Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in...

17
APPLIED BIOLOGICAL SCIENCES Drivers of Rift Valley fever epidemics in Madagascar Renaud Lancelot a,b,1 , Marina B ´ eral b,c,d , Vincent Michel Rakotoharinome e , Soa-Fy Andriamandimby f , Jean-Michel H ´ eraud f , Caroline Coste a,b , Andrea Apolloni a,b ,C´ ecile Squarzoni-Diaw g , St ´ ephane de La Rocque a,b,h,i , Pierre B. H. Formenty j ,J´ er ´ emy Bouyer a,b , G. R. William Wint k , and Eric Cardinale b,c,d a French Agricultural Research and International Cooperation Organization for Development (Cirad), Department of Biological Systems (Bios), UMR Animals, Health, Territories, Risks, and Ecosystems (Astre), Campus International de Baillarguet, 34398 Montpellier, France; b French National Agricultural Research Center for International Development, Animal Health Department, UMR Astre, Campus International de Baillarguet, 34398 Montpellier, France; c Cirad, Bios Department, UMR Astre, 97490 Sainte Clotilde, La R ´ eunion, France; d Centre de Recherche et de Veille sur les Maladies ´ Emergentes de l’Oc ´ ean Indien, 97490 Sainte Clotilde, La R ´ eunion, France; e Minist ` ere de l’Agriculture, de l’Elevage et de la P ˆ eche, Direction des Services V ´ et ´ erinaires, Ambatofotsikely, BP 530 Antananarivo 101, Madagascar; f Unit ´ e de Virologie, Institut Pasteur de Madagascar, BP 1274 Antananarivo 101, Madagascar; g Animal Production and Health Division, Food and Agriculture Organization of the United Nations, 00153 Rome, Italy; h World Organization for Animal Health, 75017 Paris, France; i International Health Regulations Monitoring Procedure and Information Team, Global Capacity and Response Department, World Health Organization (WHO), 1211 Geneva 27, Switzerland; j Emerging and Epidemic Zoonotic Diseases, Pandemic and Epidemic Disease Department, WHO, Geneva 27, Switzerland; and k Department of Zoology, Environmental Research Group Oxford, Oxford OX1 3PS, United Kingdom Edited by Burton H. Singer, University of Florida, Gainesville, FL, and approved December 8, 2016 (received for review May 18, 2016) Rift Valley fever (RVF) is a vector-borne viral disease widespread in Africa. The primary cycle involves mosquitoes and wild and domestic ruminant hosts. Humans are usually contaminated after contact with infected ruminants. As many environmental, agricul- tural, epidemiological, and anthropogenic factors are implicated in RVF spread, the multidisciplinary One Health approach was needed to identify the drivers of RVF epidemics in Madagascar. We examined the environmental patterns associated with these epidemics, comparing human and ruminant serological data with environmental and cattle-trade data. In contrast to East Africa, environmental drivers did not trigger the epidemics: They only modulated local Rift Valley fever virus (RVFV) transmission in ruminants. Instead, RVFV was introduced through ruminant trade and subsequent movement of cattle between trade hubs caused its long-distance spread within the country. Contact with cat- tle brought in from infected districts was associated with higher infection risk in slaughterhouse workers. The finding that anthro- pogenic rather than environmental factors are the main drivers of RVF infection in humans can be used to design better prevention and early detection in the case of RVF resurgence in the region. vector-borne infection | zoonosis | El Ni ˜ no | cattle trade | One Health R ift Valley fever (RVF) is a vector-borne infection caused by the RVF virus (RVFV). Mosquito species from the Aedes, Culex, and Mansonia genera are the main RVFV vectors (1). The virus persists in the environment either by vertical transmission occurring in some Aedes species or by an enzootic cycle involv- ing ruminants and mosquitoes. When environmental conditions are favorable to mosquito proliferation, this cycle is amplified by Culex populations, leading to RVF epidemics (2). In the Horn of Africa, these circumstances are met during the warm phases of El Ni˜ no southern oscillation (ENSO) and the Indian Ocean dipole zonal mode: Warm sea-surface temperatures in the equa- torial eastern-central Pacific Ocean and the western equatorial Indian Ocean result in heavy autumn rainfall and subsequent greening of the vegetation (3–5). The latter can be monitored by the remotely sensed normalized difference vegetation index (NDVI). Besides these environmental conditions triggering epi- demics, long-distance RVFV dissemination is often related to ruminant trade (6), as reported in Madagascar and Saudi Ara- bia (7–9). In domestic ruminants, the RVFV causes mass abortions and high neonatal mortality (10). Humans may get infected after the bites of infected mosquitoes or the consumption of raw milk. However, most clinical cases occur after contact with blood, aborted fetuses, and placenta of viremic animals. Farmers, vet- erinarians, slaughterhouse workers, and butchers are thus the most exposed to the risk of RVF, as well as any people attending the slaughtering of viremic animals (11): Blood aerosol produced on this occasion is highly infectious (12). The virus is not trans- mitted from person to person. In Madagascar, the first known RVF epidemic occurred in 1990–1991 on the East Coast and in the central highlands (7). A second epidemic occurred in 2008–2009 with official reports suggesting at least 700 suspected cases and 26 RVFV laboratory- confirmed fatalities. However, due to underreporting, an excess of 10,000 human cases was estimated (13–15). Madagascar is one of the poorest countries in the World, with 90% of the population living with less than USD 1.5 inhab. -1 ·d -1 (16). Because public health resources are limited in such a set- ting, it is crucial to identify the most exposed populations, based on scientific evidence. To this end, we aimed at understanding the circumstances in which RVF epidemics occurred. We first assessed the risk of RVFV introduction to ruminants via live- stock trade. Then, we described the environmental conditions associated with the start of the 1990 and 2008 epidemics and with local RVFV transmission. We then aimed to build risk indexes of RVFV human infection related to the local environment (local risk) and cattle trade (remote risk), assess their relative Significance Rift Valley fever (RVF) is an emerging, mosquito-borne viral infection of ruminants, transmissible to people, and linked to rainfall. By investigating a wider range of possible drivers this study confirms the assumption that RVF occurrence can also be dependent on nonenvironmental drivers. In Mada- gascar, human and ruminant infections were geographically distinct, with introduction and long-distance disease spread linked to livestock trading. Human infections were highest in those involved with slaughtering and handling fresh meat. The identification of possible introduction routes, major cat- tle trade hubs, and areas of high risk to ruminants has shown how multidisciplinary analyses are needed to properly under- stand disease dynamics and spread, thereby improving early detection and prevention of RVF in humans. Author contributions: R.L., V.M.R., S.-F.A., J.M.-H., C.S.-D., S.d.l.R., P.B.H.F., G.R.W.W., and E.C. designed research; R.L., M.B., V.M.R., S.-F.A., J.M.-H., C.C., A.A., C.S.-D., S.d.l.R., P.B.H.F., and E.C. performed research; R.L., M.B., C.C., A.A., and G.R.W.W. analyzed data; and R.L., M.B., V.M.R., S.-F.A., J.M.-H., C.C., A.A., C.S.-D., S.d.l.R., P.B.H.F., J.B., G.R.W.W., and E.C. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. 1 To whom correspondence should be addressed. Email: [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1607948114/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1607948114 PNAS Early Edition | 1 of 6

Transcript of Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in...

Page 1: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

Drivers of Rift Valley fever epidemics in MadagascarRenaud Lancelota,b,1, Marina Beralb,c,d, Vincent Michel Rakotoharinomee, Soa-Fy Andriamandimbyf,Jean-Michel Heraudf, Caroline Costea,b, Andrea Apollonia,b, Cecile Squarzoni-Diawg, Stephane de La Rocquea,b,h,i,Pierre B. H. Formentyj, Jeremy Bouyera,b, G. R. William Wintk, and Eric Cardinaleb,c,d

aFrench Agricultural Research and International Cooperation Organization for Development (Cirad), Department of Biological Systems (Bios), UMRAnimals, Health, Territories, Risks, and Ecosystems (Astre), Campus International de Baillarguet, 34398 Montpellier, France; bFrench National AgriculturalResearch Center for International Development, Animal Health Department, UMR Astre, Campus International de Baillarguet, 34398 Montpellier, France;cCirad, Bios Department, UMR Astre, 97490 Sainte Clotilde, La Reunion, France; dCentre de Recherche et de Veille sur les Maladies Emergentes de l’OceanIndien, 97490 Sainte Clotilde, La Reunion, France; eMinistere de l’Agriculture, de l’Elevage et de la Peche, Direction des Services Veterinaires,Ambatofotsikely, BP 530 Antananarivo 101, Madagascar; fUnite de Virologie, Institut Pasteur de Madagascar, BP 1274 Antananarivo 101, Madagascar;gAnimal Production and Health Division, Food and Agriculture Organization of the United Nations, 00153 Rome, Italy; hWorld Organization for AnimalHealth, 75017 Paris, France; iInternational Health Regulations Monitoring Procedure and Information Team, Global Capacity and Response Department,World Health Organization (WHO), 1211 Geneva 27, Switzerland; jEmerging and Epidemic Zoonotic Diseases, Pandemic and Epidemic Disease Department,WHO, Geneva 27, Switzerland; and kDepartment of Zoology, Environmental Research Group Oxford, Oxford OX1 3PS, United Kingdom

Edited by Burton H. Singer, University of Florida, Gainesville, FL, and approved December 8, 2016 (received for review May 18, 2016)

Rift Valley fever (RVF) is a vector-borne viral disease widespreadin Africa. The primary cycle involves mosquitoes and wild anddomestic ruminant hosts. Humans are usually contaminated aftercontact with infected ruminants. As many environmental, agricul-tural, epidemiological, and anthropogenic factors are implicatedin RVF spread, the multidisciplinary One Health approach wasneeded to identify the drivers of RVF epidemics in Madagascar.We examined the environmental patterns associated with theseepidemics, comparing human and ruminant serological data withenvironmental and cattle-trade data. In contrast to East Africa,environmental drivers did not trigger the epidemics: They onlymodulated local Rift Valley fever virus (RVFV) transmission inruminants. Instead, RVFV was introduced through ruminant tradeand subsequent movement of cattle between trade hubs causedits long-distance spread within the country. Contact with cat-tle brought in from infected districts was associated with higherinfection risk in slaughterhouse workers. The finding that anthro-pogenic rather than environmental factors are the main drivers ofRVF infection in humans can be used to design better preventionand early detection in the case of RVF resurgence in the region.

vector-borne infection | zoonosis | El Nino | cattle trade | One Health

R ift Valley fever (RVF) is a vector-borne infection caused bythe RVF virus (RVFV). Mosquito species from the Aedes,

Culex, and Mansonia genera are the main RVFV vectors (1). Thevirus persists in the environment either by vertical transmissionoccurring in some Aedes species or by an enzootic cycle involv-ing ruminants and mosquitoes. When environmental conditionsare favorable to mosquito proliferation, this cycle is amplified byCulex populations, leading to RVF epidemics (2). In the Hornof Africa, these circumstances are met during the warm phasesof El Nino southern oscillation (ENSO) and the Indian Oceandipole zonal mode: Warm sea-surface temperatures in the equa-torial eastern-central Pacific Ocean and the western equatorialIndian Ocean result in heavy autumn rainfall and subsequentgreening of the vegetation (3–5). The latter can be monitoredby the remotely sensed normalized difference vegetation index(NDVI). Besides these environmental conditions triggering epi-demics, long-distance RVFV dissemination is often related toruminant trade (6), as reported in Madagascar and Saudi Ara-bia (7–9).

In domestic ruminants, the RVFV causes mass abortions andhigh neonatal mortality (10). Humans may get infected after thebites of infected mosquitoes or the consumption of raw milk.However, most clinical cases occur after contact with blood,aborted fetuses, and placenta of viremic animals. Farmers, vet-erinarians, slaughterhouse workers, and butchers are thus themost exposed to the risk of RVF, as well as any people attending

the slaughtering of viremic animals (11): Blood aerosol producedon this occasion is highly infectious (12). The virus is not trans-mitted from person to person.

In Madagascar, the first known RVF epidemic occurred in1990–1991 on the East Coast and in the central highlands (7).A second epidemic occurred in 2008–2009 with official reportssuggesting at least 700 suspected cases and 26 RVFV laboratory-confirmed fatalities. However, due to underreporting, an excessof 10,000 human cases was estimated (13–15).

