Climate Suitability: For Stable Malaria Transmission in Zimbabwe Under Different Climate Change Scenarios

GLOBAL CHANGE & HUMAN HEALTH, VOLUME 3, NO. 1 (2002) 42 Kluwer Academic Publishers

Malaria continues to signifi cantly affect African health, society and economy. In sub-Saharan Africa, malaria remains the most common parasitic disease and is the main cause of morbidity and mortality among children less than fi ve years of age and among pregnant women[1]. The 1990 Global Burden of Disease study estimated that malaria accounted for approximately 10.8% of years of life lost across all sub-Saharan Africa[2]. A recent review concluded that roughly 1 million deaths (range 0.74 to 1.3 million) from the direct

effects of malaria occur annually in Africa, more than 75% of them in children[3]. This estimate may double when the indirect effects of malaria, including malaria-related anemia, hypoglycemia, respiratory distress and low birth weight, are included when defi ning the burden of malaria[4].

There has been a global resurgence of malaria over the past two decades; reasons suggested include failure of malaria control programs, population redistribution and growth, changes in land use, increasing prevalence

Climate suitability for stable malaria transmission in Zimbabwe under different climate change scenarios

As climate is one factor determining the potential range of malaria, climate change may work with or against

efforts to bring malaria under control. We developed a model of future climate suitability for stable Plasmodium

falciparum malaria transmission in Zimbabwe. Current climate suitability for stable malaria transmission is

based on the MARA/ARMA model of climatic constraints on the survival and development of the Anopheles

vector and the Plasmodium falciparum malaria parasite. We explored potential future geographic distributions

of malaria using sixteen projections of climate in 2100. The results suggest that, assuming no future human-

imposed constraints on malaria transmission, changes in temperature and precipitation could alter the geographic

distribution of malaria in Zimbabwe, with previously unsuitable areas of dense human population becoming

suitable for transmission. Among all scenarios, the highlands become more suitable for transmission, while the

lowveld and areas currently limited by precipitation show varying degrees of change, depending on climate sensi-

tivity and greenhouse gas emission stabilization scenarios, and depending on the general circulation model used.

The methods employed can be used within or across other African countries.

Jessica Hartman(A Corresponding author), Kristie Ebi(B), K. John McConnell(C), Nathan Chan(D) and John Weyant(E)

A Stanford University 205 E. 78th Apt. 19A, NY, NY 10021 tel (917) 855-9537 e-mail [email protected]

B EPRI, 3412 Hillview Ave, Palo Alto, CA 94304

tel (650) 855-2735, e-mail [email protected]

C University of California at San Francisco, Institute for Health Policy Studies, San Francisco, CA 94118


D Stanford University, 685 Mariposa Ave #6, Mountain View, CA 94041


E Stanford University, Stanford, CA 94305-4023


feature

GLOBAL CHANGE & HUMAN HEALTH, VOLUME 3, NO. 1 (2002)43 Kluwer Academic Publishers

research article

of drug and pesticide resistance, denigration of public health infrastructure and climate variability and change[5,6]. Climate change may work against efforts to bring malaria under control, with a recent model suggesting that increasing temperatures could create environmental conditions that would allow a net increase in the geographic range of areas climatically hospitable for malaria vectors[7]. With some degree of anthropogenic alteration of the climate system highly probable over the next 100 years - recent estimates suggest a global mean surface temperature increase of 1.4 to 5.8C - understanding the range of possible impacts of these climatic changes on efforts to control malaria is imperative, particularly on national and sub-national scales[8,9].

While climate is an important driver of malaria, it is not the only one. The many determinants of malaria rarely act in isolation; these determinants form a web of interconnected infl uences, often with positive feedbacks between malaria transmission and other drivers[10-12]. Determinants of malaria can be divided into climatic and non-climatic drivers. The non-climatic socioeconomic and biological drivers that have a direct impact on malaria include drug and pesticide resistance, deterioration of health care, deterioration of the public health infrastructure (including disease and vector control efforts), demographic change, and changes in land use

patterns. We focused solely on climate in this model; future models will include qualitative or quantitative representations of non-climatic drivers of malaria and their interactions.

Climate and anomalous weather events have a direct infl uence on malaria transmission by either hindering or enhancing vector and parasite development and survival. Numerous laboratory and fi eld studies have analyzed the infl uence of precipitation and temperature, and can be summarized as follows: Climate suitability is a primary determinant of

whether the conditions in a particular location are suitable for stable malaria transmission[13,14].

A change in temperatures may lengthen or shorten the season in which mosquitoes or parasites can survive.

Changes in precipitation or temperature may result in conditions during the season of transmission that are conducive to increased or decreased parasite and vector populations.

Changes in precipitation or temperature may cause previously inhospitable altitudes or ecosystems to become conducive to transmission. Higher altitudes that were formerly too cold or desert fringes that were previously too dry for mosquito populations to develop may be rendered hospitable by small changes in temperature or precipitation.

Despite more than 50 years of vector control, malaria is still a substantial public health threat[37]. More than 4 million of Zimbabwes 12 million inhabitants live in malarious areas. Even with widespread under-reporting and self-treatment, 9,696 deaths and 6.4 million cases of malaria were reported between 1989-96[37]. Three parasites have been identifi ed as causing malaria in Zimbabwe, with Plasmodium falciparum responsible for about 98% of all cases[22]. Of the 38 species of Anopheline mosquitoes in Zimbabwe, Anopheles arabiensis is the only formidable vector of malaria. A. funestus s.s. (endophilic), A. merus (breeds in salt water), and A. quadriannulatus (zoophilic) are also present, yet are not believed to be substantial threats either due to past control efforts or their ecological constraints[38].

