Global Change and Human Vulnerability to Vector-Borne Diseases

The Earth's surface is kept warm by a blanket of greenhouse gases, including water vapor, CO₂ (261), methane, nitrous oxide, and some industrial gases like chlorofluorocarbons. Each gas has different warming properties depending on its structure and persistence in the atmosphere. The composition of the atmosphere has been changing over the past century, with the concentrations of these gases, except perhaps water vapor, rising as a result of the burgeoning anthropogenic activities (144). In particular, burning of fossil fuels for power and transport is transferring millions of tons of buried fossil fuels into the atmosphere each year in the form of CO₂. Livestock, rice paddies, and leakage from coal beds are also increasing the concentration of methane. There is less uncertainty about the extent of the increase in the concentrations of these gases than there is about the consequential changes in the climate. Nevertheless, changes in the concentration of CO₂ are subject to widely different estimates that depend on assumptions about future patterns of growth in the consumption of fossil fuels. Other anthropogenic changes such as urbanization, water supply and distribution, forest clearing, agriculture, and trade issues are more predictable.

This giant undesigned and untestable human experiment with the Earth's climate and atmosphere (it is the only one that we have) has already increased the atmospheric concentration of CO₂ by almost one-third (Fig. 3a) and doubled the concentration of methane in the atmosphere (Fig. 3b). None of these changes in atmospheric composition is known to affect vector-borne diseases directly. However, the effects of higher concentrations of CO₂ on plants are to reduce their water loss through transpiration and to act as a fertilizer. The result is that plants produce more foliage with the same amount of water (261), provided that they do not exhaust the supply of other nutrients. Two consequences of this phenomenon that are relevant to the current consideration of the effects on vector-borne diseases are that the increased density of plant foliage will provide more favorable microclimates for insect vectors and that plant growth seasons will be extended in some situations, effectively increasing the duration of favorable microclimates each year. A further consequence in some habitats may be a larger residual amount of water being left in the soil after maturation of a crop. This could affect water tables and the speed with which soils saturate and produce surface water reservoirs suitable for breeding of mosquitoes. It follows that the increased water use efficiency of plants at high CO₂ concentrations may result in an expansion of the ranges of woodlands into lower rainfall areas with adequate soil nutrients, leading to spatial changes in the ranges of vectors in response to habitat changes (95).

The issue of human-induced climate change is a contentious one because it is difficult to detect a slight trend in average temperatures when it is masked by a large amount of annual variability. However, in 1996 the IPCC, an international panel of scientists commissioned to advise governments on progress in the science that underpins the understanding of climate change, stated for the first time, that “the balance of scientific evidence suggests that human activities during the last century have begun to have a discernible effect on the world's climate, causing it to warm” (140). Projected changes in the average climate depend strongly on assumptions about the relationship between the global temperature and greenhouse gases produced by human activities (144). A statistical analysis to separate solar and anthropogenic influences implied that 60% of the warming in the last 140 years was due to human activity (27). A similar analysis showed that in recent decades, the temperatures have broken out above the band of variation that has been observed over the past 1,000 years and the rate of increase is accelerating (71). This image has been referred to within the global-change community as “the image of the millennium” and provides a sobering summary of the last 200 years of human impacts on Earth. Unfortunately, one consequence of this analysis is that future rates of anthropogenic warming are expected to be greater than previously estimated because earlier analyses overestimated historical contributions from anthropogenic activity. This contributed to an update of the future scenarios for climate change (144).

These projected changes rely on extrapolation of expected trends and do not consider possible triggers that may be set off as the Earth's atmosphere warms. One phenomenon that could have overwhelming effects is a failure of the circulation of warm water from the Gulf Stream into the North Atlantic Current, which forms part of a global salt conveyer belt. The current is driven by cold, saline water sinking in giant underwater waterfalls off Labrador and Greenland, which flush the cold water back into the warm Atlantic. Global warming could cause a failure in the flushing mechanism by diluting the saline water with bursting of ice dams from melting glaciers, by increasing rainfall in the region, or by reducing the temperature gradients in the sea. In the past, this has triggered very large (up to 10°C) and very rapid (a decade or century) “flips” in climate every few thousand years, with equally rapid reversals or “flops,” leading Calvin (52) to call them climate “flip-flops.” The effect is to put the world into a cold-dry climate, with Europe's climate equating to present-day Siberia, with devastating consequences in terms of food supplies for the world as a whole. This would reduce risks of almost all vector-borne diseases, except perhaps flea-borne diseases, among a starving population. An equally rapid warming event would have catastrophic implications for vector-borne diseases, with rapid geographical expansion of tropical ranges affecting nonimmune populations. Since any sudden cooling will greatly reduce the risks of vector-borne diseases and a subsequent warming would be well beyond the time horizon that is relevant to the present topic, these flip-flops are noted for their wider risks but are not considered in this review.

Projected changes in rainfall are even more uncertain than those for temperature, with large differences in the climate projections from different global climate models (GCMs). The consensus is that most tropical areas, particularly over oceans, will receive more rainfall, with decreases in most of the subtropics and relatively smaller increases in high latitudes (144). The uncertainty about rainfall is increased by the potential for geographical and seasonal shifts in rainfall patterns, so that local outcomes are difficult to foresee. Readers are cautioned to interpret the following projections of changes in the status of vector-borne diseases with climate change with care. They are indicative only and are intended to alert the community to some of the steps that can be taken to insure against any deterioration in public health caused by vector-borne diseases.

There have been numerous reports of physical and biological changes in the environment that are consistent with current warming of the Earth's surface. These include an increase in the altitude of the freezing point in the tropics by 110 m (83) and the melting of tropical glaciers around the world (321). A wide range of biological effects of recent climate change has been noted (142, 219, 243, 282), including population and life history changes, shifts in geographical ranges, changes in species composition in communities, and changes in the structure and functioning of ecosystems. In contrast to the tropical effects noted above, increases in the altitudinal range of five species of trees by 120 to 375 m, together with an increase in the tree limit by 100 to 150 m, have been noted at high latitudes in the Swedish Scandes (170). Such widespread biological changes were noted by Epstein (89), who cited them in support of the notion that recent climate change could have contributed to some current changes in patterns of vector-borne diseases. The difficulty faced by researchers is how to separate such effects from the many other simultaneous influences that are usually involved.

The human population has increased from approximately 1 billion at the turn of the 20th century to 6 billion by the end of the century, and it is projected to grow to around 10 billion by 2050 (323). The world's urban populations increased from 1.7 billion (39%) in 1980 to 2.7 billion (46%) in 1997 (12) and are expected to reach 5 billion (60%) by 2030 (14). Over the next 25 years, urban populations in Africa are expected to more than double, those in Asia will almost double, and those in Latin America and the Caribbean are expected to increase by almost 50% (351). The density of humans will increase disproportionately because a higher proportion of those people will live in cities compared with today (10).

Drainage and water supplies are critical factors that determine the extent to which many diseases are either contained or propagate in urban communities. Poverty associated with rapid population growth leads to concentrations of people without the necessary infrastructure for the safe storage and distribution of water and drainage of wastewater. Used containers and tires provide breeding sites for mosquito vectors. In addition, the deteriorating public health infrastructure in many countries exacerbates the health problems (121). Each of these negative societal trends is expected to continue, and water-related issues in particular are expected to increase in importance in the developing world over the next few decades (352). On the other hand, in wealthy communities there is encroachment of residential or recreational populations into forested areas where natural hosts and vectors of vector-borne diseases exist.