Madagascar is one of the poorest countries in the World, with90% of the population living with less than USD 1.5 inhab.−1·d−1

(16). Because public health resources are limited in such a set-ting, it is crucial to identify the most exposed populations, basedon scientific evidence. To this end, we aimed at understandingthe circumstances in which RVF epidemics occurred. We firstassessed the risk of RVFV introduction to ruminants via live-stock trade. Then, we described the environmental conditionsassociated with the start of the 1990 and 2008 epidemics and withlocal RVFV transmission. We then aimed to build risk indexesof RVFV human infection related to the local environment(local risk) and cattle trade (remote risk), assess their relative

Significance

Rift Valley fever (RVF) is an emerging, mosquito-borne viralinfection of ruminants, transmissible to people, and linkedto rainfall. By investigating a wider range of possible driversthis study confirms the assumption that RVF occurrence canalso be dependent on nonenvironmental drivers. In Mada-gascar, human and ruminant infections were geographicallydistinct, with introduction and long-distance disease spreadlinked to livestock trading. Human infections were highest inthose involved with slaughtering and handling fresh meat.The identification of possible introduction routes, major cat-tle trade hubs, and areas of high risk to ruminants has shownhow multidisciplinary analyses are needed to properly under-stand disease dynamics and spread, thereby improving earlydetection and prevention of RVF in humans.

Author contributions: R.L., V.M.R., S.-F.A., J.M.-H., C.S.-D., S.d.l.R., P.B.H.F., G.R.W.W.,and E.C. designed research; R.L., M.B., V.M.R., S.-F.A., J.M.-H., C.C., A.A., C.S.-D., S.d.l.R.,P.B.H.F., and E.C. performed research; R.L., M.B., C.C., A.A., and G.R.W.W. analyzed data;and R.L., M.B., V.M.R., S.-F.A., J.M.-H., C.C., A.A., C.S.-D., S.d.l.R., P.B.H.F., J.B., G.R.W.W.,and E.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1607948114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1607948114 PNAS Early Edition | 1 of 6

Page 2: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

importance, identify and map high-risk areas, and assess the con-sequences for human health.

ResultsRisk of Virus Introduction by Livestock Trade. Queries in the UnitedNations (UN) ComTrade database did not reveal direct impor-tations of live ruminants from mainland Africa to Madagascar.However, they highlighted several official importations from theUnion of Comoros in 2005–2007 (Table 1). Although the num-bers are small, they confirmed the risk of RVFV introduction inMadagascar from an infected country, via livestock trade.

In addition to this official trade, illegal cattle movementsbetween the Comoros Archipelago and Madagascar were proba-bly much more frequent. Informal surveys conducted in 2009 and2010 in the main Comoros harbors and in the northwest of Mada-gascar (Mahajanga and Antsiranana) revealed the frequent pres-ence of cattle and small ruminants on board freighters andbotry (dhows) traveling from the Comoros Islands to Madagas-car and from port to port. This coastal navigation is widespreadin Madagascar, given the weakness of the terrestrial-road net-work (Fig. S1). Therefore, RVFV could have been introducedto Madagascar through ruminant trade from Comoros Islands—which were previously infected (17, 18)—and further dissemi-nated through coastal navigation between Malagasy sea ports.

Environmental Conditions and RVFV Epidemics.Triggering RVF epidemics. Conditions at the start of the two RVFepidemics in humans (March 1990 and January 2008) are shownon Fig. 1. For the three indicators [southern oscillation index(SOI), rainfall, and NDVI] and three biomes (Fig. S2A), the twoepidemics did not occur in typical RVF conditions according tothe eastern African standards (14). In March 1990, a markednegative anomaly was observed for SOI, one of the main indi-cators of ENSO (Fig. 1A). Rainfall was lower than normal inall three biomes (Fig. 1B). March 1990 was also at the end ofa period of positive NDVI anomalies and before a short periodof negative anomalies for the dry forest and xeric shrubland. TheNDVI pattern was not clear for the moist forest where the epi-demics started (Fig. 1C). Conditions during the 2008 epidemicwere almost the reverse as it occurred during a cold (positive)anomaly of SOI, when rainfall was close to normal in the moistforest and somewhat higher than normal in the two other biomes.

These general impressions were corroborated by the resultsof partial triadic analysis and hierarchical clustering (Fig. S3):The environmental conditions of RVF epidemics fell into quitedifferent clusters in 1990–1991 and in 2008–2009: With a three-class partition, the 1990 epidemics occurred in a category of rainyseasons with close-to-normal mean values for SOI (0.01), rainfall(0.16), and NDVI (−0.11). In 2008, the epidemic fell into a cat-egory with a high mean SOI (1.38) and close-to-normal meanvalues for NDVI (0.07) and rainfall (−0.16).RVFV spread at the end of 2008 epidemic. We modeled the sero-prevalence rate of immunoglobulins of type G (IgG) in rumi-

Table 1. Importation of live ruminants in Madagascar from theUnion of Comoros between 2005 and 2007 (source: UN ComTrade)

Year Species Quantity, head

2005 Cattle 30*Goats 9*

2006 Goats 14*2007 Cattle 7

Goats 88Sheep 47

*Estimated number from the reported financial value. No imports of liveruminants were reported from the Union of Comoros in 2003 and 2004.

SO

I (st

d de

v.)

−2

0

2

Mar 1990 Jan 2008

A

SO

I

Rai

nfal

l (st

d de

v.)

−2

0

2

Mar 1990 Jan 2008Xer

ic s

hrub

.

−2

0

2

Mar 1990 Jan 2008Dry

fore

st

−2

0

2

Mar 1990 Jan 2008

B

Moi

st fo

rest

ND

VI (

std

dev.

)

−2

0

1985 1990 1995 2000 2005 2010

Mar 1990 Jan 2008

Xer

ic s

hrub

.

−2

0

Mar 1990 Jan 2008

Dry

fore

st

−2

0

Mar 1990 Jan 2008C

Moi

st fo

rest

Fig. 1. Environmental conditions (monthly average) in the main Mala-gasy biomes with respect to the 1990–1991 and 2008–2009 RVF epidemics.(A) Standardized southern oscillation index (19). (B) Standardized rain-fall anomalies. Source: National Oceanic and Atmospheric Administration(NOAA)’s precipitation reconstruction over Land (PREC/L) provided by theNOAA/Office of Oceanic and Atmospheric Research/Earth Sciences ResearchLaboratory - Physical Sciences Divisions, Boulder, CO. (C) StandardizedNDVI anomalies. Source: Global Inventory Modeling and Mapping Studies(GIMMS), Advanced Very High Resolution Radiometer (AVHRR) NDVI version3 (20). On each plot, favorable conditions for RVF are shown in red.

nant sera collected at the end of the 2008 epidemic to assessthe role of environmental conditions in RVF spread in livestock.The subset of plausible models according to the available datais shown in Table S1. The importance of environmental predic-tors with respect to the selected subset of plausible models is dis-played in Table S2. Multimodel averaged coefficients are shownin Table S3. Higher rainfall, municipalities within 50 km froma sea port, and lower altitude were associated with higher sero-prevalence rate in ruminants. See SI Results and Figs. S4 and S5for details on exploratory data analysis.

The receiver operating characteristic (ROC) curve for theaveraged model had an area under the curve of 74%. The mapof predicted sero-prevalence rate showed high-risk areas onthe northwestern and northeastern coasts (lowlands). Southernregions and highlands were less affected with the exception ofsea-port municipalities (Fig. 2). See SI Results for details.

The Cattle Trade Network and RVFV Dissemination. Cattle tradedata were collected monthly from 2007 to 2011, with large varia-tions in numbers across the years. The overall network activityis presented in Fig. S6A: Each segment corresponds to a linkbetween two nodes; a segment is drawn if at least one move-ment has been recorded along that link on a given month; itscolor is related to its recorded frequency during the 5-y survey.To assess the influence of the cattle trade network on the risk ofRVFV in humans, we quantified the trade flows using the most

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1607948114 Lancelot et al.

Page 3: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

Sero prevalence rate

25°S

20°S

15°S

44°E 46°E 48°E 50°E

0 300 km

Manakara

Antsiranana

Mahajanga

Morondava

Toliara

Taolagnaro

Toamasina

Antalaha

A

0.2 0.4 0.6 0.8

Coefficient of variation

B

0.1 0.2 0.3 0.4

Fig. 2. Sero-prevalence rate of anti-RVFV IgG in ruminants (Madagascar)predicted by the averaged beta-binomial logistic regression model: (A) pre-dicted rate and (B) coefficient of variation.

comprehensive dataset, collected in 2010. This network had thelargest number of nodes (257 farms, markets, and slaughter-houses) and links (346). We could distinguish two sets of links:those active throughout the year (56% of the links) constitutingthe backbone of the network and those occasional links active for1 mo or 2 mo. Despite their large number, only a small volume(3%) of animals was traded on these occasional links. Most of thebackbone nodes and links active in 2010 were already present inthe previous year.

To assess the role of the occasional links on RVFV spread,we considered separate monthly network snapshots and we eval-uated the corresponding epidemic threshold q (21), which pro-vides an estimate of the critical probability for the virus to spreadthrough the network: The lower its value is, the higher the riskof spread. The backbone network was the smallest in January(rainy season) and then increased until reaching its plateau sizein April (early dry season) (Fig. S6B). The number of occasionallinks peaked in May and October. Despite these large variations,q was slightly affected by the presence of these occasional links.Conversely, it was strongly influenced by the variations of tradevolume in the backbone network. The whole network was proneto the diffusion of RVFV: The average epidemic threshold was¯q ' 2.1%, the period of maximum risk from March to May, withq ' 1.5%, and the minimum risk (q ' 3.9%) in January.

The distribution of in/out degrees (number of incoming/outgoing links for a given node) matched a power-law distribu-tion. Only a few nodes (Ihosy, Tsiroanomandy, Ambalavao, andMampikony) were highly connected hubs and also had the highestbetweenness. The four network hubs belonged to three differentgiant strong components (GSC)—Ambalavao and Ihosy were inthe same one. Nodes belonging to these GSC were strongly con-nected between themselves, thus increasing the risk of infectionfor other locations in their neighborhood (22). These hubs werealso cut points in the network: Their removal would disrupt thenetwork connectedness and thus limit the diffusion of an infec-tious agent like RVFV.

The pastoral areas of southwest and northwest of Mada-gascar were the major sources of the traded cattle (Fig. 3A)

and were characterized by high cattle density and low humandensity. The great majority of markets had a negative balancein cattle flows: They were sources in the network, with low localconsumption (Fig. 3B). On the other hand, two hubs (Ambalavaoand Tsiroanomandy) had a large, positive balance. Further infor-mation is provided on Fig. 3C: Cattle were collected in the mar-kets of Tsiroanomandy and Ihosy and then sent to Ambalavaofor slaughtering, Tsiroanomandy also being a consumption cen-ter. In addition, a close examination of these four hubs showedthat many small, outgoing flows offered numerous opportunitiesfor long-distance RVFV spread.Cattle Trade and the Risk of RVFV Infection in Humans. The coef-ficient for the local, environment-related index of RVFV trans-mission to humans was not significant, in contrast to the coef-ficient for the remote, cattle trade-related index (Table S4 andFig. S7). The odds ratio (OR) for the latter was ORC = 1.7 (95%confidence interval (CI): [1.1; 2.7]). The risk of humans gettinginfected with RVFV increased with the intensity of cattle tradefrom areas with infected livestock.

The map of predicted sero-prevalence rate in humans (Fig. 4)and the plot of random variations (Fig. S8) both highlightedlarge spatial variations in the infection rate, with a clear spatialpattern. The highest sero-prevalence rates were encountered inthe densely populated areas of central highlands. This pattern isemphasized in Fig. S8B: A strong additional risk (with respect tocattle trade-related risk) was found in the regions of Antsirabeand Lake Alaotra. See SI Results for details.

DiscussionCattle Trade and the Risk of RVFV Introduction. Imports of liveruminants from Comoros were the main driver for RVFV intro-duction to Madagascar livestock and subsequent trade-relatedmovements of cattle led to its spread to humans, at least in 2008.Phylogenetic studies showing that Malagasy RVFV were closelyrelated to viruses previously circulating in mainland Africa (15)suggest that RVFV was probably introduced into the ComorosIslands through cattle trade with East Africa (23). The existenceof illegal livestock importation from the Comoros Islands pro-vided opportunities for the introduction of RVFV in Madagas-car. Preventing such introductions is therefore essential to avoidfurther RVF epidemics in Madagascar. In practice, strengthen-ing communication between African, Comoros, and Malagasypublic health and veterinary services would be important to shareearly detection in the event of new RVFV circulation. Also, quar-antine measures should be reinforced for ruminants exportedfrom continental Africa to the Comoros Islands.

