Table S1. National level estimates of the area occupied by and number of people at risk of

Plasmodium falciparum malaria transmission in 2007. The data are presented for World

Health Organization (WHO) region: AFRO, African Regional Office of the WHO, AMRO,

American Regional Office of the WHO, EMRO, Eastern Mediterranean Regional Office of

the WHO, EURO, European Regional Office of the WHO, SEARO, South East Asian

Regional Office of the WHO and WPRO, Western Pacific Regional Office of the WHO. Areas

are in millions km2 and populations at risk in millions. The populations are further stratified as

those at stable (P. falciparum annual parasite incidence (PfAPI) of more than 0.1 per 1000

population per annum (pa)) and unstable (PfAPI < 0.1‰ pa).

Region/Country Area at risk (km2) Population at risk (millions)

Stable Unstable All Stable Unstable All AFRO Angola 1.26 0.02 1.28 15.41 0.28 15.70Benin 0.12 0.00 0.12 7.66 0.00 7.66Botswana 0.43 0.00 0.43 0.89 0.00 0.89Burkina Faso 0.28 0.00 0.28 14.23 0.00 14.23Burundi 0.02 0.00 0.02 5.66 0.00 5.66Cameroon 0.47 0.00 0.47 16.95 0.00 16.95Cape Verde 0.00 0.00 0.00 0.00 0.25 0.25Central African Republic 0.63 0.00 0.63 4.15 0.00 4.15Chad 0.69 0.63 1.32 9.76 0.16 9.92Comoros 0.00 0.00 0.00 0.59 0.00 0.59Congo 0.34 0.00 0.34 3.33 0.00 3.33Côte d'Ivoire 0.33 0.00 0.33 17.80 0.00 17.80Dem. Republic of the Congo 2.32 0.00 2.32 57.97 0.00 57.97Equatorial Guinea 0.03 0.00 0.03 0.52 0.00 0.52Eritrea 0.07 0.06 0.13 3.33 0.95 4.28Ethiopia 0.93 0.07 0.99 46.08 1.50 47.59Gabon 0.27 0.00 0.27 1.34 0.00 1.34Gambia 0.01 0.00 0.01 1.51 0.00 1.51Ghana 0.24 0.00 0.24 22.21 0.00 22.21Guinea 0.25 0.00 0.25 9.23 0.00 9.23Guinea-Bissau 0.03 0.00 0.03 1.43 0.00 1.43Kenya 0.00 0.03 0.03 25.61 0.18 25.79Liberia 0.10 0.00 0.10 3.43 0.00 3.43Madagascar 0.61 0.09 0.70 17.28 0.00 17.28Malawi 0.12 0.00 0.12 13.45 0.00 13.45Mali 0.65 0.67 1.32 13.46 0.47 13.93Mauritania 0.28 0.53 0.81 0.93 0.40 1.33Mayotte 0.00 0.00 0.00 0.00 0.27 0.27Mozambique 0.82 0.00 0.82 21.06 0.00 21.06Namibia 0.35 0.19 0.54 1.25 0.38 1.64Niger 0.45 0.79 1.25 13.19 0.62 13.81Nigeria 0.93 0.00 0.93 134.60 0.00 134.60Rwanda 0.02 0.00 0.02 5.03 0.00 5.03Sao Tome and Principe 0.00 0.00 0.00 0.13 0.00 0.13Senegal 0.20 0.00 0.20 10.82 0.00 10.82

Region/Country Area at risk (km2) Population at risk (millions) Stable Unstable All Stable Unstable All

Sierra Leone 0.07 0.00 0.07 5.50 0.00 5.50South Africa 0.07 0.05 0.12 3.44 2.95 6.39Swaziland 0.01 0.00 0.01 0.23 0.00 0.23Togo 0.06 0.00 0.06 5.45 0.00 5.45Uganda 0.24 0.00 0.24 27.03 0.00 27.03United Republic of Tanzania 0.92 0.00 0.92 39.84 0.00 39.84Zambia 0.78 0.00 0.78 11.84 0.00 11.84Zimbabwe 0.31 0.00 0.31 7.44 0.00 7.44Total 15.70 3.11 18.81 601.06 8.43 609.49 AMRO Belize 0.00 0.01 0.01 0.00 0.17 0.17Bolivia 0.12 0.56 0.68 0.22 2.61 2.83Brazil 4.15 0.41 4.56 12.79 16.69 29.47Colombia 0.52 0.31 0.83 5.26 7.74 13.00Dominican Republic 0.02 0.01 0.03 1.41 2.83 4.24Ecuador 0.11 0.07 0.17 4.12 1.65 5.77French Guiana 0.08 0.00 0.08 0.14 0.00 0.14Guatemala 0.05 0.03 0.08 1.02 5.27 6.30Guyana 0.17 0.04 0.20 0.14 0.52 0.66Haiti 0.03 0.00 0.03 8.58 0.02 8.60Honduras 0.05 0.03 0.08 0.88 2.63 3.51Nicaragua 0.09 0.03 0.11 1.57 2.12 3.70Panama 0.02 0.00 0.02 0.90 0.00 0.90Peru 0.50 0.06 0.57 3.87 1.70 5.57Suriname 0.02 0.11 0.13 0.01 0.05 0.06Venezuela 0.15 0.49 0.64 0.22 6.24 6.46Total 6.06 2.17 8.23 41.13 50.23 91.37 EMRO Afghanistan 0.05 0.21 0.26 4.56 12.53 17.10Djibouti 0.00 0.02 0.02 0.02 0.41 0.43Iran 0.01 0.26 0.27 0.15 2.72 2.87Pakistan 0.21 0.55 0.76 30.74 68.30 99.04Saudi Arabia 0.01 0.03 0.03 0.72 1.22 1.94Somalia 0.58 0.06 0.64 10.04 0.55 10.59Sudan 1.52 1.06 2.57 28.99 6.84 35.83Yemen 0.21 0.25 0.47 15.93 5.72 21.65Total 2.60 2.43 5.02 91.14 98.30 189.44 EURO Kyrgyzstan 0.00 0.02 0.02 0.00 1.20 1.20Tajikistan 0.00 0.02 0.02 0.00 2.16 2.16Total 0.00 0.03 0.03 0.00 3.36 3.36 SEARO Bangladesh 0.02 0.05 0.07 15.12 47.99 63.11Bhutan 0.01 0.00 0.01 0.79 0.47 1.26India 1.63 1.32 2.95 414.53 535.59 950.12Indonesia 1.31 0.43 1.74 68.59 81.93 150.52Myanmar 0.68 0.01 0.69 42.88 1.91 44.79Nepal 0.01 0.02 0.04 3.40 6.15 9.54

