Japanese encephalitis virus: epidemiology and risk-based surveillance approaches for New Zealand

ABSTRACT

The introduction and subsequent rapid spread of Japanese encephalitis virus genotype IV across all Australian mainland states and the Northern Territory since late 2021 has increased the risk of an incursion of this mosquito-transmitted zoonotic virus disease into New Zealand, with serious implications for both animal and human health. The potential modes of entry are through introduction of infected mosquitoes as hitchhikers on ships or aircraft, windborne transfer of mosquitoes, or arrival of infected reservoir bird species. A competent vector mosquito, Culex quinquefasciatus, is endemic in New Zealand and other mosquito species may also become involved. If infection becomes established in New Zealand, the scale of transmission may be considerably less than has occurred in Australia because climatic and epidemiological factors are not so favourable. Early evidence of an incursion could come from detection of clinical disease in horses or pigs, or from human cases. Targeted surveillance to confirm or refute indications of an incursion could be undertaken by antibody detection in a number of species. Dogs have been shown to be a particularly valuable sentinel species due to their cohabitation with people and high seroconversion rate. Other novel methods of surveillance could include reverse transcriptase PCR (RT–PCR) on oronasal secretions of pigs. Should evidence of the disease be detected, prompt action would be required to vaccinate at-risk human populations and clarify the epidemiological situation with respect to mammalian hosts and mosquito vector species, including whether a new mosquito species had arrived in the country.

Abbreviations: AHL: Animal Health Laboratory; JE: Japanese encephalitis disease; JEV: Japanese encephalitis virus; RT–PCR: Reverse transcriptase PCR

KEYWORDS:

• Japanese encephalitis virus
• epidemiology
• biosecurity
• surveillance
• cattle

Introduction

Japanese encephalitis virus (JEV), a mosquito-borne zoonotic flavivirus comprising five genotypes (but one serotype), occurs widely across Asia, and is a major cause of human encephalitis (Campbell et al. 2011; Mulvey et al. 2021; Wang et al. 2022). The geographical distribution of the disease and of the genotypes has expanded and varied over recent decades (Gao et al. 2019), with Australia a notable example of the range extension (van den Hurk et al. 2009). There were incursions of genotypes I and II into the Torres Strait Islands in 1995 and into Cape York in 1998 and 2004, but the virus did not appear to have become established in the Australian mainland (van den Hurk et al. 2009).

In 2021 a fatal human case of Japanese encephalitis (JE) occurred in the Tiwi Islands, 80 km north of Darwin (Waller et al. 2022), with the virus subsequently identified as genotype IV, which was first isolated in Indonesia in 1980 (Chen et al. 1992; Mackenzie et al. 2022). It is a rare genotype, and less well studied than the other genotypes. Since late 2021 there has been evidence of rapid spread of JEV around Australia associated with a virus sharing 99.8% nucleotide identity with the earlier genotype IV virus (Sikazwe et al. 2022). The virus has been detected widely across the mainland (Mackenzie et al. 2022; Zhu et al. 2022; McGuinness et al. 2023), and by the end of the transmission season in 2022/23, 45 human clinical cases and seven deaths had been recorded, with continuing updates being provided as the situation evolves. Evidence of infection has been reported in animals in all mainland states and the Northern Territory, leaving Tasmania as the only state free of the virus (DAFF 2023). JE was first diagnosed clinically in pig herds in February 2022, based on fetal deaths and birth abnormalities, in people in March 2022, and subsequently in horses (van den Hurk et al. 2022). Occupational exposure to infected animals has not been an identified risk factor in relation to human cases in Australia (Yakob et al. 2022).

These events raise concerns about possible introduction of the virus to New Zealand, where it is notifiable under the Biosecurity (Notifiable Organisms) Order 2016 and an unwanted organism under the Biosecurity Act 1993.

The epidemiology of JEV is complex and varies between locations across the range of the virus. The objective of this review is to inform veterinarians and public health specialists about the potential manifestations of JEV infection and clinical JE which could occur in New Zealand, and the potential for the virus to extend its distribution to include New Zealand. This is followed by an assessment of scanning and targeted surveillance approaches which would be appropriate to prepare for a potential incursion, and to respond to evidence that the virus had arrived. Options include a possible role for veterinarians in private practice.