Environmental Conditions and RVF Epidemics. Anyamba et al. (14)has already pointed out the contrasting behavior of the diseasein Madagascar and East and South Africa. Prolonged heavyrainfall and positive NDVI anomalies occurred after the firstknown epidemics of 1990, e.g., in 1994 with the occurrenceof the Geralda cyclone (24). More recently, torrential rainfalloccurred in February–March 2015 in Madagascar, related tothe strong El Nino conditions (25). No RVF outbreak could bedetected despite specific surveillance measures in 1994 (shortlyafter the 1990–1991 RVF epidemic) and strengthened nationaland regional surveillance in 2015 (26, 27).

RVF epidemics usually start in the arid environments of dam-bos (East Africa) and/or pans (southern Africa) (1), which aresimilar to ecosystems in the Toliara region of southwest Mada-gascar that are covered by xeric shrublands (Fig. S2A). Althoughthis sea-port municipality was hit by RVF, the sero-prevalencerate in ruminants was low in the neighboring municipalities(Fig. 2), probably because the virus was introduced in subopti-mal conditions for the vectors (no heavy rainfall, no flooding).Conversely, high sero-prevalence rates were observed in cattle inthe northwest of the island, covered by the dry forest. Moreover,

Lancelot et al. PNAS Early Edition | 3 of 6

Page 4: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

A B C

Fig. 3. Cattle flows in the Malagasy cattle trade network, 2010. (A) Municipality of origin (number of flows). (B) Balance in cattle flows at the municipalitylevel (number of heads). (C) Directed flows (number of heads).

in East Africa, Aedes mosquitoes are the primary RVFV vec-tors (1): Their biology and ecology are well adapted to arid envi-ronments. Regarding Madagascar, 23 mosquito species might beconsidered as potential RVFV vectors, including no floodwaterAedes mosquito species (28–30). A single species, Culex antenna-tus, meets all criteria for formal classification as an RVFV vec-tor (31). This mosquito is widespread in Madagascar (except inthe North), including in rice paddies that cover large areas ofthe island. The introduction of infected animals in conjunctionwith Culex hatching, during a standard rainy season profile, mighthave amplified the outbreak locally, in many places.

Two major differences are thus highlighted between Madagas-car and East Africa: (i) the lack of connection between the startof 1990–1991 and 2008–2009 RVF epidemics and El Nino eventsand, more generally, with anomalous heavy rainfall and (ii) noobvious role of Aedes mosquitoes in the primary RVFV trans-mission cycle, as well as the wide distribution of Culex (and othermosquito species).

Apparently, the climatic conditions observed during the twoepidemics are common in Madagascar: The drivers triggeringRVFV epidemics must therefore be sought elsewhere. Neverthe-less, these climatic conditions remain important for the amplifi-cation of the primary epidemiological cycle between mosquitoesand ruminants (Table S2).

Cattle Trade as a Driver of RVF Epidemics. The Malagasy popu-lation is growing fast, from 16 million to 24 million between2000 and 2015 (16), and is concentrating in Antananarivo andother large cities (Fig. S2B). Cattle are omnipresent in Malagasyagriculture, economy, and culture. This leads to an ever-growingdemand for cattle meat and draught power for crops. Conse-quently, the increasing cattle trade provides more opportuni-ties for RVF epidemics to spread (7, 8). Our description ofthe national cattle trade network strongly supports this assump-tion and further extends a previous analysis in the North of

Madagascar (8). The connectedness of the cattle trade net-work and its low percolation threshold make the risk of seedingepidemics high.

Lake Alaotra and Antsirabe (Fig. S2A) are two major cropand livestock farming regions. The former is the largest rice-production basin in Madagascar, with many paddies and swamps

A B

Fig. 4. Sero-prevalence rate in humans predicted by a mixed-effect bino-mial logistic regression model: (A) predicted rate and (B) coefficient of vari-ation. In A, numbers were placed at the centroid of districts with the highestpredicted rates (10th decile): 1, Ambohidratrimo; 2, Antananarivo Renivohi-tra; 3, Antananarivo-Sud; 4, Antsirabe Rural; 5, Antsirabe Urban; 6, Betafo;7, Faratsiho; 8, Ambatondrazaka; 9, Amparafaravola; 10, Anosibe; and 11,Moramanga.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1607948114 Lancelot et al.

Page 5: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

APP

LIED

BIO

LOG

ICA

LSC

IEN

CES

favorable to mosquito proliferation. The two regions are denselypopulated, with cities harboring big livestock markets andslaughterhouses to match local red meat needs. Therefore, thereare many opportunities for human exposure to the RVFV whenviremic animals are slaughtered.

Additionally, using the network of markets, butcheries, andhotely (cheap restaurants for travelers), farmers try to sell theirsick animals at the first clinical sign of any disease, to mitigateeconomic losses. When they do not succeed, they slaughter themand eat their meat (32). This practice was probably at the ori-gin of many unreported human infections during the 2008–2009RVF epidemics in Madagascar.

More collaboration with sociologists and anthropologists isneeded to decipher farmers’ perceptions of animal diseases andto assess the social acceptability of prevention, surveillance, andcontrol measures, such as cattle vaccination to protect peo-ple (should enough vaccine be available) or cattle movementrestrictions to avoid RVFV spread through hubs in the cattletrade network (33). If RVFV introduction into Malagasy live-stock continues, targeted (risk-based) vaccination campaigns ofruminants might be organized to protect human populations.Individual protection measures, such as vaccination when thehuman vaccine becomes available or wearing personal protec-tive equipment such as gowns, gloves, safety glasses, and maskswhen slaughtering ruminants would also be important to imple-ment in the most exposed categories of people (34, 35), togetherwith dissemination and training programs. This implementa-tion of coordinated actions between Public Health and Veteri-nary Services would represent an important advance in the so-called “One Health” joint approach of human and animal health(27, 36).

Materials and MethodsTo assess the risk of RVFV introduction, we scanned national and interna-tional databases on sea trade from 2003 to 2008. We also implementedinformal surveys in harbors of the Comoros Islands and northwestern Mada-gascar during and after this epidemic (up to 2010).

To describe the environmental conditions of the past RVF epidemics, weused global datasets available as long-term time series: (i) the SOI as a mainindicator of ENSO of major importance for the climatic conditions in EastAfrica and the southwest tropical Indian Ocean (3); (ii) rainfall data, of cru-cial importance for the epidemiology of vector-borne diseases (37); and (iii)the NDVI as an indicator of rainfall impact on the vegetation (38): forageresources for ruminants and resting sites for mosquitoes. These indicatorswere averaged over the main biomes (SI Materials and Methods). First, weplotted the time series for each biome. Second, we selected data from thedry-forest biome where the highest sero-prevalence rate was observed incattle in 2008 (39). We then used partial triadic analysis (PTA), a multitableversion of principal component analysis (PCA) (40), to identify common orcontrasted patterns in rainy seasons (November to March) from 1983 to2011 (28 y).

To assess the risks of RVFV spread within Madagascar, we used twotypes of data: (i) environmental data selected among the factors of inter-

est for mosquito-borne infections, i.e., related to the availability of rest-ing or breeding sites or the seasonal changes in conditions favoring thedevelopment of their immature stage, etc. (37) (see the list in SI Materialsand Methods), which were obtained from international databases and pro-cessed using standard methodology; and (ii) cattle trade data collected dur-ing repeated nationwide surveys in livestock markets and slaughterhouses,implemented from 2007 to 2011 for the purpose of this study.

Environmental data were used in beta-binomial logistic regression (BBLR)models to predict the RVFV sero-prevalence rate for cattle at the munici-pality level. The serological data were collected after the 2008 epidemicsand published (39). Sera were tested for the presence of anti-RVFV IgG. Weadopted a multimodel inference approach to select the best predictors. Aset of plausible models was kept for model averaging and prediction ofsero-prevalence rate in cattle at the municipality level.

Cattle trade data were used (i) to assess the risk of RVFV spread throughthe network and (ii) to assess the risk of human infection with RVFV. Farms,markets, and slaughterhouses were the nodes of a directed cattle trade net-work. A link between two nodes corresponded to a trade movement ofanimals. The direction indicated the origin and destination places, and thevolume was the number of traded animals.

We assessed the risk of RVFV infection in humans, using sera collectedduring a nationwide survey of slaughterhouse workers performed in 2008and 2009. The study was approved by the Malagasy National Ethical Com-mittee. Participants were included if they gave written informed consent.Sera were tested for the presence of anti-RVFV IgM, indicating a recentinfection, and the results were published (13). We defined two indexesfor the risk of RVFV infection in humans: (i) We created an index of localRVFV transmission risk, defined by the product of local predicted sero-prevalence rate in cattle and cattle density. This index was built to cap-ture the risk associated with the primary RVFV epidemiological cycle involv-ing mosquitoes and domestic ruminants, in the absence of known wildhosts for RVFV in Madagascar (41, 42). Humans are not involved in thiscycle. Therefore, their density was not considered in the index. (ii) Wecreated an index of RVFV transmission risk related to cattle trade. It wasdefined as the product of cattle incoming flow (number of head), predictedsero-prevalence rate in cattle at the origin, and local human density. Thislatter variable was included because meat consumption is proportionallyhigher in urban than in rural areas (43), with higher densities of slaughter-houses, butchers, and meat markets. Consequently, the risk of RVFV infec-tion related to cattle trade should be positively correlated with humandensity.

The effect of these two indexes on human sero-prevalence rate wasassessed with a BBLR model. Finally, a mixed-effect binomial logistic regres-sion model was used to study the local variations of sero-prevalence ratein humans, with the administrative district and region as the nested ran-dom effects associated with the intercept. See SI Materials and Methods fordetailed information on data sources and statistical methods, as well as linksto download the datasets used in the analyses.

ACKNOWLEDGMENTS. This study was partially funded by the World HealthOrganization and the Food and Agriculture Organization of the UnitedNations through the Central Emergency Response Fund of the UnitedNations. It was also partially funded by European Union Grant FP7-613996Emerging Viral Vector-Borne Diseases (VMERGE) and is cataloged by theVMERGE Steering Committee as VMERGE019 (http:/www.vmerge.eu). Thecontents of this publication are the sole responsibility of the authors and donot necessarily reflect the views of the European Commission.

1. Linthicum KJ, Britch SC, Anyamba A (2016) Rift Valley fever: An emerging mosquito-borne disease. Annu Rev Entomol 61:395–415.

2. de LaRocque S, Formenty P (2010) Rift Valley fever: Disease ecology and early warn-ing. Sustainable Management of Animal Production and Health, eds Odongo N,Garcia M, Viljoen G (Food and Agriculture Organization of the United Nations,Rome), pp 327–333.

3. Linthicum K, et al. (1999) Climate and satellite indicators to forecast Rift Valley feverepidemics in Kenya. Science 285(5426):397–400.

4. Saji NH, Goswami BN, Vinayachandran PN, Yamagata T (1999) A dipole mode in thetropical Indian Ocean. Nature 401(6751):360–363.

5. Anyamba A, et al. (2009) Prediction of a Rift Valley fever outbreak. Proc Natl Acad SciUSA 106(3):955–959.

6. Chevalier V, Pepin M, Plee L, Lancelot R (2010) Rift Valley fever–A threat for Europe?Euro Surveill 15(10):19506.

7. Morvan J, Saluzzo JF, Fontenille D, Rollin PE, Coulanges P (1991) Rift Valley fever onthe east coast of Madagascar. Res Virol 142(6):475–482.

8. Nicolas G, Durand B, Duboz R, Rakotondravao R, Chevalier V (2013) Description andanalysis of the cattle trade network in the Madagascar highlands: Potential role inthe diffusion of Rift Valley fever virus. Acta Trop 126(1):19–27.

9. Shoemaker T, et al. (2002) Genetic analysis of viruses associated with emergenceof Rift Valley fever in Saudi Arabia and Yemen, 2000-01. Emerg Infect Dis 8(12):1415–1420.