Region/Country Area at risk (km2) Population at risk (millions) Stable Unstable All Stable Unstable All

Sri Lanka 0.01 0.03 0.05 1.75 7.53 9.28Thailand 0.20 0.27 0.47 16.53 30.53 47.06Timor-Leste 0.01 0.00 0.01 0.96 0.00 0.96Total 3.89 2.14 6.03 564.54 712.08 1276.62 WPRO Cambodia 0.18 0.01 0.19 10.77 2.55 13.33China 0.20 0.16 0.37 17.13 20.32 37.45Lao People's Dem. Republic 0.24 0.00 0.24 5.30 0.01 5.31Malaysia 0.23 0.00 0.23 6.29 16.17 22.46Papua New Guinea 0.40 0.00 0.40 4.11 0.00 4.11Philippines 0.16 0.06 0.22 26.95 20.41 47.35Solomon Islands 0.02 0.00 0.02 0.43 0.00 0.43Vanuatu 0.01 0.00 0.01 0.22 0.00 0.22Viet Nam 0.16 0.18 0.33 19.31 53.31 72.62Total 1.60 0.41 2.02 90.51 112.77 203.28 Global Totals 29.85 10.29 40.14 1388.39 985.17 2373.56

Protocol S1. Methods and sources of medical intelligence used to describe Plasmodium falciparum annual parasite incidence (PfAPI) globally

We defined areas of extremely low, unstable transmission risk as spatial units reporting a

PfAPI of less than 0.1 per 1000 population per annum (pa). This is because this criterion

was found to be a reliable indicator for the cessation of indoor residual spraying during the

consolidation phase of the Global Malaria Eradication Programme [1-4]. During this period,

the limit was reduced from 0.5‰ as it became recognized that surveillance, including

passive and active case detection of incident cases, was often less accurate and reliable

than nations thought: malaria often resumed after the cessation of spraying from 0.5‰, but

rarely from 0.1‰. This more conservative categorization of unstable transmission around

one clinical P. falciparum case 10,000 population (0.1‰ pa), thus, helps compensate for the

vagaries of district or provincial level reporting of parasitologically confirmed cases [5-7].

Using a higher threshold between ‘stable’ and ‘unstable’ transmission would have required a

greater confidence in the fidelity of the source information and the underlying surveillance.

PfAPI data sourcing

Table SI A1 summarizes the PfAPI data collection process and availability for all P.

falciparum malaria endemic countries. Complete PfAPI data were not available for Ecuador,

Bolivia, Haiti, Suriname and Venezuela. Risk maps presented at a regional malaria meeting

in Cartagena, Colombia [8], were used, showing high, medium, low and no risk areas for P.

falciparum and P. vivax. These maps were digitized and then combined with first-level

administrative unit species-specific API risk data reported in 2002 [9] to constrain the Pf-

limits within each country and map areas of either <0.1 or ≥0.1 cases per 1000 people pa.

When these first-level administrative unit case reports were less than 10 in 2002 [9], it was

assumed that contemporary risks in these areas were <0.1 ‰ pa. In Venezuela, the high risk

areas reported in the Cartagena meeting [8] corresponded with descriptions of international

travel advisories which state that risk in Amazonas and Bolivar is mainly restricted to the

Orinoco River basin and its tributaries in the former, and along the same river in the latter but

also in areas bordering the states of Apure and Guarico [10]. These descriptions were used

to create a 60 km distance buffer of high risk along the Orinoco and its tributaries. The map

of Venezuela in the Cartagena report shows high risk areas towards the north mainly in the

state of Sucre [8]. This disagrees with published reports [9,11] of API in Sucre being

attributed only to P. vivax, however, and probably refers to the local high risk of P. vivax

malaria. Sucre was, therefore, classified at no risk for P. falciparum. For Suriname, similar

descriptions of high risk restricted to the Maroni river are available [8] and, therefore, a

similar 60 km buffer was created along this river and classified as high risk. The latter buffer

did not capture two small high risk areas west of the river shown in the Cartagena map [8]

and these were left as unstable risk due to lack of geographic information to digitize them

with precision.

Plasmodium falciparum API data were not available for most countries within the WHO

Africa region, largely because these are not regarded as priority malaria metrics for this

region and reporting systems are probably the least reliable when compared to other regions

of the world [5-7,12,13]. Southern Africa defined as Swaziland, South Africa, Zimbabwe,

Botswana and Namibia, represents an exception with a combination of published risk data

[14,15] and expert opinion from national malaria control programmes [16] used to estimate

the limits of risk through matched administrative boundaries or digitized maps in these

countries. For island populations risks were reviewed from published sources to confirm

current P. falciparum transmission status including Cape Verde [17], Comoros [18], Mayotte

[19], and Mauritius [19]. The islands of Reunion and Seychelles were excluded as malaria

endemic as confirmed through published literature [19].

Malaria programme managers in EMRO confirmed that the small foci of malaria transmission

in Ihrir, Illizi Department, Algeria and the Fayoum Governate, Egypt, are entirely due to P.

vivax. Similarly, the limited cases reported in Syria were all due to P. vivax. These countries

were, therefore, confirmed as P. falciparum free. Tajikistan and Kyrgyzstan were regarded

as the only P. falciparum endemic countries in the WHO’s European region. Both countries

have reported P. falciparum malaria cases since 2003 [20], with very few cases reported in

2006, however, and with elimination scheduled by 2010 [21].