Transmission is highly seasonal with the bulk of cases recorded between February and May, and with peak transmission typically occurring in March[22]. Stable transmission is partially defi ned by

temperature and precipitation, with altitude a proxy indicator of temperature. Some areas of Zimbabwe, particularly the northern and southern lowveld regions, have year round malaria transmission with peaks in the austral summer months. Like much of southern Africa, this transmission season corresponds to the onset of summer and the single rainy season. Also similar to many southern and eastern African countries, Zimbabwe has dramatic elevation ranges (Figure 1). The marked gradient of elevation correlates with maximum and minimum temperatures. Figures 2 and 3 show annual mean temperature and total precipitation. This heterogeneity coupled with interannual climatic variability results in constantly shifting fringe areas that are prone to malaria outbreaks, similar to many countries in southern and eastern Africa[38]. High altitude regions, which also are the areas of densest human population, are currently malaria free due to climatic constraints and to a long policy of barrier spraying in the transitional elevation zone[39].

Box The epidemiology of malaria in Zimbabwe


Death, disease and deformityClimate change and malaria in Zimbabwe

Our study investigates the impact of various climate change scenarios on the projected spatial distribution of climatic suitability for stable Plasmodium falciparum malaria transmission. We chose to initially populate our model with data from Zimbabwe because of its heterogeneity of climatic suitability for malaria transmission (see Box for a discussion of the epidemiology of malaria in Zimbabwe). Specifi cally, there are areas in Zimbabwe where the climate is suitable for endemic malaria transmission, areas where transmission is absent, and areas representing gradations of transmission in between. The benefi t of using a country with multiple levels of climatic suitability is that, for each scenario of climate change, we can observe the range of projected fl uctuations in the levels of suitability and we can identify key areas that appear the most vulnerable.

MethodsWe utilized the COSMIC program[15] to generate Zimbabwe-specifi c scenarios of climate change, which we then used as inputs for a model of climate suitability for stable Plasmodium falciparum malaria transmission to generate maps of future transmission potential. Our model was based on the MARA/ARMA decision rules[13,14] and developed using the ArcView 3.2 Geographic Information System[16].

Climate suitability for stable malaria transmission

The MARA/ARMA climatic suitability model for malaria transmission was one tool we used to create projections of future climate suitability in Zimbabwe. This biological model defi nes a set of decision rules based on minimum and mean temperature constraints on the development of the Plasmodium falciparum parasite and the Anopheles vector, and on precipitation constraints on the survival and breeding capacity of the mosquito. MARA/ARMA determined the decision rules by reviewing laboratory and fi eld studies throughout Africa and by evaluating current malaria distribution maps[13,14]. This biological model approximates the current boundaries of malaria distribution across the continent quite well.

MARA/ARMA uses three variables to determine climatic suitability for a particular geographic location: mean monthly temperature, the monthly mean of daily minimum temperatures, and total cumulative monthly precipitation. An important distinction between this model and others is that the MARA/ARMA decision rules were developed using fuzzy logic to resolve the uncertainty in defi ning distinct boundaries to divide malarious from non-malarious regions. Rather than using a Boolean designation of climatically suitable or

not, an area can be assigned a fuzzy logic suitability value ranging from zero (not suitable) to one (suitable). A value of one means that malaria transmission is most likely stable, and zero means that transmission is very unstable, with malaria either absent or with rare epidemics. Values between zero and one (0.1 to 0.9) represent a gradient from unstable to increasingly stable transmission. For all variables, assignment of fuzzy logic values between zero and one were based on a sigmoid curve, as shown in Figure 4. Utilizing fuzzy logic over Boolean sets allows for a more fl uid spatial and temporal defi nition of the geographic extent of malaria transmission. The MARA/ARMA set of decision rules does not project transmission intensity.

Temperature is a major factor determining the distribution and incidence of malaria. Temperature affects both the Plasmodium parasite and the Anopheles mosquito, with thresholds at both temperature extremes limiting the survival or development of the two organisms[13,14]. Anopheles must live long enough to bite an infected person, allow the parasite to develop and then bite a susceptible human. MARA/ARMA determined the lower mean monthly temperature threshold of 18C from parasite development time and mosquito survivorship. The lower limit resulted from the fact that below 18C few mosquitoes live long enough for the parasite to complete development. At 22C, parasite development time is suffi ciently short and mosquito development suffi ciently high for stable Plasmodium falciparum malaria transmission. Mean temperature conditions are suitable until 31C, when the survivorship rate peaks. At this point, less than 40% of the mosquitoes survive long enough for the parasite to complete its development cycle. As temperatures rise above 32C, the mosquitos probability of survival decreases. However, higher temperatures enable the mosquitoes to digest blood meals more rapidly, which in turn increases the rate at which they bite. This increased biting rate coupled with faster development of the parasite leads to increased infective mosquito bites for those mosquitoes that do survive[13,14]. 40C is the upper temperature threshold for both mosquitoes and larvae to survive.

Based on these biological constraints on the vector and parasite, the MARA/ARMA decision rules for assigning fuzzy logic values to mean temperature were as follows: The fuzzy logic value increased from zero to 18C

to one at 22C; 22C to 32C was considered 100% suitable for

stable transmission and a fuzzy logic value of one was assigned; and

Death, disease and deformity


Climate change and malaria in Zimbabwe

Fuzzy logic values decreased from one at 32C to zero at 40C.

The second climate variable was the monthly average of the daily minimum temperatures during the winter season. Ground frost (or ambient temperatures of 4 - 5C) kills mosquito populations and prevents the persistence of a stable year round mosquito population[13,14]. In the MARA/ARMA decision rules, if the monthly average of the daily minimum temperatures reached low enough for frost to form during a given month, then it was assumed that vector populations could not become established. The lowest monthly average of daily minimum temperature conducive for frost was set at 4C (fuzzy value of zero), with the upper limit designated 6C (fuzzy value of one).

Total monthly precipitation was the third climate variable used in the MARA/ARMA decision rules. Mosquitoes require water for their eggs to develop into larvae and ultimately emerge as winged adults. Given the ill-defi ned relationship between precipitation and mosquito survivorship and abundance, the MARA/ARMA team examined current precipitation and temperature requirements in areas where the epidemiology of malaria was well understood[13,14]. Based on these examinations, they established 80mm of precipitation per month as suitable for stable transmission in southern Africa. Where there was no precipitation, a fuzzy value of zero was assigned. Fuzzy values from zero to one were assigned across the continuum of precipitation levels from 0mm to 80mm.