Host range

Maintenance hosts

The vertebrate maintenance hosts are generally considered to be ardeid wading birds (van den Hurk et al. 2009) and bats (Mackenzie et al. 2002), although there may be other hosts which can maintain infection in particular ecosystems. In the initial epidemiological studies in Japan (Buescher and Scherer 1959; Scherer et al. 1959a, 1959b) the ardeid birds identified as involved in transmission of infection were the black-crowned night heron (Nycticorax nycticorax), plumed egret (Ardea alba plumifera) and little egret (Egretta garzetta). While only a partial range of ardeid waders have since been evaluated, several appear to be capable of acting as maintenance hosts (Soman et al. 1977; Banerjee et al. 1979; Boyle et al. 1983). The little egret appears to be a significant species in endemic environments in Asia and was shown to be susceptible to infection in Australia (Boyle et al. 1983).

Ten ardeid species have been recorded in New Zealand, some only rarely (https://nzbirdsonline.org.nz). The little egret is an annual migrant to New Zealand (presumably from Australia, where it is common), usually arriving in small numbers in autumn, overwintering, and departing in spring, although some remain over summer. In 2013–14 a short-term influx of the species occurred, so larger numbers can arrive sometimes. Variable numbers of the cattle egret (Ardea ibis) migrate from Australia to New Zealand each year in autumn, and overwinter here. The white-faced heron (Matukumoana) arrived first in New Zealand in the 1940s and is now common in both New Zealand and Australia. The extent of annual migration of this species from Australia is unknown. The Nankeen night heron (Nycticorax caledonicus) recently became locally established in New Zealand (on the Whanganui River) and is an occasional vagrant elsewhere. Occasional vagrants of the Pacific heron (Ardea pacifica) and plumed egret arrive in the country. Therefore, it appears that ardeid birds arrive in New Zealand from Australia in most or all years. If infection becomes endemic in ardeid birds in Australia, infected birds may reach New Zealand. Australia has 22 ardeid species, at least several of which may be competent maintenance hosts – some may never have been exposed to JEV until the range extended into southern Australia.

While ardeid birds are considered central to the maintenance of JEV infection in vertebrate hosts, they are far from the only family that can become infected. There may be up to about a hundred different species across a wide range of bird families which can produce a viraemia following bites by infected mosquitoes. In some species the viraemia may be at a level high enough to be infectious for mosquitoes. However, the information available is less than definitive for most non-domestic species, which makes it unrealistic to make any advance prediction of the possible role of particular New Zealand resident species in transmission following an incursion of JEV, other than those in the ardeid family.

Domestic poultry play probably, at most, a minor role in the epidemiology of JEV. From the available evidence, chickens are on the borderline between maintenance and dead-end hosts. In serological studies conducted in endemic areas they have a low seroprevalence of antibodies, variable between test methods (Kumar et al. 2018b; Auerswald et al. 2020; Ladreyt et al. 2020), but there is some evidence from Asia that high poultry density is associated with high-risk areas (Walsh et al. 2022), not necessarily in a causal role. A study of young chicks (Cleton et al. 2014) found that when infected at a very young age chicks may produce viraemia adequate to transmit infection to mosquitoes, but the peak viraemia following challenge declines to a low level by 42 days of age. Infected chicks suffered weight loss relative to controls. The situation for level of viraemia following exposure of older birds remains unclear. Domestic ducks show slightly greater propensity to act as maintenance hosts. If chickens would play any part in transmission of JEV within New Zealand, it would almost certainly be through mixed age, backyard poultry flocks, especially those with multiple species. Australia has used sentinel chickens to detect the presence of arboviruses, including JEV, but there are better sentinels for JEV.

In Asia bats of both the Microchiroptera and Megachiroptera sub-orders appear capable of acting as the principal maintenance hosts in areas of low pig density (Pearce et al. 2018; Diptyanusa et al. 2021). The two bat species found in New Zealand (Chalinolobus tuberculatus and Mystacina tuberculata) belong to the Microchiroptera, but are unlikely to be of any significance in relation to JEV because of their small numbers and localised habitat (Meduna 2007).

Among New Zealand wildlife that could act to some degree as maintenance hosts, feral pigs are an important candidate, discussed below with domestic pigs. The only other species for which there is evidence that it could potentially be infected by and then infect mosquitoes is the brushtail possum (Trichosurus vulpecula). A study of susceptibility of Australian marsupials (Daniels et al. 2000) showed that possums developed a higher and more extended viraemia than other Australian marsupials evaluated and might be capable of acting as maintenance hosts.

Amplifying hosts

The principal amplifying host for JEV is the pig, both domestic and feral. In situations with infected vector mosquito populations, pigs reach a high prevalence of infection, with high viraemia which persists for several days. There is a propensity for mosquitoes to feed on pigs, and this together with a high birth rate ensures a continuing supply of susceptible pigs (van den Hurk et al. 2009).

