The incidence of zoonotic diseases, transmitted to humans from wild or domestic animals, has noticeably increased during the past few decades and currently represents 70% or more of emerging diseases [1]. Japanese encephalitis virus (JEV), an arbovirus of the Flavivirus genus, family Flaviviridae, is transmitted by mosquitoes from animals to humans. In humans, this zoonotic disease is the largest worldwide cause of epidemic viral encephalitis [2], [3]. The pathogenesis of the JEV and the clinical manifestations of the disease, including severe neurological syndromes, depend on several factors that have been deeply reviewed elsewhere [4], [5]. The incubation period ranges from 5 to 15 days; JE infections are lethal in about 25–30% of cases, mostly in infants, and lead to permanent sequelae in about 50% of cases.

JE was first described in 1871 in Japan, and first characterized in 1935 [6]. Despite an effective vaccine developed in 1941, and the subsequent national immunization campaigns that have greatly reduced the incidence of JE in several countries, sporadic cases continue to be reported, and JEV continues to spread widely in South, East, and Southeast Asia and Australasia [7]. Currently, more than three billion people live in JE-endemic countries [5]. There, though people of all ages may be exposed to JEV, the JE incidence is higher in children because most adults are immune [8]. Now, about 68,000 JE cases are estimated to occur annually, causing at least 10,000–15,000 deaths in more than 20 Australasian countries [8], [9].

This review aims at offering an overview of the factors affecting JEV epidemiology. It discusses the current state of knowledge on JEV diversity, molecular epidemiology, and ecology, and examines how environmental variables and modifications, especially those associated with agriculture, affect the landscape characteristics and JEV ecology. The literature review was achieved by using both general web search engines and scientific web search engines such as PubMed, Springerlink, ScienceDirect, and Web of Science. It focused on the literature available on virology, molecular and spatial epidemiology, entomology, and landscape and behavioral ecology.

Five JEV genotypes have been distinguished by nucleotide sequencing of the C/PrM and E (capsid, precursor membrane, and envelope) genes [10]. Genotypes I, II, and III are distributed throughout Asia, while genotype IV is restricted to Eastern Indonesia [11]. Genotype V was long restricted to Malaysia but isolates were recently found in China and in the Republic of Korea [12], [13]. Genotypes IV and V form the oldest JEV lineage, which appears to have originated from an ancestral virus in the Indonesian-Malaysian region. JEV therefore probably originally spread from this region [4], [11]. Genotypes I, II, and III are the most prevalent, having accounted for 98% of the strains isolated from 1935 to 2009 [14].

Until recently, genotypes I and III were mostly associated with epidemic diseases in temperate regions of Asia, while genotypes II and IV were associated with endemic diseases in tropical regions. It was therefore postulated that the JEV epidemiology differed with genotypes [15]. However, more recent observations show that genotypes can be found indifferently within epidemic or endemic areas. Indeed, genotype I has been isolated in Australasia within the tropical region [16], genotype II has been shown to have circulated in the Republic of Korea within the temperate region before the 1970s [17], [18], and genotype III, the only one in India so far, has circulated both in the south Indian peninsula (characterized by endemic activity) and in the north (characterized by epidemic activity) [7], [19].

Differences in the JEV genome might explain phenotypic variations of the disease. Experimental changes in the E gene have indeed been shown to be associated with loss or gain of neuroinvasiveness and neurovirulence in mice and monkeys [20]–[23]. Even so, experimental studies of the neurovirulence of wild JEV strains in mice showed no clear-cut genotype-phenotype relationship [24], and there is still no firm evidence that the JEV genotypes circulating differ in their virulence (Figure 1). However, the dengue virus type 2, another closely related Flavivirus, has been shown to possess interesting genotype-dependent characteristics such as human viral pathogenicity or midgut viral replication in Aedes aegypti [25], [26]. The JEV molecular markers previously and currently studied might not be sufficient and need further investigation.

Rather than by viral genetic determinants solely, the JEV epidemiological pattern appears to be actually influenced by a complex set of variables, including several environmental, ecological, and immunological factors [7] (Figure 1).

The distribution area of JEV has consistently enlarged. Sometimes, the genotypes of JEV isolates changed locally. In northern Vietnam and Thailand, a shift from genotype II to genotype I was reported [27], [28]. In northern temperate regions (i.e., Japan, South Korea, and Northeast China), genotype I progressively replaced genotype III and became the main genotype [29]. It is now very probably the most widespread genotype in Asia, which was otherwise found to be lethal for humans [9], [29]–[33]. However, to date, vaccines are only derived from genotype III strains; whereas protective levels of cross-reactive neutralizing antibodies of these vaccines were found against the various circulating genotypes, variations between genotypes call for further studies [34], [35]. In particular, since the immune response against genotype I was less pronounced, its duration should be addressed [34].

The (re)emergence of vector-borne and zoonotic diseases as well as variations in disease risk and incidence are strongly driven by environmental factors. Among the most relevant factors are climate but also land cover and land use, respectively the biophysical attributes of the earth's surface and the human purpose or intent applied to these attributes [68]. However, these variables occur at several different geographic ranges, and hence multi-scale studies are a major stake for the understanding of the spatial epidemiology of diseases [69].