Medical intelligence confirmed the P. vivax only status of the Korean Peninsula [22,23],

Argentina [24], Paraguay [24], Iraq [25], Uzbekistan [26], Turkey [25], Azerbaijan [25] and

Mauritius [27]. The United Arab Emirates have not reported any autochthonous cases of P.

falciparum malaria for over five years [28] and satisfied the criteria for the certification of the

eradication of malaria in 2007 [29]. The Chiapas region of Mexico, Canton Matua, in Costa

Rica and Ahuachapa in El Salvador have all had only 1 to12 P. falciparum cases reported

over the last five years, and these countries are regarded as largely P. falciparum free [24].

Mapping PfAPI data

In order to map PfAPI data consistently, digital boundaries of first and second level units

were obtained from the United Nations' Second Administrative Level Boundaries (SALB)

dataset (n=24) [30] and the Global Administrative Unit Layers (GAUL) developed by the

Food and Agriculture Organization (FAO; n=82) [31]. In addition, PfAPI classifications were

reconciled at third-level administrative units for Nepal (Global Administrative Areas database

of the University of California at Berkeley, http://biogeo.berkeley.edu/gadm) and restricted

areas in Peru through the Peruvian Dirección General de Epidemiología

(http://www.oge.sld.pe). In South Africa, third-level administrative data [16] were digitized

against boundary data provided by FAO Geonetwork [32].

Table 1. Sources and spatial/temporal resolutions of reported P. falciparum annual pa rasite incidence (PfAPI) data for 87 countries identified

as endemic for P. falciparum malaria. The data are grouped by World Health Organization (WHO) region (AFRO, African Regional Office of the

WHO, AMRO, American Regional Office of the WHO, EMRO, Eastern Mediterranean Regional Office of the WHO, EURO, European Regional

Office of the WHO, SEARO, South East Asian Regional Office of the WHO and WPRO, Western Pacific Regional Office of the WHO) and

summarised globally. Additional data sources consulted to confirm and/or refine the risk margins are also presented. ADMIN1, 2 or 3 refers to

administrative division at the first, second or third levels where Admin0 is the national boundary.

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

AFRO

Angola NA NA

Benin NA NA

Botswana ADMIN1 NA [33-35]

Burkina Faso NA NA

Burundi NA NA

Cameroon NA NA

Cape Verde ADMIN1 NA [17,33]

Central African Rep. NA NA

Chad NA NA

Comoros ADMIN1 NA [18]

Congo NA NA

Côte d'Ivoire NA NA

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

Dem. Rep. of Congo NA NA

Equatorial Guinea NA NA

Eritrea NA NA

Ethiopia NA NA

Gabon NA NA

Gambia NA NA

Ghana NA NA

Guinea NA NA

Guinea-Bissau NA NA

Kenya NA NA

Liberia NA NA

Madagascar NA NA [36]

Malawi NA NA

Mali NA NA

Mauritania NA NA [33]

Mayotte ADMIN1 NA [37]

Mozambique NA NA

Namibia ADMIN2 NA [33,35,38]

Niger NA NA

Nigeria NA NA

Rwanda NA NA

Sao Tome and Principe NA NA

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

Senegal NA NA

Sierra Leone NA NA

South Africa ADMIN2 NA [15,16,35]

Swaziland ADMIN2 NA [35]

Togo NA NA

Uganda NA NA

United Rep.of Tanzania NA NA

Zambia NA NA

Zimbabwe ADMIN2 NA [33,35,39,40]

AMRO

Belize ADMIN2 (6) 2004

Bolivia ADMIN2 (112) 2002 [8,9,33,41]

Brazil ADMIN2 (5310) 2004-2006 [8]

Colombia ADMIN2 (297) 2005 [8,33]

Dominican Rep. ADMIN1 (32) 2004

Ecuador ADMIN2 (220) 2002 [8,9,33]

French Guiana ADMIN2 (21) 2006 [33]

Guatemala ADMIN1 (22) 2004

Guyana ADMIN1 (10) 2004

Haiti ADMIN2 (41) 2002 [33]

Honduras ADMIN1 (19) 2004-2006

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

Nicaragua ADMIN1 (18) 2004

Panama ADMIN2 (68) 2006

Peru ADMIN2 (191) and ADMIN3 (14)

2004 [8,33]

Suriname ADMIN2 (62) 2002 [9,33] A 60km buffer around the Maroni river was assumed to correspond to high risk

Venezuela ADMIN2 (318) 2002 [9,33,41] Assumptions of risk confined mainly near the Orinoco and its main tributaries (60km buffer) were deemed sensible to avoid overestimating risk in Bolivar and Amazonas

EMRO

Afghanistan ADMIN1 (32) 2005 [42]

Djibouti ADMIN1 NA [43]

Iran ADMIN2 (251) 2004-2006 [44,45]

Pakistan ADMIN2 (119) 2004-2006 [46] No information available for six ‘tribal areas’ in the Fata region of Pakistan and 15 ADMIN1 units in the disputed territory of Jammu Kashmir.

Saudi Arabia ADMIN1 (13) 2005-2006 [47]

Somalia ADMIN2 NA

Sudan ADMIN1 NA

Yemen ADMIN1 (19) 2002, 2005-2006

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

EURO

Kyrgyzstan ADMIN2 (40) 2002-2005 [20,48]

Tajikistan ADMIN2 (56) 2005-2006 [20,48]

SEARO

Bangladesh ADMIN2 (64) 2003-2004 [49]

Bhutan ADMIN1 (20) 2002-2004 [33,49]

India ADMIN2 (538) 2002-2004 Twenty-two ADMIN1 units could not be reconciled or matched to any reported PfAPI data in India and were left as missing data.