The MARA/ARMA decision rules further stipulate that both temperature and precipitation have to be favorable at the same time of the year to allow transmission, and suitable conditions have to continue long enough for the transmission cycle to be completed. A fi ve-month period was considered a suffi cient length of time for conditions to be suitable for stable transmission. Consistent with the MARA/ARMA decision rules, for each month and geographic location, we assigned a fuzzy logic value for each of the three climate variables. We determined the lowest precipitation or mean temperature fuzzy value for each month and location. Then we established the minimum fuzzy value of a moving 5-month period. For example, the lowest fuzzy logic value for mean temperature or precipitation for the months January through May was determined, then the 5-month minimum fuzzy value for February through June was determined, et cetera. Finally, we compared the minimum 5-month fuzzy value of precipitation and temperature with the fuzzy value corresponding to the lowest monthly average of daily minimum temperature in the winter period and assigned the lowest fuzzy value

(i.e. the most constraining variable) to the geographic location. This approach of choosing the lowest fuzzy value of all variables produced a conservative climatic suitability value.

Baseline climatology

Our baseline climate scenario was a mean monthly climatology in a 0.05 latitude by 0.05 longitude raster format, compiled from data spanning 1920-1980 for all of Africa[17]. We extracted monthly total precipitation, monthly mean temperature and the mean range between minimum and maximum daily temperatures in all months for Zimbabwe.

As an average climate representing over sixty years of data, this baseline climatology is not necessarily representative of any given year or even the most recent decade. This is important to note as Zimbabwe experiences signifi cant interannual and decadal variability. In particular, total annual precipitation in Zimbabwe is characterized by strong variability, with global oceanic-atmospheric processes, such as the El Nio Southern Oscillation (ENSO), the dominant cause of precipitation variability (low rainfall is associated with El Nio years)[18,19].

Zimbabwe experienced a warming and drying trend over the past 100 years that overshadows the strong decadal and interannual variability. The instrumental surface air temperature suggests an increase of up to 0.8C in some parts of Zimbabwe, with the winter months (June to November) showing a slightly larger warming than the summer months (December to May)[20,21]. This warming trend was accompanied by a 10% reduction in rainy season precipitation.

The baseline climatology of Zimbabwe is infl uenced by this variability, as well as by the topography of Zimbabwe and its geographic placement within the African continent. Zimbabwe lies at the southern distribution limit of malaria in Africa[22]. A central watershed lying above 1200 meters that runs from northeast to southwest divides Zimbabwe. Rivers arising from this watershed drain into the Zambezi system in the north and the Sabi-Limpopo systems in the south. Low-lying areas where malaria transmission normally occurs fl ank the central plateau. The central plateau rises to peaks as high as 2592 meters along the border with Mozambique. The middle and lowveld regions descend from the central plateau to an elevation of 162 meters at the lowest point in the country (Figure 1). Annual mean temperatures correspond to the elevation gradient and range from 12C to greater than 25C (Figure 2).

The climate of Zimbabwe can be broadly divided into three seasons: the cool dry season (April to



August); the hot dry season (August to October); and the hot wet season (November to March). Typically, precipitation ranges from less than 400 mm during the average rainy season in the southern lowveld region to upwards of 800 mm in the highveld regions near the capitol, Harare, and in the eastern highlands (Figure 3).

Climate projections

Future climate projections were created in COSMIC, the output of which was the change in mean temperature (C) and monthly precipitation from 1990 to 2100, on a countrywide basis. These changes in mean temperature and precipitation were added to the baseline climatology (which has 14,000 grid cells) to present different scenarios of climate in the year 2100.

Analyses of the regional impact of projected global climate change in southern Africa suggest an increase in temperature of 0.7 - 4.7C by the 2050s (dependent on emissions scenario, climate sensitivity and general circulation model (GCM) used), relative to the average 1961-1990 climate[20,21]. The magnitude and direction of projected changes in precipitation is less consistent, with some GCMs suggesting an increase and others a decrease. This variability is primarily due to the current inability of GCMs to represent interannual and decadal variability, and dynamic land cover-atmospheric interactions; both of these have and will most likely continue to infl uence precipitation in southern Africa. Additionally, GCMs respond differently to greenhouse gas forcing, resulting in a range of possible outcomes[20,21].

To represent this range of possible future climates, we chose four of the fourteen GCM-based climate projections generated from the COSMIC program[15,23,24] to create Zimbabwe specifi c projections of monthly precipitation and mean temperature starting in 1990 and carried out to 2100. COSMIC uses an energy-balance-climate/upwelling-diffusion-ocean model to calculate the annual changes in global mean temperature under different radiative forcing scenarios. COSMIC then scales the projected geographic and seasonal climate patterns produced by the fourteen GCMs to individual countries using an intertemporal scaling procedure. The four GCMs chosen were: the Canadian Centre for Climate Research (CCC)[25], United Kingdom Meteorological Offi ce (UKMO)[26], Goddard Institute for Space Studies (GISS)[27], and the Henderson-Sellers model using the CCM1 at NCAR (HEND)[28]. We used climate sensitivities of 4.5C (high) and 1.4C (low), where climate sensitivity is the estimated increase in the global mean surface air temperature after a doubling of atmospheric CO2 concentrations relative to the atmospheric concentrations before the industrial revolution. Finally, we chose the equivalent carbon dioxide (ECD) analogues to the 350 ppmv and 750 ppmv greenhouse gas emission stabilization scenarios of the IPCC Second Assessment Report (ECD concentration is the amount of CO2 that gives the same radiative forcing as the radiative forcing by all greenhouse gases combined, i.e. CO2, N2O, CH4 and CFCs). Figure 5

elevation (meters)

150 - 300300 - 450450 - 600600 - 750750 - 900900 - 10501050 - 12001200 - 13501350 - 15001500 - 16501650 - 18001800 - 20002000 - 2300

HarareHarare

KwekweKwekwe MutareMutare

HarareKadomaKadoma

Kwekwe

GweruGweruGweru

Mutare

BulawayoBulawayoBulawayo

Figure 1 Elevation (Zimbabwe)

annual mean temperature (degrees c)