The burgeoning world population (326) is driving an intensification of agriculture and its spread into new areas where forests now exist. With such development comes the need for deforestation and for the storage, supply, and distribution of water for human consumption and for irrigation. World rates of deforestation in 1990 to 1995 averaged 101,724/km², or 0.3% per annum (12). Future projections of agricultural expansion indicate that there will be more dependence on intensive agriculture and use of irrigation where water supplies permit, with the associated habitat modification, water storage, and sanitation issues (94, 96, 118, 119, 139, 173, 323). Each of these developments affects the opportunities available for breeding by vectors of diseases.

Future land cover change will occur mostly in the tropics and subtropics. It is likely to result in increases in surface temperatures of up to 2°C, with drier conditions where the land cover is reduced. In contrast to the reduced diurnal temperature range expected with greenhouse warming, future land cover change in the tropics may increase the temperature range by decreasing evaporative cooling during the day. The sensitivity of surface temperatures to future anthropogenic land cover change in the tropics is up to 1.5°C warmer than the range induced by decadal-scale interannual variability in vegetation density (80).

With changes in land use comes fragmentation of habitats, loss of biodiversity and alteration of existing vector-host-parasite relationships. The rate of species loss is now higher than at any time since the period when the dinosaurs went extinct (175). Fragmentation of habitats isolates populations with low mobility but still provides access to mobile species, thus altering the species balance in undisturbed areas. This can lead to changes in the physical environment for vectors and hence affect patterns of disease.

The burgeoning industrial and agricultural chemical industries are producing thousands of new products each year. Very few are assessed for health risks, and many are being found to either mimic or disrupt the activity of hormones, especially estrogen and thyroid hormones, in humans and animals (60). Such endocrine-disrupting chemicals (EDCs) include alkyl phenols, dioxins, organochlorine pesticides, phthalates, polychlorinated biphenyls, polybrominated diphenyl ethers, and synthetic pyrethroids (167). They are found in plastics, herbicides, and pesticides that are distributed widely around the world. Apart from the potentially devastating effects on human sexual development and neurological function, there is evidence that many such chemicals also degrade immunological function (22, 69). From the perspective of risk assessments and prevention, an even greater concern is that such effects are sometimes associated only with mixtures with each other or with nitrate fertilizers in groundwater at concentrations found in groundwater across the United States (255). The side effects of these chemicals could reduce options for human adaptation to vector-borne diseases either by degrading immune responses or by withdrawal of vector control products from the market.

Increasing affluence in the developed world, combined with more economical mass transport and liberalization of international trade, is accelerating the increase in the numbers of people and the quantity of materials that are being transported around the world. Humans travel on scales from local to global. One million people are reported to travel internationally each day, and one million people travel from developed to developing countries (and vice versa) each week (106). A more recent report quoted a figure of 700 million tourist arrivals per year (115).

There were more than 14.5 million refugees worldwide at the end of 2000, 1 million more than 2 years previously (Fig. 5). From the perspective of vector-borne diseases, the most significant problems are likely to occur in Africa, with more than 3.3 million refugees, South and Central Asia, with 2.6 million, East Asia and the Pacific, with 0.8 million, and the Americas, with 0.6 million (13). In addition, a similar number of people were internally displaced within countries, usually from rural areas into cities.

Each of these types of movements has the potential to spread disease pathogens and their vectors over long distances.

International trade in merchandise has increased three- to fourfold over the period from 1980 to 2000, with most of the increase occurring in Asia, where there was a fivefold increase in the value of exports (World Trade Organization, http://www.wto.org/english/res_e/statis_e/webpub_e.xls ) (Fig. 6). Inevitably, this must lead to an increase in the incidence of vectors being transported to other countries.

In this review, the risks and opportunities arising from each of these drivers of global change in relation to the vector-borne diseases are discussed, both in isolation and in combination when appropriate. The potential environmental effects of each global change driver and their additive or interactive combinations that could affect vector-borne diseases are summarized in Table 1. The effects are proximal causes of changes in transmission rates of vector-borne diseases. As can be seen in the second column of Table 1, combinations of global change drivers tend to increase the suitability of the environment compared with the effects of each driver separately. These include combinations of climate change, increased trade and travel, increased urbanization in poor countries, and intensification of agriculture with irrigation, deforestation, and loss of biodiversity. Few interactions reduce the risks, but any that do are likely to be related to the availability of surface water for breeding of mosquitoes under any regional drying due to changed rainfall patterns and the intensification of the hydrological cycle.

Before impacts can be reviewed, it is necessary to establish a framework for assessing the risks.

A change in the geographical distribution of a vector-borne disease has a quantal effect on the exposure of naive hosts to that disease. Such a change can come about from the movement of either a vector or pathogen to a new environment through trade, human movement, or natural means. Alternatively, it can result from climate change allowing a vector or pathogen to shift its geographical range into environments that become more suitable.

At the global level, assessment of the risks associated with different sources, pathways, and destinations of vectors and pathogens can be assessed by following the quarantine procedures that operate under international “pest risk analysis” agreements for plant health (146). Linking of global change risk assessment approaches with those used by the quarantine community has been proposed (18, 305). The procedure attempts to identify and manage risks by targeting weaknesses at the source of an infection, along the transport pathway and at the destination (Fig. 7). We will see below, when considering adaptation options, that it is more difficult to manage risks when humans are the host of a vector-borne disease than when plants or animals are involved.

In the case of introduction of an exotic vector or pathogen from another continent or its spread to an area from an adjoining zone of endemicity, it is necessary to establish whether the pathogen or vector is able to persist in that new environment. The local habitat forms the template on which ecological processes operate (298). In a risk assessment, the first feature of the habitat that is usually considered is climate, partly for pragmatic reasons—meteorological data are more readily available than other types of data—but other features need to be included systematically. In the context of detection of risks of new exposure, the question is whether the species will be able to persist. This requires analytical tools with which to estimate the response of the species to the new environmental conditions. These are discussed below as part of a review of impact assessment tools. The next question relates to the sensitivity of the local population to the disease and that requires measures of the abundance of the vector and pathogen in the habitat on one hand and the immune status of the host population on the other.

Sensitivity is the degree to which a system responds to an external perturbation, such as a change of temperature (144). It is essential to define clearly which attribute of a system is being used as the measure of sensitivity because different parts of any system can respond by different amounts. For example, for vector-borne diseases, the potential transmission rate may be very sensitive to a temperature change but the incidence of clinical disease may not alter if the host population is already immune. In this review, when referring to sensitivity of vector-borne diseases, we are referring to potential transmission rates. Host immunity is treated as an adaptation to infection.