10. Gerdes G (2004) Rift Valley fever. Rev Sci Tech 23(2):613–623.11. Anyangu AS, et al. (2010) Risk factors for severe Rift Valley fever infection in Kenya,

2007. Am J Trop Med Hyg 83(2 Suppl):14–21.12. Reed C, et al. (2013) Aerosol exposure to Rift Valley fever virus causes

earlier and more severe neuropathology in the murine model, which hasimportant implications for therapeutic development. PLoS Negl Trop Dis 7(4):e2156.

13. Andriamandimby SF, et al. (2010) Rift Valley fever during rainy seasons, Madagascar,2008 and 2009. Emerg Infect Dis 16(6):963–970.

14. Anyamba A, et al. (2010) Prediction, assessment of the Rift Valley fever activity in Eastand Southern Africa 2006-2008 and possible vector control strategies. Am J Trop MedHyg 83(2 Suppl):43–51.

15. Carroll SA, et al. (2011) Genetic evidence for Rift Valley fever out-breaks in Madagascar resulting from virus introductions from the EastAfrican mainland rather than enzootic maintenance. J Virol 85(13):6162–6167.

Lancelot et al. PNAS Early Edition | 5 of 6

Page 6: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

16. The World Bank Group (2015) Countries: Madagascar (The World Bank Group,Washington, DC). Available at www.worldbank.org/en/country/madagascar. AccessedDecember 28, 2015 .

17. Sissoko D, et al. (2009) Rift Valley fever, Mayotte, 2007-2008. Emerg Infect Dis15(4):568–570.

18. Cetre-Sossah C, et al. (2012) Prevalence of Rift Valley fever among ruminants,Mayotte. Emerg Infect Dis 18(6):972–975.

19. NOAA (2015) Southern Oscillation Index. Available at https://www.ncdc.noaa.gov/teleconnections/enso/indicators/soi/. Accessed December 30, 2015.

20. Pinzon JE, Tucker CJ (2014) A non-stationary 1981–2012 AVHRR NDVI3g time series.Remote Sens 6(8):6929–6960.

21. Volkova VV, Howey R, Savill NJ, Woolhouse MEJ (2010) Sheep movement networksand the transmission of infectious diseases. PLoS One 5(6):e11185.

22. Danon L, et al. (2011) Networks and the epidemiology of infectious disease. Interdis-cip Perspect Infect Dis 2011:1–28.

23. Maquart M, et al. (2016) Phylogeographic reconstructions of a Rift Valley fever virusstrain reveals transboundary animal movements from eastern continental Africa tothe Union of the Comoros. Transbound Emerg Dis 63(2):e281–e285.

24. Fitchett JM, Grab SW (2014) A 66-year tropical cyclone record for south-east Africa:Temporal trends in a global context. Int J Climatol 34(13):3604–3615.

25. Monastersky R (2016) Monster El Nino probed by meteorologists. Nature 529(7586):267–268.

26. Zeller H, Andriamanana R, Quirin R (1998) Rift Valley Fever Surveillance in Mada-gascar, Tech Rep Project IPM/PSE/BM, January 1996–March 1998 (Institut Pasteur deMadagascar, Antananarivo, Madagascar).

27. Cardinale E, Rasamoelina-Andriamanivo H, Razafimandimby A, Lepec R, Flachet L(2015) A concrete regional “One Health” surveillance system and management ofepidemics: A success story in the South West Indian Ocean. ISVEE 14, Yucatan 2015-Veterinary Epidemiology & Economics: Planning Our Future, Session 2: One Health,ed De Anda JH (ISVEE, Merida, Mexico) .

28. Ratovonjato J, et al. (2011) Detection, isolation, and genetic characterization ofRift Valley fever virus from Anopheles (Anopheles) coustani, Anopheles (Anopheles)squamosus, and Culex (Culex) antennatus of the Haute Matsiatra region, Madagascar.Vector Borne Zoonotic Dis 11(6):753–759.

29. Nepomichene TJJN, Elissa N, Cardinale E, Boyer S (2015) Species diversity, abundance,and host preferences of mosquitoes (Diptera: Culicidae) in two different ecotypes ofMadagascar with recent RVFV transmission. J Med Entomol 52(5):962–969.

30. Tantely LM, Boyer S, Fontenille D (2015) A review of mosquitoes associated with RiftValley fever virus in Madagascar. Am J Trop Med Hyg 92(4):722–729.

31. Reeves WC (1958) Arthropods as vectors and reservoirs of animal pathogenic viruses.Handbuch der Virusforschung [Handbook of Virus Research], eds Hallauer C, MeyerKF (Springer, New York), pp 177–202.

32. Ribot JJ, Coulanges P (1988) Les zoonoses malgaches [Malagasy zoonoses]. Rev ElevMed Vet Pays Trop 41(1):9–22.

33. Goutard FL, et al. (2015) How to reach the poor? Surveillance in low-income countries,lessons from experiences in Cambodia and Madagascar. Prev Vet Med 120(1):12–26.

34. Zeller H, Rakotoharinadrasana H, Rakoto-Andrianarivelo M (1998) Rift Valley fever inMadagascar: Infection risks for the abattoir staff in Antananarivo. Revue Elev MedVet Pays Trop 51(1):17–20.

35. Bausch DG, Senga M (2017) International Encyclopedia of Public Health (Elsevier, Ams-terdam, The Netherlands), 2nd Ed, Vol 2, pp 396–409.

36. de LaRocque S, Formenty P (2014) Applying the One Health principles: A trans-sectoral coordination framework for preventing and responding to Rift Valley feveroutbreaks. Rev Sci Tech 33(2):555–567.

37. Kalluri S, Gilruth P, Rogers D, Szczur M (2007) Surveillance of arthropod vector-borne infectious diseases using remote sensing techniques: A review. PLoS Pathog3(10):1361–1371.

38. Anyamba A, Small JL, Tucker CJ, Pak EW (2014) Thirty-two years of Sahelian zonegrowing season non-stationary NDVI3g patterns and trends. Remote Sens 6(4):3101–3122.

39. Jeanmaire E, et al. (2011) Prevalence of Rift Valley fever infection in ruminants inMadagascar after the 2008 outbreak. Vector Borne Zoonotic Dis 11(4):395–402.

40. Thioulouse J, Simier M, Chessel D (2004) Simultaneous analysis of a sequence of pairedecological tables. Ecology 85(1):272–283.

41. Olive M-M, Goodman SM, Reynes JM (2012) The role of wild mammals in the mainte-nance of Rift Valley fever virus. J Wildl Dis 48(2):241–266.

42. Olive M-M, et al. (2013) Absence of Rift Valley fever virus in wild small mammals,Madagascar. Emerg Infect Dis 19(6):1025–1027.

43. Institut Nationalde la Statistique de Madagascar (2015) Policy Brief - Les PrincipauxResultats sur le tTheme “Alimentation et Securite Alimentaire” [Main Results for Foodand Food Security], Technical Report (INSTAT, Antananarivo, Madagascar). Availableat instat.mg/. Accessed May 6, 2016.

44. Olson DM, et al. (2001) Terrestrial ecoregions of the world: A new map of life onEarth. BioScience 51(11):933–938.

45. Pepin M, Bouloy M, Bird BH, Kemp A, Paweska J (2010) Rift Valley fever virus (Bun-yaviridae: Phlebovirus): An update on pathogenesis, molecular epidemiology, vectors,diagnostics and prevention. Vet Res 41(6):61.

46. Madani TA, et al. (2003) Rift Valley fever epidemic in Saudi Arabia: Epidemiological,clinical, and laboratory characteristics. Clin Infect Dis 37(8):1084–1092.

47. The CGIAR Consortium for Spatial Information (CGIAR-CSI) (2015) SRTM 90m Dig-ital Elevation Data, Update Version 4 (International Centre for Tropical Agricul-ture, Palmira, Colombia). Available at srtm.csi.cgiar.org/. Accessed December 28,2015.

48. Vito (2015) DevCoCast: The GEONETCast for and by Developing Countries [FlemishInstitute for Technological Research NV (Vito), Mol, Belgium]. www.vgt4africa.orgAccessed December 28, 2015.

49. NASA (2015) MODIS - Moderate Resolution Imaging Spectroradiometer(National Aeronautics and Space Administration, Washington, DC). Available athttps://modis.gsfc.nasa.gov/. Accessed December 28, 2015.

50. International Research Institute for Climate and Society, Earth institute, ColumbiaUniversity (2015) Malaria Early Warning System (MEWS). Dekadal (10-Day) Pre-cipitation (Earth Institute. Columbia University, New York). Available at ridl.ldeo.columbia.edu/maproom/Health/Regional/Africa/Malaria/MEWS/. Accessed December28, 2015.

51. Tatem A, Linard C (2011) Population mapping of poor countries. Nature 474(7349):36.

52. WorldPop Project (2015) WorldPop: Open Access Demographic Data (GeoData Insti-tute. Univ of Southampton, Southampton, UK). Available at www.worldpop.org.uk.Accessed December 28, 2015.

53. Robinson TP, et al. (2014) Mapping the global distribution of livestock. PLoS One9(5):e96084.

54. United Nations Statistics Division (2016) UN Comtrade Database (United Nations.New York). Available at comtrade.un.org/data/. Accessed January 1, 2016.

55. Munyua P, et al. (2010) Rift Valley fever outbreak in livestock in Kenya, 2006-2007.Am J Trop Med Hyg 83(2 Suppl):58–64.

56. Nderitu L, et al. (2011) Sequential Rift Valley fever outbreaks in eastern Africa causedby multiple lineages of the virus. J Infect Dis 203:655–665.

57. Scetauroute International (2004) Plan national de transport 2004–2020.Evaluation des principaux flux de transport maritime. Rapport final pro-visoire [National plan for transportation 2004–2020. Assessment of themain sea-traffic flows. Draft final report], (Republique de Madagascar, Vice-primature chargee des programmes economiques, Ministere des TravauxPublics, des Transports et de l’amenagement du Territoire, Programme secto-riel des transports, Secretariat executif, Antananarivo, Madagascar), TechnicalReport.

58. Wood SN (2003) Thin plate regression splines. J R Stat Soc Series B Stat Methodol65(1):95–114.

59. Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Prac-tical Information-Theoretic Approach (Springer, New York), 2nd Ed.

60. Wasserman S, Faust K (1994) Social Network Analysis: Methods and Applications,Structural Analysis in the Social Sciences (Cambridge Univ Press, Cambridge,UK) Vol 8.

61. Ortiz-Pelaez A, Pfeiffer DU, Soares-Magalhaes RJ, Guitian FJ (2006) Use of socialnetwork analysis to characterize the pattern of animal movements in the initialphases of the 2001 foot and mouth disease (FMD) epidemic in the UK. Prev Vet Med76(1-2):40–55.

62. RCore Team (2016) R: A Language and Environment for Statistical Computing (R Foun-dation for Statistical Computing, The R Foundation, Vienna).

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1607948114 Lancelot et al.

Page 7: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Supporting InformationLancelot et al. 10.1073/pnas.1607948114SI Materials and Methods

Data.Environmental conditions of RVF epidemics. Three variables wereused to characterize the environmental conditions of RVF epi-demics in Madagascar: (i) the SOI, (ii) rainfall, and (iii) NDVI.SOI is a standardized index based on the observed sea-level pres-sure differences between Tahiti and Darwin, Australia. Its nega-tive (warm) anomalies are often associated with RVF epidemicsin eastern Africa (3). Other indicators related to the sea-surfacetemperature (SST) in the western tropical Indian Ocean, suchas the dipole mode index (4), account for a lower proportion ofSST variability; they were not considered here. We used timeseries of standardized values (data expressed in standard devi-ations) for each of these indicators: Southern oscillation indexprovided by National Centers for Environmental Information(NOAA) (19); for rainfall, NOAA’s PREC/L provided by theNOAA/OAR/ESRL PSD, Boulder, CO; and NDVI: GIMMSAVHRR NDVI version 3 (20).