12 - 1414 - 1616 - 1818 - 2020 - 2222 - 2424 - 2626 - 28

HarareHarare


HarareKadomaKadoma

Kwekwe

GweruGweruGweru

Mutare


Figure 2 Annual mean temperature 1920-1980 (Zimbabwe)

October - April precipitation (mm)

250 - 350 350 - 450450 - 550550 - 650650 - 750750 - 850850 - 950950 - 10501050 - 11501150 - 12501250 - 13501350 - 14501450 - 15501550 - 1700

HarareHarare


HarareKadomaKadoma

Kwekwe

GweruGweruGweru

Mutare


Figure 3 Cumulative rainy season precipitation (October April)

1920-1980 (Zimbabwe)




fuzz

y va

lue

0

0.2

0.4

0.6

0.8

1

1.2

17.5

19.5

21.5

23.5

25.5

27.5

29.5

31.5

33.5

35.5

37.5

39.5

mean temperature (C)

fuzzy logic value for mean temperature

fuzz

y va

lue

0

0.2

0.4

0.6

0.8

1

1.2

0 3 6 9 12

15 18

21

24

27

30

33 36

39

42

45

48

51

54

57

60

63

66

69

72

75

78

81

84

precipitation (mm)

fuzzy logic value for precipitation

fuzz

y va

lue

0

0.2

0.4

0.6

0.8

1

1.2

3.5

3.7

3.9

4.1

4.3

4.5

4.7

4.9

5.1

5.3

5.5

5.7

5.9

6.1

6.3

6.5

minimum temperature (C)

fuzzy logic value for minimum temperature

Figure 4 Assignment of fuzzy logic values to climate variables

conc

entr

atio

n (p

pmv)

0

1950 2000 2050 2100 2150 2200 2250

200

400

600

800

1000

1200

CO2/350

ECD/350

CO2/750

ECD/750

year

Figure 5 Equivalent Carbon Dioxide (ECD) and CO2 Emissions

under the IPCC 350 ppmv and 750 ppmv stabilization scenarios

shows the greenhouse gas emission trajectories for the ECD equivalent and the IPCC 350 ppmv and 750 ppmv stabilization scenarios in which CO2 is the sole greenhouse gas. Due to the many uncertainties related to modeling the radiative forcing of aerosols on the climate system, we excluded them from the climate change scenarios, focusing solely on the infl uence of a suite of greenhouse gases[9].

Creating maps of future malaria climate suitability for stable

malaria transmission in Zimbabwe

COSMIC projected the change in monthly mean temperature and monthly total precipitation for each month from the 1990 baseline year to 2100. These changes on a national level were added to the baseline 0.05-latitude by 0.05-longitude gridded climatology of Zimbabwe. The MARA/ARMA decision rules were then applied to the future climate scenarios and were used to determine a fuzzy logic climate suitability value for each grid cell in Zimbabwe.

In order to incorporate the climate projection from the GCMs into the climatic suitability model, we made three assumptions. The fi rst was that the range between minimum and maximum temperatures would not decrease over the next 100 years. While the global trend has suggested a decrease in the diurnal temperature range, with nighttime minimum temperatures rising at double the rate of daytime maximum temperatures, the trend in Zimbabwe has been an increase in the diurnal range due to increases during the rainy season[20,21]. Because of this uncertainty, we chose to keep the range between minimum and maximum temperatures during the baseline period (1920-1980) constant for the climate projections. A second assumption was that permanent water bodies do not negate the MARA/ARMA precipitation requirements for a given geographic location. Finally, we assumed that climate did not change between the baseline and our starting year of 1990.

ResultsFigure 6 shows the baseline fuzzy logic climate suitability for stable Plasmodium falciparum transmission in Zimbabwe. Areas where malaria is endemic in the average year are denoted with a dark red coloring (fuzzy value of >0.9). White designates locations where climatic conditions inhibit malaria transmission (fuzzy value of zero). For fuzzy logic values between 0.9 and zero, malaria transmission likely varies between years, with epidemics or strongly seasonal malaria present. The highland areas are clearly identifi ed by their low fuzzy logic values. Due to the inaccessibility and inconsistent



nature of historical malaria data for Zimbabwe, we did not validate this map against reported malaria cases.

Figure 7 shows the change in temperature from the baseline climatology to the year 2100 for the climate change scenarios evaluated. The scenarios represented a range of possible future temperatures, from marginal increases in temperature to an increase of 4 to 5C in the HEND 750 ppmv stabilization scenario with a climate sensitivity of 4.5C.

Figure 8 shows the change in precipitation from baseline climatology to the year 2100 for the climate change scenarios evaluated. Note that projected changes in precipitation differ both in magnitude and direction of change. While all scenarios projected an increase in temperature, the GISS and UKMO scenarios projected a net increase in precipitation while the HEND and CCC scenarios suggested a net decrease. The HEND 750 ppmv stabilization scenario with a climate sensitivity of 4.5C projected the most dramatic variation in precipitation, with an increase of over 30 mm in the spring months (November to January) and a decrease of over 60 mm in the summer (February to April).

The results of our model are shown in Figure 9 and Table 1. In the baseline climate suitability scenario, 40% of Zimbabwe was suitable for stable transmission (fuzzy value >0.9). The magnitude and direction of change varied for the four GCMs, following the direction and amplitude of projected precipitation change. Under the scenario with the smallest change, with a climate sensitivity of 1.4C and a stabilization of greenhouse gases at 350 ppmv, the projected net change was small

chan

ge in

deg

rees

cel

cius

0sep oct nov dec jan feb mar apr may jun jul aug sep oct nov dec

CCC_350_1.4

CCC_350_4.5

CCC_750_1.4

CCC_750_4.5

GISS_350_1.4

GISS_350_4.5

GISS_750_1.4

GISS_750_4.5

HEND_350_1.4

HEND_350_4.5

HEND_750_1.4

HEND_750_4.5

UKMO_350_1.4

UKMO_350_4.5

UKMO_750_1.4

UKMO_750_4.5

1

2

3

4

5

6

month

CCC_350_1.4

CCC_350_4.5

CCC_750_1.4

CCC_750_4.5

GISS_350_1.4

GISS_350_4.5

GISS_750_1.4

GISS_750_4.5

HEND_350_1.4

HEND_350_4.5

HEND_750_1.4

HEND_750_4.5

UKMO_350_1.4

UKMO_350_4.5

UKMO_750_1.4

UKMO_750_4.5

chan

ge in

deg

rees

cel

cius

-60

-50

-20

-30

-40

0

10

-10

20

30

40

sep oct nov dec jan feb mar apr may jun jul aug sep oct nov dec

month

Figure 7 Change in temperature to the year 2100 from baseline climate (1920-1980) under various climate change scenarios in Zimbabwe