Before proceeding further, it is also necessary to appreciate the nonlinear nature of many biological responses to changes in the environment. These arise from a number of features of biological systems, including thresholds such as developmental or behavioral temperature thresholds, discontinuities at the edges of the ranges of species, nonlinear responses to temperature and moisture, multiplicative effects of population growth in vectors with multiple generations each year, negative feedback associated with competition or predation as population densities increase, interactions between variables resulting in nonadditive effects, and the disproportionate effects of changes in the frequency of extreme events with small changes in the value of the mean. Awareness of such nonlinear effects helps to prevent surprises and so leads to more sustainable adaptation to global change. One consequence of these behaviors is the need to augment empirical and descriptive approaches, such as statistical models, with mechanistic computer simulation models (307).

The sensitivity of a human population to a given disease under global change depends on the combined responses of the pathogen, vector, and host populations. This combination has the potential to generate significant complexity. The geographical location is also a key determinant of the sensitivity of a species to environmental change. A change in the suitability of the environment within the current geographical distribution of the disease will alter the development, survival, and reproductive rates of vectors and pathogens and so affect the intensity of disease transmission and resultant exposure of the population to the disease. The extent to which the exposure changes, following a new introduction of a vector or pathogen or a change in the density of an endemic vector with climate change, depends on the position of the particular habitat relative to the potential geographical distribution of the vector or pathogen in relation to climate (Fig. 8).

Computer models help to identify places where changes in the values of climatic variables are in the sensitive range of a pathogen or vector (307). This is illustrated schematically by a “climatic envelope” in Fig. 8, which shows why populations at different locations within the geographical distribution of a species respond differently to a given change in the value of a climatic variable. Populations within the core of the distribution, A, in Fig. 8, defined as the zone with minimal stresses on the species (314), are relatively insensitive to a given change in temperature or moisture because the change occurs in a part of the response curve that is relatively flat. In contrast, populations in marginal parts of the distribution, B, occur where response curves are likely to be steeper and thresholds exist for some physiological processes and so are more responsive to an equal degree of change. Additionally, at or near the edges of the distribution, there is great potential for high variability in the occurrence of limiting conditions because a given change in conditions can flip a population between persistence and extinction very readily. Thus, it is evident that the underlying pressures driving the transmission dynamics of vector-borne diseases will vary more around the edges of the area of endemicity in response to climatic variability (9, 67).

Recognition of variation in the susceptibility of humans to vector-borne diseases, based either on genetic (e.g., sickle cells and malaria) or acquired immunity developed in response to exposure, is needed. The concept of herd immunity leading to endemic stability of a disease is crucial to understanding the dynamics of diseases, the likely susceptibility of populations to such diseases, and the likely consequences of interventions to reduce transmission rates (296). Since the nature of clinical symptoms differs with different transmission rates, it is essential to understand the relationships between pathogen infection, morbidity, and disease outcomes in order to plan interventions that avoid undesirable consequences. These relationships vary greatly with different pathogens, and so a case-by-case approach is necessary.

The sensitivity of a human population to a change in exposure depends on the immune status of the population. Diseases affect human health most severely when there is initial contact resulting from humans entering new habitats, from a spread of disease organisms and their vectors, or from temporary surges in transmission rates during abnormal seasonal conditions, for example. In such cases, naive, nonimmune populations are especially susceptible to epidemics of acute disease.

The rate of disease transmission leads to different disease patterns, such as that for pathogens to which humans can acquire immunity, like malaria (Fig. 9). On a spatial scale, the graph represents a cross-section of Fig. 8. At high transmission rates, a condition known as endemic stability is created, whereby most hosts are immune, creating herd immunity (C). Around the edges of the area of endemicity there are areas with intermittent epidemics of disease (B), which are referred to as endemically unstable areas. These frequently cause the greatest concern because they involve nonimmune hosts that are particularly susceptible to infection in a high-risk environment. The disease is absent from areas (A) that do not provide suitable environmental or socioeconomic conditions for transmission. These areas may be too cold or too dry for the pathogen or vector to develop and survive during the unfavorable season, or they may have human living conditions, such as insect screens or air conditioning, that prevent contact with the vectors. In areas with very high transmission rates, the hosts may succumb to acute forms of the disease if their health is in any way impeded, creating an “overload” condition (D). In the case of dengue, subsequent infection with a different serotype greatly increases this latter risk by causing additional clinical effects with hemorrhagic symptoms (153).

Once we have defined the extent to which a population is likely to be exposed to a vector-borne disease under global change and have also assessed its sensitivity, we are able to derive a measure of the likely impact on the population, in the absence of any intervention to manage that risk as part of an adaptation strategy. This provides us with a baseline measure of risk against which we can estimate the benefits of different options to adapt to the risks.

Two approaches have been recommended for analyses of likely impacts of climate change used by the IPCC (244). These are referred to as the top-down and bottom-up approaches. Many authors have used climate change scenarios generated by GCMs as a means of investigating likely impacts of vector-borne diseases (147, 208, 209, 212, 280). The alternative approach relies on sensitivity analyses of a range of climatic variables, and was preferred by Sutherst (304, 307) on the grounds that the climate change scenarios are too immature and changeable to be of lasting value. The use of sensitivity analyses covering the broadest range identified by the global climate modelers avoids the problem of rapid dating that occurs with scenarios. On the other hand, when calculating the net present value of future costs, it is necessary to attach a date to the sensitivity analyses so that discounting can be applied. The two approaches have been combined into a risk management approach (151).

Climate change and trade and transport in particular have the potential to affect the geographical distribution of vectors. While climate change alters the immediate environment of a vector, translocation to a new region presents similar and probably greater changes for vectors. Such changes need to be studied by using geographical-scale approaches. Sutherst (305) presented a conceptual framework for studying the effects of climate on the distribution and abundance of species and how they are affected by climate. As shown in Fig. 8, populations respond differently to a given change in a climatic variable depending on where they are situated in relation to their climatic envelope.

Integrated assessment frameworks to bring some of these elements together are under development (53, 205, 248) but do not yet incorporate local environmental circumstances (220). A framework for analyzing impacts on parasites has been described (307) based on theoretical ecology and the framework outlined above. The adoption of political ecology as a framework for analysis of emerging infectious diseases has also been advocated (217). The discipline incorporates social, economic, environmental, and biological components to present a holistic approach. Each of these approaches needs a set of analytical tools.

Tools for risk assessment under global change have been reviewed elsewhere (220, 312). A wide range of approaches have been applied to the assessment of risks for humans to vector-borne diseases under global change. They have ranged from the use of historical analogues (121, 267, 269, 270), and geographical analogues (264) to a variety of statistical (193, 194, 280) and modeling (189, 205, 210, 212, 248, 304, 307, 312, 315) tool and models combined with fuzzy logic or rules (68). Historical events are helpful in providing pointers to potential causes of changes in status of diseases (269) but usually suffer from a lack of firm data, sometimes leading to uncertainty in interpretation.

Each technique has limitations, but the most important considerations are the need to apply basic scientific principles (55, 304, 311). Using independent data to test the resultant models, complying with the limitations of chosen analytical or descriptive tools and data sets, and avoiding parameterization of models based on data sets with very narrow ranges of variation can achieve compliance. Often, apparently accurate statistical models fail to demonstrate predictive ability when tested against independent data rather than data derived by dividing a set of data and using one set for fitting and the other for testing the models (262, 278, 279). In addition, popular statistical models, like logistic regression and discriminant functions, rely on pattern matching with meteorological data and so cannot cope with novel climates or with extrapolation to other locations with different patterns of temperature and rainfall, even though they may describe the current range of a species quite accurately. Hence, such descriptive techniques are suited only to answering questions that involve small incremental changes in conditions within the existing ranges.

