Mosquito-borne disease and climate change in Australia: time for a reality check

Abstract

Will warming climate increase the risk or prevalence of mosquito-borne disease in Australia, as has been projected in a number of scientific publications and governmental reports? Unfortunately, most of these ‘predictions’ do not adequately consider the current and historical distribution of the vectors and diseases, their local ecology and epidemiology and the impact of societal features and the capacity for public health interventions in Australia. Overall, a strong case can be made that we are unlikely to see significant changes in the distribution of transmission of the exotic pathogens causing malaria and dengue, and while activity of endemic arboviruses such as Murray Valley encephalitis and Ross River viruses may possibly increase in some areas, it is likely to decrease in others. The ecologies of mosquito-borne diseases can be complex and difficult to predict, and any evaluation of potential effects of changes in climate will need a detailed examination of site-specific vector, host and other factors likely to influence the outcomes on human health. Of itself, climate change as currently projected, is not likely to provide great cause for public health concern with mosquito-borne disease in Australia.

INTRODUCTION

Mosquito-borne disease in Australia (Russell & Dwyer 2000; Russell & Kay 2004) continues to be a significant concern, and it is now 20 years since the potential impact of climate change was first seriously discussed (Liehne 1988). In the decade that followed, there were projections of increased activity associated with climate change in Australia for the exotic pathogens malaria and dengue (Murray-Smith & Weinstein 1993; Jackson 1995), and the endemic pathogens Murray Valley encephalitis virus (MVEV) and Ross River virus (RRV) (Mackenzie et al. 1993; Lindsay & Mackenzie 1997), although some relating to malaria and dengue were challenged (Russell 1998; Walker 1998).

During the past 10 years, there has been increasing concern for health impacts of global warming in Australia, and continuing projections and predictions for increasing mosquito-borne disease as a result of climate change (Martens et al. 1999; Hales et al. 2002; McMichael et al. 2003, 2006; Hennessy et al. 2004; Woodruff et al. 2005, 2006; Preston & Jones 2006; AGO 2007; Horton & McMichael 2008). Although, in a general sense, the proposition that climate change (through temperature, rainfall and sea-level influences) will affect mosquito populations and pathogen activity has some validity, many such claims of increasing mosquito-borne disease problems for Australia are relatively simplistic, and do not take adequate account of the current or historic situations of the vectors and pathogens, and the complex ecologies that might be involved.

This brief commentary is designed to provide an overview of the vector situations for a range of mosquito-borne diseases in Australia, with respect to predictions of such increases that have been raised for projected climate changes as proposed and modelled by CSIRO (2001, 2007) and IPCC (2001, 2007). Such projections are based on various greenhouse gas and sulphate aerosol emission scenarios within different models but, for example, according to the recent CSIRO (2007) projections for mid-range emissions, ambient temperatures are likely to increase by approximately 1^οC by 2030 (relative to 1990) and warming will be a little less in coastal areas and a little more inland, although rainfall is likely to decrease in southern and eastern areas (particularly in winter and spring, although there may be increases in some eastern regions in summer) and be little changed in the far north overall. Later in the century, both temperature and rainfall projections become more dependent on emission scenarios, but warming is predicted to further increase overall and rainfall is likely to continue decreasing in southern regions.

MALARIA

Malaria is the most important mosquito-borne disease in a global sense, and is caused by any of four human Plasmodium parasites transmitted by various Anopheles species in different parts of the world. It has been endemic in Australia (Black 1972), but was declared eradicated in 1981 by the World Health Organization. However, imported malaria remains a concern and in recent times there have been up to 1000 persons annually arriving in Australia with malaria infections, with local transmission occurring occasionally because there are competent vectors in many regions.

With respect to climate change, the recent arguments for a climate-induced increase in malaria transmission in Australia typically have rested on two arguments: first that the malaria receptive zone would be extended southwards with increases in temperatures, and second that the distribution of the ‘Australian vector’Anopheles farauti Laveran sensu lato (s.l.) would be similarly extended southwards.