There is a recent detailed review of the role of pigs in the epidemiology of JEV, which also summarises clinical and immunological responses to infection (Ladreyt et al. 2019). Viraemia in the pig occurs early after exposure, and is short-lived, typically from Day 1 to about Day 5 post-infection. Antibodies then develop from about Day 7. However, infected pigs secrete virus in oronasal fluid for considerably longer than the period of viraemia (Lyons et al. 2018; Park et al. 2022). Confirmation of infection in a pig herd can be by reverse transcriptase PCR (RT–PCR) on oronasal fluid (Chiou et al. 2021), or serological testing (Chai et al. 2018). If prompt detection by serology is desirable, such as when a recent incursion is suspected, tests which detect IgM rather than IgG should be used (Dhanze et al. 2020). Infected pigs not only have a viraemia high enough to infect mosquitoes, but they can also cause direct pig-to-pig transmission by the oro-nasal route (Ricklin et al. 2016; Lyons et al. 2018; Park et al. 2022). JE has traditionally been thought of as mainly a rural disease because of the role of pigs, but in Asian countries peri-urban pigs are now seroconverting as much as rural pigs (Di Francesco et al. 2018).

Clinical signs of sow exposure (Ladreyt et al. 2019) are mainly the birth of stillborn, mummified or hydrocephalic piglets if infection occurs during pregnancy. Limited data is available, but the evidence shows that typically 15–30% of sows have abnormal litters. As with other causes of similar reproductive loss, the effect on a litter depends on the stage of pregnancy when infection occurs. Inapparent infection can also occur in sows. Boars may suffer testicular enlargement and impaired sperm production and motility. Young piglets can also develop encephalitis, but older ages of pigs typically have inapparent infection, other than the effects in breeding animals described above. Both domestic and feral pigs are susceptible to infection, and seroprevalence can be quite high.

Dead-end animal hosts

Horses are the dead-end animal host most likely to be detected with clinical disease (Ellis et al. 2000). Although inapparent infection occurs in a majority of cases, pyrexia and non-suppurative encephalitis occur in a proportion of cases, causing death in some affected individuals and persistent effects on brain function in a larger number (Pant 2006; Mansfield et al. 2017; Kumar et al. 2018a). They are not considered infectious to mosquitoes (or directly to mammalian hosts) because the level of viraemia is too low.

A wide range of other animal species can become inapparently infected with JEV (Mackenzie et al. 2002; van den Hurk et al. 2009) producing a viraemia which is too low to allow transmission, but allows an immune response to be mounted, so that past infection can be detected by serological tests. Of these species, the highest prevalence of seroconversion is found in dogs (Shimoda et al. 2010; Kumar et al. 2018b; Ladreyt et al. 2020).

Cattle also have inapparent infection, and while prevalence of seroconversion is usually low it can be up to about 30% following spread of the virus through a region, if the vector species favour cattle as a blood source (Sakai and Horimoto 1989; Kumar et al. 2018b).

Human hosts

Infection in people results from bites by infected vector mosquitoes. Less than 1% of people who become infected develop encephalitis, but while the typical ratio of symptomatic to asymptomatic cases is 1:300, the ratio varies from 1:25 to 1:1,000 in different situations (Karthikeyan et al. 2017). Up to 30% of clinical cases are fatal, with a further 30–50% of survivors suffering permanent neurological sequelae. About 75% of clinical cases occur in children and adolescents (Pearce et al. 2018). In endemic areas, a high proportion of people become infected at some point (Mulvey et al. 2021). Because of its epidemiology and wide distribution, JE is the single largest cause of viral encephalitis at a global level (Wang et al. 2022). Human beings are dead-end hosts, the level of viraemia is too low to infect mosquitoes (Moore 2021) and infection is not transmitted directly between people.

There is a valuable overview of the history of JE in Malaysia, where periodic epidemics of variable magnitude have occurred, indicating the variability in scale and frequency of outbreaks in a country where the virus is endemic (Kumar et al. 2018a). In Asian environments, close contact with pigs and processing of pig products was a risk factor for seropositivity in people (Holt et al. 2016).

Insect vectors

The literature on this subject is not easy to interpret reliably, in part because nomenclature of the various mosquito species has changed over time, and there are synonyms for the same species and variability in naming. A recent review of the vector competence of mosquito species for JEV (Auerswald et al. 2020) has identified 14 species as confirmed vectors of JEV (11 Culex spp., two Aedes spp. and one Armigeres sp.). These authors identified 11 additional species as potential vectors based on experimental vector competence studies, and 26 species in five genera from which JEV had been detected in field-caught mosquitoes. A second recent review (Van den Eynde et al. 2022) concluded that 17 species were confirmed vectors and 10 additional species were potential vectors. The two lists are broadly consistent but disagree on some species.