The ranges and relative abundance of two important African malaria vectors, Anopheles gambiae sensu stricto and A. arabiensis, have been related to temperature and a ratio of potential evaporation to rainfall as a measure of moisture availability (193). The population at risk from lymphatic filariasis in Africa was also related to these climatic variables (194). Using a combined model and rule-based approach, thresholds for temperature and rainfall have been inferred by using observations from areas of Africa with different malaria transmission patterns (68). The CLIMEX model (304, 311, 313) incorporates a hydrological model to describe the availability of moisture and so accounts for the effects of changes in temperature, rainfall, and evaporation when assessing risks from climate change. It automatically takes any temperature-moisture interactions into account when provided with appropriate meteorological data that include relative humidity or equivalent readings. CLIMEX has been used widely for risk assessments under climate change (303, 304, 307). The malaria transmission factors, biting and entomological inoculation rates, were predicted in Kenya using a soil moisture model (250). The model substantially improved the prediction of biting rates compared to rainfall, explaining up to 45% of A. gambiae biting variability and 32% of the variability of A. funestus when given different time lags.

Each species of vector has characteristic climatic requirements (194, 304) and vector competence for a given biotype of parasite (344) (29). This flags the need to keep each element of the disease triangle (Fig. 1) in mind because the climatic requirements of each vector-pathogen combination needs be taken into account in order to develop a realistic measure of risk (48, 304). Incorporation of species-specific vectorial capacity into global risk assessments has started (208). Integrated assessment frameworks are being developed to bring some of these elements together, with applications to malaria, dengue, and schistosomiasis (53, 205, 220, 248, 315). The use of spatial tools, such as geographical information systems and landscape ecology, in studies of vector management and global change effects on vector-borne diseases has been advocated (43, 97, 159).

To detect impacts of global change, we need monitoring data for a number of environmental and disease related variables covering long time series. While historical records of disease incidence provide a valuable basis for detecting a change in transmission patterns, we also need to monitor concurrent environmental and social conditions. This provides benchmarks against which to measure the likely impacts of changes in any given variable.

Temperature is one such variable that varies systematically with latitude and altitude. The effect of altitude on temperatures is approximately equivalent to 5.7°C/km of elevation. The expected altitudinal range shift with increasing temperatures can be calculated using the formula (183)where T is the screen temperature at height h (meters) and T_c is the equivalent at sea level. Thus, ignoring regional variations in sensitivity to global warming, for each 1°C increase in the global temperature there will be a potential increase of ∼170 m in the elevation of a given transmission rate of vector-borne diseases. Warming over the past century has been ∼0.6°C (144). Thus, reported changes must be within the range of 100 m to have credibility. For example, one credible report is of an increase in height of the freezing point in the tropics of ∼110 m (83).

Similarly, expected latitudinal shifts in species ranges with global warming can be expected to be broadly consistent with effects of latitude on temperatures. The approximate formula (183)where T_c is the long-term sea level temperature at latitude A, describes the difference in temperature between that latitude and the equatorial temperature. Thus, on average, there is an approximate difference in temperature of 0.6°C per degree of latitude. Hence, range shifts of about 1.7° of latitude or 200 km can be expected for each 1°C increase in global temperatures. Again, a rise of 0.6°C in the last century can be expected to have been accompanied by range shifts of ∼118 km. These figures provide a benchmark against which to relate reported changes in the distribution of vector-borne diseases with respect to global warming, but they need to be related to local modifying factors that create different microclimates and so change the actual area at risk.

Similar benchmarks need to be provided for all other environmental and social variables, with rigorous interpretation of historical events that have often been poorly documented or misinterpreted.

Climate change has the potential to change the intensity of transmission of vector-borne diseases in addition to altering the exposure to the diseases by shifting the geographical distributions, as shown above. The extent to which the disease is sensitive to changes in transmission rates depends on a number of variables, including the responses of vector and pathogen to changes in the particular range of temperature or moisture concerned, and the immune status of the host population. Ideally, it would be useful to have a readily measurable benchmark against which to assess reported claims of involvement by climate in observed changes in the status of established vector-borne diseases. Unfortunately, the species-specific nature of the transmission dynamics of each pathogen renders that more difficult than the task of benchmarking changes in geographical distributions. However, it is still possible to match geographical locations with similar intensity of transmission by using models such as CLIMEX. Alternatively, mathematical models of the biological processes in the transmission cycle (also called mechanistic models) can be used to infer the sensitivity of the transmission rates of vector-borne diseases to changes in climate (189, 210, 212, 307). The major advantage of dynamic models over nonmathematical approaches or statistical models is that they are able to detect surprises or so-called emergent properties of systems that arise from discontinuities, such as thresholds or nonlinearities in biological processes.

An indirect effect of increasing concentrations of CO₂ in the atmosphere is that the water use efficiency of plants is increased. Under good climatic and nutritional conditions, this will result in larger plants that provide more humid shelter for insect vectors and for plant pathogens. However, since plant growth under field conditions is often limited by factors other than water, such as nutrients, there is likely to be a generally higher water table and soil moisture content than occurs at present, unless global warming results in reduced rainfall or substantially increased evaporation (261). Such conditions are conducive to an increase in the frequency of pools of open water that provide suitable habitat for mosquito breeding. This would have consequences for both freshwater- and fouled-water-breeding mosquitoes and the diseases that they transmit.

Climate change has the potential to alter the average exposure of human populations to vector-borne diseases by changing the geographical distribution of conditions that are suitable for the vectors and disease pathogens. An increase in global temperatures will result in an expansion of warm temperature regimens into higher altitudes and latitudes. Any associated changes in rainfall in tropical and subtropical zones will also render habitats more or less suitable for vectors. In addition, the implications of the asymmetrical increases of temperature with global warming (154) for the epidemiology of vector-borne diseases need to be clarified. Greater effects can be expected from the extended relaxation of limiting effects of low temperatures on vector survival, behavior, and disease transmission in cold-limited climates than from smaller and less frequent increases in extreme maximum temperatures. These changes would make temperate environments more receptive to many tropical vector-borne diseases while having less negative effects on tropical environments.

Extreme climatic events have major effects on the transmission rates of vector-borne diseases. In the light of expectations that climate change will increase the frequency of such events disproportionably, such extreme events may emerge as a more important feature of climate change than are changes in average climatic conditions. They are therefore considered in more detail below.

There have been a large number of studies and reviews of the sensitivity of vector-borne diseases to climate change (37, 40, 49, 50, 61, 62, 82, 84, 88-90, 110, 121, 123, 125, 126, 134, 164-166, 169, 186-189, 191, 192, 206, 209, 215, 220-223, 225, 228, 230, 245, 246, 249, 264, 266, 267, 269, 270, 295, 303, 307, 309, 310, 335, 350). The results have led to quite different perceptions of the role of climatic change and other factors in historical patterns of disease incidence.

