Malaria and infectious diseases in West Africa during the 21st Century

Malaria in Burkina Faso

Yves M. Tourre (LDEO of Columbia University, USA),Cécile Vignolles (CNES, Toulouse), Christian Viel (Météo-France, Toulouse), Fazlay Faruque (University of Mississipi Medical Center, USA). John Malone (Louisiana State Univesity)


An unexpected result was obtained from the Paluclim project on Malaria dynamics in Burkina Faso. Rainfall is the confounding factor for the density of malaria vectors, but environmental conditions are to change during the upcoming years. Using an impact model for malaria risks and which included climate components, several predictions for rainfall and temperature indices were made for different -temporal scales (i.e., seasonal, inter-annual, low-frequency and climate change scenarii). If a definite link exists between low-frequency rainfall variability in the Sahel and the Atlantic Multi-decadal Oscillation, (AMO), the predicted extreme temperature increase during the 21st century is to reduce malaria risks there. Temperature increase could thus be seen as becoming the new limiting factor for malaria (and other vector borne diseases) in the Sahelian regions. This new result should be taken into account by stakeholders involved with public health and geographic information systems (HIS and GIS).

Keywords: Malaria, Atlantic Multi-decadal Oscillation, Sahelian rainfall, Climate Change

I. Introduction

Figure 1

The impact of climate variability and climate change in Burkina Faso (Figure 1) during the 21st century have been investigated through the Paluclim project (Vignolles et al., 2016). The project used the tele-epidemiology approach, and was applied to rural malaria in the Nouna region (northern Burkina Faso, Sahel). One of its objectives has been to study the multi-scale climate variability (including climate change), conditions and impacts on malaria risks. After having investigating the natural rainfall variability and climate indices for the last 30 years (from historical data), the important parameter which conditioned malaria outbreaks is the monthly total rainfall. It became clear one of the key climate signals modulating Sahelian rainfall amount is the low-frequency oscillation called the Atlantic Multi-decadal Oscillation or AMO The latter changed phases in the mid 1990s (i.e., from negative to positive values) were associated with rainfall increased over the Sahel (see also Zhang et al. 2006; and Paz et al., 2008).

Climate prediction associated with rainfall variability and other parameters during the 21st Century had to be evaluated. For example it is important to know what could be the impacts of Climate Change on AMO variability, environmental data such as rainfall output, temperature ranges, malaria risks and more generally on Sahelian infectious diseases.

II. Data and Method

Indices (IND) for environmental conditions and malaria risks had been computed (see Vignolles et al, 20016 for details). IND for malaria diffusion are computed from the a seasonal impact model following Craig et al., (1999) and Tanser et al., 2003. Index values are 0 (for unsuitable conditions) and 1 (for suitable conditions). IND are based upon rainfall, temperatures (min. and max.) and RH. The suitable conditions, linked with temperatures, are obtained from studying the relationship between the temperature and sporogony, vectors survival, and length of the larval cycle once again. The IND are, of course, an over simplification, since other parameters are also involved with infectious diseases, such as vector densities, and human susceptibility. The indices thus cannot be directly compared to in-situ data and, in addition, do not take into account the vectors’ aggressiveness. Nevertheless their relative simplicity allow for diagnostics of the malaria diffusion under different climate conditions and temporal scales.

Then observed historical data and simulated/predicted data from different models (see hereafter) were adjusted from the quantile-quantile method during the 1983-2005 period. Finally the adjusted values were converted into the IND indices for rainfall (INDp) and temperature (INDt) for malaria risks.

The prediction and climate scenarii for the 21st Century were subsequently obtained through the global simulations from the Coupled Model Inter-comparison Project Phase 5 or CMIP-5 ( (Taylor et al. (2012). More than 300 projections or Representative Concentration Pathway (RCP) have been analyzed. For the Paluclim project scenarii RCP45 and RCP85 were picked. The RCP45 scenario is for a radiative forcing of 4.5 W.m2 in 2100, equivalent to 660 eqCO2 and leading to a plateau at the end of the period. The RCP85 scenario, is for a radiative forcing larger than 8.5 W.m2 on and after 2100, equivalent to 1370 eqCO2 with a continuous increase. The latter scenario is been seen as the most pessimistic and extreme scenario (i.e., with very little regulations on green house gases emissions).

