Study area
Four major river basins, located across different sub-regions of SSA, were selected for this study: Limpopo, Omo-Turkana, Volta, and Zambezi (Fig. 1). These basins were selected to (i) foster inclusion of enable different African regions and (ii) ensure focus on basins with sufficient data availability.
source malaria data23 on ArcGIS software (version 10.5. 1, Environmental Systems Research Institute Inc, Redlands, CA, USA, 2016)].
Distribution of large and small dams in Limpopo, Volta, Zambezi and Omo-Turkana basins by malaria stability zone. [The figure was made using open-
The Limpopo River basin is located in southern Africa. Draining an area of approximately 408,000 km2, the Limpopo River basin is distributed among South Africa (45%), Botswana (20%), Zimbabwe (15%) and Mozambique (20%). About 14 million people live in this basin. The climate of the Limpopo River basin varies along the path of the river from a temperate climate in the west to a subtropical climate at the river mouth in Mozambique. The hydrology of the Limpopo River basin is influenced by the highly seasonal distribution of rainfall over the catchment. About 95% of rain falls between October and April with a peak normally in February. Temperature varies from 30 to 34 °C in summer and 22–26 °C in winter15.
The Volta River basin is located in West Africa with a population of over 23 million. Draining an area of 409,000 km2 the basin is spread across six countries: Benin (4%), Burkina Faso (42%), Cote d’Ivoire (3%), Ghana (41%), Mali (4%) and Togo (6%). Average annual rainfall varies across the basin from approximately 1600 mm in the southeast, to about 360 mm in the north. Annual mean temperatures in the basin vary from 27 to 30 °C16. The main rainy season is between March and October.
The Zambezi River basin is located in southern Africa. Draining an area of 1.34 million km2, the basin is spread across eight countries: Angola (19%), Botswana (1%), Namibia (1%) Benin (4%), Zimbabwe (16%), Zambia (42%), Tanzania (2%), Malawi (8%) and Mozambique (12%). The population of the Zambezi basin is estimated to be about 32 million. Annual rainfall in the basin ranges from 550 mm in the south to 1800 mm in the north. The annual mean temperatures ranges from 18 °C at higher elevations in the south of the basin to 26 °C for low elevations in the delta in Mozambique17.
The Omo-Turkana Basin covers approximately 131,000 km2, stretching from southern Ethiopia to northern Kenya. Hydrologically, the basin is dominated by Lake Turkana, with the Omo River, which drains the Ethiopian portion of the basin, supplying 90% of the inflow to the lake. The basin is home to approximately 15 million people, the majority of whom live in the Ethiopian highlands, in the north. The annual mean temperature ranges from 24 °C in the north to 29 °C in the south. The mean annual rainfall ranges from 250 mm in the south to 500 mm in the north18.
Data sources
Dam data
Small dams
Data on location and size of small dams are not readily available in either global or regional data sets. The European Commission’s Joint Research Center (JRC) Yearly Water Classification History v1.0 data set was used to identify water bodies in each of the four basins19. Water bodies less than 100 ha and greater than 2 ha were identified. All were checked with Google Earth images to distinguish between reservoirs and natural water bodies (Supplementary Fig. S1). Ultimately, a total of 4907 small dams located in the four basins were identified and included in the analyses.
Large dams
For large dams, the FAO African Dams Database20, International Commission for Large dams (ICOLD)21 and the International Rivers Database22, which together contain 1286 georeferenced African large dams, were utilized. The accuracy of dam locations was first verified with Google Earth. When the location of a dam did not precisely match the coordinates stipulated in either of the two databases, manual corrections were made by adjusting the coordinates of a dam to its location as shown in Google Earth (see Supplementary Information). Dams for which precise locations could not be determined, as well as dams without reservoirs (i.e., run-of-river schemes), were removed. Ultimately, across the four basins, a total of 258 large dams with confirmed georeferenced locations were identified and included in the analyses.