Madagascar is a large island (1,580 km long, 590,000 km2)with contrasted climatic and other environmental conditions. Wecomputed standardized rainfall and NDVI for each of the threemain biomes in terms of land cover: (i) tropical and subtropicalmoist broad-leaf forests (referred to as “moist forest”), locatedon the eastern coast and highlands of Madagascar; (ii) tropicaland subtropical dry broad-leaf forests (referred to as “dry for-est”), located in the northwestern part of the island; and (iii)deserts and xeric shrublands (referred to as “xeric shrublands”),covering the southwestern region of Madagascar (Fig. S2A). Thedefinition and geographical distribution of these biomes weretaken from Olson et al. (44) and associated datasets.Sero-prevalence data in ruminants. Data were kindly provided byMalagasy Veterinary Services, Pasteur Institute of Madagascar,and Food and Agriculture Organization of the United Nations.A partial analysis of these data has been published (39). Theywere collected during the last RVF epidemic in 2008 in Mada-gascar. Animals were sampled in local livestock market placesor before slaughtering. Their birth places were recorded at themunicipality level. In total, 316 municipalities were involved inthis survey. Blood was taken from 4,427 ruminants, i.e., 3,438 cat-tle, 227 sheep, and 762 goats. Sera were tested for the presence ofanti-RVFV antibodies—immunoglobulins of type G (IgG) andM (IgM), with commercial enzyme-linked immunosorbent assays(ELISA). IgG appear soon after the rise of IgM and persist muchlonger: Animals surviving a RVF epidemic can develop a life-long immunity to RVF infections (45). In our sample, the medianage of animals was 4 y, ranging from 3 mo to 19 y; 357 ani-mals were born during the 2008–2009 RVF epidemic, and only8 animals were born during or before the 1990–1991 RVF epi-demic. Therefore, the risk of sampling animals infected duringthis earlier epidemic was limited. In addition, results of phyloge-netic studies of viruses sampled during epidemics, of event-basedsurveillance, and of cross-sectional and prospective serologicalsurveys in cattle did not provide strong evidence of long-term andwidespread persistence of RVFV circulation between successiveepidemics in Madagascar (15, 34). Therefore, we used IgG fordata analysis, considering that their sero-prevalence was a goodindicator of RVFV infection during the 2008–2009 epidemic.Sero-prevalence data in humans. We used the dataset describedand analyzed by Andriamandimby et al. (13). A nationwidecross-sectional serological survey was performed in 2008 and2009, which was approved by the Malagasy National EthicalCommittee. It covered people at risk for RVFV exposure in all

111 districts of Madagascar. Participants were included if (i) theyhad been working in slaughterhouses located in the district since2007 or had been exposed to fresh meat or blood of ruminantsand if (ii) they had been residents of the district when they startedthis work. Five milliliters of blood was then collected from eachparticipant and samples were tested for anti-RVFV IgG andIgM, using ELISA (46). Because IgG may persist more than 10 yin humans after a natural infection with RVFV, we used IgMserological prevalence rate as the response in this analysis.Environmental data. We used remotely sensed data as detailedbelow and computed their average value over the area of eachmunicipality.

We used elevation data with a 90-m resolution, as a generalindicator of climatic conditions and agro-ecosystem. This datasetwas available from the Consortium for Spatial Information ofthe Consultative Group for International Agricultural Research(CGIAR-CSI) (47).

Water bodies and vegetation are key variables for the devel-opment of mosquitoes and the abundance of their ruminanthosts. Surface water is necessary for the larval development ofmosquitoes and for watering cattle and small ruminants in pas-toral areas. We calculated the maximum proportion of waterbodies in each municipality during the study period.

Live green vegetation, characterized by NDVI, is importantfor adult mosquito mating and resting. Also, NDVI is an indi-cator of relative humidity, which is a major factor for mosquitosurvival. And finally, NDVI is related to the amount of vege-tal biomass that is used by ruminants to cover their nutritionalneeds. We computed minimum, average, maximum, and SD ofNDVI, together with the proportion of decades with an averagevalue greater than the national average.

Water bodies and NDVI datasets were available from theGEONETCast project and its data repository (48).

Temperature is a major factor for mosquito activity, develop-ment, and survival, as well as for the kinetics of the extrinsicvirus cycle in RVF-infected mosquitoes (37). We used two mod-erate resolution imaging spectroradiometer (MODIS) 8-d com-posite datasets provided by the National Aeronautics and SpaceAdministration (NASA) (49): nighttime land surface tempera-ture (NLST) and daytime land surface temperature (DLST). Weused minimum, maximum, and average NLST and DLST values.

Rainfall is important for the larval development of mosquitoesand for the development of herbaceous vegetation (37). We used10-d composite data provided by the Malaria Early WarningSystem (MEWS) (50). We extracted minimum, average, max-imum, and SD of rainfall precipitation, together with propor-tion of decades with an average value greater than the nationalaverage.Population data. We used a raster dataset ('100 m spatial reso-lution) of the 2010 Malagasy human population (51) kindly pro-vided by the program WorldPop (52) (Fig. S2B). We computedthe total human population in each municipality.

For ruminant population, we used the Gridded Livestock ofthe World (GLW) database, which provides modeled livestockdensities of the world, adjusted to match official national esti-mates for the reference year 2010, at 1-km spatial resolution (53)(Fig. S2C).Ruminant trade. We assumed RVFV might have been intro-duced in Madagascar via international livestock trade. Look-ing for data to support this assumption, we queried the UnitedNations’ ComTrade database (54), a repository of officialtrade statistics. It contains annual trade statistics starting from1962 (and monthly trade statistics since 2010). Because (i) a

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 1 of 11

Page 8: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

retrospective serological survey using a cattle serum bankshowed that RVFV has been circulating in Comoros Archipelagoat least since 2004 (18) and (ii) a large RVFV epidemic started inSomalia in 2006 and reached northern Tanzania 1 y later (14, 55,56), we started the queries in 2005, looking for cattle and smallruminant importations.

We also investigated sea trade data, particularly coastal tradewith small freighters and botry (dhows). Because many coastalareas are poorly served by terrestrial roads, a lot of goods—including live animals—are shipped by sea. First, in 2010, weconducted interviews with dhow captains in the main harborsof Comoros Archipelagos and northwestern Madagascar (Maha-janga, Nosy Be, Antsiranana) to assess whether live ruminantsare shipped from the Comoros Islands to Madagascar andbetween Malagasy harbors. Second, we used published data onsea trade traffic between Malagasy harbors to identify a possibleRVFV entry point (57).

Nationwide cattle movements were collected by field veteri-nary officers from 2007 to 2011. Data were taken from traderegistries held for tax purposes in primary and secondary cat-tle markets. Also, a questionnaire survey was implemented byfield veterinary officers during the same period with farmers andlivestock traders bringing their animals into slaughterhouses. Allof the data were cleaned, checked, geo-referenced, and storedin a PostgreSQL database (The PostgreSQL Global Develop-ment Group). Elementary data (description of a cattle move-ment) encompassed the date, origin, destination, source type(farmer, livestock market, or slaughterhouse), and volume (num-ber of animals involved in the movement).Data availability. All of the environmental and population dataare freely available, following indications given in SI Materi-als and Methods. The borders of Madagascar administrativeunits needed for the analyses (municipalities, districts, andregions) are freely available from the GADM database ofGlobal Administrative Areas. Other datasets are available fol-lowing these links: Coastal navigation data are available athttps://figshare.com/s/73ffd8ff3d3011c8e741. Cattle trade dataare available at https://figshare.com/s/7a1c14c8f0a5ccb7e7b8.Animal sero-prevalence data are available at https://figshare.com/s/f09650b3309a1a944f15. Correspondence between IDsused in animal sero-prevalence data and in vector layers ofadministrative units is available at https://figshare.com/s/84924c2fbc37f03f4ab1. Human sero-prevalence data are avail-able at https://figshare.com/s/a0a8ec4be4d78dbd53ac.

Data Analysis.Partial triadic analysis of climatic variables. We jointly analyzeda series of 28 tables, each representing a single rainy season.Each table had three columns (SOI, rainfall, and NDVI) andfive rows (months from November to March). From these inputs,data were reorganized into new table with the 5 mo× three vari-ables (five groups of three) as the rows and 28 rainy seasons asthe columns. A PCA was then run on this new table to get theinterstructure of the 28 rainy seasons. The correlation circle wasdrawn to visualize the correlations between the rainy seasons.Finally, their row numbers were used in a hierarchical clusteranalysis, and the resulting dendrogram was used to visualize theclusters of rainy seasons.Modeling ruminant serological prevalence rate. The goals were(i) to assess whether municipalities located close to a harbor weremore at risk for RVFV transmission in cattle than others, (ii) toidentify the most important environmental predictors of RVFVtransmission in cattle, and (iii) to estimate the population-levelsero-prevalence rate in cattle and to map this rate at the munic-ipality level to characterize the risk of RVFV spread throughthe cattle trade network. For this purpose, we built a statisticalmodel for the sero-prevalence rate of antibodies against RVFVin municipality i , computed as the proportion πi of cattle sera

testing positive with ELISA. Predictors were the environmentalvariables described above. To assess the risk of RVFV introduc-tion and spread related to sea trade, we defined circular buffersof various radii (10 km, 25 km, and 50 km) around sea portsand built a municipality-level indicator, taking the value of 0 ifthe buffer intersected with the municipality (reference categoryin regression analyses) and 1 elsewhere. We kept the buffer sizeminimizing the Akaike information criterion corrected for small-sample size (AICc: see below) in statistical models.

To assess the shape of the relationship between each predic-tor and the serological prevalence rate, we estimated and plot-ted the trend in the logit of this rate and a smoothing splinefunction of the data (58). Those predictors showing a nonlin-ear trend with the sero-prevalence rate on the logit scale werecategorized into three classes according to the 1/3 quantiles.We also checked between-predictor correlations, using PCA onthe table of predictors, and hierarchical clustering on the PCAscores. We discarded highly correlated variables (ρ > 0.8), keep-ing those with the most straightforward bio-ecological meaning.Sero-prevalence rate was then modeled using BBLR to accountfor spatial correlation of these rates and their possible overdis-persion with respect to the binomial distribution.

For model selection and inference, we adopted an approachbased on model comparison with the AICc (59),

AICc = −2× L+ 2× k × n/(n − k − 1)

with L the maximized likelihood, k the number of model coef-ficients, and n the sample size. A set of R BBLR models wasdefined using all possible combinations of the selected predic-tors. Goodness of fit of the most complex model was assessedusing Pearson’s χ2 test. Models were then ranked according tothe difference in AICc with respect to the AICc-best model (∆i)and weighted according to their Akaike weight:

wi = exp(−0.5×∆i)

/ R∑r=1

exp(−0.5×∆r ).

Averaged coefficients and standard deviations were computed,as well as confidence intervals for model coefficients and predic-tions. Averaged BBLR model’s predictive capacity was assessedusing the ROC curve, and the corresponding area under thiscurve (AUC). The averaged model was then used to predict sero-prevalence rates and coefficients of variations for the 1,434 Mala-gasy municipalities.Ruminant trade network. We used the framework of social net-work analysis (SNA) to describe the links between a set of inter-connected so-called nodes, i.e., farmers, livestock markets, orslaughterhouses (60). They were sources or destinations in thecattle movement network. Links were directed: movements fromsource to destination nodes.

We focused on the 2010 cattle trade network, for which themost comprehensive data were available. These data are repre-sented in Fig. S6A, where each line corresponds to a trade link. Asegment was drawn if the link was active in a specific month. Thecolor corresponds to the frequency of activity of the link over 5 y.In 2010, besides the links that were present before, a new set oflinks appeared and persisted over all the year. Thus, almost 60%of the links were present through all of the years. The remaininglinks were active only for 2 mo in 2010, at the most.

We described the seasonal and annual distribution of the num-ber of traded cattle, and then we used SNA tools to find flowpatterns in the network. Network-centrality measures were com-puted, such as degrees (in- and out-degrees), betweenness, orcloseness, to identify the major nodes with respect to livestockmovements and the consecutive risk of disseminating RVFV(61). The average path length provided information about thenumber of hops to reach a given node. Consequently, it is ameasure of the average time before a node can be infected.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 2 of 11

Page 9: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Correlation between connected nodes and the in-degree (j )and out-degree(k) of the nodes were also estimated. Follow-ing Volkova et al. (21), we computed the epidemic threshold q ,which is an estimate of the minimum transmission probability fora newly introduced virus to infect all of the network:

q =〈k〉〈jk〉 .

Building indexes of human risk of RVFV infection. The goals were(i) to build risk indexes of RVFV human infection related tothe local environment (local risk) and cattle trade (remote risk),(ii) to assess their relative importance, and (iii) to identify andmap high-risk areas.

The local risk was assumed to be proportional to the numberof locally infected cattle. Therefore, the local index was definedas the product of the cattle density (heads per square kilometer)and the sero-prevalence rate in cattle predicted by the BBLRmodel.

The remote risk index was the product of the following threevariables:

• The number of incoming cattle (estimated from the trade datain 2010), aggregated at the district level.