New Zealand has 12 native species of mosquitoes on the two main islands, and three introduced species (Walrond 2007; https://www.smsl.co.nz/NZBEL/New + Zealand + Mosquitoes.html), although the three introduced species arrived between the nineteenth and early to mid-twentieth centuries, and are well established (Weinstein et al. 1997).

Two of the introduced species (Cx. quinquefasciatus and Ochlerotatus (previously Aedes) notoscriptus) are the most likely vectors for JEV in New Zealand, although some of the other species cannot be excluded, for example Cx. pervigilans, which is distributed throughout the country and includes both birds and mammals as blood meal sources (Holder et al. 1999). This species does not appear to have been tested for transmission of JEV, either using the New Zealand specimens (Kramer et al. 2011) or those in other regions (Van den Eynde et al. 2022). Some of the other New Zealand species have also not been tested as vectors for JEV.

Culex quinquefasciatus is widely distributed globally and is among the eight most important confirmed vector species for JEV (Van den Eynde et al. 2022). In the only study using New Zealand-derived specimens (Kramer et al. 2011), a proportion of both New Zealand (17%) and US-derived specimens (86%) became infected with JEV, but none disseminated virus, and none transmitted the virus. Incubation temperature was 24°C. This may have been a consequence of the incubation temperature, because in studies using incubation at 28°C (van den Hurk et al. 2003; Huang et al. 2016), transmission occurred. In a study in Brazil, transmission was more effective at 28 than at 23°C (Van den Eynde et al. 2022). The JEV isolate used with the New Zealand insects was genotype III (Nakayama), and a comparative study of different genotypes found that genotype I had higher infection, dissemination and transmission rates in Cx. quinquefasciatus than genotype III, and a shorter extrinsic incubation period (Karna and Bowen 2019). When the New Zealand population of Cx. quinquefasciatus was held at 10°C to simulate overwintering, producing an extended extrinsic incubation period of 120 days, infection and dissemination occurred, but transmission could not be evaluated. Genotypes I and II have occurred in Australia previously (Pyke et al. 2001), but genotype IV currently circulating in Australia has not yet been evaluated. There appear to be strain differences within Cx. quinquefasciatus in capacity to transmit JEV (Huang et al. 2016; Hernández-Triana et al. 2022; van den Hurk et al. 2022), which complicates assessment of its potential as a vector under New Zealand conditions.

On the basis of this evidence, at a minimum it seems that if JEV entered New Zealand, Cx. quinquefasciatus should be considered a likely competent vector. There is evidence that it is currently extending its range within New Zealand (Kasper et al. 2022).

The next most likely vector is Ochlerotatus notoscriptus, which was separated from the Aedes genus in 2000, but the change was contentious, and the former name is still widely used. A study in Australia at 28°C showed it was capable of infection, dissemination and transmission of JEV (van den Hurk et al. 2003). The New Zealand specimens were incubated at 24°C and failed to infect mosquitoes at that temperature (Kramer et al. 2011). No other studies could be found, so the vector competence of this species remains as possible, at least when the ambient temperature allows.

The New Zealand salt pool mosquito (Opifex fuscus) was capable of infection and dissemination but not transmission at 24°C (Kramer et al. 2011), but its habitat is limited to seaside pools above the high tide line, so it is unlikely to be an influential vector. No other study of its vector competence could be found.

It is not yet clear which species are contributing to transmission in Australia and their relative importance. It has been proposed that the principal vector is Cx. annulirostris, but that other species such as Cx. quinquefasciatus and Cx. gelidus may also be involved (van den Hurk et al. 2022). Ecological niche modelling has shown that both Cx. annulirostris and Cx. quinquefasciatus are capable of maintaining distribution across the areas of Australia in which JEV infection has been diagnosed (Furlong et al. 2022).

There have been two relatively recent incursions of new mosquito species into New Zealand, so further introductions are possible. There was a substantial introduction of Ochlerotatus (Aedes) camptorhynchus, which has not been identified as a prospective vector of JEV, and a much smaller introduction of Cx. sitiens, which is a confirmed vector of JEV (Disbury and McGinn 2018, 2019). Both incursions were eradicated by the New Zealand Ministry for Primary Industries (Yard 2010; Kay and Russell 2013). A variety of species are detected periodically by current border surveillance, including Cx. annulirostris. Increasing summer temperatures may facilitate establishment of new species, which may then overwinter successfully (Brinkhoff 2023).