The coherent pattern of the retreat of tropical glaciers, an upward shift in the freezing isotherm in the tropics, increases or decreases in the geographical ranges of temperate or Arctic species, respectively, at higher latitudes, earlier spring migration and breeding by birds, and earlier seasonal activity of insects have been cited as examples of impacts of gradual global warming (90, 125, 126, 243, 282). The observations were claimed to be consistent with increasing global temperatures and with model predictions. Taken together, these phenomena provide strong evidence that climatic changes in recent decades are already affecting up to 50% of the species examined in a survey of the literature (243, 282). There is no reason to believe that vector-borne diseases are exceptions to this experience of small and gradual changes in seasonal activity and expansion of ranges to higher altitudes and latitudes. The issue for workers in the field is to establish adequate baseline data on seasonal transmission patterns, prevalence of disease and geographical distributions, benchmarks to monitor and assess the consistency of changes with known physiological processes, and sufficiently accurate monitoring data in strategic locations to be able to detect the changes as they occur around species range boundaries initially. Only then will the medical community be able to separate the subtle effects of climate change on vector-borne diseases from the more obvious effects of other factors.

A number of authors have raised the possibility that global warming may have played some part in the recent range expansions and outbreaks of vector-borne diseases (37, 40, 89, 90, 125, 163, 186, 188, 197). Two types of climatic effects could theoretically have been involved. First, changes like the ones observed for other species of plants and animals referred to above will gradually increase the transmission rates in cooler climates and so extend transmission into previously disease-free areas while slowing transmission in areas that become too hot. In some cases, realization of this expectation with vector-borne diseases will be different for other species because often the vectors are already there but the pathogens have been eliminated. Second, changes in the intensity of extreme climatic events will alter the patterns of epidemics. Any change in the intensity and frequency of extreme climatic events may take decades to detect against a background of insufficiently long historical records and high climatic variability with ENSO-like events, and it is probably too early to invoke this effect in any historical events.

Other authors (34, 121, 136, 266, 270, 294), citing historical outbreaks and identifying the largest current signals in a set of data, concluded that other factors were more important than climate and questioned the veracity of claims (90, 197, 206, 223) that anthropogenic climate change could have contributed to the epidemics. For example, it has been suggested (34) that replacement of forest by agriculture provided new habitats for breeding of mosquitoes at Usambara in Tanzania, rather than local temperatures rising as a result of land clearing, as was previously suggested (214). The role of climate change in an outbreak of malaria in Rwanda (197) was questioned because it coincided with a change in detection methods that was more likely to explain the jump in reported incidence of infections (270). Opposing interpretations of historical temperature data in East Africa, based on interpolated meteorological data (235), led to different interpretations of the role of global warming in the observed increase in incidence of malaria in recent decades (136, 247; J. A. Patz, M. Hulme, C. Rosenzweig, T. D. Mitchell, R. A. Goldberg, A. K. Githeko, S. Lele, A. J. McMichael, and D. Le Sueur, Letter, Nature 420:627-628, 2002). A negligible role was seen for global warming in the resurgence of vector-borne disease in Latin America, Africa, and Asia in the past two decades (121, 266, 270, 294). It was claimed to be inappropriate to use climate-based models to predict future prevalence because climate plays a small role in disease outbreaks (270). That view does not take sufficient account of the role of climate in determining the underlying seasonal phenology or geographical distributions of vectors and pathogens. Neither does it recognize the contribution of climate-based modeling in allowing researchers to extrapolate results from one geographical location to another. What has been missing from the global change studies is adequate inclusion of nonclimatic variables in the analyses, which reflects the difficulty in building and maintaining global databases of local environmental changes.

In a more pragmatic view, global warming was considered to be unlikely to cause major epidemics of tropical mosquito-borne disease in the United States (121) and Australia (48) as long as the public health infrastructure and living conditions remain the same. It is not surprising that the role of climate change in historical events has been contentious because the extent of the changes is only just beginning to be large enough to be distinguished from natural variability and the events were not sufficiently well documented to ensure that all variables could be accounted for. Despite the differences in emphasis and focus on historical or future events, all authors agree that there are multifactorial causes of change in the incidence of vector-borne diseases. Land use events will be more important in the short term, but climate change has the potential to be important in the longer term.

Climate change will affect both the invertebrate vectors and the development of pathogens in those vectors. Basic biological considerations indicate that with global warming, the duration of the growth season will increase, allowing more generations of vectors each year in cooler areas. Development is prevented at low temperatures, but as temperatures rise, a race develops between parasite development and accelerated mortality of the vector. The winters will be shorter and less severe and so will reduce the mortality rates of species of vectors that are currently limited by low temperatures. Other temperate species of vectors are insensitive to winter conditions in some environments (270). In some environments, temperatures will rise to levels where the mortality rates of the vectors are so high that they die before being able to transmit a pathogen. Changes in future moisture regimens are much more uncertain.

The higher human population densities and lack of the necessary urban infrastructure in tropical regions in particular have profound effects on the transmission potential of diseases (123, 221). Dense urban development with poor infrastructure is widespread in the developing world and leads to increases in the incidence of human diseases that need large human populations to persist and wastewater for the vectors to breed (123). The deteriorating public health infrastructure in many countries exacerbates the problem. Vectors of dengue fever and yellow fever are able to exploit artificial sources of water such as water storage pots, tires, or old containers in garbage. Polluted wastewater also provides a suitable breeding environment for Culex mosquitoes that transmit lymphatic filariasis (199) and arboviruses such as SLE virus in the United States and Rift Valley fever virus in Africa (292). Meanwhile, rural-urban migration is credited with leading to a new pattern of infection referred to as urban schistosomiasis (329). On the other hand, the reduction in the incidence of Chagas' disease with urbanization has been alluded to above in relation to climate change.

The effects of environmental change on emerging parasitic diseases have been reviewed recently (249). Examples were cited of upsurges in malaria with deforestation in Africa, Asia, and Latin America, where the diversity of vector species ensures that there is continuing transmission, because habitat changes favor different species of mosquitoes. The types of vegetation and ground cover determine the vector species that occurs following deforestation, with natural and human-built water storage facilities playing a major role. Agriculture has pervasive local effects on vector-borne diseases by affecting the availability of breeding sites for different species of vectors (229). A number of land use issues that affect the incidence of vector-borne diseases have been identified, including deforestation, land cultivation, and various water storage, distribution, and irrigation structures and practices (64, 105, 229, 231, 303). They can be expected to continue to affect the future patterns of those diseases as the land use changes accelerate. For example, clearing of cattail marshes in Africa for cultivation removes breeding sites for Anopheles funestrus, but cultivation of papyrus swamps and deforestation provide open, sunlit water that is suitable for breeding by A. gambiensis (231). Cultivation of irrigated rice can either increase the incidence of malaria greatly or have no effect when the vector is A. funestrus in mainland Africa (66, 174). Similar differences in the responses of different species of mosquito to irrigation were observed with malaria vectors in Sri Lanka (7). Forecasting the consequences of changes in environmental management requires the involvement of multidisciplinary teams in the planning and implementation of the projects (230).

Deforestation combined with industrial, residential, and agricultural development in the Amazon has had widespread effects on the species of anopheline mosquitoes present and on the rates of transmission of malaria (316). The most significant effects were altered breeding grounds and shelter for the mosquitoes, which changed the intensity of contact with humans. Notable was the great diversity of effects in different regions; however, changes that favoured Anopheles darlingi, the most efficient malaria vector, were the most important.