For rainfall output, 8 models were used, whilst 6 models were used for T min. and T max. temperatures, all obtained from the Institute Pierre Simon Laplace (IPSL-CM5A-LR and MR). Some models are from the Canadian Centre for Climate Modeling and Analysis (CCma-CanESM2), the Centre National de Recherches Météorologiques (CNRM-CM5), Hadgem2-ES, INM-CM4, Atmosphere and Ocean Research Institute (The University of Tokyo) or MIROC5 and NCC. The grid-point systems for all models is of 2.5°. The multi-model (or ensemble) approach was used, to assess environmental climate conditions associated with malaria riks. Four gridpoints were taken around Nouna, from which time-series for averaged rainfall, minimum and maximum temperatures were extracted. Since only one simulation is used per model, the uncertainties could not be included in the results. Nevertheless the multi-model approach allows assessment of environmental conditions associated with malaria risks during the 21st century (Knutti et al., 2010). III. The evolving climate in the Nouna region and its impacts on malaria

Figure 2

In Figure 2 the capacity for models to reproduce (after the quantile-quantile adjustment) the indices values for favorable malaria conditions, and from rainfall and temperature indices are shown. The multi-model ensemble displays a large spread for the rainfall INDp index (Figure 2, top). The blue curve is the mean for historical simulation, the red curve is the mean from observations. In spite of the large dispersion the index is outside the possible values particularly during the positive phase of the AMO, post 1996. This strongly suggests an underestimation of rainfall amount by the models, or shorter rainy seasons. This is not the case for the INDt index (Figure 2, bottom), where simulated and observed values are much more coherent, i.e., with a much smaller spread.

A. Climate evolution and malaria risks

The lack of the AMO climate signal is obvious when the probability density function of the INDp (according to the observed and simulated AMO phases), are displayed for historical values (Figure 3, top). During the negative phase of the simulated AMO the density peak is clearly established for an index value of 2 (light blue curve). On the contrary during a simulated positive AMO (dark blue curve), the density curve displays a plateau with index values of 2-3. For the observed values (red curve) the index values are 3-4. It is can be seen that there is a lag between simulated and observed values. In general the favored hypothesis from the probability density functions is that the simulated rainy seasons are in general shorter during positive phase of AMO. Since the AMO is a multi-decadal climate signal, variability of the mean, averaged over 30-year periods, was computed for the 2010-2100 period using RCP 45 and RCP85 scenarii. The mean for the density functions along with their spread (grey shading) are displayed in Figure 3 (RCP45 middle, and RCP85 bottom).

Except for the 2040-2070 period and for the RCP85 scenario, the distribution is somewhat equivalent to the one obtained during the AMO negative phase (historical data). In general the rainfall conditions are not favorable but the spread around the means is quite large Under the RCP85 scenario, the 2040-2070 conditions more similar to that of a positive AMO are shown.


Figure 4, the same as for Figure 3 is shown but for INDt. At the top of Figure 4 the historical values of the index are shown also as a function of AMO phases, whilst at the bottom the density functions are displayed every 30 years starting in 2010 (middle for the RCP45 scenario, bottom for the RCP 85 scenario).

Only the red curve (Figure 4, top) displays a maximum value of 7 for the density function during the AMO positive phase. For the other curves, the peak is for an index of 6 whilst rainfall taking the AMO phasing into account is not totally evident, the overall structures are well differentiated.

For the evolutions through 30-years period starting in 2010 (Figure 4, bottom) and for the two scenarii RCP45 (middle) and RCP85 (bottom) the conditions during the 2010-2040 periods are already less favorable than during the historical period with a peak of 5 for the index. In the average the RCP45 does not show major changes before 2070, the time during which the temperature conditions start to make it difficult for survival of mosquitoes and thus of malaria diffusion.. On the contrary the RCP85 scenario display changes as soon as 2040 with very low peak value of 3 for the index.The 2070-2100 period display an extremely low peak value of 1 for the index.. Thus whatever the rainfall conditions between 2070 and 2100 might be the environmental conditions could be strongly diminished for malaria transmission.


Figure 5 a large interannual variability is seen for INDp (Figure 5, left) with a small tendency leading to unfavorable conditions when compared to the historical period. Before 2040, the RCP85 scenario seems more favorable than the RCP45 scenario. But due to the large spread of INDp no significant conclusions for malaria risks can be drawn.

On the contrary the tendency of the INDt (Figure 5, right) indicates less and less favorable conditions for malaria in the Nouna region, for both RCP45 and 85 scenarii. This large tendency is associated with the evolution of the annual maximum temperature.

