funestus and Anopheles gambiae are the major vectors around large dams in SSA (Fig. 1). Anopheles arabiensis frequently predominate in the vicinity of dams in unstable areas, whereas An. arabiensis, An. gambiae and An. funestus often coexist in the vicinity of dams in stable areas.

Overall, the population living close to (i.e., < 5 km) large dam reservoirs in SSA increased by about 4.3 million (i.e., from 14.4 million to 18.7 million) between 2000 and 2015 (Table 1). During this period, the number of dams for which geo-referenced data were available increased from 884 in 2000 to 919 in 2015. A total of 18.7 million people lived close to the large dams investigated in 2015. Nearly 75% of the population resides in malarious areas. There was a substantial increase in the population in unstable areas where the number of people living around the dams doubled between 2000 and 2015, as a result of both growing population and growing area of unstable transmission. Nonetheless, large dams in areas of stable transmission tend to be located in more densely populated regions, resulting in greater population at risk around dams in areas of stable transmission.

Overall, recent years have seen a growing number of dams situated in areas with unstable transmission (Fig. 2). The number of dams situated in areas of stable transmission reduced from 264 in 2000 to 234 in 2015 while the number of people living close to these reservoirs increased from 7.9 million to 8.3 million (Table 1). The number of dams located in unstable areas grew during the same period (199 in 2000 and 265 in 2015; a 33% increase), as did the population living close to them (2.6 million in 2000 to 5.6 million in 2015; a 218% increase). In contrast, the number of dams situated in no malaria areas remained virtually the same between 2000 (n = 421) and 2015 (n = 420).

Annual malaria incidence has declined markedly between 2000 and 2015 in both reservoir and control communities (Table 2). In communities close to (< 5 km) large reservoirs in stable areas, incidence decreased from 43.3 cases per 1000 person-year to 29.7 cases per 1000 person-year between 2000 and 2015 while the absolute number of cases increased from just under 3.5 million in 2000 to just over 3.5 million in 2015. In communities close to (< 5 km) reservoirs in unstable areas, the incidence of malaria roughly halved from just over 25 cases per 1000 person-year in 2000 to just under 12 cases per 1000 person-year in 2015 while the absolute number of cases increased from just over 481,000 in 2000 to just over 639,000 in 2015. Absolute numbers of malaria cases increased over this period, due mainly to population growth.

Against a background of declining incidence, the impact of dams appears to decrease in stable areas and increase in unstable areas, between 2000 and 2015 (Fig. 3). In stable areas, in 2000, the average annual malaria incidence in communities close to (< 5 km) reservoir shorelines were more than 1.7 times higher (P < 0.05) than communities farther away (5–10 km). In 2015, the Relative Risk (RR) of malaria incidence in communities living close to the reservoirs (< 5 km) was only about 1.3 when compared to those living farther away (P = 0.05). By contrast, in unstable areas, analogous RR increased from just over 1.5 in 2000 to more than 1.9 in 2015 (P < 0.05).

The absolute number of malaria cases associated with large dams was more than 1.6 million in 2000. The magnitude of impact may partially result from relatively low transmission in control communities in 2000. Subsequent years followed a more logical trend: 679,000 cases in 2005, 798.000 cases in 2010, and 1.2 million cases in 2015 (Fig. 4). The majority of cases were found around dams in stable areas, due to the greater population and higher incidence in these areas.

Different environmental factors were assessed to determine their relationship with malaria transmission around dams (Table 3). The correlation between malaria incidence and slope and seasonally-submerged shoreline area were highly significant (P < 0.001). Rainfall and minimum temperature were also significantly (P < 0.05) correlated with malaria incidence. Slope had the highest negative correlation; the steeper the slope, the lower the malaria incidence.

Slope alone was found to be the most important factor explaining 46.8% (P < 0.001) of the variation in malaria incidence around large dams (Table 4). A unit degree increase in slope was associated with a 9.4 and 12.5 unit decrease in malaria incidence in stable and unstable areas, respectively. Slope, rainfall and minimum temperature together explained 74.2% and 81.3% of the variation in malaria transmission around reservoirs in stable and unstable areas, respectively.

Gentler, more gradual slopes in the seasonally-submerged areas of reservoirs are associated with much greater malaria transmission (Fig. 5). Considering communities around dams (< 5 km) in both stable and unstable zones, malaria incidence dropped from 68 cases per 1000 person-year to 45 cases per 1000 person-year when slope increased from < 1° to 1–5°. Slopes between 5 and 10°, and greater than 10°, were associated with around 30 cases per 1000 person-year. Ultimately, therefore, slopes less than 5 degrees, and especially less than 1° seem extremely conducive for malaria transmission.

In the present study, the findings indicated that 659,000–1.6 million annual malaria cases were attributable to large dams and that 14.4–18.7 million people were at risk of contracting malaria due to large dams in Africa each year. These numbers generally trended upwards over the years of analysis. The study also found that the relative impacts of large dams in areas of unstable transmission are not just statistically greater but such impacts are trending upwards even more than in stable areas. Finally, the present findings reveal that topography around a specific dam site (slope) is the most important factor affecting malaria incidence. The steeper the slope the lower the chance to create transient shallow puddles for mosquito breeding. Reservoir water levels typically fluctuate through a year and during periods of drawdown (i.e. falling water-levels) flat shorelines often create numerous long-lived puddles disconnected from the main water body, which are often ideal for mosquito breeding [12].