The implications of a number of environmental changes associated with rural development have been investigated. Forest penetration has involved humans in viral and leishmanial zoonoses and has introduced Onchocerca volvulus into Central and South American forests. Clearing of forests for agricultural developments such as rubber plantations in Malaysia has resulted in increases in malaria, and there has been a spread of the snail vectors of schistosomiasis into forests. In Indonesia the incidence of lymphatic filariasis was reduced by forest clearing and development, which exposed the mosquito breeding sites to sunlight (182). Similarly, clearing is used to control malaria in Asia by reducing the breeding of Anopheles balabacensis (229). However, deforestation can increase the breeding of sun-loving Anopheles vectors of malaria in Africa, as mentioned above (229).

An “edge effect” changed the patterns of transmission of trypanosomiasis and loiasis in West Africa. Occupation of land, with the resultant change in vegetation cover, affected the transmission dynamics of triatomine vectors of T. cruzi, phlebotomine transmission of leishmaniasis, and tsetse fly transmission of T. rhodesiense (64). Each of these effects are destined to continue as deforestation continues apace around the tropics.

Water storage reservoirs, from jars to dams, and irrigation practices provide breeding sites for mosquito vectors of several viral and protozoan diseases including malaria, while large, still water bodies are particularly well suited to the breeding of snail vectors of schistosomiasis. There has been a changing incidence and geographical distribution of schistosomiasis, partly due to human migration, new irrigation and water impoundments, particularly in Africa, and successful eradication in some countries (30, 56). Ecological changes resulting from the building of the Aswan Dam in Egypt have resulted in a shift from predominantly S. haematobium to S. mansoni. There was a massive outbreak of S. mansoni in northern Senegal, due to intense transmission, after the construction of the Diama dam on the Senegal river and the Manantali dam on the Bafing river, Mali (297). The Three Gorges Dam in southern China is expected to increase the transmission of S. mansoni over a vast area by preventing annual flushing of the snail vectors (180).

Some of the effects of various water-related activities on vector-borne diseases have been summarized (229). They include reduction in the incidence of dracunculiasis by controlling the Cyclops intermediate hosts in drinking water; artificial lakes that drown the breeding habitats of Simulium spp. (vectors of onchocerciasis) while creating new habitat for Anopheles vectors of malaria and filariasis and snail hosts (Biomphalaria and Bulinus spp.) of schistosomiasis; dam spillways providing larval habitats for Simulium spp., resulting in artificial foci of onchocerciasis in West Africa; heavily vegetated irrigation canals that provide habitats for Anopheles spp. and snails; and irrigated rice fields that provide habitats for Anopheles and Culex spp., including C. tritaeniorhynchus, an important vector of Japanese encephalitis virus.

A complex relationship exists between irrigation and malaria incidence in Africa (143). In environments with small populations of mosquitoes, increased numbers of vectors following the introduction of irrigation can increase the incidence of malaria in areas of unstable transmission, such as the African highlands and on the fringes of deserts. However, irrigation has little impact on malaria transmission for most of sub-Saharan Africa, where it is stable. The lower incidence of malaria in some communities with irrigation than in surrounding areas was explained by a combination of changes in the species composition of the mosquitoes and better protective measures made possible by increased wealth of the community.

Irrigation in two large-scale agricultural areas of western Kenya facilitated the transmission of malaria throughout the year by providing habitats for the two main vectors (111). At Ahero, Anopheles arabiensis was most abundant when the rice crop was immature, followed by A. funestus when the crop was mature. At Miwani, populations of A. gambiae peaked during the long rains whereas the proportion of A. arabiensis was greatest during the dry season.

Malaria incidence increased in the Usambara Mountains in Tanzania, following forest clearing to make way for tea plantations and later for agriculture (34, 214). Immigration of malaria-infected laborers from the surrounding lowlands spread the disease in the highlands. Interpretation of these historical changes in the incidence of malaria around Amani has been the subject of continuing debate, and the data have been extensively quoted as an example of the effects of higher local temperatures, resulting from land clearing, on the transmission of malaria. Reiter (270) favored an explanation based on the social changes associated with immigration of infected workers. Reports of the history of malaria in the area have recently been reinterpreted (34). The earlier increases in incidence were explained by intensive agriculture providing more breeding sites for the mosquito vectors that were previously uncommon in this highland habitat. A recent resurgence of malaria in the area was attributed to widespread development of resistance of the parasites to chloroquine. The same explanation was provided by Shanks et al. (294) for the dramatic increases in malaria cases since 1965 at tea plantations in the western highlands of Kenya.

The effect of land use change on malaria transmission was investigated at Kabale in the southwestern highlands of Uganda from December 1997 to July 1998 (185). Mosquito density, biting rates, sporozoite rates, and entomological inoculation rates were compared between eight villages near natural papyrus swamps and eight villages near swamps that had been drained and cultivated. Associated microclimatic conditions were also compared. There was a nonsignificant tendency for all malaria indices to be higher near the cultivated swamps. The numbers of A. gambiae sensu lato were associated with higher temperatures near the cultivated swamps, which resulted from the change in land cover.

The increasing incidence of malaria in Africa has been attributed to the growing reliance on corn as a staple food crop (356). In the absence of corn pollen, the silt-laden puddles of water, in which the A. arabiensis mosquitoes breed, are very low in nutrients. Corn pollen provides a rich and widespread source of nutrients for mosquito larvae. This could contribute significantly to increasing vector population growth during the flowering season of the crop, which corresponds to the increase in the numbers of mosquitoes in Ethiopia. It was suggested that the force of transmission of malaria in sub-Saharan Africa might be reduced if maize plantings were removed from the immediate vicinity of homes, as apparently was required in Rhodesia some decades ago.

Lyme disease in the northeastern United States and tick-borne encephalitis in Europe are increasing problems as human settlements encroach on forested areas (123). Emergence of Lyme disease, caused by the spirochete Borrelia burgdorferi, has been attributed partly to reforestation of outer suburban and agricultural areas in the northeastern United States (21). There were concurrent moves to conserve native fauna, particularly the white-tailed deer, Odocoileus virginianus, which are suitable hosts for adults of the tick vector, Ixodes dammini (21). White-footed mice, Peromyscus leucopus, provide suitable hosts for burgeoning populations of the tick larvae (100), which are efficient vectors of B. burgdorferi (172). However, habitat fragmentation also reduces biodiversity and so leads to a preponderance of these mice. The disease “dilution effect,” from the tick larvae feeding on less competent hosts, is lost, with the result that a large proportion of the larval population become infected (240). Infected nymphs then pose the major threat to humans. The ticks have recently been found to transmit other bacteria that may cause similar symptoms, such as Babesia microti and unnamed Borrelia, Ehrlichia, and Bartonella species (91, 337).

As just observed, natural biodiversity of hosts can reduce the transmission of vector-borne diseases, and there is also substantial information indicating the importance of natural enemies in controlling vectors of disease (57). This topic has been studied widely, and biological agents have been exploited for biological control (176, 285). Most of the effort has been directed against aquatic Diptera. The natural-enemy component was shown to be responsible for significant population reduction and to be indispensable to integrated control approaches. Outbreaks of malaria in Venezuela 1 year after ENSO-related droughts may have been caused by the death of the natural enemies of the mosquito vectors during the droughts (36).