Finally the temperature evolution itself is displayed in

Figure 6 for the period 1980-2100, with the dashed orange line at the bottom representing the temperature mean for the so-called historical period (1983-2005), whilst the dashed red line is the limit for temperature at which the conditions will become extremely unfavorable (2090). The RCP45 scenario in light blue is for an increase of the mean temperature in Nouna of 3°C while the RCP85 scenario in dark blue is for more than 5°C. The temperature is thus attaining 40 °C (RCP85) at the end of the 21st century which explains the rapidly decreasing values of the INDt index in Figure 5.

B. Malaria risks evolution using time slices of 30-year periods

From the Figures 2, 3, and 4, it is seen that no obvious signs for malaria risks increase are linked to rainfall. Indeed the rainfall distribution is somewhat similar to that during the 1983-2005 period. Thus over the Nouna region the climate tendency, without the natural variability (including that of the AMO), does not change significantly the rainfall index for changes in malaria risks.

For the maximum temperature evolution during the 2010-2040 period conditions are already less favorable than during the historical period with a peak of 5 for the temperature index. In the mean the RCP45scenario does not reflects major changes before the 2070-2100 period, when temperature is already highly limiting the malaria risks. On the contrary the RCP85 scenario leads to major changes as soon as the middle of the 21st Century with a peak of 3. During the 2070-2100 period temperature conditions are associated with a very low index of 1 (Figure 5).

IV. Conclusion

It has been shown that the important parameter for malaria risks is the monthly total rainfall with a 80 mm threshold for the first month of a given 3-month sequence. Moreover the knowledge of the AMO phases and its low-frequency variability is thus very important for adaptation procedures by the regional Health Information Systems (HIS).

According to the climate projections for the 21st Century, and preliminary results from models also used for the CORDEX ( experiment, environmental conditions are going to change, particularly insofar as temperature is concerned. Indeed it is shown that the large temperature increase in the Nouna region should lead to malaria risks reduction. Temperature becomes then the limiting factor for at least the malaria risks in the Nouna region.

‘Malaria is of great public health concern, and seems likely to be the vector-borne disease most sensitive to long-term climate change. Malaria varies seasonally in highly endemic areas’ (from Patz et al, 2003). Clear effects of climate change have already been established for several human infectious diseases, such as malaria, cholera, dengue. Nevertheless the complexities of these systems offer challenges for the distinction between climatic and non-climatic effects (Ostfeld, 2008). In this paper the above and new results highlight the potential for malaria and other infectious diseases in Africa to invade southern Europe where conditions might be more suited (i.e., lower maximum temperatures than in the Sahel) for the mosquitoes ecological and meteorological niches. For example an assessment in Portugal projected an increase in the number of days per year suitable for malaria transmission (Casimiro et al., 2006). Thus, climatic change factors may favor transmission, increased vector density, re-emergence of malaria in Europe (Rogers and Randolph, 2000). It is recognize that socioeconomic, building codes, land use, treatment could also slow down the likelihood of climate-related epidemics. In any case whilst the rainfall amount and threshold certainly determines the mosquito abundance, the temperature will affect the malaria parasites. Malaria risks should thus be considerably reduced at the end of the 21st Century in the Sahel, when compared to actual risks. To that effect Li et al. (2007) have developed a Web-based real-time syndromic surveillance system with GIS disease mapping capabilities. This Web-based tool of value to epidemiologists and public health officials for the interpretation and analysis of both routine and new outbreak-related health data, which can be linked to climate change impacts (see also Healthy Futures inTaylor et al., 2016).

IV. Acknowledgement

The authors would like to thank CNES for managing the Paluclim project, Drs. Philippe Dandin, Phippe Bougeaut former directors of Meteo-France Climatology and CNRM divisions of Météo-France. Tourre would like to thank Dr. Sean C. Solomon the new director of Lamont-Doherty Earth Observatory (LDEO) for supporting his research. This is LDEO contribution # XXXX.

V. References

Casimiro E, Calheiros J, Santos FD, Kovats S. National assessment of human health effects of climate change in Portugal: approach and key findings. Environ Health Perspect. 2006;114(12):1950-6. Craig, M. H., R. W. Snow, and D. le Sueur, 1999, A climate-based distribution model of malaria transmission in sub-Saharan Africa: Parasitology Today, v. 15, no. 3, 105-111. Knutti R., Abramowitz G., Collins M., Eyring V., Gleckler P.J., Hewitson B. and L. Mearns, 2010, Good Practice Guidance Paper on Assessing and Combining Multi Model Climate Projections. In: Meeting Report of the Intergovernmental Panel on Climate Change Expert Meeting on Assessing and Combining Multi Model Climate Projections. IPCC Working Group I Technical Support Unit, University of Bern, Bern, Switzerland. Li, T., F. Faruque, W. Williams, and R. Finley, 2007, Application of Geographic Information Systems to Syndromic Surveillance. Geospatial Health Ostfeld, R. 2008, Infectious Disease Ecology : Effects of Ecosystems on Disease and of Disease on Ecosystems. Edited by Richard S. Ostfeld , Edited by Felicia Keesing , Edited by Valerie T. Eviner. Princeton U. Press. 520 p. Patz, J. A., A. Githeko, J-P., Mc Carthy.S. Hussein, U. Confalonieri, and N. De Wet, 2003, Climate Change and Human Health: Climate change and infectious diseases. WHO report, Chap 6, 103-132