Indonesia ADMIN1/2 (281) 2005 [49] The following rules were used to interpolate ADMIN1 data to some ADMIN2 polygons: i) in 5 cases, no ADMIN2 data were available for the whole ADMIN1 and this was defaulted to 2005 ADMIN1 data; ii) there were 50 ADMIN2s for which no data were available but other ADMIN2 units within the same ADMIN1 had data; these missing ones were assigned the overall ADMIN1 value.

Myanmar ADMIN1 (14) and ADMIN2 (11)

2003-2004 [33,49]

Nepal ADMIN3 (175) 2002-2003 [33,49]

Sri Lanka ADMIN2 (25) 2004 [33,50]

Thailand ADMIN1 (76) 2003-2004 [33,51]

Country ADMIN level for PfAPI* Time period

Sources consulted

Notes

Timor Leste ADMIN1 (13) 2004-2005

WPRO

Cambodia ADMIN1 (24) 2003-2005

China ADMIN1 (32) ADMIN2 (16)i

2003-2005 [33,52-54]

Lao Dem. Peoples Rep. ADMIN1 (18) 2003-2005 [33,55]

Malaysia ADMIN1 (15) 2003-2005 [56]

Papua New Guinea ADMIN1 (20) 2003-2005 [57]

Philippines ADMIN2 (79) 2003-2005 [33]

Solomons ADMIN1 (10) 2003-2005

Vanuatu ADMIN1 (6) 2003-2005

Vietnam ADMIN1 (61)i 2003-2005 [33,55]

*Missing data: Apart from data missing in Pakistan and India (specified in the table), data were not available for the following territories: The

Santanilla islands of Honduras, D.I. Yogyakarta of Indonesia, the district of Kilinochchi in Sri Lanka and the province of Bac Nihn in Vietnam.

References

1. Black R (1968) Manual of epidemiology and epidemiological services in malaria programmes. Geneva: World Health Organization. 223 p.

2. Hay SI, Smith DL, Snow RW (2007) Measuring malaria endemicity from intense to interrupted transmission. Lancet Infect Dis: in press.

3. Pampana E (1969) A textbook of malaria eradication. London: Oxford University Press. 593 p.

4. Yekutiel P (1980) Eradication of infectious diseases: a critical study. 2 Contributions to epidemiology and biostatistics. Basel, Switzerland: Karger. pp. 164.

5. Erhart A, Thang ND, Xa NX, Thieu NQ, Hung LX, et al. (2007) Accuracy of the health information system on malaria surveillance in Vietnam. Trans R Soc Trop Med Hyg 101: 216-225.

6. Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI (2005) The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature 434: 214-217.

7. Sharma VP (2007) Battling the malaria iceberg with chloroquine in India. Malar J 6: 105. 8. PAHO (2000) The Roll Back Malaria Initiative in the Rainforest Region of South America.

Cartagena: Pan American Health Organization. 9. PAHO (2003) Report on the Status of Malaria Programs in the Americas (Based on 2002

data) CD44/INF/3. Washington, D. C.: Pan American Health Organization / World Health Organization - 44th Directing Council/55th Session of the Regional Committee.

10. IAMAT (2004) World Malaria Risk Chart. Available: http://www.iamat.org/pdf/WorldMalariaRisk.pdf. Accessed: July 2007.

11. Caceres JL, Pizzo N, Vela FA, Perez W, Rojas JG, et al. (2005) Impacto de la cura radical masiva sobre la incidencia malaria del estado de Sucre, Venezuela. Bol Malar Sal Amb 45: 27-36.

12. Chilundo B, Sundby J, Aanestad M (2004) Analysing the quality of routine malaria data in Mozambique. Malar J 3: 3.

13. Gething PW, Noor AM, Gikandi PW, Ogara EA, Hay SI, et al. (2006) Improving imperfect data from health management information systems in Africa using space-time geostatistics. PLoS Med 3: e271.

14. Kleinschmidt I, Sharp B, Mueller I, Vounatsou P (2002) Rise in malaria incidence rates in South Africa: a small-area spatial analysis of variation in time trends. Am J Epidemiol 155: 257-264.

15. Kleinschmidt I, Sharp BL, Clarke GP, Curtis B, Fraser C (2001) Use of generalized linear mixed models in the spatial analysis of small-area malaria incidence rates in Kwazulu Natal, South Africa. Am J Epidemiol 153: 1213-1221.

16. Malaria in Southern Africa (2005) Malaria risk update 2005. http://www.malaria.org.za/malaria_risk/update2005.

17. Alves J, Roque AL, Cravo P, Valdez T, Jelinek T, et al. (2006) Epidemiological characterization of Plasmodium falciparum in the Republic of Cabo Verde: implications for potential large-scale re-emergence of malaria. Malar J 5: 32.

18. Blanchy S, Julvez J, Mouchet J (1999) Stratification epidemiologique du paludisme dans l'archipel des Comores. Bull Soc Pathol Exot 92: 177-184.

19. Tchen J, Ouledi A, Lepere JF, Ferrandiz D, Yvin JL (2006) Epidemiologie et prevention du paludisme dans les iles du sud-ouest de l'Ocean Indien. Med Trop (Mars) 66: 295-301.

20. Ejov M (2005) Scaling up the response to malaria in the WHO European Region: Progress toward curbing an epidemic 2000-2004. Copenhagen: World Health Organization Regional Office for Europe.

21. WHO Regional Office for Europe (2006) Regional strategy: from malaria control to elimination in the WHO European Region 2006-2015. WHO-EUR/06/5061322. Copenhagen: World Health Organization Regional Office for Europe. 41 p.

22. Yeom JS, Kim TS, Oh S, Sim JB, Barn JS, et al. (2007) Plasmodium vivax malaria in the Republic of Korea during 2004-2005: changing patterns of infection. Am J Trop Med Hyg 76: 865-868.