Perimeters of large and small dam reservoirs
Reservoir perimeters of both large and small dams were extracted from the European Commission’s Joint Research Center (JRC) global surface water datasets19, published through the Google Earth Engine. This dataset includes maps of the location and temporal variability in maximum perimeter records of the global surface water coverage from 1984 to 2015. In this study, the maximum perimeter records were used in each year of 2000, 2005, 2010 and 2015. The data were exported to ArcGIS.
Data on anopheles mosquito distribution
Data for vector distribution were obtained from the Malaria Atlas Project (MAP) database23. The MAP database contains a georeferenced illustration of the major malaria vector species in different malaria-endemic areas in Africa.
Malaria data
Annual malaria incidence data were obtained from the MAP database. We acquired data for the years 2000, 2005, 2010 and 2015. These years were selected to align with updates to Worldpop population data24, which are recomputed every five years. MAP produced a 1 km resolution continuous map of annual malaria incidence for Africa based on 33,761 studies across the region. We imported these data to ArcGIS for analyses. Annual malaria incidence was determined as the number of cases per 1000 population. To ascertain the impact of dams on malaria incidence rates as a function of distance from the reservoir perimeter, we created two distance zones: 0–5 km (at risk) and 5–10 km (control). When distance zones were overlapping for two or more nearby dams, areas were assigned to the closest distance cohort. Populations residing more than 5 km from a reservoir perimeter (large or small) were considered to be free of risk from dam induced malaria transmission because the maximum mosquitoes’ flight range is considered to be < 5 km25. Hence, the 5–10 km zone served as a control.
Population data
Annual population data of SSA were obtained from the Worldpop database24. A 1 × 1 km gridded population map was imported to ArcGIS for analyses. The total number of people living in each distance cohort was determined for each reservoir every 5 years for the period 2000–2015.
Data analyses
Mapping vector distribution around small and large dams
To illustrate the locations of reservoirs with respect to different Anopheles species, we superimposed malaria vector distribution obtained from the MAP database on the small and large dams in the four basins to show the risk of malaria transmission around reservoirs in areas with different vector compositions.
Distribution of small and large dams in areas of different stability
Different studies use different approaches to describe malaria stability26,27. We followed Gething et al.26 where areas were categorized as stable (> 0.1 malaria cases per 1000 population), unstable (≤ 0.1 malaria cases per 1000 population) and no malaria (zero malaria incidence) based on the level of malaria incidence in each of the four years: 2000, 2005, 2010, and 2015. The number of dams in each of the three stability categories for each of the four years was determined, as well as the population at-risk of dam-related malaria (i.e., < 5 km from reservoir shorelines).
Malaria incidence around small and large dams
The number of annual malaria cases was estimated for the two distance cohorts (< 5 km and 5–10 km) by multiplying malaria incidence rates by the population in each zone. Repeated analysis of variance (ANOVA) was applied to determine differences in malaria incidence between the two cohorts, followed by post hoc HSD Tukey’s test28.
Incidence per km of reservoir shoreline
Entomological investigation of dam-associated malaria transmission29 suggests that reservoir shoreline constitutes the most important breeding habitat and malaria risk factor in dam-affected geographies. As such, comparison of the relative malaria impact of small and large dams was determined by computing the average number of malaria cases per km of reservoir shoreline. This was calculated for each reservoir by dividing malaria incidence in the < 5 km cohort by the reservoir perimeter. The average incidence per km was computed separately for small and large reservoirs in each basin for each of the four years 2000, 2005, 2010 and 2015.
Malaria cases attributable to dams
The annual number of malaria cases attributed to dams was determined for areas of both unstable and stable transmission. To do so, the rate of transmission in communities 5–10 km from reservoirs to the population living within 5 km of reservoirs was applied to gauge the number of malaria cases that could be presumed to occur in the absence of the dam. This was calculated as (I1–I2) P, where I1 is malaria incidence in communities living within 5 km of reservoirs, I2 is malaria incidence in communities living between 5 and 10 km from reservoirs, and P is the total human population in communities living within 5 km of reservoirs.
Source: Ecology - nature.com