Paz, S., Y. M. Tourre, and J. Brolley, 2008, Multitemporal climate variability over the Atlantic Ocean and Eurasia: linkages with Mediterranean and West African climate. Atmos. Sci. Lett., 9, 4, 196-201 pp. doi: 10.1002/asl.181 Rogers D. J, and S. E. Randolph 2000, The global spread of malaria in a future, warmer world. Science, 289, 1763-1766.

Tanser, F. C., B. Sharp, and D. le Sueur, 2003, Potential effect of climate change on malaria transmission in Africa. The Lancet, 362, 1792-1798 pp.

Taylor K. E., Stouffer R. J. and G. A. Meehl, 2012, An overview of CMIP5 and the experiment design. BAMS, 485-498 pp. Taylor, D., Kienberger, S, Malone, J. B., and A. M. Tompkins, 2016, Health, environmental change and adaptive capacity; mapping, examining and anticipating future risks of water-related vector-borne diseases in Eastern Africa. Geospatial Health, doi: 10.4081/gh.2016.464. Vignolles, C., Sauerborn, R., Dambach, P., Viel, C., Jean-Michel Soubeyroux, J-M., Sié, A., Rogier, C., Yves M. Tourre. Y. M., 2016, Re-emerging Malaria vectors in rural Sahel (Nouna, Burkina Faso): The Paluclim project . ISPRS-Prague Report. Commission VI, WG VI/4.

Zhang, R., and T. L. Delworth, 2006, Impact of Atlantic multidecadal oscillations on India/Sahel rainfall and Atlantic hurricanes, Geophys. Res. Lett., 33, L17712, doi:10.1029/2006GL026267.

List of Figures and Captions

Figure 1. Localization of the Nouna region in Burkina Faso (4.1/3.5° W and 12.4/13° N) in West Africa.

Figure 2. Rainfall index INDp (top) and temperature index INDt (bottom) during the historical period after the quantile-quantile adjustment for the simulated data. In red are the observations in the Nouna region, in blue are the mean simulated values. In grey is the spread from the models aroundthe mean. After 1996, the INDp values are not only larger than observed values but are getting out of the statistical range from 2000 until 2004.

Figure 3. (top) Probability density function for INDp values (abscissa) during the 1983-2005 period, taking into account the influence of the AMO. In red (orange) are observations during years when the AMO is positive (negative). Dark (light) blue curves are for mean simulations when the AMO is positive (negative). The dark/light grey shaded zones are for the spread from the mean during positive (negative) phase of the AMO. (middle) The density distribution of the index is displayed for RCP45 every 30 years starting in 2010. (bottom) The density distribution of the index is displayed for the RCP85 secneario also starting in 2010. The blue curve is for the mean, the grey shading represents the spread around the mean. Peak values of the index are identified by a line for each 30-year period.

Figure 4. Same as in Figure 3 except for the probability density function of INDt

Figure 5. Mean evolution (with spread) of the yearly index values for favorable conditions of the rainfall index INDp (left) and the temperature index INDt (right) for the 1983-2100 period. Post-2005 (highlighted by the black rectangle), the evolutions are for the climate scenarii RCP45 (top) and RCP85 (bottom). The dark blue (light blue) shaded areas are for values above (below) the means computed for the 1983 – 2005 historical period. Within the black rectangle results until 2005 are very similar for both scenario (RCP45 and RCP85) and parameters.

Figure 6. Evolution of mean maximum temperature from the multi-models ensemble with two different RCP scenarii as described in the text. Once again until 2005, evolutions are similar (within grey-shaded rectangle), and then spread as a function of RCP45 (light blue) and RCP85 (dark blue) scenario, particularly after year 2035. The red dashed line is for the upper limit of Tx for malaria diffusion, and which is attained at the end of the 21st Century (around 2085) by using the RCP85 scenario.

In the same section…

Back to top