Consistent with broader evidence on the results of the Roll Back Malaria (RBM) program in Africa [28], the present study highlighted how malaria incidence is decreasing in both dam and non-dam communities. Unlike previous work [25], the results herein point to a greater aggregate effect of dams in areas of stable transmission, due to the considerably greater population in such areas. Generally, the impact of dams on relative risk of malaria tends to be greater as transmission levels decrease. According to the World Health Organization report [27], annual malaria incidence decreased by 52% between 2000 and 2015. Around African dams, a similar magnitude reduction of malaria incidence was estimated during the same period in unstable areas (54%) but only a 31% reduction was estimated in stable areas. Intensive malaria control is thus required to reduce the malaria impact of dams in stable and unstable areas.

The present study found a growing number of dams in areas with unstable transmission while the number of dams situated in areas of stable transmission reduced between 2000 and 2015. The changing estimates of number of dams in different transmission zones evidenced between 2000 and 2015 is explained primarily by shifting boundaries of malaria transmission stability. Changes in boundaries of stability may result from improved malaria control efforts and climate changes that have lowered transmission in some areas such that they have moved from stable to unstable transmission. Ultimately, expansion of unstable areas and contraction of stable areas has resulted in an increase in the number of dams in unstable areas.

Juxtaposing central findings from this study—(i) rates of malaria are declining, near and far from dams, (ii) the number of dams in unstable zones of malaria transmission is increasing, and (iii) the relative impact of dams in areas of unstable transmission intensifies with decreased transmission—suggests that dams may pose a significant barrier to malaria reduction and eradication in Africa. Indeed, it would appear that the more progress is made to reduce malaria transmission through national, regional and international programs, the more the confounding impact of dams will become. Therefore, intense, direct focus on reducing dams’ impacts is called for, and the time for sidelining dams’ impacts as a relatively peripheral issue to malaria control may soon be over.

Direct focus on dams’ impacts likely requires a central focus on reservoir shoreline slope. Hydrologically, slope acts as a proxy for the likelihood of an area to retain surface water. Gentle slope generally corresponds to poor drainage, thereby promoting persistence of surface water bodies and the formation of stable pools convenient for mosquito breeding. In contrast, steeper slope facilitates drainage and reduces the likelihood that pools will form for periods of sufficient duration for mosquitoes to complete their aquatic stages. The findings presented in this paper are consistent with scant evidence from elsewhere. A study in Zambia [39] showed a small change in slope affects the risk of malaria; households with malaria positive residents were on flatter areas (slope of 0.024° ± 0.014°) than malaria negative households (slope of 0.031° ± 0.023°; P = 0.04). In western Kenya, valleys with gentle slopes were characterized by slow moving rivers, poor drainage and with large surfaces to hold water suitable for larval breeding while valleys with steeper slopes were characterized by fast running rivers in the valley bottoms, that were unsuitable for mosquito breeding [40]. Jobin [23] acknowledged the importance of the shoreline topography on malaria transmission around dams as gauged from professional judgement, and Endo et al. [24] explored this issue in laboratory experiments.

Reservoirs also affect the groundwater table. The presence of the reservoir induces higher groundwater levels close to the reservoir than further away. Surface pools that can serve as mosquito breeding habitats tend to be stable in areas where drainage may be inhibited because the groundwater level is high and close to the surface. Depending on micro-topography, sometimes groundwater will seep into depressions creating longer-lived, semi-permanent pools [40]. Such stable habitats are often rich in nutrients and maybe partly covered by vegetation thereby creating ideal breeding habitat for malaria mosquitoes. Moreover, agricultural practices are more common on gently sloping shorelines and such activities tend to increase the availability of breeding habitat for malaria mosquitoes. For example, livestock grazing on the shoreline of the Koka reservoir in Ethiopia created numerous water filled hoof prints: ideal habitat for mosquito larvae [12].

The findings regarding shoreline slope are particularly relevant for impact assessment, mining water storage facilities and environmental control dams. Scuddler [41] indicated the critical need for impact assessment studies to determine social, environmental and institutional costs of large dams. The present work further adds the public health challenges associated with large dams if left unchecked during planning and implementation.

The present study has several limitations. First, average shoreline slope was used in our calculations. In reality shoreline slope will vary around a reservoir and may affect where communities are located which will in turn affect the overall impact of any dam. Second, 25% of the existing supposedly georeferenced dams could not be correctly located so were excluded in this study. Overall, application of a rigorous approach for reservoir delineation led to substantial modifications to dam locations and considerable reductions in the number of dams utilized relative to previous work [25]. Considering that the initial set of georeferenced dams represents only about two-thirds of the total number of dams that are known to exist in Africa [29], findings generated from this work may ultimately comprise less than half of the total number of currently existing large dams in Africa. The impacts identified herein, therefore, reflect a conservative estimation of large dams’ impact. Third, MAP data fills in “no-data” areas with model extrapolation using, amongst others, rainfall and temperature data. Thus, it is not surprising that our results showed significant correlation with these variables. However, slope is not an environmental factor used in the MAP modelling, so the fact that slope emerged as such a strong explanatory variable for variation in malaria incidence—despite potentially inflated model predisposition toward climate variables—underlines the significance of this paper’s major finding. Because this study was a statistical investigation, rigorous exploration of the biophysical processes that explain these relationships should be the central focus of follow-on work.

As global malaria efforts continue to shrink malaria distribution, dam associated reservoirs in sub-Saharan Africa will continue to offer foci for malaria transmission. In particular, large reservoirs with shallow shoreline slope seem to facilitate malaria transmission by enhancing breeding habitat for malaria vector mosquitoes. This calls for more rigorous impact assessments to determine likely malaria impacts while planning African dams. Explicit consideration of the presence and degree of topographic characteristics at potential dam sites could enable water managers, in collaboration with health planners, to better address the pervasive water–malaria challenge.