23. Yeom JS, Ryu SH, Oh S, Lee WJ, Kim TS, et al. (2005) Status of Plasmodium vivax malaria in the Republic of Korea during 2001-2003. Am J Trop Med Hyg 73: 604-608.

24. PAHO (2006) Regional Strategic Plan for Malaria in the Americas 2006-2010. Washington DC: Pan American Health Organization.

25. Korenromp E (2005) Malaria incidence estimates at country level for the year 2004 - Proposed estimates and draft report. Geneva, Switzerland.

26. Severini C, Menegon M, Di Luca M, Abdullaev I, Majori G, et al. (2004) Risk of Plasmodium vivax malaria reintroduction in Uzbekistan: genetic characterization of parasites and status of potential malaria vectors in the Surkhandarya region. Trans R Soc Trop Med Hyg 98: 585-592.

27. WHO (2005) International Travel and Health: Situation as on 1 January 2005. Geneva: World Health Organization.

28. WHO Regional Office for the Eastern Mediterranean (2007) Strategic plan for malaria control and elimination in the WHO Eastern Mediterranean Region 2006–2010. WHO-EM/MAL/340/E. Cairo: World Health Organization Regional Office for the Eastern Mediterranean. 41 p.

29. WHO (2007) United Arab Emirates certified malaria-free. Wkly Epidemiol Rec 82: 25-32. 30. SALB (2007) Second Administrative Level Boundaries.

http://www3.who.int/whosis/gis/salb/salb_home.htm. 31. FAO (2005) The Global Administrative Unit Layers (GAUL). Rome: Food and Agriculture

Organization of the United Nations. 32. FAO (2007) FAO GeoNetwork. http://www.fao.org/geonetwork/srv/en/about. Accessed:

June 2007. 33. CDC (2007) Travelers' Health. http://www.cdc.gov/travel/regionalmalaria/. Accessed:

July 2007. 34. Ministry of Health of the Republic of Botswana (2001) Botswana Roll Back Malaria

Strategic Plan 2002-2005. National Malaria Control Programme. 7 p. 35. SAMC Malaria risk maps in Southern Africa. http:/www.malaria.org.zw/countries.

Accessed: July 2007. 36. Pasteur Institute of Madagascar, Roll Back Malaria (2004) Madagascar Malaria Map.

Malaria Reserach Unit. 37. Julvez J, Galtier J, Halidi MA, Henry M, Mouchet J (1987) Actualites epidemiologiques

de paludisme. Epidemiologie du paludisme et lute antipaludique a Mayotte (Archipel des Comores, Ocean Indien). Bull Soc Pathol Exot Filiales 80: 505-519.

38. Ministry of Health & Social Sciences of the Republic of Namibia (1995 ) National policy and strategy for malaria control. Ministry of Health & Social Sciences, Republic of Namibia. 95 p.

39. Root G (2006) Malaria prevalence and community-based survey. Ministry of Health and Child Welfare of Zimbabwe and World Health Organization.

40. Shiff C (2002) Integrated approach to malaria control. Clin Microbiol Rev 15: 278-293. 41. PAHO. Meeting of national directors of epidemiology and malaria programs; 2005 7–10

November 2005; San José. Pan American Health Organization. 42. Asha AW (2004) Afghanistan country report at 4th RBM-EMRO meeting at Isfahan, Iran

25-28 May 2004. Isfahan, Iran. 43. National Malaria Control Programme Manager (2007). Djibouti. 44. Raesi A (2007) National Malaria Control Programme Manager. Iran. 45. Zamani G (2007) RBM Monitoring and Evaluation coordinator for EMRO. 46. Mansoor F (2007) National Malaria Control Programme Manager. Pakistan. 47. Saudi Arabia National Malaria Control Programme Manager (2007). 48. WHO European Region (2006) Centralized Information System for Infectious Diseases

(CISID). http://data.euro.who.int/cisid/. 49. SEARO (2007). www.searo.who.int/EN/section10.

50. Briet OJ, Gunawardena DM, van der Hoek W, Amerasinghe FP (2003) Sri Lanka malaria maps. Malar J 2: 22.

51. Zhou G, Sirichaisinthop J, Sattabongkot J, Jones J, Bjornstad ON, et al. (2005) Spatio-temporal distribution of Plasmodium falciparum and P. vivax malaria in Thailand. Am J Trop Med Hyg 72: 256-262.

52. Zhou SS, Tang LH, Sheng HF (2005) Malaria situation in the People's Republic of China in 2003. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi 23: 385-387.

53. Zhou SS, Tang LH, Sheng HF, Wang Y (2006) Malaria situation in the People' s Republic of China in 2004. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi 24: 1-3.

54. Zhou SS, Wang Y, Tang LH (2006) Malaria situation in the People's Republic of China in 2005. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi 24: 401-403.

55. Pratap S (2003) Update on malaria, multidrug resistance and economic development in the Mekong region of South East Asia. Southeast Asian J Trop Med Public Health 34: 105 pages.

56. WPRO (2005) Malaria Epidemiology, Cambodia. www.wpro.who.int/sites/mvp/epidemiology/malaria/.

57. Muller I, Bockarie M, Alpers M, Smith T (2003) The epidemiology of malaria in Papua New Guinea. Trends Parasitol 19: 253-259.

Protocol S2: Developing global biological limits for Plasmodium falciparum transmission

The temperature mask

Temperature affects many aspects of mosquito physiology [1]. One aspect critical for malaria

transmission is the temperature dependence of sporogony: the time it takes for sporozoites

to develop and become infectious in Anopheles. A method for estimating the duration of

sporogony has been proposed, based on the number of degree-days required by the

parasite to complete development [2,3], for malaria the sum of the number of degrees in a

day by which the mean temperature exceeds the minimum required for the development of

sporozoites. Nikolaev [4] showed that the degree-days required for the maturation of

sporozoites in an An. maculipennis population from Russia were 105 for P. vivax, 111 for P.

falciparum and 144 for P. malariae and that parasite development ceased below 16oC for P.

falciparum and P. malariae and below 14.5oC for P. vivax. The duration of sporogony (DS) in

days can thus be calculated as:

MINMEAN TTDS

−=

DD [1]

where DD are the parasite species-specific degree-days, TMEAN is the mean ambient

temperature and TMIN is the minimum temperature required for parasite development.

