Rding to their putative influence on mosquito vector populations. Our MFA classification is in accordance with the known Malagasy ecosystems [18]. The risk of transmission and respective roles of direct and vectorial transmission are probably different among eco-climatic areas. In the case of direct transmission the force of infection is expected to depend on the number of potentially infectious contacts that a susceptible individual experiences over a time unit. This contact rate is expected to depend, among others, on cattle density. A positive association between cattle density and IgG seroprevalence rate in cattle and/or humans would thus suggest a direct transmission of RVFV. In the case of direct cattle to human transmission, the force of infection in the human population is also expected to depend on the frequency of human behaviors resulting in exposure to ruminant fluids or products. By contrast, in the case of vectorial transmission, due to the so-called “dilution effect” and for a fixed vector density, increased cattle density would decrease the probability for a susceptible individual to be bitten by an infectious vector over a time unit [50,51]. Therefore, a negative association between cattle density and IgG seroprevalence rates in cattle and/or in humans would rather suggest a vectorial transmission. The force of infection is also expected to increase with vector density. In Madagascar the density of vectors mostly depends on climatic and landscape factors [1,25,28]. High cattle densities are not systematically associated with high vector densities, as the main RVF vectors breed rather in large water bodies [25] than in artificial containers created from livestock-related activities. Thus, a positive association between cattle and/ or human seroprevalence, local environmental and climatic conditions favorable to mosquitoes is expected under the I-BRD9 web hypothesis of vectorial transmission. According to our analysis, cattle seroprevalence increased with age suggesting an enzootic circulation. Cattle seropositivity was positively associated to humid environment (large surface of permanent wetlands, marshlands and irrigated lands) each of these factors being favorable to Culex and Anopheles mosquitoes [25]. Actually, during the 2008?9 epidemics, 3 mosquito species were found to be naturally infected by RVFV: Anopheles coustani, An. squamosus and Culex antennatus [52]. Cx. antennatus is considered a RVFV vector and both Anopheles species as candidate vectors [25]. Cattle seroprevalence was also positively associated with cattle density suggesting the existence of a direct transmission between cattle, as suggested by order NVP-QAW039 Nicolas et al [3,19]. However, in our study cattle density and environmental factors were not independent (correlation with Factor 1, Factor 2 and Factor 3). Because of such associations it was impossible to disentangle the influence of cattle density from the influence of environmental conditions and thus to thoroughly assess the relative importance of vectorial and direct transmission. The prediction map of cattle seroprevalence highlighted the eastern-coast, western and north-western parts as high-risk areas. Surprisingly, some districts affected by RVFV outbreaks are located in the predicted low risk area [13?5]. The last outbreaks were mostly reported in the highlands, which are highly connected by road to the capital city, Antananarivo. Outbreaks occurring in isolated areas may have been missed explaining why a low.Rding to their putative influence on mosquito vector populations. Our MFA classification is in accordance with the known Malagasy ecosystems [18]. The risk of transmission and respective roles of direct and vectorial transmission are probably different among eco-climatic areas. In the case of direct transmission the force of infection is expected to depend on the number of potentially infectious contacts that a susceptible individual experiences over a time unit. This contact rate is expected to depend, among others, on cattle density. A positive association between cattle density and IgG seroprevalence rate in cattle and/or humans would thus suggest a direct transmission of RVFV. In the case of direct cattle to human transmission, the force of infection in the human population is also expected to depend on the frequency of human behaviors resulting in exposure to ruminant fluids or products. By contrast, in the case of vectorial transmission, due to the so-called “dilution effect” and for a fixed vector density, increased cattle density would decrease the probability for a susceptible individual to be bitten by an infectious vector over a time unit [50,51]. Therefore, a negative association between cattle density and IgG seroprevalence rates in cattle and/or in humans would rather suggest a vectorial transmission. The force of infection is also expected to increase with vector density. In Madagascar the density of vectors mostly depends on climatic and landscape factors [1,25,28]. High cattle densities are not systematically associated with high vector densities, as the main RVF vectors breed rather in large water bodies [25] than in artificial containers created from livestock-related activities. Thus, a positive association between cattle and/ or human seroprevalence, local environmental and climatic conditions favorable to mosquitoes is expected under the hypothesis of vectorial transmission. According to our analysis, cattle seroprevalence increased with age suggesting an enzootic circulation. Cattle seropositivity was positively associated to humid environment (large surface of permanent wetlands, marshlands and irrigated lands) each of these factors being favorable to Culex and Anopheles mosquitoes [25]. Actually, during the 2008?9 epidemics, 3 mosquito species were found to be naturally infected by RVFV: Anopheles coustani, An. squamosus and Culex antennatus [52]. Cx. antennatus is considered a RVFV vector and both Anopheles species as candidate vectors [25]. Cattle seroprevalence was also positively associated with cattle density suggesting the existence of a direct transmission between cattle, as suggested by Nicolas et al [3,19]. However, in our study cattle density and environmental factors were not independent (correlation with Factor 1, Factor 2 and Factor 3). Because of such associations it was impossible to disentangle the influence of cattle density from the influence of environmental conditions and thus to thoroughly assess the relative importance of vectorial and direct transmission. The prediction map of cattle seroprevalence highlighted the eastern-coast, western and north-western parts as high-risk areas. Surprisingly, some districts affected by RVFV outbreaks are located in the predicted low risk area [13?5]. The last outbreaks were mostly reported in the highlands, which are highly connected by road to the capital city, Antananarivo. Outbreaks occurring in isolated areas may have been missed explaining why a low.