This sampling of the large literature on the effects of land use on vector-borne diseases gives an indication of the diversity of changes and of the wide range of effects that they have on a number of vector-borne diseases. It also illustrates the difficulty facing researchers in trying to extract generalizations from such disparate observations and to create global databases in order to predict future global trends and impacts.

It would be remise not to at least mention the risk that is posed to human health in general, and threats from vector-borne diseases in particular, by the global contamination of even the most remote habitats with EDCs. Of particular concern in the present context is the listing of synthetic pyrethroids as active EDCs (113) because they are being used so widely in agriculture and in impregnated bednets to prevent malaria. At present there have been no records of reduced immunity involving vector-borne diseases (22, 76), but observations of the impairment of the immune system by currently recorded groundwater concentrations of mixtures of agricultural chemicals across the United States is a warning sign (255). Health and safety issues receive less attention in developing countries, and such pollution with industrial and agricultural chemicals is widespread. The traditional reliance on natural immunity to diseases such as malaria and schistosomiasis in the developing world may be under threat. The paramount importance of herd immunity to malaria, in particular, in protecting populations from acute disease makes any such decline of immunity dangerous.

There are a number of different dimensions to the movement of people and materials around the world that affect the distribution and incidence of vector-borne diseases. People can either act as carriers of pathogens into new environments or accidentally translocate vectors in transport vehicles. People can also become victims of vector-borne diseases when they travel to new countries where they are exposed to diseases for the first time. Such people are usually naive to the disease, and thus morbidity and mortality rates are relatively high. Some examples of past incidents are listed as an indication of the potential for future spread of vector-borne diseases as trade and travel increase with globalization of trade and increasing wealth.

Movements of people, materials, or vehicles have been responsible for short- and long-distance transfer of several disease vectors (109) (Table 4). Aedes albopictus, a vector of dengue, was transported from Japan into the Americas and Europe (135, 161, 287), Ochlerotatus japonicus, a vector of arbovirus, was carried from Japan into the United States (253), and Ochlerotatus camptorhynchus, a known vector of Ross River virus, was moved from Australia into New Zealand (156). The worldwide migration of insecticide resistance genotypes of mosquitoes is an excellent example of the extent to which vectors have been spread around the world (263).

The global spread of vector-borne diseases is being driven not only by human activities but also by natural forces. Vector-borne diseases have been spreading for centuries, and the shifts in their geographical distributions are an integral part of the epidemiology of the diseases. Human sleeping sickness, caused by Trypanosoma brucei gambiense and T. brucei rhodesiense, entered Kenya in about 1901 and the 1950s, respectively. The former disease was eradicated by attacking Glossina fuscipes with DDT, but the latter disease still persists in the Lambwe Valley in Western Kenya (338). Japanese encephalitis virus and related mosquito-borne viruses have caused outbreaks in Malaysia and India, but the source of the virus is unknown (86, 320, 333). The virus also spread into north Queensland, Australia, in 1995, apparently either by natural wind-borne transfer of infected mosquitoes, perhaps Culex annulirostris, from Papua New Guinea (273) or in migrating birds (132).

Increasing travel opportunities, population pressures in rural areas, and natural disasters and civil unrest are increasing the numbers of people traveling and being exposed to vector-borne diseases (65, 293). Each year, 20 million people visit malarious areas, and there are 10,000 cases of imported malaria in the European Community alone, most of which are caused by the more virulent P. falciparum (115). The types of human movements that influenced the incidence of malaria in different regions are summarized in Table 5. They show the variety and geographical range of such movements. Travelers have carried pathogens into other countries (258), including malaria into several countries (44, 47, 203, 207, 214, 259) and dengue viruses of all four serotypes into many countries but particularly the western Pacific, South Asia, and South America (124), and Ross River virus into Fiji and other islands in the Pacific (114, 284) and into New Zealand (121, 202).

Genetic variation in both disease competence in vectors and virulence of pathogens adds another dimension to the observed spread of vector-borne diseases around the world. In particular, dengue virus and its vector, Aedes aegypti, exhibit major variation in virulence and competence, respectively, in different geographical strains (301). Two molecular groups of the mosquito, with different geographical origins, varied in susceptibility to the dengue virus. A domestic form, A. aedes aegypti, from Southeast Asia, the South Pacific islands, and South America, had high dengue virus infection rates compared with a sylvan form, A. aedes formosus, from West Africa and some Indian Ocean islands. This is resulting in an expansion of the geographical range and changes in the epidemiology of the disease (93).

The potential impact of the spread of vectors and pathogens around the world greatly exceeds that of potential range expansions under climate change. While the latter are incremental increases, the former can change the risk level of a whole continent (304) by introducing a competent vector for the first time. Similarly, the above example of dengue fever illustrates how the introduction of a pathogen into new areas where competent vectors already exist has caused massive disease epidemics.

While each of the drivers of global change will have effects on the transmission of vector-borne diseases, the combined effects have the potential in some cases to multiply the risks. The portion of the world that is becoming climatically suitable for tropical vectors with global warming is increasing. Possible higher tropical rainfall combined with higher CO₂ concentrations may offset greater evaporation at higher temperatures, so that both factors will tend to increase the amount of surface water available for breeding by mosquitoes in the expanded area that is made suitable for breeding by the increase in temperatures. Higher temperatures will not only increase evaporation but also accelerate development of pathogens and reduce the longevity of vectors. These counteracting climatic changes will have a significant effect on the demand for irrigation, which itself will be driven by the need to intensify agriculture to feed the burgeoning human population in the developing world. While increased urbanization will raise the density of human populations, global population growth will also increase the total numbers of people at risk. Globalization of trade and movement of goods and people are accelerating, and the intensity of contacts between humans is increasing in the emerging megacities with inadequate water supply, sanitation, and public health infrastructures.

These combined forces suggest a tendency toward increased risks of spread and transmission of vector-borne diseases. The greatest risks from global change appear to be associated with mosquito-borne diseases, particularly malaria and dengue fever. The greatest potential impacts probably lie in the interactions between invasions of vectors or pathogens from tropical and subtropical regions into an expanded receptive zone and increasing urbanization and poverty in developing countries. This latter effect is illustrated by the spread of dengue into the Americas since 1980 (121). On the other hand, as discussed below, social changes and adaptive management have the potential to counter many of these negative influences, provided that resources are redirected into public health.