Figure 1 plots the results of the above equation for P. falciparum, P. vivax and P. malariae

development in An. maculipennis. It shows that P. vivax is able to develop at the lowest

temperatures, followed by P. falciparum and P. malariae, thus helping explain the species

specific latitudinal limits of the parasites globally [5,6]. The figure also shows that the length

of sporogony of P. falciparum increases rapidly when temperatures drop below 22oC. The

curve never reaches a true asymptote on the y-axis; but the duration of sporogony becomes

so extended that few anophelines will survive long enough to inoculate humans and at 16oC

parasite development ceases entirely. Conversely, as temperature increases the duration of

sporogony decreases, so that at 30oC it can take less than ten days. Obviously, the duration

of sporogony then becomes limited by parasite and vector survival, which plummet as

temperatures rise above 32oC [7]. The duration of sporogony is dependent fundamentally on

enzyme kinetics [8] and thus widely assumed to be relatively independent of vector species.

It is the interplay between the duration of sporogony and the species-specific longevity of the

Anopheles vector that forms the basis of the temperature mask.

1

Protocol S2, Figure 1. The relationship between the duration of sporogony and temperature

for P. falciparum (dark red), P. vivax (blue) and P. malariae (green). The open circle

indicates the temperature below which the length of sporogony for P. falciparum is not

outlived by most vectors, which corresponds to 31 days at circa 19.58oC (dotted orange

lines). The dashed red line indicates the absolute temperature below which P. falciparum

development ceases at 16oC.

Kiszewski et al. [9] provide a comprehensive review of the longevity of Anopheles from

natural and experimental studies of the bionomics of 33 malaria vectors worldwide that were

viewed to be dominant contributors to local transmission. The resulting summary of daily

survival rates for adult anophelines was found to be so variable that the authors used a

median value for all vectors for their malaria stability index. The mean daily survival rate was

0.846 for all Anopheles, ranging from 0.682 for An. albimanus to 0.950 for An. sergentii.

We have further limited the criteria for dominant vector species by insisting that they were

indicated as dominant in local malaria transmission by all four of the following authors [9-13].

2

The daily survival rate for the 19 main/dominant vectors by region was used to determine the

fraction of the population surviving over successive days (Figure 2). Although temperature

will affect other parameters of the basic reproduction rate of infection [14], including biting

and resting vector habits, it seems reasonable to consider the proportion of the population

surviving 31 days as the critical point of interruption of P. falciparum transmission. To a close

approximation, most vector populations would have been reduced to 99% of their original

population size within 31 days. The regional variation encountered in this approximation was

then explored further.

In AMRO, the mean daily survival probability for the five regionally dominant vectors is

0.782. This means that 99.95% of an initial population would not survive 31 days (Figure 2,

top left). By far the longest-lived vector in the region is An. pseudopunctipennis for which

98.09% of the population will have perished in 31 days; this makes this vector resistant to

higher altitudes across its distribution in the Andean slopes [15]. Similarly, the mean daily

survival probability for the three regionally dominant vectors in AFRO is 0.780. This means

that 99.95% of an initial population would not survive 31 days (Figure 2, top right). In

SEARO/WPRO the mean daily survival probability for the eight regionally dominant vectors

is 0.836. This predicts that 99.62% of an initial population would not survive 31 days (Figure

2, bottom right). The longest lived vector in the region is An. dirus, for which 93.41% of the

population will have perished in 31 days. This species is mainly a forest dweller [16,17] and

therefore is restricted geographically and ecologically to a niche that does not suffer extreme

temperature limits. Excluding areas where average temperatures were such that sporogony

would not complete in 31 days was thus considered a biologically plausible and conservative

limit of P. falciparum malaria transmission in these regions.

In EMRO the mean daily survival probability for the five regionally dominant vectors is 0.846.

This indicates that 99.45% of an initial population would not survive 31 days (Figure 2,

bottom left). This mean disguises two relatively long-lived vectors in the region: An. sergentii,

which is an oasis breeder, and An. superpictus, which is another mountain foothill breeder

[11,12,18]. These species have longevities that predict that only 79.61% and 82.69% of their

populations would have died after a 31-day period. The 31 days time limit was applied for

this region, except where these species are dominant, where it was doubled to 62 days. This

duration would result in a population mortality of 95.84% and 97.00% for An. sergentii and

An. superpictus, respectively. The distributions of these vectors described by White [13]

were digitized and the 62-day rule applied within these species ranges.

3

Protocol S2, Figure 2. The fraction of the original population of Anopheles surviving by day

for each dominant vector species of the World Health Organization (WHO) geographical

regions. (AFRO, African Regional Office of the WHO, AMRO, American Regional Office of

the WHO, EMRO, Eastern Mediterranean Regional Office of the WHO, EURO, European

Regional Office of the WHO, SEARO, South East Asian Regional Office of the WHO and

WPRO, Western Pacific Regional Office of the WHO). In all panels, the red line is the

average survival fraction for the region and each species is identified by a unique colour

shown in the top right of the panel. The black dotted lines mark the 31 days duration of

sporogony limit. The green dotted line in the EMRO/EURO panel marks the 62 days criterion

applied for An. sergentii and An. superpictus.

4

In summary, with the exception of An. sergentii and An. superpictus, it is rare for adult

dominant vectors of malaria to survive longer than a month, with more than 99% of the

average population dying after 31 days. The longer-lived vectors are generally those

adapted to survive at higher altitudes or harsher conditions, such as is the case of An.

superpictus and An. sergentii. Despite the fact that a relatively small proportion of the

populations of these vectors are normally able to survive longer than one month, the

numbers of individuals surviving might still pose a significant risk for malaria transmission by

being able to support parasite development at lower temperatures. After two months,

however, most individuals of both species (>95%) would also have succumbed.