On the first point, the so-called malaria receptive zone, the region lower than 19^οS, was declared during the 1940s as a logistical contingency for deployment of troops during the Second World War (Ford 1950). It was supposedly based on the distribution of An. farauti, identified as the vector of malaria in the 1942 Cairns, Queensland (Qld) outbreak and known to be a vector in Papua New Guinea; however, in reality, it reflected the historic distribution of malaria cases in northern regions of Qld, Northern Territory (NT) and Western Australia (WA), within a temperature zone suitable for maintenance of Plasmodium falciparum. In fact, the distribution of An. farauti s.l. is very much restricted to northern coastal regions and bears no relationship to the general distribution of historic malaria transmission in Australia. Other Anopheles species across northern Australia must also have been responsible for malaria transmission and so, with respect to the second point, any extension of the distribution of An. farauti s.l. is likely to be of relatively little significance or consequence for malaria risk.

It is possible, even likely, that the distribution of some or all northern Anopheles species will be extended southerly by climate warming (providing other critical ecological factors are favourable) and, because endemicity of malaria is temperature dependent for Plasmodium parasites, warmer conditions will likely extend the potential endemic zone for P. falciparum and P. vivax further south, but it needs to be kept in mind that very high temperatures can kill the mosquito or the parasite or both and thus reduce potential for malaria transmission (e.g. Ijumba & Lindsay 2001).

However, this can be argued to be of little significance because, historically, malaria transmission in Australia has not been restricted to the northern tropics, occurring as far south as Melbourne in eastern Australia and Perth in western Australia. There are Anopheles species widely spread in temperate southern Australia (including Tasmania) responsible for transmission and, based on circumstantial evidence, An. annulipes Walker s.l. has often been thought the most likely vector. However, a confounding factor for understanding transmission by local Anopheles species is that An. farauti s.l. and An. annulipes s.l. (at least) are known to be species complexes, and virtually nothing is known of the respective competence of the various siblings for transmission of the Plasmodium species and strains that may enter Australia with travellers, although the evidence of cases in both northern and southern Australia stands as testament to local transmission.

Overall, local transmission in Australia has been rare, even within the receptive zone, and Bryan et al. (1996) mentioned various possible reasons for this, related to parasite, vector and human factors. Future warmer temperatures from climate change may well provide for increases in vector populations in certain areas, as well as enhanced pathogen incubation in the mosquitoes. However, the generally drier conditions that are projected for many areas, particularly in southern Australia, may provide for a decrease in aquatic habitat, and the concomitant lower humidity could result in reduced mosquito longevity that is critical for the incubation/development of Plasmodium in the mosquito, and therefore actually reduce vectorial capacity for even the most competent vectors.

As noted by Sutherst (2004), based on malaria parasite requirements only, almost all of Australia can be said to be at some risk of transmission and, as mentioned above, malaria has been (and still can be) locally transmitted in many regions of Australia. However, this has been dependent on the introduction of the infective stages (gametocytes) of the parasites with travellers. Although the distribution of local competent vectors may change marginally with changing climate, and therefore some human populations may be at slightly greater risk, if there is a real threat of increased transmission of malaria in Australia in the future, it will be because of an increase in the importation of the parasite with travellers and refugees from countries with high levels of transmission.

With malaria increasing in many areas of the world, for reasons mostly associated with failure of local control programs, we may see an increase in numbers of infected persons arriving in Australia and thus an increased risk of local transmission, particularly in northern regions but also in southern regions during the warmer months. However, our health services have the experience and capability to deal with imported and local infections, have been doing so effectively for decades, and there is little likelihood that malaria will become re-established in Australia. The appropriate maintenance of these health services is the critical issue for questions of increased malaria risk – not whether climate warming will bring minor extensions of the receptive zone or the distribution of vectors.

THE ARBOVIRAL ARTHRITIDES

Ross River virus and Barmah Forest virus (BFV) are widely distributed alphaviruses that cause debilitating polyarthritis with significant personal and community impacts in rural regions (Russell & Dwyer 2000). These are the most common mosquito-borne pathogens causing disease in Australia and, although reports are generally thought to under-represent actual incidence, an annual average of 4244 cases of RRV and 1105 cases of BFV were registered with Commonwealth health authorities for the period 1995–2007 (http://www9.health.gov.au/cda/Source/CDA-index.cfm).