Figure 8 Change in precipitation to the year 2100 from baseline climate (1920-1980) under various climate change scenarios in Zimbabwe

Figure 6 Baseline fuzzy logic climate suitability for stable malaria

transmission in Zimbabwe

fuzzy climatic suitability (range 0-1)*

* 0 = not suitable (transmission is very unstable)

1 = suitable (transmission is most likely stable)

0>0 - 0.10.1 - 0.20.2 - 0.30.3 - 0.40.4 - 0.50.5 - 0.60.6 - 0.70.7 - 0.80.8 - 0.90.9 - 11

HarareHarare

KwekweKwekweMutareMutare

HarareKadomaKadoma

KwekweGweruGweruGweru

Mutare





Table 1 Percentage of Zimbabwe with climate suitability for stable malaria transmission in 2100

general circulation model

scenario CCC UKMO GISS HEND

ppmv/C

Baseline 40%

350/1.4 * 39% 45% 45% 41%

350/4.5 * * 37% 68% 66% 34%

750/1.4 * * * 34% 77% 76% 25%

750/4.5 * * * * 10% 96% 86% 3%

fuzzy logic value of 0.9

* emissions stabilization scenario of 350 ppmv and a climate sensitivity of 1.4C CCC general circulation model of the Canadian Centre for Climate Research

* * emissions stabilization scenario of 350 ppmv and a climate sensitivity of 4.5C UKMO general circulation model of the United Kingdom Meteorological Offi ce

* * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 1.4C GISS general circulation model of the Goddard Institute for Space Studies

* * * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 4.5C HEND the Henderson-Sellers general circulation model using the CCM1 at NCAR

350/1.4 *

CCCscenarioppmv/C UKMO


GISS HEND

350/4.5 * *

750/1.4 * * *

750/4.5 * * * *

0>0 - 0.10.1 - 0.20.2 - 0.30.3 - 0.40.4 - 0.50.5 - 0.60.6 - 0.70.7 - 0.80.8 - 0.90.9 - 11

fuzz

y cl

imat

ic s

uita

bilit

y (r

ange

0-1

)*

* emissions stabilization scenario of 350 ppmv and a climate sensitivity of 1.4C* * emissions stabilization scenario of 350 ppmv and a climate sensitivity of 4.5C* * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 1.4C* * * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 4.5C

CCC general circulation model of the Canadian Centre for Climate ResearchUKMO general circulation model of the United Kingdom Meteorological OfficeGISS general circulation model of the Goddard Institute for Space StudiesHEND the Henderson-Sellers general circulation model using the CCM1 at NCAR

* 0

= n

ot s

uita

ble

(tra

nsm

issi

on is

ver

y un

stab

le)

1 =

sui

tabl

e (t

rans

mis

sion

is m

ost

likel

y st

able

)

Figure 9 Climate suitability for stable malaria transmission in 2100 under various climate change scenarios in Zimbabwe



-1 -0.9-0.9 -0.8-0.8 -0.7-0.7 -0.6-0.6 -0.5-0.5 -0.4-0.4 -0.3-0.3 -0.2 -0.2 -0.1-0.1 00 0 0.10.1 0.20.2 0.30.3 0.40.4 0.50.5 0.6 0.6 0.70.7 0.80.8 0.90.9 11

Cha

nge

in fu

zzy

clim

ate

suita

bilit

y fr

om b

asel

ine

peri

od

350/1.4 *

CCCscenarioppmv/C UKMO


GISS HEND

350/4.5 * *

750/1.4 * * *

750/4.5 * * * *

* emissions stabilization scenario of 350 ppmv and a climate sensitivity of 1.4C* * emissions stabilization scenario of 350 ppmv and a climate sensitivity of 4.5C* * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 1.4C* * * * emissions stabilization scenario of 750 ppmv and a climate sensitivity of 4.5C

CCC general circulation model of the Canadian Centre for Climate ResearchUKMO general circulation model of the United Kingdom Meteorological OfficeGISS general circulation model of the Goddard Institute for Space StudiesHEND the Henderson-Sellers general circulation model using the CCM1 at NCAR

with a maximum increase of 5% in the area of stable malaria transmission in the UKMO and GISS models, and a decrease of 1% in the CCC GCM. Under the scenario with the highest change, where the climate sensitivity was set at 4.5C and the stabilization scenario was equivalent to 750 ppmv, the projected net change varied from the HEND model projecting that only 3% of Zimbabwe may be suitable for malaria transmission to the UKMO model suggesting that 96% of Zimbabwe may have climate suitable for stable transmission.

As seen in Figure 10, even with little net change in climate suitability, as in the CCC 350_1.4C scenario, there is the potential for a redistribution of areas that are suitable, with the central plateau becoming more suitable for transmission (green) and the lowveld areas becoming slightly less suitable (bright blue). The central plateau, where the human population is currently concentrated, potentially becomes more suitable for malaria transmission in all but the HEND 750_4.5C

scenario. The anomalous change for this scenario can be accounted for by the dramatic decrease in precipitation, particularly in the crucial month of March, rendering all but the highest elevations in Zimbabwe unsuitable for malaria transmission.