Diagnostic methods

Clinical disease or pathology in pigs or horses where JE is suspected is notifiable in New Zealand. Virus detection by RT–PCR is available at the Animal Health Laboratory (AHL; Wallaceville, NZ), as is sequencing and virus isolation. A range of post-mortem tissues are suitable, but particularly central nervous system and fetal tissue (WOAH 2019). Blood and cerebrospinal fluid can be used from acutely ill animals. As the viraemic period is typically short (Mansfield et al. 2017), serology is a useful addition to virus detection. Serosurveillance is useful to delimit the areas at risk and paired acute and convalescent sera can be tested in individuals where recent infection is suspected. A generic Flavivirus competitive ELISA is available at the AHL. There is close antigenic relatedness between flaviviruses, and cross-reactivity between different flaviviruses must be considered in interpreting serological findings (Chan et al. 2022). Confirmation of which member of the JE serogroup is present would be required in any index case. The plaque reduction neutralisation test (PRNT) is a specific method of differentiating between group reactive antibodies (Maeda and Maeda 2013) and is available in Australia. Diagnostic sensitivity and specificity of the various diagnostic tests should be evaluated in choosing testing procedures (Pham et al. 2022). Additional diagnostic methods which may be of value because they have novel features are becoming available (Varghese et al. 2023).

Routes of Japanese encephalitis incursion to New Zealand

The risk of JEV introduction from Australia to New Zealand will depend partly on the scale of infection in eastern Australia. The current expansion of the distribution of JEV genotype IV in Australia has been facilitated by La Niña conditions which produce favourable weather conditions in the eastern half of Australia. Also, in 2021 the vector Cx. tritaeniorhynchus was discovered and confirmed in Australia for the first time (Lessard et al. 2021), although its possible role in the spread of genotype IV is unclear. From 2020 onwards, reported Cx. annulirostris counts appeared to be unusually high in eastern Australia and South Australia, from state mosquito surveillance reports. This species requires temperatures above 17.5°C for population maintenance (McDonald et al. 1980). There were no reports of Cx. tritaeniorhynchus outside the Northern Territory in 2022.

It is clear from 2 years of evidence that JEV is now endemic in Australia, although the size of the affected area will fluctuate with climatic conditions. Incursion of JEV into New Zealand must therefore be considered a significant risk to occur at some time, depending on climatic conditions and chance factors. Whether infection becomes established or is eliminated will depend on speed of detection, the nature of the response, and particularly climatic conditions during the period following introduction of the virus.

Possible routes of entry and spread of exotic mosquitoes have previously been reviewed (Frampton 2005). Vertical transmission of JEV can occur in various vectors including Cx. quinquefasciatus (Rosen et al. 1989), facilitating maintenance of infection between generations (van den Hurk et al. 2009). Hence infection could potentially arise from importation of larvae, not just adults. Of 244 mosquito interceptions detected through the national surveillance programme 2001–18 (Ammar et al. 2019), they included six confirmed JEV vectors, and 27% of interceptions were larvae, including 12 for vector species. Entry by sea occurred for 58% of adult interceptions and 88% of larvae, the rest occurring by air. Transport took place in a variety of food and non-food items, but for 41% of detections the transport item could not be identified (Ammar et al. 2019).

The main vector species in Australia is hypothesised to be Cx. annulirostris, which has so far never established a population in New Zealand and may require temperatures above past temperatures recorded in New Zealand, but not necessarily current and future temperatures. Population sizes of this species have been unusually high across the eastern half of Australia since 2021, so populations of infected insects could potentially establish in New Zealand and might maintain in areas which meet the minimum temperature requirement of the species. Importantly, infection of maintenance or amplifying hosts in New Zealand from a temporary population of C. annulirostris could result in infection of Cx. quinquefasciatus, which can maintain its populations long term in New Zealand and can transmit the virus to susceptible hosts following overwintering of the virus in mosquitoes (van den Hurk et al. 2009). The geographical distribution and populations of this species have both increased in recent years (Kasper et al. 2022).

The second mechanism of introduction could be through migration of ardeid birds, principally the cattle egret and little egret, as discussed earlier. It is likely that breeding success in these birds has been high due to weather conditions in Australia, and young naïve progeny are likely to become infected with JEV. High numbers may increase the likelihood of additional migration to New Zealand, as occurred with the little egret in 2013–14. Duration of viraemia in birds is too short for a single bird to cause persistence of virus for long (van den Hurk et al. 2009), but there is a low probability that a new arrival could still be viraemic and transfer virus to Cx. quinquefasciatus, which would then establish infection in local mammals. In Australia, it seems most likely that the very rapid range expansion of JEV since 2021 was due at least in part to movement of viraemic birds, which infected new mosquito populations, rather than just from long-distance transfer of mosquitoes over a few weeks (van den Hurk et al. 2022; Brinkhoff 2023).