Ross River virus appears to be endemic and annually active in most states and many regions throughout Australia, with macropods generally accepted as native vertebrate hosts and a wide range of mosquito species as vectors, dependent on locality, season and circumstance (Russell 2002). The natural history of the vector/virus association, and thus the epidemiology of the disease, varies with region and even locality – mostly because of temporal and seasonal effects on the range of potential vector species. Notwithstanding this, it can be said that the ‘freshwater’Culex annulirostris Skuse, Coquillettidia linealis (Skuse)/sp. nr. linealis (of E.N.Marks) and various ‘floodwater’Aedes (e.g. Ae. normanensis (Taylor)) are important vectors in inland areas, the ‘saline’Ae. vigilax (Skuse) and Ae. camptorhynchus (Thomson) and ‘brackish’Ae. procax (Skuse) likewise in coastal regions, and the ‘container’Ae. notoscriptus (Skuse) is strongly suspected for urban areas. In a national sense, BFV is more sporadic in distribution and activity, and its natural history appears to differ from that of RRV; the vertebrate host is unknown, but genetic studies (Poidinger et al. 1997) indicate a more mobile host than is the case with RRV and thus bats and/or birds may be involved. Although the important vectors for BFV are similar to those for RRV, activity of the viruses does not commonly coincide and, as with RRV, competence for transmission varies with vector population and virus strain.

With respect to climate change, the thrust of arguments for increasing incidence and distribution of these viruses is associated with thoughts that increasing temperature will facilitate increased distribution southwards of vectors, longer seasons of peak vector and virus activity, increased rapidity of virus cycles in vectors, and the creation of more inland habitat with increased rainfall and coastal habitat with rising sea levels.

There is no evidence to suggest that RRV or BFV have moved from northern to southern regions of Australia in recent times. The first reports of RRV disease were from southern inland New South Wales (NSW) in 1928, and the first reports of BFV infections were from southern coastal NSW in the 1980s. It is a moot point as to whether projected increases in temperature will bring any southerly extensions of any consequence for vectors/viruses, given the overall number and range of vector species and the evidence of virus activity in all states and territories, including southern Tasmania. However, an extension of the activity season with warming is a likely result in cooler regions, in both inland areas where rainfall will not be seriously reduced so as to restrict larval habitat or limit adult survival (through low humidity), and in coastal areas where sea levels may rise and extend saltmarsh habitat. The rate of extrinsic incubation of RRV and BFV might be expected to increase with temperature, but it is actually known to decrease at higher temperatures for RRV, and there is laboratory evidence that adult mosquitoes reared (as larvae) at higher temperatures may be less competent vectors (Kay & Jennings 2002). If this occurs in nature, then warmer temperatures may result in reduced efficiency of transmission for some combinations of vectors and viruses; furthermore, increases in environmental temperatures have been shown to correspond to increases in mosquito mortality that reduce vectorial capacity (Reeves et al. 1994).

From the most recent models (CSIRO 2007), there appears to be little likelihood of increased rainfall extending inland habitat, other than in northern Australia and perhaps northeastern NSW. In northern Australia, increased rainfall could result in prolongation of seasonal RRV activity, but questions of susceptibility/immunity in vertebrate host populations cloud the issue, as they would wherever increased activity was promoted by more favourable conditions. For coastal eastern Australia, any increases in summer rainfall could enhance vector populations, but frequent rain can provide for habitat persistence that allows for predator populations to mitigate vector populations. If vertical transmission is an important survival strategy for the viruses, then projected decreasing winter and spring rainfall in southern regions may delay or preclude initiation of virus activity dependent on floodwater Aedes species. In southwestern WA, with projected reduced rainfall providing fewer habitat opportunities for both freshwater (Cx. annulirostris) and saltmarsh (Ae. camptorhynchus) vectors, outbreaks of RRV and BFV may become fewer.

For coastal areas, saltmarsh may be expected to increase with sea level rise and thus extend habitat for Ae. vigilax and Ae. camptorhynchus in northern and southern regions, respectively. However, much will depend on local topography and natural drainage patterns, with marshes possibly being created in some areas but probably decimated in others; complete inundation and frequent flooding can lead to saltmarshes becoming less suitable for the above-mentioned vectors but more favourable for the saline Culex sitiens Wiedemann, which is not thought to be an important vector. Additionally, increased sea levels could bring encroachments to urban communities that would remove saltmarsh habitat and also result in less periurban freshwater natural habitat being available.

