Risk assessment with regard to the occurrence of malaria in Africa under the influence of observed and projected climate change

Malaria is one of the most serious health problems in the world. The projected climate change will probably alter the range and transmission potential of malaria in Africa. In this study, potential changes in the malaria transmission are assessed by forcing three malaria models with bias-corrected data from ensemble scenario runs of a state-of-the-art regional climate model. The Liverpool Malaria Model (LMM) from the Geography Department of the University of Liverpool is utilised. The LMM simulates the spread of malaria at a daily resolution using daily mean temperature and 10-day accumulated precipitation. The simulation of some key processes has been modified in the model, in order to reflect a more physical relationship. An extensive literature survey with regard to entomological and parasitological malaria variables enables the calibration and validation of a new LMM version. Comparison of this version with the original model exhibits marked improvements. The new version demonstrates a realistic simulation of entomological variables and of the malaria season, as well as correctly reproduces the epidemic poten tial at fringes of endemic malaria areas. Various sensitivity experiments reveal that the LMM is fairly sensitive to values of its required parameters. Effects of climatic changes on the malaria season are additionally verified by the MARA Seasonality Model (MSM). The Garki model finally enables the completion of the malaria picture in terms of the immune status and the infectiousness of different population groups, as well as relative to the age-dependent prevalence structure. In every case three ensemble runs were performed on a 0.5° grid. The LMM was driven for the present-day climate (1960-2000) by bias-corrected data from the REgional MOdel (REMO), with a land use and land cover specified by the Food and Agriculture Organization (FAO). Malaria projections were carried out for 2001-2050 according to the climate scenarios A1B and B1 as well as FAO land use and land cover changes. Garki model runs were subsequently forced by the Entomological Inoculation Rate (EIR) from the LMM. Finally, additional results relative to the malaria season were produced by MSM. For the present-day climate (1960-2000), the highest biting rates are simulated for Equatorial Africa. The malaria runs show a decrease in the malaria spread from Central Africa towards the Sahel. The length of the malaria season is closely related to monsoon rainfall. The model simulations show a marked influence of mountainous areas causing a complex pattern of the spread of malaria in East Africa. The malaria infected population reveals the expected peak in children below an age of about five years. Regions of epidemic malaria occurrence, as defined by the coefficient of variation of the annual parasite prevalence maximum, are found along a band in the northern Sahel. Farther south, malaria occurs more regularly and is therefore characterised as endemic. Epidemic-prone areas are additionally identified at various highland territories, as well as in arid and semi-arid zones of the Greater Horn of Africa. No adequate immune protection of the population was found for these areas. Largely due to land surface degradation, REMO simulates a prominent surface warming and a significant reduction in the annual rainfall amount over most of tropical Africa in either climate change scenario. Assuming no future human-imposed constraints on malaria transmission, changes in temperature and precipitation will alter the future geographic distribution of malaria. In the northern part of sub-Saharan Africa, the precipitation decline will force significant decreases of the malaria transmission in the Sahel. In addition to the withdrawal of malaria transmission along the fringe of the Sahara, the frequency of malaria occurrence will be reduced for several grid boxes of the Sahel. As a result, epidemics in these more densely populated areas will become more likely, in particularly as adults lose their immunity. The level of malaria prevalence farther south will remain stable for most areas. However, the start of the malaria season will be delayed and the transmission is expected to cease earlier. Most pronounced changes in Africa are found for East Africa. Significantly higher temperatures and slightly higher rainfall cause a substantial increase in the season length and parasite prevalence in formerly epidemic-prone areas. Territories formerly unsuitable for malaria will become suitable under the warmer future climate. The simulations indicate changes in the highland epidemic risk. At most grid boxes malaria transmission will stabilise below about 2000 m. At these altitudes the population will improve their immune status. In contrast, malaria will climb to formerly malaria-free zones above these levels enforcing the probability of malaria epidemics