DiscussionAlthough climate is a major factor determining the potential range of malaria, there is a scarcity of detailed multivariate maps and models describing its distribution[29]. Determining the baseline distribution of malaria is the fi rst step in understanding how the future distribution might be altered with changing climate; this is one of the achievements of the MARA/ARMA project. The maps created using the MARA/ARMA model approximate the edge of malaria distribution across the African continent quite well[13]. Our model is a fi rst step in developing a range of possible scenarios of the potential future geographic distribution of malaria in

Figure 10 Change in climate suitability for stable malaria transmission from baseline climate (1920-1980) to 2100 in Zimbabwe




and deaths for severe malaria; this trend is expected to continue over at least the short-term[31].

Demographic changes, such as increased human mobility, population growth and redistribution, affect malaria risk[6]. Population redistribution is a result of urbanization, labor migration and confl ict, and allows the transport of both vector and parasite, often exposing immunologically nave populations to malaria[32]. Dramatic climatic changes could create environmental refugees as populations are forced to migrate from areas where resources are few to urban areas or regions where they may have a means of support[8]. Environmental refugees present an enormous opportunity for parasite and vector migration.

Pressures placed on environmental resources, which are often linked to population pressures, can also alter the landscape of malaria transmission. Altered vegetation coverage, surface water fl ows and water storage methods as a result of agricultural intensifi cation or population needs may alter the quantity and quality of breeding habitats for mosquitoes and the species composition of the community[33]. In addition, climate change is likely to alter land use patterns (agricultural use, urban areas, de- or afforestation) and so could potentially infl uence the mosquito population composition and size, resulting in changes in malaria transmission[8].

The objective of the Roll Back Malaria (RBM) partnership is to halve the malaria burden in participating countries by the year 2010 through interventions that are adapted to local needs and by reinforcement of the health sector[34]. Intensifi ed national action will take place within a framework of country-level partnerships supported by a global partnership, with technical support networks available for assistance. Success of these and other intervention programs, including the development of an effective and inexpensive malaria vaccine, would reduce the burden of malaria both today and in the future.

Other factors that could infl uence malaria mortality in Africa include civil unrest, deterioration of public health systems and HIV proliferation[7].

National expenditures on such public goods as vector control, health care delivery and infrastructure, disease surveillance, public education and basic malaria research might dominate the future determinants of malaria risk and are typically limited by a countrys GDP. Just as a countrys GDP infl uences malaria risk, malaria has been shown to decrease economic growth in severely malarious countries by 1.3% per year[35]. The result is that a disease such as malaria creates something of a vicious cycle for poorer countries. Malaria is likely to persist without the creation and maintenance of an appropriate public health infrastructure; however, the economic

Zimbabwe. The methods employed can be used within or across other African countries.

The climate scenarios chosen produce a range of possible climate suitability futures where stable malaria transmission might either increase or decrease depending on the climate sensitivity, the greenhouse gas emission stabilization scenario and the general circulation model used. Despite the wide range in the resulting percent of the area that may become suitable, the predominant pattern across the scenarios suggests that climate suitability for malaria transmission in Zimbabwe is more likely to increase than decrease in the central plateau, where the population is currently concentrated. Evaluation of the results by decade (not shown) suggests that most of Zimbabwe could have near complete climate suitability for stable malaria transmission by 2050 under the scenario showing the greatest change (UKMO_750_4.5C). Within this scenario, precipitation and temperature increase linearly and there is no interannual or decadal variability.

An important question for decision makers is how quickly climate suitability might change, because abrupt climate change might require different mitigation and adaptation options than a gradual change in climate[9]. The uncertainty in the rate and geographic range of change emphasizes the importance of establishing the capacity to monitor changing temperature and precipitation patterns to determine when and where climate suitability for malaria transmission might change; and to establish surveillance programs to determine changes in malaria incidence or intensity along the edges of its established range.

It is important to emphasize that the potential geographic distribution modeled in this study does not directly translate into actual cases of malaria. Areas of the United States and Europe have climates suitable for malaria transmission, yet only a limited number of cases occur (and those are generally imported) because of the strength of the public health infrastructure and other factors[30]. In addition to climatic variables, a number of other factors will infl uence whether or not the future geographic distribution of malaria is different from today, including parasite drug resistance, demographic change, changes in land use patterns and the success of intervention programs such as Roll Back Malaria.

A factor of concern over the short term is increasing parasite drug resistance. Although the current malaria control and prevention measures used in Africa include personal protection, drug use and vector control strategies, most control is based almost exclusively on chemotherapy[4]. Increasing drug resistance has resulted in a two- to three-fold increase in malaria admissions



development necessary for these improvements is directly hindered by the presence of malaria itself. Changes in climate have the potential to make it even more diffi cult for poor countries to reduce the burden of malaria. A key research goal is to identify which of these many factors are most likely to drive the future distribution of malaria and to include modules for these factors into models such as ours.

A number of other research groups have modeled the potential for spread of malaria as a consequence of global climate change; these models include both climate suitability and data on populations at risk.

Martens and colleagues developed a partial integrated mathematical model[7]. In this model, climate change scenarios were linked with a module that uses the formula for the basic reproduction rate to calculate the transmission or epidemic potential for a population of mosquitoes that can transmit malaria. The basic reproductive rate is the number of new cases of a disease that will arise from one case introduced into a non-immune host population during a single transmission cycle. Included within the basic reproductive rate are a number of variables that are sensitive to temperature, including mosquito density, feeding frequency, survival and extrinsic incubation period (the development of the parasite in the mosquito). Incorporating knowledge about the current distributions and characteristics of the main mosquito species of malaria resulted in global projections of 300 million additional people at risk of falciparum malaria in 2080 under the HadCM2 scenario and of an additional 260 to 320 million at risk under the HadCM3 scenario. Most of the changes in potential transmission were projected to occur in temperate zones in areas where vectors are present, but it is currently too cold for transmission.

In contrast to this biological approach, Rogers and Randolph developed a statistical model of the present-day global distribution of malaria that they used to establish the current multivariate climatic constraints[36]. The recorded global limits of malaria were digitized and turned into a raster grid with a longitude and latitude resolution of 0.5. The statistical model was based on the mean, maximum and minimum of temperature, precipitation and saturation vapor pressure. The covariance between variables also was important for setting distributional limits. This multivariate model agreed relatively well with the actual present-day global distribution. The model was then applied to the HadCM2 scenarios for future climate. The high scenario projected mean global land surface changes of +3.45, +3.63 and +3.29C in mean, minimum and maximum temperatures by the year 2050. The statistical model predicted no

signifi cant net change in the future global population living in malarious zones. Small differences between the current and the projected malaria distributions occurred at the edges of the current range.