Although activity patterns of the two viruses do not coincide, and it is difficult to know what influence climate change may have on vertebrate hosts (which may differ substantially between regions), the viruses share a similar range of vectors and thus climate factors might be expected to impact similarly on them through the mosquitoes. However, the catholicity of the range of vector species confounds pronouncements. In inland areas, Cx. annulirostris will likely continue to be the major vector, increasing its impact in tropical regions with increased seasonal rainfall but becoming of less importance in areas with reduced rainfall not augmented by irrigation practices (and decreased rainfall has led to less water being available for irrigation in many areas), and this may see a shift to Aedes species that take advantage of the more temporary habitats. In coastal areas, increasing temperatures may see the northern saltmarsh species Ae. vigilax extend its distribution southwards and share more habitat with Ae. camptorhynchus, which it might replace in some situations. However, as both species are important vectors currently within their overlapping distributions there might be little likelihood of any major overall change in viral activity.

There has been a number of attempts in recent years to model the relationships of RRV/BFV activity with climatic variables; however, in general, these efforts have repeatedly showed how important it is to approach this issue on a ‘small regional’ basis, and how little use is a ‘broad area’ approach that encompasses too much variability with the critical biological and environmental factors of these viruses, which display many regional and local epidemiologies (Jacups et al. 2008).

THE ARBOVIRAL ENCEPHALITIDES

Murray Valley encephalitis virus and Kunjin virus (KUNV) are flaviviruses that cause life-threatening illnesses (Russell & Dwyer 2000). They are uncommon to rare outside the northern (particularly northwestern WA and the Top End of the NT) endemic zones, where almost annual activity is detected and occasional deaths occur in local or visiting persons (Mackenzie et al. 1994; Broom et al. 2003; Kurucz et al. 2005). Historically, epidemics have occurred mostly in the Murray Darling basin of southeastern Australia; however, the last was in 1974 and activity of either virus has been detected on relatively few occasions subsequently. The viruses have wading birds as vertebrate hosts and, although there is some evidence for vertical transmission in Aedes floodwater mosquitoes, various Culex mosquitoes are most likely to be the enzootic vectors between birds, and Cx. annulirostris is generally accepted to be the major bridge and epidemic vector for humans.

Epidemic activity in southeastern Australia has been linked to flooding conditions that promote enhanced breeding of the bird hosts and mosquito vectors, but the question of whether the virus is introduced by birds from northern endemic regions of seasonal activity, or arises with such conditions in local cryptic foci associated with wetlands in the Darling/Murrumbidgee/Murray river systems of NSW, Victoria (Vic.) and South Australia (SA) remains unanswered. Nonetheless, the riverine wetlands in southeastern Australia have been declining with respect to providing favourable circumstances for virus cycles and, if this trend continues, there will be further decreasing likelihood of future epidemic activity of MVEV and KUNV in southeastern Australia.

With respect to climate change, and its impact on the dynamics of the natural cycles of the viruses and the likelihood of amplification and overflow transmission to humans, the two principal areas of concern are the endemic regions of northern Australia and the epidemic regions of southeastern Australia. An extension of seasonal wetlands and related conditions in northern Australia will likely enhance the cycles that sustain both viruses, and this may provide for more intensive and extensive activity in the Kimberley and Pilbara regions of WA, the Top End of the NT and inland northern Qld, and perhaps allow for geographical extensions of activity towards southern regions. However, the predicted drier conditions for spring seasons in the southeast will militate against increased virus activity, particularly if critical bird and mosquito breeding in the Murray Darling basins is prevented by the predicted reduction in the extensive rainfall required to sustain the natural wetlands and the predicted reduction in the high humidity conditions required to sustain adult mosquito longevity for viral proliferation.