It is also possible for mosquitoes to be transported on the wind for substantial distances, as has been demonstrated in Australia (Kay and Farrow 2000; Mackenzie et al. 2002; van den Hurk et al. 2009) and shown to be a likely precipitating factor for earlier incursions into Australia (Ritchie and Rochester 2001). Mosquitoes of the Culex genus have been collected up to 500 km offshore in the Pacific Ocean, at heights up to 380 m (van den Hurk et al. 2009). The estimated travel distance over a single night has been estimated at 152 km over land in one study (Kay and Farrow 2000), and longer nightly distances may occur over sea. Unusual wind patterns such as cyclones may also facilitate long-distance movements.

Introduction with animal imports is considered very unlikely, because all the species that are imported from Australia (mainly horses, dogs and cats) are dead-end hosts, which would not be capable of infecting resident mosquitoes. However, these animals could be antibody-positive on importation, and their status should be determined, to ensure that they do not complicate later serosurveillance.

Since only about 1% of human JEV infections become clinical (Campbell et al. 2011), infected but asymptomatic people will undoubtedly arrive from Australia as part of normal travel movements. However, the level of viraemia in people is too low for this to be a realistic method of introduction, even if new arrivals are bitten by resident mosquitoes while infected.

Development of a risk-based surveillance strategy

The complex epidemiology of JE, the uncertain influence of short-term weather patterns and long-term climate changes, and the unknown probability of one or more of the potential transfer mechanisms causing JEV to reach New Zealand, and then become established, makes it difficult to use standard methods of risk assessment as a way of choosing a risk-based surveillance strategy. The probability of an introduction of JEV is heavily influenced by the geographical distribution and scale of transmission occurring in Australia, which is likely to fluctuate substantially over time. In principle, surveillance in New Zealand should be proportional to the risk of introduction of JEV at a particular time. However, as has been apparent in the current Australian outbreak, it is possible to go from zero to widespread infection across a large part of the country very rapidly if conditions are favourable, and the virus could be transferred to New Zealand at the same time that it is at high prevalence in Australia. The mean annual probability of an incursion is low based on past climatic data, but the cumulative probability of an incursion within the next 20 years is considerably higher, especially if climatic circumstances change over that time.

Preparedness for a possible incursion of Japanese encephalitis

An appropriate preparedness surveillance strategy would be to enhance awareness of the risk of an incursion, undertake immediate differential diagnosis of unexpected cases of human encephalitis, and investigate suggestive events in horses and pigs. This scanning surveillance could be supported by annual low-intensity targeted surveillance during the warmer parts of the year to confirm continuing freedom, escalating if this triggers concern of the possible presence of the virus.

The medical community should be made aware that any unexplained cases of encephalitis require urgent investigation. However, human cases are likely to occur after the first evidence is detectable in animals. All veterinarians should be asked to report to Biosecurity New Zealand neurological cases in horses and above normal occurrence of abortions, stillbirths and hydrocephalus/encephalitis in piglets. However, it is most likely that early incidents of this nature will occur in small outdoor herds on lifestyle properties, where a few sows may farrow per year. Owners may not recognise that a single litter showing these abnormalities is important enough to justify seeking advice or presenting possible affected piglets for assessment. This may delay identification of an incursion. Infection could also establish quite early in feral pigs, which will further disseminate JEV with low probability of early detection.

An estimated 4,000–6,000 smallholders keep pigs for some or all of the time in New Zealand and are much more likely to be involved in maintenance of JEV than commercial herds (< 200). Smallholder pig herds are widely distributed throughout the country and are more likely to be in close proximity to feral pigs, which occur over at least a third of New Zealand (McIlroy 2005).

Therefore, while all these methods of scanning surveillance should be used, there is a serious risk that detection would be significantly delayed after JEV established in New Zealand, especially if the incursion came through a port and spread through small peri-urban pig herds.