Using average monthly temperature records estimated from a global climate surface [19],

the duration of P. falciparum sporogony was estimated for each month. Those pixels where

the duration of sporogony was 31 days or less were identified in each month. This provided

12 images with a binary outcome of whether P. falciparum sporogony could be completed in

more or less than 31 days. The images were combined to identify the number of temperature

suitable months available in a synoptic year (Figure 3). All pixels where the 31 days limit was

not achieved for any single month (i.e. grey pixels in Figure 3), or two consecutive months in

the geographic range of An. sergentii and An. superpictus, were used as a conservative

mask to exclude areas where transmission is highly unlikely to occur.

Protocol S2, Figure 3. An overlay of 12 monthly images of where the duration of sporogony

exceeds 31 days, restricted to P. falciparum Malaria Endemic Countries (PfMECs). Pixels

where the temperature did not reach 19.58oC in any single month of a synoptic year (here

5

shown as grey areas within the PfMECs) were used to mask in all areas (except within the

range of An. sergentii and An. superpictus were two consecutive months were required).

The aridity mask

Upper temperature limits were not defined on the basis of physiological tolerances of vectors

measured in laboratories [20-23] as these were so high as to be rarely achieved in nature

and often subject to behavioural avoidance [20]. We preferred a partial surrogate for extreme

aridity: a hybrid measure encompassing both high temperature and low water availability.

The ability of adult vectors to survive long enough to contribute to parasite transmission and

their eggs and larvae to survive in sufficient numbers to sustain transmission is dependent

on the level of humidity and the species-specific ability to withstand arid conditions [24-26].

Hyper-aridity is the main criterion used to define a desert biome [27] and, therefore,

identifying desert extents was assumed to be an accurate proxy for the extreme mask to limit

the risk of malaria transmission.

These potentially limiting conditions prevail in deserts and their fringes found in malaria

endemic countries, notably the Sahara (and the Sahel), the Namib, the Arabian and the Thar

deserts, as well as the northern arid areas of East Africa and Peru (Figure SI B4). Since in

these areas optimum growth of most plants is hindered, a proxy for vegetation cover can be

used to classify arid areas [28]. Such a proxy can be derived from satellite sensors by

combining the information of different channels of the electromagnetic spectrum to derive

vegetation indices [29]. One of the most commonly used is the normalized difference

vegetation index (NDVI) [30], available as a multitemporal series from the Advanced Very

High Resolution Radiometer sensor [31] and, more recently, from the MODerate-resolution

Imaging Spectroradiometer (MODIS) sensor on board the Terra and Aqua satellites [32-34].

In addition to NDVI, MODIS products include the enhanced vegetation index (EVI) [34]. EVI

is calculated similarly to NDVI, which is derived from two channels of the electromagnetic

spectrum (red and near-infrared). EVI incorporates a third channel (blue) and corrects for

some of the distortion caused by atmospheric particles and ground cover beneath the

vegetation. This makes EVI a more robust index by offering improved sensitivity, particularly

in areas with high biomass content where it saturates less than NDVI, but also reduced

contamination throughout by particles in the air [32-34].

Temporal Fourier processed, monthly, bi-directional reflectance distribution function

corrected EVI images [32,35,36] were reclassified using ArcView GIS 3.2 (ESRI 1999) to

give a binary output of areas where EVI ≤0.1 and EVI >1. These reclassified images were

6

then overlaid in pairs to produce 12 new images. The 12 pairs were then combined to

identify pixels where conditions were suitable for transmission (i.e. where EVI pixel values in

a synoptic year were higher than 0.1 for at least 2 consecutive months).

Despite strict quality control, these data are affected by atmospheric contamination in the

form of clouds and aerosols, although these effects are less frequent in arid low rainfall

areas with infrequent cloud cover. To avoid the introduction of these errors in the final mask,

a sub-mask was used of only those territories within P. falciparum endemic countries that

were defined as being at some level of risk according to PfAPI data and that overlapped with

the arid areas defined by the EVI threshold. This included whole or partial territories of 20

PfMECs as follows: Afghanistan, Angola, Chad, Djibouti, Eritrea, Ethiopia, India (northwest),

Iran, Kenya, Kyrgyzstan, Mali, Mauritania, Niger, Pakistan, Peru (northwest), Saudi Arabia

(southwest), Somalia, Sudan, Tajikistan and Yemen.

The aridity sub-mask was applied in a step-wise fashion by which risk was down-regulated

one class (i.e. stable to unstable and unstable to no risk). Therefore, the only areas where

risk was excluded were those where PfAPI had already defined limited risk of malaria. The

sub-mask was then applied on top of the PfAPI and temperature masks (Figure 1, bottom

panel, main paper).

Protocol S2, Figure 4. Overlay of reclassified monthly EVI images (≤0.1 and >0.1). The

scale shows the number of ‘arid’ months occurring in each pixel in a synoptic year. Despite

quality control, cloud contamination was still evident in some humid areas (e.g. Gulf of

7

8

Guinea) as fine speckle. The red outline indicates P. falciparum risk areas where the aridity

mask was applied to avoid introducing cloud-contaminated pixels in the final image.

References 1. Clements AN (1999) The Biology of Mosquitoes. Wallingford, UK: CABI Publishing. 740 p. 2. Detinova TS (1962) Age grouping methods in Diptera of medical importance with special

reference to some vectors of malaria. Geneva: World Health Organization. 3. Moshkovsky SD (1946) [The dependence upon temperature of the speed of development

of malaria plasmodia in the mosquito]. Med Parazitol (Mosk) 15: 19. 4. Nikolaev BP (1935) The influence of temperature on the development of malaria

plasmodia in the mosquito. Tr Pasteur Inst Epidem Bakt (Leningr) 2: 108-109. 5. Lysenko AY, Semashko IN (1968) Geography of Malaria: a medico-geographic profile of

an ancient disease. In: Lebedew AW, editor. Medicinskaja Geografija. Moscow: Academy of Sciences. pp. 25-146.