In the endemic region of northern WA, MVEV activity has been shown to be only partially correlated with rainfall (Broom et al. 2003), whereas in other northern regions of WA and the NT, widespread activity of MVEV has been shown to follow excessive rainfall. In earlier climate models (CSIRO 2001), increased rainfall for northwestern endemic regions had been predicted; however, heavy rainfall events can be initially detrimental for Culex larvae, which can be flushed away from protected areas, and a subsequently slower seasonal build-up of vector populations may delay transmission. Conversely, heavy rainfall on arid grasslands may result in extensive activity of floodwater Aedes species (such as Ae. normanensis), which can be followed by Cx. annulirostris using newly vegetated and persisting habitat, and there may be a concomitant increase in virus transmission. The stronger tropical cyclones projected for northern WA may bring southerly extensions of MVEV and KUNV activity. In the NT, increased or earlier summer rainfall could bring a southerly extension of annual MVEV/KUNV activity to south of Tennant Creek and towards Alice Springs (Whelan et al. 2003). The most recent predictions (CSIRO 2007), however, have indicated little change in rainfall patterns for the endemic zones, and although there might be more intense tropical cyclones, the overall number might decrease. Overall, this may mean the endemic activity of MVEV and KUNV in northern Australia will change little from the current patterns.

The predicted general decrease in winter and spring rainfall in inland southern Australia will have a continuing deleterious effect on important wetland habitat, negatively influencing water bird populations and the abundance of spring enzootic and bridge mosquito vectors. The concomitant reduced humidity might also impact negatively on the overwintering adults that normally initiate the population growth of Culex species that culminates in the summer/autumn transmission period, and during the summer months the low humidity will result in reduced survival of the adult vectors and thus reduced viral transmission rates. Overall, there is little or nothing in the climate projections to indicate any significant increase in MVEV and KUNV activity in southeastern Australia.

DENGUE

Dengue viruses are the most important mosquito-borne arboviral pathogens internationally; dengue disease is caused by any of the four closely related virus serotypes, and the viruses are spread mostly by the urban mosquito Aedes aegypti L. The disease reappeared in Australia in 1981 (Kay et al. 1984) and, since 1990, all four serotypes have been imported and been locally active. There has been almost annual transmission, with almost 3000 confirmed cases overall, and two deaths in 2004 (McBride 2005). All dengue transmission in Australia for the past 50 years at least has been confined to Qld and the vector Ae. aegypti is currently not established elsewhere; cases have occurred only in northern coastal Qld where populations of the vector are greatest, with Charters Towers being the most southerly point of transmission. Although surveillance has been relatively limited, there appears to have been no substantial change in the distribution of Ae. aegypti in Qld over the past 25 years (Kay et al. 1983; Sinclair 1992; Qld Health Department unpublished recent survey data – P Mottram pers. comm. 2008).

With respect to climate change, the arguments for an extension of dengue southwards with climate change appear to hinge on proposals that increases in temperature will allow the vector to move southwards, and that future climate conditions will allow for maintenance of virus transmission in more southerly areas. It has been stated that a temperature increase of 2–3^οC would see dengue spread to Brisbane and with 3–4^οC to Sydney (Preston & Jones 2006).

Such projections arise from modelling based on a global association of recent (1975–1996) dengue activity with vapour pressure as a measure of humidity (Hales et al. 2002); however, this modelling did not take into account the present and past distributions of either vector or virus in Australia. The subsequent predictions for increased distribution of dengue (Woodruff et al. 2005) and Ae. aegypti (McMichael et al. 2006) in Australia were based on estimated ‘current’ ranges that are not consistent with the known recent distributions for either the vector or virus transmission and, as they also ignore the past distributions in Australia that extend well beyond those being projected, it is difficult to reconcile the ‘predictions’ with ‘actualities’ and thus accept these future scenarios.

An analysis of the historic distribution of Ae. aegypti (from various published and unpublished sources, including O'Gower 1956; Russell et al. 1984; Lee et al. 1987) shows it previously occurred in many parts of Qld (including Brisbane until the mid-1950s), NT, WA and NSW. In the NT, it existed until the late 1950s to early 1960s with southernmost records of Anthonys Lagoon and Newcastle Waters. In Western Australia, it existed at least until 1970 and was established widely, at least as far south of Perth as Harvey (33^οS) but possibly Busselton (34^οS) (an unconfirmed record). In eastern Australia, it was widely distributed in NSW (including Sydney) before 1950, reaching inland as far as Bourke and south as far as Culcairn (36^οS and 40 km north of the Vic. border), and there are unconfirmed records from Vic. as far south as Melbourne (38^οS) (Russell et al. 2009).