For these reasons and because the majority of JEV infections in animals are inapparent, there is a need for targeted surveillance to support the scanning surveillance activities. The question is then, which host species are most accessible and therefore likely to become infected and seroconvert, assuming the main vector is Cx. quinquefasciatus. Evidence about blood meals taken by this species is quite variable (Takken and Verhulst 2013), in part because the data comes from different countries and environments, and none is from New Zealand. In some North American (Garcia-Rejon et al. 2010; Greenberg et al. 2013) and Caribbean studies (Fitzpatrick et al. 2019) birds were the main species bitten, while in other North American studies (Loftin et al. 1997; Janssen et al. 2015) and Asian studies (Yeo et al. 2020; Boyer et al. 2021) mammals were the main target. Australian studies found birds were the main target in one Queensland location (Kay et al. 1985) and mammals in another (Kay et al. 1979). There is also some evidence that the target species are mammals early in the season, and birds later (Greenberg et al. 2013; Lujan et al. 2014). Overall, it would appear that these mosquitoes are fairly opportunistic feeders, so it is reasonable to assume that in New Zealand a range of species would seroconvert, and therefore be suitable for use in targeted surveillance. Dogs were prominent target species in several studies (Kay et al. 1979; Janssen et al. 2015; Boyer et al. 2021).

Mosquito surveillance is routinely undertaken by the Ministry of Health and Ministry for Primary Industries (Ammar et al. 2019; Acosta et al. 2020). The purpose of this surveillance is early detection of exotic species. Several trap types and other sampling methodologies are available for capturing different species and life stages of mosquitoes (CDC 2023). Light traps are less effective for detecting Cx. quinquefasciatus than for other mosquitoes and are not suitable for detecting specimens that have had a blood meal and hence may contain virus. Gravid traps are needed for this purpose (Thornton et al. 2016). As part of preparedness for a possible future incursion, it would be valuable to assess the relative frequency of different host species as sources of blood meals for this mosquito as this would assist in allocating surveillance resources.

While virus can be detected in mosquitoes either by virus isolation or PCR, the probability that an incursion would be detected first by PCR testing of mosquitoes seems very low. However, a useful option would be to regularly freeze mosquitoes collected from gravid traps and store them, so that species-differentiated samples from a relevant area of the country could be examined by PCR if other evidence of infection was detected. This may assist in determining which species have become infected. Larvae can also be collected and speciated by visual or PCR methods (Beebe et al. 2007) but cannot be used to detect virus.

Dogs commonly have the highest seroprevalence among the dead-end hosts (Kumar et al. 2018b; Ladreyt et al. 2020) and have been recommended as sentinels (Shimoda et al. 2010; Pham-Thanh et al. 2021). They would seem to be a very appropriate first line of detection in New Zealand, for several reasons. Firstly, they occur throughout the country, including close to potential entry sites such as ports. Secondly, dogs are seen by veterinarians, and blood samples can be easily collected. Thirdly, the level of viraemia is low, so even if a dog was viraemic at the time of collection, the risk to a veterinarian from an accidental needle stick would be minimal. A study in South Korea (Yeh 2022) found that seropositivity peaked at around 10% in stray and farm dogs in the year before resurgence of JE in the study area, but levels were lowest in pet dogs.

Therefore, a suitable target population for annual summer serological sampling would be dogs obtained by dog pounds and Society for the Prevention of Cruelty to Animals clinics. These will typically have been wandering prior to collection and therefore have been exposed to mosquito bites, and in general are unowned, so do not require owner consent for blood collection. This would be the most cost-effective method for undertaking low-intensity surveillance in the absence of any perceived immediate risk of an incursion. If there were indications of a possible incursion, private practices in the area could be asked to provide samples from at-risk dogs seen in their clinics.

Pigs would be the second most valuable species for targeted surveillance. Spread of JEV in Australia was first detected through clinical recognition of unusual disease in newborn piglets on commercial pig farms (van den Hurk et al. 2022). As noted, New Zealand has low density of commercial pig herds, and while passive surveillance of these farms would be a valuable component of surveillance, it is likely to have low sensitivity to detect infection early in an incursion. Surveillance in feral pigs would not add sufficient value to justify its use, considering the cost and practical problems of sampling pigs captured by hunters.

A possible alternative could be to collect meat juice samples at abattoirs (Yonemitsu et al. 2019), as was done for Aujeszky’s disease eradication from New Zealand in the 1980s. Because a very high proportion of pigs slaughtered at abattoirs come from large commercial herds, like clinical reporting, it would have low sensitivity for detecting an incursion in its early phase and therefore is not justified. Higher sensitivity could be achieved by meat juice sampling of pigs slaughtered by itinerant services, which kill pigs for smallholder owners.

Pigs secrete JEV in oro-nasal secretions beyond the period of viraemia, and it has now been shown that contact transmission can occur between pigs, without mosquito involvement (Ricklin et al. 2016). Oronasal fluids can be collected from chewing ropes placed in pig pens and tested by RT–PCR for presence of virus (Lyons et al. 2018; Chiou et al. 2021). This would be a suitable low-cost method (with low operator risk) of conducting surveillance in cooperating smallholder and commercial pig herds to confirm or deny presence of infection. A structured sampling plan based on this technique could be used to determine the infection status of pig herds if an incursion is suspected.