Although using very different methods, both modeling efforts reached similar conclusions; that is, that the future spread of malaria is likely to occur at the edges of its geographical distribution where current climate constrains transmission. These results are consistent with our fi ndings in Zimbabwe.

Our results also are consistent with a model for Zimbabwe developed by Lindsay and Martens that included three scenarios: an increase of 2C; an increase of 2C with a 20% increase in precipitation; and an increase of 2C with a 20% decrease in precipitation[12]. The results showed that the effect of a temperature increase would be greatest on malaria transmission potential at higher altitudes. In the relatively drier lower altitudes, a temperature increase of 2C combined with a 20% decrease in precipitation resulted in areas becoming too dry for malaria transmission, while a 20% increase in precipitation increased transmission potential. We found a similar variation with changes in temperature and precipitation across the climate scenarios used.

Both Martens et al. and Rogers and Randolph used climate scenarios based on simulations from the Hadley Centre (HadCM2 for both and HadCM3 for Martens et al. 1999)[7,36]. The UKMO scenario used in our model is an older version of these scenarios. Because of large number of differences among the simulations, it is not possible to directly compare our results with those from Rogers and Randolph, and Martens et al.

Our results illustrate the importance of using several climate scenarios to illustrate the range of possible future geographical distributions of malaria. The net change in area suitable for malaria transmission varies dramatically depending on which scenario is used, with a 37% decrease in suitability using the HEND_750_4.5C scenario in contrast to a 56% increase with the UKMO_750_4.5C scenario. Our model also utilized a high-resolution climate baseline capable of showing the climatic heterogeneity that exists in Zimbabwe. This allowed us to examine the fl uctuations in suitability under different scenarios and enables decision makers to identify localized areas vulnerable to climate change. Given spatial information on areas where climate suitability for malaria transmission may change, decision makers can take these possible changes into account when prioritizing actions to enhance the capacity of future generations to effectively handle malaria risk.Climate change and malaria in Zimbabwe

Key words climate change, malaria, methodology, scenarios, Africa




AcknowledgementThis works was supported by the Electric Power Research Institute (EPRI) under contract numbers WO009242 and WO043518, and by the California Energy Commission under contract 500-97-043. This paper does not neces-sarily represent the views and policies of the California Energy Commission or the State of California. The authors gratefully acknowledge Drs. Williams and Schlesinger for the support they provided on COSMIC. The authors also would like to acknowledge Dr. David LeSueur, whose untimely death prevented him from completing his review. Davids practical expertise in malaria, gentle good humor and wonderful stories will be sorely missed.

References[1] Anonymous. Blair Research Institute 1996 Annual Report: Epidemiology of Plasmo-

dium falciparum drug resistance in Zimbabwe. Harare, Zimbabwe: Blair Research Institute, 1995.

[2] Murray, C and A Lopez. The Global Burden of Disease. Boston: Harvard Univer-sity Press, 1996.

[3] Snow, RW, M Craig, U Deichmann and K Marsh. Estimating mortality, morbid-ity, and disability due to malaria among Africas non-pregnant population. Bul-letin of the World Health Organization 1999; 77(8): 624-40.

[4] Breman, J. The ears of the hippopotamus: manifestations, determinants and esti-mates of the malaria burden. American Journal of Tropical Medicine and Hygiene 2001; 64(suppl 1): 76-84.

[5] Githeko, AK and W Ndegwa. Predicting malaria epidemics in the Kenyan high-lands using climate data: a tool for decision makers. Global Change and Human Health 2001; 2(1): 54-63.

[6] Greenwood, B and T Mutabingwa. Malaria in 2002. Nature 2002; 415:670-2.[7] Martens, P, et al. Climate change and future populations at risk of malaria.

Global Environmental Change 1999; 9: S89-S107.[8] IPCC. Climate Change 2001: Impacts, adaptation, and vulnerability. Contribution of

Working Group II to the Third Assessment Report. eds. McCarthy J, Canziani OF, Leary N, Dockken D, White KS. Cambridge: Cambridge University Press, 2001.

[9] IPCC. Climate Change 2001: The scientifi c basis. Contribution of Working Group I to the Third Assessment Report. eds. Houghton JT et al. Cambridge: Cambridge University Press, 2001.

[10] Janssen, M and P Martens. Modeling malaria as a complex adaptive system. Artifi cial Life 1997; 3:213-36.

[11] Chan, NY, KL Ebi, F Smith, TF Wilson and AE Smith. An Integrated Assess-ment Framework for Climate Change and Infectious Diseases. Environmental Health Perspectives 1999; 107(5): 329-37.

[12] Lindsay, SW and WJM Martens. Malaria in the African highlands: past, present and future. WHO Bulletin OMS 1998; 76:33-45.

[13] Craig, MH, RW Snow and Dl Sueur. A climate-based distribution model of malaria transmission in sub-Saharan Africa. Parasitology Today 1999; 15(3): 105-11.

[14] MARA/ARMA. Towards an atlas of malaria risk in Africa. Durban: MARA, 1998.[15] Schlesinger, ME and LJ Williams. COSMIC - Country specifi c model for intertem-

poral climate. Palo Alto: EPRI, 1997.[16] ESRI. ArcView 3.2. Redlands, CA: ESRI, 2001.[17] Hutchinson, M, H Nix, J McMahon and K Ord. Africa: a topographic and climatic

database. Canberra: Centre for Resource and Environmental Studies Australian National University, 1995.

[18] Lindesay, JA. South African rainfall, the Southern Oscillation and a Southern Hemisphere semi-annual cycle. Journal of Climatology 1988; 8:17-30.

[19] Ropelewski, CF and MS Halpert. Global and regional scale precipitation pat-terns associated with the El Nio-Southern Oscillation. Monthly Weather Review 1987; 115:1606-26.