An analysis of the historic distribution of dengue transmission shows more extensive southern activity than in recent times. It previously occurred not only in southeast Qld (including Brisbane where there were deaths from dengue in the early years of last century), but also in NSW and WA where local transmission of dengue occurred as far south as Gosford (80 km north of Sydney) and Carnarvon (900 km north of Perth), respectively (Russell et al. 2009). Indeed the coastal belt north of Carnarvon, WA around through Darwin, NT and down to Townsville, Qld was considered an endemic region (Lumley & Taylor 1943). More than 60 years ago, the international distribution of dengue was recognised to extend between 36^οN and 35^οS, nearly equalling the geographic limits of Ae. aegypti, which has been able to establish in areas between the January and July isotherms of 10^οC in the northern and southern latitudes, respectively, roughly equivalent to 45^οN and 35^οS (Christophers 1960), and approximating the Australian historic distribution mentioned above.

The reasons for the retreat of Ae. aegypti in Australia appear to be complex, and it is not appropriate here to attempt to discuss them in detail; however, various public health initiatives, and community and household advancements, are thought to have contributed (Russell et al. 2009). As noted by Sutherst (2004), because dengue is an urban disease and outbreaks are driven largely by availability of virus and container habitats for vectors, there appears to be less climatic influence than with other mosquito-borne diseases. With respect to the current distribution of dengue in Qld, it is likely more associated with a greater incidence of importation of virus with viremic travellers to localities which have greater vector populations (such as Townsville and Cairns, and other towns to the north of Cairns), whereas other localities further south have fewer such importations and have fewer vectors.

Dengue is increasing in many parts of the world and we are likely to see a greater number of dengue importations to Australia, perhaps with concomitant increases in local outbreaks in northern Qld where the vector is abundant and where temperature increases might enhance virus replication in the vectors. The threat to more southern Qld localities that already have Ae. aegypti might also increase, particularly with increases in water storage and hoarding for domestic purposes that can promote abundance of the vector. However, it should be very clear that, currently, the distribution of the species is not being governed by climate in eastern Australia, otherwise Ae. aegypti would be more widely established in southeastern Qld and northeastern NSW where climatic conditions that sustained the species in the past have continued to be highly favourable. Additionally, although mean eastern Australian temperatures have increased since 1950 (CSIRO 2007), as mentioned above there is no evidence for any significant southward increase in the distribution of Ae. aegypti during the past 25 years. Therefore, it is hard to accept the arguments that there will be southward extensions of the vector and the disease to Brisbane and Sydney driven simply by the increases in ambient temperatures predicted with climate change.

Increasing urbanisation in tropical countries to our north, and increasing trade contact with Australia, is likely to bring an increasing risk of introductions of Ae. aegypti to ports in the NT and northern WA, which will have to be combated as currently. However, the recent arrival of the exotic species Aedes albopictus (Skuse) into Australia, with its establishment in many of the Torres Strait islands, is of greater concern for southern Australia. Ae. albopictus is a known secondary vector of dengue (and a potential vector of other viruses) and is far more tolerant of temperate environments. If Ae. albopictus becomes established on the mainland, it could very likely spread to all the southern states (Russell et al. 2005), providing a vector for introduced dengue in urban regions not currently at risk of local transmission, and the southern spread of this species could be enhanced by warmer temperatures.

CONCLUSION

There is a wealth of data on mosquito-borne disease epidemiology in Australia that can inform predictions of future vector and disease distributions. Although the complexity of the natural cycles of the endemic arboviruses tend to hinder and confound predictions of future activity under climate change, the dynamics of malaria and dengue transmission are comparatively simple and the historical data can elucidate efforts to assess the potential for increased risks in a changing climate if properly accessed and used.

Overall, it is likely there might be some increases in mosquito-borne disease in Australia with a warming climate, but with which mosquitoes and which pathogens, and where and when, cannot be easily discerned. Additionally, the influence of climate on other components of transmission cycles (such as vertebrate hosts), and the impact of human lifestyle factors and public health services, must also be considered when contemplating the likelihood and consequences of increased contact between humans and mosquitoes and the pathogens they may carry.

ACKNOWLEDGEMENTS

This brief commentary presents a précis of my personal views on what is a very complex issue, which has generated many discussions with expert colleagues from around Australia, including Bart Currie, Brian Kay, Michael Lindsay, John Mackenzie, Scott Ritchie, John Walker and Peter Whelan and, inasmuch as what I have written here may reflect their personal views on different aspects of the issue as well as my own, I acknowledge their contributions to the debate.