• The average sero-prevalence rate po in cattle at the origin.For a given destination (district), we enumerated for the J dif-ferent origins (i) the observed numbers of exported cattle nj

and (ii) the predicted sero-prevalence rates pj (j = 1, . . . , J ).Then po was computed as

po =

∑Jj=1 njpj∑Ji=1 nj

.

• The average human density (number of inhabitants per squarekilometer) at the district level. This variable was included inthe index to account for its anthropogenic nature. As a mat-ter of fact, the socio-economic pattern is quite different inurban and rural populations. Meat consumption is propor-tionally higher in the former: According to a national surveyimplemented by the National Institute of Statistics in 2015(43), meat, fish, and animal products were consumed on aver-age 3.1 d−1·wk−1 (2.5 d−1·wk−1) in urban (rural) households(n = 1,499, Wilcoxon test, P < 10−4). The higher human den-sity combined with higher meat consumption results in ahigher density of slaughterhouse workers, butchers, and peo-ple exposed to infected blood, blood aerosols, or fresh meat.Therefore, the risk of RVFV infection related to cattle tradeshould be positively correlated with human density.These two indexes were included in a BBLR model to assess

their adjusted, relative importance. To validate the model, weused the confirmed cases of RVF in humans reported by thenational surveillance system (13). The confirmed clinical casesof RVF occurred in 20 districts, mostly located on the high-lands. We summed the predicted sero-prevalence rate in humansfor these districts, thus forming the b0 value of our statistic ofinterest. Then, we randomly sampled 20 districts from the listof 105 districts with a predicted sero-prevalence rate and com-puted the b1 value of the statistic. We iterated the process Btimes, combined the observed and simulated values, and com-

puted P =P(bi ≥ b0, i = 1, . . . ,B + 1), with large values of B(e.g., B = 9,999). The model was validated if P < 0.05.

To study the local variations of sero-prevalence rate in humansaround the population mean predicted by the two risk indexes,we used a mixed-effect binomial logistic regression (MBLR)model with two random effects u and v : the administrative dis-trict and the administrative region (grouping level for districts).

The R language and environment for statistical computingwas used for data analysis, including plots and maps (62). Spe-cific tasks were accomplished with the following add-on pack-ages: ade4 for PCA and PTA; sna for social network analy-sis; aods3 and MuMIn for BBLR and multimodel inference;lme4 for MBLR; and sp, rgeos, gdistance, and raster for spatialanalyses and mapping. All these packages and their documen-tation are available on the Comprehensive R Archive Network(https://cran.r-project.org).

SI ResultsRuminant Sero-Prevalence Rate. The estimated sero-prevalencerate in cattle was 25.9% (n = 4,427), 95% CI [24.6; 27.2]. PCAand correlation plots on the set of explanatory variables showedhigh correlations between many of these variables (Fig. S4). Outof 18 possible predictors, 5 were kept for modeling: altitude,average DLST, average NDVI, average monthly rainfall, andproportion of municipality area covered by water bodies. Furtheranalyses on the selected predictors revealed nonlinear relation-ships between the predictors and the response on the logit scale(Fig. S5). Therefore, we cut altitude and DLST into three cate-gories defined by their 1/3 empirical quantiles. Also, we trans-formed the proportion of municipality area covered by waterbodies into its square root. To account for the risk of RVFVintroduction through sea trade, we added a predictor taking thevalue 0 if a given municipality was within 50 km of a sea port and1 elsewhere.

Pearson’s test for the full BBLR model with the six pre-dictors did not allow rejecting the null hypothesis of a goodfit: χ2 = 133.8, df = 118, P = 0.15. We selected the subset ofmodels with AICc difference <7 with respect to the AICc-bestmodel. This subset included 12 models that were used for modelaveraging (Table S1).

Human Sero-Prevalence Rate. The observed sero-prevalence rateof IgM was 10.6% (n = 1,975, 95% CI [8.4%; 12.9%]). Resultsof exploratory data analysis are shown in Fig. S7. These indexeswere the predictors in a BBLR model of human sero-prevalencerate. Pearson’s χ2 test for the overall goodness of fit did notallow rejecting the null hypothesis that this model correctlyfitted the data (χ2 = 97.6, df = 101, P = 0.58). The external val-idation test was successful (P = 0.02, B = 9,999). Model coeffi-cients are reported in Table S4. The coefficient for the local indexwas not significant (P = 0.67), conversely to the coefficient forthe remote, cattle trade-related index (P = 0.02).

A MBLR model was then used to study the variance compo-nents of sero-prevalence rate. It had the same fixed effects as theBBLR (local and cattle trade-related indexes) and two nestedrandom effects associated with the intercept: u (district: n = 111)and v (region: n = 22). The standard deviations of u and v werelarge and significantly >0 (α= 0.05): σu = 0.61, 95% CI [0.35;0.90], σv = 0.53, 95% CI [0.21; 0.91].

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 3 of 11

Page 10: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

42°E

44°E

46°E

48°E

50°E

52°E

25°S

20°S

15°S

10°S

Long

itude

Latitude

030

0 km

Car

goD

how

AN

TSIR

AN

AN

A

VO

HE

MA

R

MA

HA

JAN

GA

TOA

MA

SIN

A

TOLI

AR

A

TAO

LAN

AR

O

AN

TSO

HIH

Y

NO

SY

BE

AN

TALA

HA

MA

NA

KA

RA

MO

RO

ND

AVA

BE

SA

LAM

PY

PO

RT

SA

INT

LOU

IS

SO

ALA

LA

MA

NA

NJA

RY

BE

LO

MA

RO

AN

TSE

TRA

MA

INTI

RA

NO

MA

NA

NA

RA

MO

RO

MB

E

VIN

AN

IVA

O

AN

ALA

LAVA

AN

TON

IBE

(A)

TOAMASINA

MAH

AJAN

GA

ANTS

IRAN

ANA

NAMAKIA

TAOLA

NARO

PORT

SAI

NT L

OUIS

ANTALAHA

MO

RO

ND

AVA

TOLI

ARA

VOHEM

AR

NOSYBE

ANTSOHIHYMOROMBE

MAROANTSETRA

ANANALARTI HOB MA

BELO

RATSIANARANA AMBODIRAFIA SOALALAMANANARA BESALAMPY

TAMBOHORANO

ANTONIBE

MARAMBO

VINAN

IVAO

BEM

ARIV

O

MAI

NTIR

ANO

AMPA

NAV

OAN

A

VILA

MAT

SA

MAN

AKAR

A

MAN

ANJA

RY

AN

ALA

LAVA

(B)

Fig

.S1.

Sea

trad

etr

affi

cin

Mad

agas

car:

coas

taln

avig

atio

nb

yca

rgo

esan

dd

ho

ws.

(A)

Pseu

do

rou

tes

wer

eco

mp

ute

du

sin

ga

leas

t-co

std

ista

nce

alg

ori

thm

app

lied

on

afr

icti

on

surf

ace

defi

ned

by

bat

hym

etri

cd

ata

(Th

eSa

telli

teG

eod

esy

rese

arch

gro

up

atSc

rip

ps

Inst

itu

tio

no

fO

cean

og

rap

hy,

Un

iver

sity

of

Cal

ifo

rnia

,San

Die

go

).R

ou

tew

idth

isp

rop

ort

ion

alto

the

traf

fic

inte

nsi

ty(t

on

s·y−

1in

2004

).(B

)C

ho

rdd

iag

ram

sho

win

gth

etr

ade

flo

ws

(nu

mb

ero

ffl

ow

s).D

ata

for

Aan

dB

:Sce

tau

rou

teIn

tern

atio

nal

(57)

.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 4 of 11

Page 11: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Fig. S2. Biomes (A) and human and ruminant population densities (B and C), Madagascar. Sources: Olson et al. (44) for the biomes, Tatem et al. (51) for thehuman population, and Robinson et al. (53) for ruminant populations.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 5 of 11

Page 12: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

1983−1984

1984−1985

1985−1986

1986−1987

1987−1988

1988−1989

1989−1990

1990−1991

1991−1992 1992−1993

1993−1994

1994−1995 1995−1996

1996−1997

1997−1998

1998−1999

1999−2000

2000−2001

2001−2002 2002−2003

2003−2004

2004−2005

2005−2006

2006−2007

2007−2008

2008−2009

2009−2010

2010−2011

Eigenvalues

(A) 1988−19891999−20002008−20092007−20082010−20112005−20061998−19992000−20012002−20031997−19981986−19871991−19921987−19881989−19902009−20101985−19862004−20051992−19931993−19941984−19851995−19961983−19842006−20071996−19972003−20041994−19951990−19912001−2002

(B)Fig. S3. Typology of 28 rainy seasons (1983–2011) in the dry-forest biome of Madagascar, according to anomalies of the SOI, of rainfall and of NDVI.(A) Correlation circle of the 28 rainy seasons. (B) Dendrogram of a hierarchical cluster analysis on rainy seasons.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 6 of 11

Page 13: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

0

1

2

3

4

5

6

(A)

mxDlst

mnDlst

avDlst

mxNlst

mnNlst

avNlst

mxNDVI

mnNDVI

avNDVI

sdNDVI

propNDVI

Alt

mxPLUVIO

avPLUVIO sdPLUVIO

propPLUVIO

RapSurfWB

(B)

mxDlstsdNDVImxNlstavNlstmnNlstmnDlstavDlstAltavPLUVIOpropPLUVIOmxNDVIavNDVIpropNDVImxPLUVIORapSurfWBmnNDVIsdPLUVIO

(C)