[20] Hulme, M, R Doherty, T Ngara, M New and D Lister. African Climate Change: 1900-2100. Climate Research 2001; 17:145-68.

[21] Unganai, LS. Historic and future climatic change in Zimbabwe. Climate Research 1996; 6:137-45.

[22] Taylor, P and SL Mutambu. A review of the malaria situation in Zimbabwe with special reference to the period 1972-1981. Transactions of the Royal Society of Tropical Medicine and Hygiene 1986; 80:12-9.

[23] Williams, L, D Shaw and R Mendelsohn. Evaluating GCM output with impact models. Climatic Change 1998; 39:111-33.

[24] Mendelsohn, R, M Schlesinger and L Williams. Comparing impacts across cli-mate models. Integrated Assessment 2000; 1:37-48.

[25] McFarlane, NA, GJ Boer, J-P Blanchet and M Lazare. The Canadian Climate Centre Second-Generation General Circulation Model and its equilibrium cli-mate. Journal of Climate 1992; 5:1013-77.

[26] Wilson, CA and JFB Mitchell. A doubled CO2 climate sensitivity experiment with a global climate model including a simple ocean. Journal of Geophysical Research 1987; 92(D11): 13315-43.

[27] Hansen, J, et al. Global climate changes as forecast by Goddard Institute for Space Studies Three-Dimensional Model. Journal of Geophysical Research 1988; 93(D8): 9341-64.

[28] Henderson-Sellers, A, et al. Tropical deforestation: modeling local - to regional-scale climate change. Journal of Geophysical Research 1993; 98(D4): 7289-315.

[29] Watson, ER, M Zinyowera, R Moss and D Dokken. The Regional Impacts of Cli-mate Change: An Assessment of Vulnerability. A special report of IPCC Working Group II. Cambridge: Cambridge University Press, 1998.

[30] Reiter, P. Climate change and mosquito-borne disease. Environmental Health Perspectives 2001; 109:141-61.

[31] Trape, J. The public health impact of chloroquine resistance in Africa. American Journal of Tropical Medicine and Hygiene 2001; 64(suppl 1): 12-7.

[32] Trape, J, et al. Malaria in and urbanization in Central Africa: the example of Brazzaville. Part B: Pernicious attacks and mortality. Transactions of the Royal Society of Tropical Medicine and Hygiene 1987; 81(suppl 2): 34-42.

[33] Lindblade, K, E Walker, A Onapa, J Katungu and M Wilson. Land use change alters malaria transmission parameters by modifying temperature in a highland area of Uganda. Tropical Medicine and International Health 2000; 5:263-74.

[34] Remme, J, F Binka and D Nabarro. Toward a framework and indicators for monitoring Roll Back Malaria. American Journal of Tropical Medicine and Hygiene 2001; 64(suppl 1): 1-11.

[35] Gallup, J and J Sachs. The economic burden of malaria. American Journal of Tropical Medicine and Hygiene 2001; 64(suppl 1): 85-96.

[36] Rogers, DJ and SE Randolph. The global spread of malaria in a future, warmer world. Science 2000; 289:1763-9.

[37] Makono, R and S Sibanda. Review of the prevalence of malaria in Zimbabwe with specifi c reference to parasite drug resistance (1984-96). Transactions of the Royal Society of Tropical Medicine and Hygiene 1999; 93(5): 449-52.

[38] Freeman, T. Malaria: Zimbabwe 1995. A review of the epidemiology of malaria trans-mission and distribution in Zimbabwe and the relationship of malaria outbreaks to preceding meteorological conditions. Harare, Zimbabwe: GTZ, 1995.

[39] Freeman, T and M Bradley. Temperature is predictive of severe malaria years in Zimbabwe. Transactions of the Royal Society of Tropical Medicine and Hygiene 1996; 90:232.



The authorsJessica Hartman has been involved with several research projects investigating the rela-tionships between vector-born disease and anthropogenic cli-mate change or natural climate variability. Jessica is at the

New York Academy of Medicine working with the NYC Department of Health to develop new methods for early warning and detection of infectious diseases outbreaks for bioterrorism surveillance and West Nile Virus. Her research focuses on creating early warn-ing systems that integrate disease, climate, weather, and environmental data on multiple spatial-temporal scales for the early detection of vulnerable populations and disease outbreaks.

Nathan Chan has worked on a number of research areas involving climate change science and impacts. While a consultant at Decision Focus, Inc., he performed integrated assessments for EPRI and the U.S. Environmental Protection

Agency, and analytical modeling of heat stress and vector disease impacts of climate change. His research interests include health effects, land-use changes, and economic impacts of global change.

John P. Weyant is Professor of Management Science and Engineering and director of the Energy Modeling Forum (EMF) at Stanford University. Prof. Weyant earned a B.S./M.S. in Aeronautical Engi-neering and Astronautics, M.S.

degrees in Engineering Management and in Opera-tions Research and Statistics all from Rensselaer Poly-technic Institute, and a Ph.D. in Management Sci-ence with minors in Economics, Operations Research, and Organization Theory from University of Cali-fornia at Berkeley. His current research focuses on analysis of global climate change policy options, and models for strategic planning.

Kristie L. Ebi is an epidemiol-ogist who heads the Global Cli-mate Change and Health Pro-gram at EPRI and is a Scien-tist in the Global Change and Health Program at the WHO European Centre for Environ-ment and Health. Kris was a

Lead Author for the Health Sector of the United States National Assessment of the Potential Health Consequences of Climate Variability and Health, and a Contributing Author to the Health Chapter of the Third Assessment Report of the Intergovernmental Panel on Climate Change. One focus of her research is integrated assessment modeling.

K. John McConnell is an economist at the Institute for Health Policy Studies at the University of California at San Francisco. John has worked on the analysis of predictive information for public health, from an evaluation of an early

warning system for dengue fever to an assessment of US national guidelines for screening for cervical neoplasia.

Documents

Climate Suitability: For Stable Malaria Transmission in Zimbabwe Under Different Climate Change Scenarios