For a long time, JEV epidemiology has exclusively been broached at the global scale and primarily on the basis of climatic variables (i.e., rainfall and temperature) because they indeed strongly affect vector density [67]. Recently, Miller et al. [70] identified an optimal range of temperature during the wet season that is favorable to the Cx. tritaeniorhyncus biology. As they also found that most of JE cases were located in areas of high probability of vectors, they underlined the link between climatic covariables, vector ecology, and human health. Moreover, temperature might affect the competence of a vector-genotype couple over another, as shown for the West Nile virus [71].

In our previous study [14], six Asian and Australasian regions were identified according to anthropogenic, biological, geographic, and physical factors including not only climatic conditions, but also biomes and land use. The regional history of JEV genotypes was analyzed (i.e., the number of JEV isolates and the proportion of each genotype). Then we emphasized the two regions of intensive JEV circulation: the Paleartic biogeographic realm [72] comprising Eastern Russia, Japan, Korea, and Northern China, and a tropical region characterized by large cover of rice fields comprising the Indochinese peninsula (Cambodia, Lao PDR, Myanmar, Thailand, and Vietnam), South China, and Taiwan [73].

Actually, land cover and land use changes have been driving the JE risk and incidence. In Asia, paddy field surfaces have constantly extended since the early 1960s (http://faostat3.fao.org/). Given paddy fields provide long-term Culex sp. breeding sites and attract many wading birds for foraging and resting, they enhance the circulation and expansion of mosquito and wading bird populations [74]. Thus, in areas where rice-irrigated farming is widespread, the JEV transmission might become less dependent on rainfall [75], [76]. Likewise, because the amount of live pig heads has increased since the 1960s (http://faostat3.fao.org/), both backyard and industrial pig farming have provided a continuous increasing, outstanding potential source of blood meals for mosquitoes.

Additionally, a finer-scale spatial analysis (i.e., local or regional landscape) helps to better picture how landscape structure affect the JEV epidemiology. At fine scale, JEV ecology is indeed largely affected by both land cover and land use as these factors affect vectors' and hosts' ecology (Figure 1). Once the JEV is introduced, its ecological sustainability indeed relies on its growth within the susceptible host population and on its transmission between hosts through vectors. Then the risk of incidental infection to humans depends on the likelihood of disease transmission within the surrounding landscape [77]. To be transmitted to humans, JEV needs: 1) competent mosquito vectors, 2) reservoir-amplifying hosts, and 3) nonimmune humans within the range of the JEV-competent vector and animal hosts. In this process, the three major landscape elements are paddy fields, pig farms, and human habitations. In the landscape, they are distributed in habitat patches. Their density, size, and spatial arrangement are key factors determining the degree to which the landscape facilitates or impedes movement of vectors among them (i.e., the landscape connectivity) [78] (Figure 1). Within a given landscape, this biological connectivity may strongly affect the JEV transmission dynamics.

Vaccination is the most effective way to reduce the incidence of JE in humans, yet it has no effect on the JEV transmission cycle. Vaccination of livestock, especially pigs, would in contrast reduce the amplification of the virus, the rate of mosquito infection, and subsequently the risk of transmission to humans. Yet, vaccination of pigs is generally not used to prevent JE because it is costly, hardly feasible logistically, and not necessarily effective in piglets (they must be immunized after the disappearance of maternal antibodies) [6], [79]. Moreover, pigs represent a relevant sentinel model, the surveillance of which could predict a potential JE outbreak in a human population nearby [54]. Immunizing sentinel pigs would also impede the detection of such a threat. Other factors have strong effects on the JEV and genotype circulation within infected landscapes and their emergence in nonendemic landscapes and regions. There are, among them, the movements of either naive or infected animals such as the migrations of flying vertebrates and trading and transportation of domestic animals. These movements modify human exposure to JEV and thus require the particular attention of the health system. In southeastern Asia, trade of live animals occurs between farms, local markets, and more importantly within the Indochinese peninsula and China, and also to Hong Kong and Singapore [80]. Viremic birds, for instance, may have been responsible for JEV spread and introductions [4], [27] into India [81], Taiwan [82], or Papua New Guinea [83]. Over long distances, migratory birds are the most likely spreader of the JEV as some species, such as Egretta garzetta and Nycticorax nycticorax, have a complex migration system over a large geographical area (http://maps.iucnredlist.org). However, the large-scale movement patterns of the main wading birds species implicated in JEV transmission are little known. Additionally, JEV may disperse through wind-blown infected mosquitoes, as is assumed to be the cause of the JEV introduction from Papua New Guinea to north Australia [84] or from China to Japan [85].

Moreover, there is no evidence that JEV requires adaptation to shift between the bird-associated wild cycle and the pig-associated rural domestic cycle [2], [3]. The JEV may easily be transferred from one to another if competent vectors feeding on both wild and domestic hosts are present in JEV-infected areas where both hosts are in close proximity.

For more than a half century, human activities have led to changes of land use and land cover and subsequent deep environmental changes of habitats. These modifications have modified the risk of infection for humans. The ecology of arthropod-borne diseases, noticeably JE, remains complex since they are highly dependent on various biotic and abiotic environmental factors and on the spatial scale of study. To assess and predict JE emergence therefore remains difficult. Multidisciplinary research focusing on virology, molecular and spatial epidemiology, landscape and behavioral ecology, history, and socioeconomic studies are essential to shed light on the biological mechanisms involved in the emergence, spread, reemergence, and genotypic changes of JEV.