V1

10 25

0.52 0.78

16 26

0.46 0.41

10 18

0.42 0.08

40 140

0.16 0.43

15 40

0.57 0.39

0 1500

0.31 0.21

15 40

0.39 0.045

0.10 0.40

0.35

3555

0.071

1025

��

� �

� �

��

��

��

���

� �

��

��

��

��

��

��

��

�� �

V2 0.85 0.65 0.74 0.73 0.094 0.24 0.12 0.37 0.11 0.61 0.11 0.56 0.096 0.48 0.13

� �

� �

��

��

� �

��

��

��

� �

��

��

��

��

��

��

��

��

��

��

��

� �

��

� �

��

��

� �

��

� �

� �

��

��

��

��

��

��

��

� �

��

��

��

� �

��

��

��

V3 0.69 0.72 0.75 0.12 0.11 0.30 0.49 0.27 0.60 0.16 0.52 0.044 0.49

22320.096

1628 �

��

��

��

��

��

��

��

� �

�� �

��

�� �

� ��

��

� ��

��

���

��

� �

� ��

� �

��

��

� �

��

��

��

� �

��

� �

��

� �

� � �

��

� ��

���

��

���

��

� ��

��

��

� ��

� �

��

��

��

���

��

��

��

��

��

��

� �

� � �

��

�� �

� ��

��

� ��

��

���

��

� �

� ��

� �

��

��

��

V4 0.66 0.93 0.35 0.14 0.20 0.28 0.19 0.95 0.13 0.49 0.16 0.49 0.32�

��

��

� �

� �

� �

� �

��

���

��

��

��

��

� �

��

��

��

��

���

� �

��

� �

��

��

��

� �

� �

� �

��

�� �

��

��

��

��

� �

��

��

��

��

��

��

� �

� �

��

��

��

� �

� �

� �

��

���

��

��

��

��

� �

��

��

��

��

��

��

� �

� �

��

��

��

� �

� �

� �

� �

���

��

��

��

��

��

��

��

��

��

��

��

��

��

� V5 0.79 0.16 0.21 0.089 0.46 0.068 0.67 0.087 0.48 0.056 0.48

015

0.22

1020 �

� �

���

��

��

��

��

�� �

��

��

��

� �

� �

��

��

��

� �

��

��

���

��

��

� �

��

��

��

��

���

��

��

��

�� �

��

��

��

��

�� �

��

��

��

� �

��

��

��

��

� �

��

��

�� �

��

��

��

��

��

��

��

��

��

��

� �

���

��

��

��

��

�� �

��

��

��

� �

��

��

��

��

� �

��

��

���

��

��

� �

��

��

��

��

��

��

��

� �

���

��

��

��

��

���

��

� �

��

��

��

��

��

��

��

��

��

���

��

��

��

��

��

��

��

��

��

��

� �

�� �

��

��

��

��

���

��

��

��

� �

��

��

� �

��

� �

��

��

�� �

��

� �

� �

��

��

� �

��

��

��

��

V6 0.29 0.23 0.20 0.34 0.17 0.92 0.17 0.40 0.12 0.42 0.32��

��

��

��

��

��

��

��

��

�� ���

��

���

��

�� �

��

� ��

��

��

��

��

��

��

��

��

��

�� ���

��

� ��

��

�� �

��

� ��

��

��

��

��

��

� �

��

��

��

�� ���

��

���

� �

�� �

��

� ��

��

��

��

� �

��

� �

��

��

��

�� ���

��

���

��

�� �

��

� ��

��

��

��

� �

��

� �

��

��

��

�� ���

��

� ��

��

�� �

� �

���

��

��

��

� �

��

� �

��

��

��

�� ���

��

���

� �

�� �

� �

� ��

V7 0.19 0.79 0.27 0.78 0.48 0.43 0.064 0.17 0.16

1800.17

40

��

��

��

��

��

��

���

��

��

��

��

��

� �

��

��

��

��

� �

��

��

��

���

��

��

��

��

��

��

��

� �

��

��

� �

��

��

��

���

��

��

��

��

��

� �

��

��

��

��

��

��

��

��

� ��

��

��

��

��

��

� �

��

� �

��

��

� �

��

��

��

� ��

��

��

��

��

��

��

��

��

��

���

��

��

��

��

� ���

��

��

��

��

��

��

��

� �

��

��

��

��

��

����

��

��

��

��

��

��

��

��

�� V8 0.46 0.054 0.42 0.23 0.043 0.21 0.23 0.18 0.042

��

��

��

��

��

��

� �

��

��

��

� �

��

���

��

��

��

��

��

� ��

��

��

��

��

��

� �

��

��

��

��

��

��

���

��

��

��

��

��

���

��

��

��

��

��

��

��

��

��

��

� �

��

���

��

��

��

��

��

���

��

��

��

��

��

��

� �

��

��

��

��

��

���

��

��

��

��

��

�� �

��

��

� �

��

��

��

� �

��

��

��

� �

� �

���

��

��

��

��

��

� � �

��

��

��

��

��

��

� �

��

��

��

� �

��

���

��

��

��

��

��

� � �

��

��

��

��

��

��

� �

��

��

��

� �

� �

� ��

��

��

��

��

��

���

��

��

��

��

��

��

��

��

��

��

� �

��

�� �

��

��

��

��

��

� ��

V9 0.078 0.97 0.38 0.30 0.06 0.17 0.078

1200.16

1545

��

��

���

� �

��

��

��

� ��

�� �

��

��

��

��

��

��

��

��

��

��

��

��

� ��

� �

��

��

��

���

� ��

��

��

��

��

��

��

��

��

��

��

��

��

� ��

� �

��

��

��

� ��

�� �

��

��

��

��

��

��

��

��

��

� �

��

��

� ��

� �

��

��

��

� ��

� ��

��

� �

��

��

��

��

��

��

��

� �

��

� �

� ��

��

� �

��

��

���

�� �

��

� �

� �

��

��

��

��

��

��

� �

��

��

� ��

� �

��

��

��

� ��

�� �

��

� �

��

��

��

��

��

���

��

� �

��

��

� ��

��

� �

��

��

� ��

� � �

��

� �

��

��

��

��

��

��

��

��

��

��

� ��

� �

� �

��

��

���

�� �

��

� �

� �

��

��

��

��

��

��

��

��

��

� ��

� �

� �

��

��

� ��

� ��

��

� �

��

��

��

��

��

��

��

��

V10 0.019 0.27 0.016 0.33 0.011 0.35 0.13��

��

��

���

��

��

� ��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

� ��

��

��

� ��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

���

��

��

� ��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

� ��

��

��

� � �

��

��

��

��

��

���

��

��

��

��

��

��

��

��

��

��

� �

� ��

��

��

� ��

��

��

��

� �

��

��

��

��

��

��

��

��

��

��

��

��

��

���

��

��

�� �

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

���

��

��

� ��

��

��

��

� �

��

��

��

��

��

� �

��

��

��

��

��

��

��

���

��

��

���

��

��

��

��

��

��

��

��

��

��

��

��

��

��

� �

��

��

� ��

��

��

�� �

��

��

��

��

��

��

��

��

��

� �

���

��

��

��

� �

��

��

���

��

��

� ��

��

��

��

� �

��

��

��

��

��

� �

��

��

��

��

V11 0.37 0.26 0.084 0.13 0.08

0.0

1.0

0.091

0 �

��

�� �

��

���

��

���

� �

��

��

���

��

��

��

� ��

��

��

��

� �

��

� �

��

��

��

���� �

��

��

��

��

��

��

��

���

��

� ��

��

���� �

��

��

� � �

��

��

��

�� �

��

��

��

� �

��

� �

� �

��

��

�����

� �

��

��

��

��

� �

��

�� ��

��

���

��

���� �

��

��

���

��

��

��

� ��

��

��

��

� �

��

� �

��

��

��

���� �

��

��

��

��

��

��

��

�� �

���

���

��

���

��

��

��

� ��

��

� �

��

�� �

��

����

��

��

��

��

��

��

���� �

� �

��

��

��

��

� �

��

�� �

��

� ��

��

���

� �

��

��

� ��

��

� �

��

� ��

��

��

��

� �

��

��

� �

��

��

�� ���

� �

��

���

��

��

��

���

��

���

��

���

� �

��

��

���

��

� �

��

���

��

��

��

��

��

��

��

��

��

��� ��

� �

��

��

��

��

� �

��

�� �

��

� ��

��

� ��

��

��

��

� � �

��

��

��

�� �

��

��

��

��

��

��

��

��

��

�� ���

� �

��

��

��

��

��

��

���

��

� ��

��

� ��

� �

��

��

� � �

��

� �

��

���

��

��

��

��

��

� �

� �

��

���

�� ���

� �

��

��

��

��

��

��

�� �

��

� ��

��

� ����

��

��

� � �

��

� �

��

�� �

��

��

��

��

��

��

� �

��

��

�� ���

��

��

��

��

��

��

��

�� �

��

� ��

��

���

��

��

��

� ��

��

��

��

� ��

��

��

��

��

��

��

��

��

��

��� ��

��

��

��

��

��

��

��

�� �

��

� ��

��

�����

���

� � �

��

� �

��

� � �

��

��

��

��

��

��

� �

��

��

�� ���

��

��

��

��

��

��

V12 0.22 0.45 0.21 0.47 0.34�� ��� �

� � ��� � � ��

�� �

��� �

���� �

��

��

���

��

��

� �� � ��

��

���

��

��

�� �� � ��� �

� �

��

� � �� �� � �� �

� �� ��� ���

�� �

��

� ��

�� �� �

� �

��

� � �

� �

��

�� �� ��

��

� � �

� �

��

�� ��� ����

��

��

� �� � �� � ���

� � ���� ���

�� �

��

� ��

���� �

��

��

���

� �

��

� �� � � �

��

� ��

��

��

�� ��� ��� �

��

��

� � �� �� � ���

� � ��� � ���

�� �

��� �

�����

��

��

� ��

��

� �

�� ��� �

��

� ��

��

��

���� ���� �

��

��

�� � � �� � �� �

�� �� � � ���

�� �

��

���

���� �

� �

��

� ��

� �

��

� �� � � �

��

� � �

� �

��

� �� ��� ���

��

��

���� �� � �� �

� �� ��� ���

�� �

��� �

���� �

��

��

���

� �

� �

��� �� �

��

� ��

��

��

��� ���� ��

��

��

�� � � ���� � �

�� � �� �� ��

���

��

���

� � ���

� �

��

� � �

��

� �

�� � ���

��

� ��

� �

��

� ��� �� ���

��

��

� � �� �� �� ��

��� ���� ��

�� �

��

� ��

� ��� �

��

��

� � �

� �

��

��� � ��

��

�� �

� �

��

� � � ��� ���

��

��

� ��� � ��� � �

�� � �� �� ��

� ��

��

� ��

� � ���

� �

��

� � �

��

� �

�� �� ��

��

� � �

� �

��

� �� ��� ���

��

��

� ��� � ���� �

�� ��� �� �����

��

���

�� ���

� �

��

� ��

� �

��

� ���� �

��

� ��

��

��

�� � � � �� ��

� �

��

� � �� ���� ��

�� � �� �� ��

� ��

��

� ��

�� ���

� �

��

� � �

��

� �

� � �� ��

��

� � �

� �

��

� �� ��� ���

��

��

� ��� � ��� ��

���� ��� ��

� ��

����

�� ���

� �

��

���

��

��

��� �� �

��

���

� �

��

�� � ������

� �

��

� ���

V13 0.36 0.12 0.12

1400.27

1545

� �

��

��

��

���

� �

����

�� �

����� � �

��

��

��

��

� ��

� �

��

��

���

���

��

� ��

��

�� �

� ���

��

��

��

��

� �

��

��

��

���

��

� �� �

�� �

� ��� �� �

��

��

��

��

� ��

� �

��

��

���

���

��

� ��

��

���

� ���

��

��

��

��

� �

��

��

��

���

��

����

�� �

������ �

��

��

��

��

� ��

� �

��

��

���

���

��

� ��

��

�� �

� ���

��

��

��

��

� �

��

��

��

���

��

� ���

�� �

����� ��

��

��

��

��

� ��

� �

��

��

�� �

���

��

���

��

���

� ���

��

��

��

��

� �

��

��

��

�� �

��

� �� �

� ��

�� ��� � �

��

��

��

��

���

��

��

� �

�� �

���

��

�� �

��

�� �

����

��

��

��

��

� �

��

��

���

��

��

����

�� �

����� � �

��

��

��

���

� ��

� �

��

��

�� �

���

��

���

��

���

� ���

��

��

��

��

��

��

��

��

���

��

��� �

�� �

�� � � � ��

��

��

��

��

���

��

��

��

���

���

��

�� �

��

�� �

��� �

��

��

��

��

� �

��

��

��

���

��

��� �

�� �

�� � ��� �

��

��

��

���

���

��

��

� �

���

� ��

��

� ��

��

�� �

��� �

��

��

��

��

��

��

��

��

���

��

� �� �

�� �

�� � � ���

��

��

��

��

���

��

��

��

���

���

��

�� �

��

�� �

��� �

��

��

��

��

��

��

��

��

���

��

��� �

�� �

���� � ��

��

��

��

��

���

��

��

��

���

���

��

���

��

�� �

� �� �

��

��

��

��

��

��

��

��

���

��

��� �

�� �

�� �� � ��

���

��

��

���

��

��

��

���

���

��

�� �

��

�� �

��� �

��

��

��

��

��

��

��

���

� �

� �

����

� ��

�� �����

��

��

��

��

� ���

��

��

� �

���

� ��

� �

�� �

��

���

��� �

��

��

��

��

��

��

��

�����

��

��� �

���

� � �����

��

��

��

��

� ��

� �

��

� �

���

� ��

��

� � �

��

���

� ���

��

��

��

��

V14 0.35 0.90 0.04

���

��

��

� �

���

��

��

��

��

��

��

� �

� �

��

��

��

��

��

� �

��

��

��

��

���

��

��

� �

�� �

��

��

��

��

��

��

� �

� �

��

��

� �

��

��

��

��

��

��

��

���

��

��

� �

���

��

��

��

��

��

��

� �

��

��

��

��

��

� �

� �

��

��

��

��

���

��

��

� �

���

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

� �

��

�� �

��

��

��

���

��

��

��

��

��

��

� �

��

��

��

� �

��

��

� �

��

��

� �

��

���

��

��

� �

����

��

��

��

��

��

��

��

��

��

��

��

� �

��

��

��

� �

��

�� �

��

��

� �

�� �

��

��

��

��

��

��

��

� �

��

��

��

��

��

� �

��

��

��

��

�� �

��

��

� �

���

��

��

��

��

��

��

� �

��

��

��

� �

��

��

� �

��

��

��

��

�� �

��

��

� �

�� �

��

��

��

��

��

��

� �

� �

��

��

� �

��

��

� �

��

��

��

��

���

��

��

� �

�� �

��

��

��

��

��

��

��

� �

��

��

��

��

� �

� �

��

��

��

��

�� �

��

��

� �

�� �

��

��

��

��

��

��

� �

� �

��

��

� �

��

��

� �

��

��

��

��

���

��

��

��

�� ���

��

��

��

��

��

��

� �

��

��

��

��

� �

��

��

��

��

��

���

��

��

��

���

��

��

��

��

��

��

��

� �

��

��

��

��

� �

��

��

��

� �

��

���

��

��

��

�����

��

��

��

��

��

� �

� �

��

��

� �

��

� �

��

��

��

� �

��

V15 0.51

20

0.059

0.10

��

� �

��

��

��

� �

��

��

��

� �

� ���

��

��

� ��

��

��

��

��

���

���

��

��

��

� �

��

��

��

� �

��

��

��

��

� �

��

��

��

� � ��

��

��

���

��

��

��

��

���

���

��

��

��

��

� �

� �

��

��

��

��

����

��

��

��

��

� ���

��

��

���

��

��

��

��

���

���

��

��

��

� �

��

��

��

��

��

��

��

��

� �

��

��

� �

� ���

��

��

���

��

��

��

� �

���

����

��

��

��

� �

� �

� �

��

� �

� �

��

����

� �

��

� �

� �

�� ��

��

��

���

��

��

� �

� �

���

�� �

��

��

��

��

� �

��

��

� �

��

��

��

��

��

��

��

� �

� ���

���

��

���

��

��

��

� �

���

���

��

��

��

��

� �

� �

��

� �

� �

��

��

��

��

��

� �

� �

�� ��

��

��

� ��

��

��

� �

��

���

�� ��

��

��

��

��

� �

��

��

��

� �

��

��

��

��

��

� �

��

� �� �

���

��

���

��

��

��

��

� ��

���

��

��

��

��

� �

��

��

� �

� �

��

��

��

� �

��

� �

��

�� � �

��

��

���

���

��

��

��

���

�� �

��

��

��

��

��

��

��

� �

��

��

��

��

��

��

��

� �

�� ��

��

��

� ��

���

��

� �

��

���

���

��

��

��

��

��

��

��

��

� �

��

��

��

��

��

� �

� �

�� � �

��

��

���

��

��

��

��

���

�� �

��

��

��

��

��

��

��

��

��

��

��

� �

��

��

� �

��

�� � �

��

��

� ��

��

��

� �

��

� ��

�� �

��

��

��

��

��

��

��

��

��

��

����

��

��

� �

��

� ���

��

��

���

���

��

� �

��

� ��

�� �

��

��

��

��

��

��

��

� �

��

��

��

��

��

��

��

��

� ���

���

��

���

��

��

� �

��

���

���

��

��

��

� �

� �

��

��

� �

��

��

��

� �

��

��

� �

��

� ���

��

��

���

��

��

��

��

� ��

�� �

��

��

��

� �

��

� �

V16 0.16

���� � �

�� � ��

��

� �

�� �

��

��

��

��

�� �

�� �

��

��

��

� ��� �� �

��

���

� ��� � �� � ���

��

� � �� �

� �

� �

��

� ���

� �� ����

��

�� �� � �

�� �� �

��

��

�� �

��

��

��

��

� � �

�� �

��

� �

��

� ����� �

��

���

� ��� � ��� ����

��

�� ���

� �

� �

��

� � ��

� � �� ���

��

�� ��� �

�� � ��

��

��

�� �

��

��

��

��

� � �

�� �

��

��

��

� � ���� �

��

���

� ��� � ��� ���

���

�� �� �

� �

� �

��

� ���

� �� � � ��

��

�� ��� �

�� � ��

��

� �

�� �

��

��

��

��

���

�� �

��

��

��

�� ���� �

��

���

���� � ������

��

�� � ��

��

��

��

� � ��

� � ��� � �

��

�� �� ��

��� ��

��

��

� ��

� �

��

��

��

� ��

���

���

��

��

� � ��� ��

� �

���

� ��� � ���� ��

��

�� �� �

��

��

��

�� ��

�� �� � ��

��

�� �� � ��

� �� ��

��

�� �

��

��

��

��

� ��

�� �

���

� �

��

�� ���� �

��

���

���� � ������

���

�� � ��

��

��

��

� � ��

� � ��� � �

��

��� � ���

�� � ��

� �

�� �

� �

��

��

��

���

���

��

� �

��

���� � ��

��

���

�� ��� �� �� ��

���

� ��� �

��

��

��

�� � �

�� �����

��

��� �� �

���� �

��

��

�� �

� �

��

��

��

� � �

���

��

� �

��

� ��� ���

� �

� ��

��� �� ��� ����

��

� �� � �

� �

��

��

�� � �

�� �� �� �

��

��� � ��

��� � �

��

��

�� �

� �

��

��

��

���

���

��

� �

��

� ��� ���

��

� ��

����� �� �� ��

��

� ��� �

��

��

��

��� �

�� �� �� �

��

���� ��

��� ��

��

� �

�� �

��

��

��

��

� � �

���

��

��

��

�� �� � ��

��

� ��

���� � ������

��

� � �� �

� �

� �

��

� �� �

�� �����

��

��� ���

��� � �

��

��

�� �

� �

��

���

��

���

���

��

� �

��

� ��� ���

��

� ��

����� �� �� ��

��

� � �� �

��

��

��

��� �

�� �� ���

��

��� ���

�����

��

� �

� ��

� �

���

��

��

�� �

�� �

��

��

��

�� ��� ��

� �

� ��

�� � ��� � �� ��

��

� ����

� �

��

��

��� �

� � � � �� �

��

������

�������

� �

���

� �

��

��

��

�� �

�� �

��

� �

��

���� � ��

� �

� ��

�� � �� ��� � ��

��

�� ���

��

��

��

� ���

����� ��

��

�� � � ��

��� � �

��

� �

���

��

���

��

��

� � �

���

��

� �

��

� ���� � �

��

���

�� ��� ������

��

� ����

� �

� �

��

� � ��

� � � �� � �

��

���� � �

�� �� �

��

��

� ��

� �

��

��

��

���

�� �

��

� �

��

�� �� ���

� �

� ��

� � � ��� ��� ��

���

� �� ��

� �

��

��

� ���

� ��� � � �

��

�� � �� �

��� � �

��

��

�� �

��

��

��

��

� � �

�� �

��

� �

��

� ����� �

��

���

�� �� � ��� ���

��

� �� ��

� �

� �

��

� �� �

� � � �� � �

��

0.0

0.7

V17

(D)

Fig. S4. Exploratory data analysis for the 18 predictors of sero-prevalence rate in cattle. (A–C) PCA (A) eigenvalues, (B) correlation circle for the variables,and (C) hierarchical clustering on the PCA scores. (D) Matrix of scatter plots: variable names on the diagonal: V1, mxDlst; V2, mnDlst; V3, avDlst; V4, mxNlst;V5, mnNlst; V6, avNlst; V7, mxNDVI; V8, mnNDVI; V9, avNDVI; V10, sdNDVI; V11, propNDVI; V12, Alt; V13, mxPLUVIO; V14, avPLUVIO; V15, sdPLUVIO; V16,propPLUVIO; V17, RapSurfWB. Below the diagonal, red curves are loess smoothing lines. Above the diagonal, numbers are the linear correlation coefficients,in red for ρ > 0.80.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 7 of 11

Page 14: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Temperature (°C)

logi

t(p)

−2.0

−1.5

−1.0

22 24 26 28 30 32 34

(A)

NDVI

logi

t(p)

−2.0

−1.5

−1.0

−0.5

0.0

0.5

0.4 0.5 0.6 0.7

(B)

Rainfall (mm)

logi

t(p)

−2.0

−1.5

−1.0

−0.5

20 30 40

(C)

Altitude (m)lo

git(p

)

−2.5

−2.0

−1.5

−1.0

−0.5

0.0

0 500 1000 1500

(D)

Ratio

logi

t(p)

−2

−1

0

1

2

0.0 0.2 0.4 0.6 0.8

(E)

Fig. S5. Trend in anti-RVFV IgG sero-prevalence rate in cattle (logit scale) as a smoothing-spline function of the predictor. (A) Average DLST. (B) AverageNDVI. (C) Average monthly rainfall. (D) Altitude. (E) Square root of the surface water ratio. Observed data are indicated by small tick marks on the x axis; thesolid and dashed lines show the trend and its 95% CI.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 8 of 11

Page 15: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Fig. S6. (A) Monthly activity (number of links) in the cattle trade network, Madagascar. Each colored line on the graph corresponds to a link. The color ofthe line corresponds to the frequency of movements on that particular link during the 5 y of the survey. (B, Top) Number of links for the monthly networks.(B, Bottom) The corresponding epidemic threshold. The color corresponds to the type of network considered: “Backbone link” is for the backbone structurealone; “All links” also includes the occasional links.

Local index

logi

t(p)

−2.5

−2.0

−1.5

−1.0

−0.5

0.0

0.5

0 5 10

(A)

Remote index0 1 2 3

(B)

Fig. S7. Trend in anti-RVFV IgM sero-prevalence rate in humans (logit scale) as a smoothing-spline function of the RVF risk index. (A) Local, environment-related index. (B) Remote, cattle trade-related index. The y scale is common to A and B. Observed data are indicated by small tick marks on the x axis; thesolid and dashed lines show the trend and its 95% CI.

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 9 of 11

Page 16: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Fig. S8. Random variations around the population means predicted by the mixed-effect logistic regression model of sero-prevalence rate in humans. (A)Estimates and 95% CI for the 20 highest random variations. (B) Geographical variations. Named districts are those with a random variation significantly >0(α = 0.05).

Table S1. Binomial logistic regression models of anti-RVFV IgGsero-prevalence rate in cattle, Madagascar (n = 4,427 sera)selected according to AICc

Variables df* log(likelihood) AICc ∆AICc Weight

1,3,4,6 7 −315.9 646.7 0.00 0.2521,2,3,4,6 8 −315.0 647.2 0.49 0.1981,2,3,4,5,6 10 −313.1 648.1 1.41 0.1251,3,4 6 −318.0 648.7 1.96 0.0951,2,3,4,5 9 −314.7 649.0 2.26 0.0811,2,3,4 7 −317.0 649.0 2.28 0.0813,4,6 5 −319.7 649.9 3.16 0.0522,3,4,6 6 −319.0 650.6 3.93 0.0351,3,4,5,6 9 −315.6 650.7 4.02 0.0341,3,6 6 −319.4 651.4 4.72 0.0241,3,4,5 8 −317.6 652.4 5.64 0.0151,2,3,6 7 −319.3 653.6 6.91 0.008

Labels of variables: 1, average altitude; 2, 10-d average NDVI; 3, 10-daverage rainfall; 4, municipality within 50 km of the nearest sea port; 5,10 d average DLST; 6, municipality-level water/surface ratio.*Degrees of freedom.

Table S2. Importance of predictors in the averaged beta-binomiallogistic regression model of anti-RVFV IgG sero-prevalence ratein ruminantsPredictor Importance No. of models

Rainfall 1.000 12Within 50 km of a seaport 0.968 10Altitude 0.913 10Water surface ratio 0.728 8NDVI 0.529 6DLST 0.255 4

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 10 of 11

Page 17: Drivers of Rift Valley fever epidemics in Madagascar · Drivers of Rift Valley fever epidemics in Madagascar ... France;´ dCentre de Recherche et de Veille sur les Maladies Emergentes

Table S3. Coefficient of the averaged beta-binomial logisticregression model, selected according to AICc, to explain theRVFV IgG sero-prevalence rate in cattle, Madagascar

Term Estimate SE z value P(> |z|)

Intercept −1.474 1.095 −1.347 0.1780Alti2 −0.719 0.258 −2.782 0.0054Alti3 −0.885 0.359 −2.466 0.0137Rainfall 0.058 0.012 5.040 0.0000Buffer 50 km −0.858 0.306 −2.802 0.0051Water surface ratio 1.219 0.601 2.028 0.0425NDVI −1.963 1.386 −1.416 0.1566Daytime temp2 −0.306 0.249 −1.230 0.2186Daytime temp3 −0.557 0.352 −1.581 0.1138

Altitude (Alti) and DLST were split into three equal-size categoriesdefined by their 1/3 quantiles numbered 1 (reference category), 2, and 3.The municipality-level water/surface ratio was transformed into its squareroot. Municipalities located within 50 km from a seaport were coded as 0(reference category); others were coded as 1.

Table S4. Fixed-effect coefficients for the beta-binomial logisticregression model of sero-prevalence rate of anti-RVFV IgM inhumans (n = 1,975 sera)

Term Estimate SE z value P (> |z|)

Intercept −2.351 0.186 −12.608 < 10−4

Local risk index 0.017 0.039 0.425 0.671Cattle trade-related index 0.540 0.229 2.354 0.019

Lancelot et al. www.pnas.org/cgi/content/short/1607948114 11 of 11