Cattle are used for the existing national arbovirus surveillance programme (Acosta et al. 2020), and JEV could be included in the surveillance activities. Cattle are not the favoured target species for Cx. quinquefasciatus, and in countries where JEV is endemic the prevalence of JEV antibodies in cattle is lower than in dogs or pigs (Loftin et al. 1997; Janssen et al. 2015; Kumar et al. 2018b). Surveillance using cattle would therefore have lower detection sensitivity, but since samples are already collected JEV could be added to the regular screening programme at low cost.

Surveillance in wild birds may be useful in confirming the presence of infection in particular species following confirmation of an outbreak but is unlikely to be a cost-effective method of detecting early evidence of an incursion at national level.

Further surveillance to determine distribution of Japanese encephalitis following a confirmed incursion

If there was putative evidence from scanning or targeted surveillance that there may have been an incursion of JEV, then both scanning and especially targeted surveillance would need to be promptly enhanced.

Surveillance of dogs would need to be intensified, especially in the putative outbreak area, and has the additional advantage that dogs live in close proximity to people and therefore represent a good early warning system of risk areas for human health. Investigation of possible clinical disease could be undertaken by practising veterinarians in smallholder pig herds in areas considered at risk of circulating virus, and RT–PCR testing of chewing ropes from a wide range of pig herds would be particularly valuable. Blood of susceptible species submitted to diagnostic laboratories for any reason could be tested for JEV as an additional activity. Historical collections of mosquitoes could be tested by PCR to detect past evidence of infection if they have been stored appropriately.

After initial epidemiological characterisation of the outbreak, it would be necessary to determine the spatial extent of virus distribution, and the composition of the mosquito population in affected areas. In addition to the use of surveillance in dogs and smallholder pig herds already in operation, sampling of mosquitoes would become necessary, to determine the species mix and the spatial distribution of any newly arrived species. Wild bird sampling may help to clarify which species are becoming infected.

Response to a confirmed incursion

The nature and scale of the response should depend principally on the geographical region of the country in which the infection is detected, with implications for the pattern of spread and probability of infection becoming endemic. As well as investigating endemic mosquito species to assess their vector competency, it would be crucial to quickly determine whether any new species of mosquitoes were present, most importantly Cx. tritaeniorhynchus, Cx. gelidus and Cx. annulirostris.

Vaccination of people considered at high risk of exposure to infected mosquitoes should commence promptly (Vannice et al. 2021), and prompt access to initial supplies of vaccine would be desirable. Decisions concerning possible control measures in animal populations and mosquito populations would depend on early epidemiological evidence about spatial distribution and other features of the outbreak. Pig farms in affected areas would need to implement mosquito control to minimise the risk of an amplification event.

Conclusion

The arrival of the previously rare genotype IV of JEV into Australia and its very rapid spread across the mainland, causing human cases and deaths, has substantially raised the risk that at some point in the future the virus may enter New Zealand. While the annual probability of incursion is low, the risk is likely to fluctuate considerably, as the climatic and epidemiological situation in Australia varies over time. Entry would most likely occur through hitchhiking adult or larval vector mosquitoes arriving on sea or air transport, but arrival of the virus on wind-blown mosquitoes or migratory ardeid waterbirds is also possible. If the virus arrived, pigs would be the main amplifying host, although possums may possibly also contribute. It is probable but not certain that one or more resident mosquito species may become the vector. It is also possible that a new vector mosquito species may establish as summer temperatures rise, bringing the virus with it.

Veterinarians should be aware of the clinical syndromes caused by the virus in horses and pigs, and doctors should consider JE as a possible cause of any unexpected cases of encephalitis they see. The disease is notifiable. It is however possible that the disease may not be recognised initially and may have become well established in mosquito populations before it is recognised as present. This risk could be reduced by supplementing existing mosquito surveillance with serological testing of dogs, particularly unowned dogs, since dogs are an effective sentinel species, although infection is clinically inapparent. The existing arbovirus surveillance in cattle could be extended to include JEV.

If the risk of an incursion becomes elevated, serological testing of dogs could be used more widely, and PCR testing of oronasal secretions of pigs could be added to the surveillance programme. If the presence of the virus is confirmed, it would be necessary to determine the spatial distribution of infection in pigs, dogs and possibly cattle, depending on which vector mosquito was involved. It would also be necessary to determine which mosquito species were present and which were vectors. Personnel involved in investigating the nature and extent of the outbreak should be vaccinated. Based on this information, an appropriate response could be determined.

Disclosure statement

No potential conflict of interest was reported by the authors.