6. Lysenko AJ, Beljaev AE (1969) An analysis of the geographical distribution of Plasmodium ovale. Bull World Health Organ 40: 383-394.

7. Muirhead-Thompson RC (1951) Mosquito behaviour in relation to malaria transmission and control in the tropics. London: Edward Arnold & Co.

8. Sharpe PJ, DeMichele DW (1977) Reaction kinetics of poikilotherm development. J Theor Biol 64: 649-670.

9. Kiszewski A, Mellinger A, Spielman A, Malaney P, Sachs SE, et al. (2004) A global index representing the stability of malaria transmission. Am J Trop Med Hyg 70: 486-498.

10. Mouchet J, Carnevale P, Coosemans M, Julvez J, Manguin S, et al. (2004) Biodiversité du paludisme dans le monde. Paris: John Libbey Eurotext. 428 p.

11. Service MW (1993) Mosquitoes (Culicidae). In: Lane RP, Crosskey RW, editors. Medical Insects and Arachnids. London: Chapman & Hall. pp. 120-240.

12. Service MW (1993) The Anopheles vector. Bruce-Chwatt's Essential Malariology. London: Edward Arnold. pp. 96-123.

13. White GB (1989) Malaria. In: Sloof R, editor. Geographical distribution of arthropod-borne diseases and their principal vectors WHO/VBC/89967. Geneva: World Health Organization, Division of Vector Biology and Control. pp. 7-22.

14. Rogers DJ, Randolph SE (2006) Climate change and vector-borne diseases. Adv Parasitol 62: 345-381.

15. Rubio-Palis Y, Zimmerman RH (1997) Ecoregional classification of malaria vectors in the neotropics. J Med Entomol 34: 499-510.

16. Guerra CA, Snow RW, Hay SI (2006) A global assessment of closed forests, deforestation and malaria risk. Ann Trop Med Parasitol 100: 189-204.

17. Obsomer V, Defourny P, Coosemans M (2007) The Anopheles dirus complex: spatial distribution and environmental drivers. Malar J 6: 26.

18. Zahar AR (1984) Vector control operations in the African context. Bull World Health Organ 62 Suppl: 89-100.

19. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25: 1965-1978.

20. Kirby MJ, Lindsay SW (2004) Responses of adult mosquitoes of two sibling species, Anopheles arabiensis and A. gambiae s.s. (Diptera: Culicidae), to high temperatures. Bull Entomol Res 94: 441-448.

21. Bayoh MN, Lindsay SW (2003) Effect of temperature on the development of the aquatic stages of Anopheles gambiae sensu stricto (Diptera: Culicidae). Bull Entomol Res 93: 375-381.

22. Bayoh MN, Lindsay SW (2004) Temperature-related duration of aquatic stages of the Afrotropical malaria vector mosquito Anopheles gambiae in the laboratory. Med Vet Entomol 18: 174-179.

23. Jepson WF, Moutia A, Courtois C (1947) The malaria problem in Mauritius: the bionomics of Mauritian anophelines. Bull Entomol Res 38: 177-208.

9

10

24. Shililu JI, Grueber WB, Mbogo CM, Githure JI, Riddiford LM, et al. (2004) Development and survival of Anopheles gambiae eggs in drying soil: influence of the rate of drying, egg age, and soil type. J Am Mosq Control Assoc 20: 243-247.

25. Omer SM, Cloudsley-Thompson JL (1970) Survival of female Anopheles gambiae Giles through a 9-month dry season in Sudan. Bull World Health Organ 42: 319-330.

26. Omer SM, Cloudsley-Thomson JL (1968) Dry season biology of Anopheles gambiae Giles in the Sudan. Nature 217: 879-880.

27. UNEP (2006) Global Deserts Outlook. Nairobi: Division of Early Warning and Assessment (DEWA), United Nations Environment Programme. 184 p.

28. Suzuki R, Xu JQ, Motoya K (2006) Global analyses of satellite-derived vegetation index related to climatological wetness and warmth. Int J Climatol 26: 425-438.

29. Myneni RB, Maggion S, Iaquinto J, Privette JL, Gobron N, et al. (1995) Optical remote-sensing of vegetation - modeling, caveats, and algorithms. Remote Sens Environ 51: 169-188.

30. Tucker CJ (1979) Red and photographic infrared linear contributions for monitoring vegetation. Remote Sens Environ 8: 127-150.

31. Hay SI (2000) An overview of remote sensing and geodesy for epidemiology and public health application. Adv Parasitol 47: 1-35.

32. Hay SI, Tatem AJ, Graham AJ, Goetz SJ, Rogers DJ (2006) Global environmental data for mapping infectious disease distribution. Adv Parasitol 62: 37-77.

33. Guerra CA, Hay SI (2005) Remote Sensing: Generalities and Data Products for Malaria Risk Mapping in the Americas. In: Confalonieri UEC, Marinho DP, editors. Remote Sensing and the Control of Infectious Diseases: Proceedings from an Interamerican Workshop. Rio de Janeiro: ENSP/FIOCRUZ. pp. 71-89.

34. Tatem AJ, Goetz SJ, Hay SI (2004) Terra and Aqua: new data for epidemiology and public health. Int J Appl Earth Obs 6: 33-46.

35. Scharlemann J, Benz D, Hay SI, Purse B, Tatem AJ, et al. (2008) Global data for ecology and epidemiology: a novel algorithm for temporal Fourier processing MODIS data. PLoS One: in press.

36. Rogers DJ (2000) Satellites, space, time and the African Trypanosomiases. Adv Parasitol 47: 128-171.