Perceptions of the risks of the different global change drivers affecting vector-borne diseases vary according to the method used for assessment of those risks. Biologists often adopt a reductionist approach and try to isolate each variable and define its influence on the system concerned, even if the effect is small. Economists assess the relative importance of each of the variables and are not concerned with small or future effects (243). Biologists are concerned with small effects that may become larger in the future. They are also concerned with nonlinear and often disproportionate responses to small changes in (say) temperature around threshold developmental temperatures, for example (307). This dichotomy of views is also evident in the community investigating global change and risks of vector-borne diseases. Theoreticians and modelers have tended to emphasize the effects of climate change because it is part of a large global activity, it is tractable, and it can be supported by sound physiological data from the laboratory. Systems scientists and ecologists are anxious to take holistic approaches to global change issues, as evidenced by the The Amsterdam Declaration on Global Change (http://www.sciconf.igbp.kva.se/Amsterdam_Declaration.html ), but are frustrated by a lack of tools to integrate physical and social systems, data on local effects, and social structures that facilitate that type of research. Public health practitioners, on the other hand, have been more concerned with weighing the risks from each source under current conditions. While all agree that there are many factors affecting the status of each disease, it is difficult to generalize across regions, as we saw in Fig. 8, and the more subtle effects of small changes in climate, for example, are difficult to isolate and quantify. However, that does not mean that climate change will not be a major concern in the future, since numerous modeling studies forecast significant changes in phenology and vector numbers with increasing temperatures in cooler environments. Practitioners also argue that the vectors are usually already present in areas adjacent to current zones of endemicity but the pathogens are not, suggesting that more emphasis needs to be placed on analysis of the pathogen part of the triangle in Fig. 1. Practitioners may also be dismissive of modeling because they do not appreciate the significance of nonlinearity of responses such as that shown in the field observations in Fig. 11. We need to remember that the ecology of vector-borne diseases is highly complex and defies simplistic analyses (270).

It is evident that changes in human population growth, modification of the environment for agriculture, and trade and travel are important current influences on vector-borne diseases. The development of resistance in both vectors and pathogens and the deterioration of public health infrastructure and sanitation are major contributing factors to a deteriorating global problem of dengue and malaria in particular. In future, accumulating residues of EDCs may also emerge as a major issue in the maintenance of human immunity to vector-borne diseases. These latter factors represent failures of management rather than direct increases in risks from environmental changes. They portend problems with future efforts at adaptation. It is sobering to find that more people are dying from malaria today than 30 years ago (254). Future risks may arise from much more subtle effects from uncertain directions such as anthropogenic climate change and variability, and EDCs, but only time and more research will clarify those risks.

Many deliberate policy changes could be implemented to enable us to adapt to the impacts of vector-borne diseases under global change. They involve legislation or administrative actions to respond to changes in distributions of vectors, diseases, and human hosts. The measures include changes of monitoring, case detection, diagnosis and reporting, public information and education, knowledge capture and management using information technologies such as remote sensing and mapping with geographical information systems (138, 239), climate forecasting, computer modeling, quarantine surveillance, early-warning systems, and mass vaccination programs. Most of these activities are integral components of any effective preventative health system and so ought to be implemented to meet current needs. While eradication is an idealistic goal, it is usually either unattainable or unsustainable, judging from past efforts with malaria in particular (349).

The failure of some vector or disease eradication programs has been attributed to a lack of funds (242). Efforts to eradicate malaria with insecticides failed in many regions due to the withdrawal of DDT, the development of resistance to the pesticides, or the reintroduction of the mosquitoes from other areas (349). Partial success was achieved with eradication of the main vector of dengue fever virus, Aedes aegypti, with successes in southern, temperate zones of Australia but failure in the Americas after initial success in Central America and Brazil (121).

There have, however, been some impressive results from concerted and coordinated efforts to minimize the transmission of vector-borne diseases over large areas. These have relied on suppression of the vector populations or saturation treatment of the pathogens. Eradication of malaria was successful in most of Europe by using DDT, removing breeding sites by draining swamps, and encouraging the keeping of cattle in proximity to homesteads in order to provide an alternative host for the mosquitoes of the malaria pathogens (191, 269, 358). However, in Ethiopia there was an increased incidence of malaria in children sharing the same house with cattle, therefore, there needs to be some distance between the livestock and human housing (108). In South America, eradication of the Anopheles gambiae mosquito from northeast Brazil in the 1930s was a significant watershed in the history of malaria control (241). More recently the global Roll Back Malaria campaign, being conducted by agencies of the United Nations during the decade from 2001 to 2010, has launched a concerted effort to halve the incidence of malaria around the world by using a combination of vector control and early treatment of infections in vulnerable populations (271).

Pesticide-based vector control programmes greatly reduced the incidence of onchocerciasis (or river blindness) in West Africa and Chagas' disease in South America (228) by reducing populations of the Simulium and Triatoma vectors, respectively. In West Africa, eradication of onchocerciasis by aerial spraying of the breeding sites of Simulium in fast-flowing rivers was successful in several countries. Further attempts, involving mass treatments of several million people with ivermectin, were initially hindered by a lack of coordination until the advent of the African Programme for Onchocerciasis Control in 1995 (92). Widespread treatment of the human population with ivermectin also reversed anterior-chamber lesions, but there were some adverse reactions to ivermectin treatment when people were heavily coinfected with Loa loa (102). There is optimism that similar approaches can be applied to eradicate lymphatic filariasis in India by using diethylcarbamazine and ivermectin (10).

Long-term success in eradication of the reduvid bug, Triatoma infestans, has been achieved in the State of Sao Paulo, Brazil (275). Control of Chagas' disease, caused by Trypanosoma cruzi, was achieved by suppressing the vectors with insecticides and providing improved housing. In contrast, eradication of domestic T. infestans failed in an area of Santiago del Estero in Argentina (289). It is hoped that Chagas' disease can be controlled by using a combination of insecticides and control of blood banks in South America (10).

In Africa, eradication of the tsetse fly (Glossina spp.) vectors of human and animal trypanosomiasis has met with mixed success because the fly tends to reinvade areas that have been cleared unless development follows and permanent vegetation-free buffer zones are maintained (133, 338). Stronger links between tsetse control policy and rural development, with intersectorial participation, are essential (64). This means that eradication of vectors such as the tsetse fly must be a slow and cautious process (24). Failures to eradicate Glossina pallidipes from parts of Africa, including the Lambwe Valley in Kenya, has led to persistent outbreaks of human trypanosomiasis. In Uganda, a resurgence of human sleeping sickness was associated with encroachment of the dense, exotic shrub Lantana camara. The epidemic was stopped by the use of insecticide-impregnated tsetse traps. Social resistance to the clearing of bush in valleys and later civil war has led to a catastrophic increase in the incidence of epidemics of sleeping sickness caused by Trypanosoma rhodesiense in Zaire and Central Africa, as public health programmes have failed (228, 260). Environmental concerns related to land degradation and loss of wildlife following successful eradication of the tsetse fly in regions of Africa reduce the attractiveness of eradication of the fly from unfarmed land (133). Where human trypanosomiasis is concerned, the case is much more compelling. Recent attempts have been made to promote area-wide eradication of tsetse flies in Africa following success on a small island in Zanzibar (158). These initiatives have been questioned on historical, ecological, logistical, and financial grounds (281).

The usual engineering or behavioral response by affected individuals to vector-borne disease is to change their behavior at crucial times or to use one or more of a wide range of options to control either the vector or the pathogen or both. Some personal behavioral responses include avoidance of exposure to vectors by staying indoors at dawn and dusk and wearing protective clothing. Routine responses by communities, as part of everyday operations in response to a perceived increase in climatic variability, could include tuning tactics in a surveillance program to detect earlier infections that may initiate outbreaks. Engineering options for vector control include habitat modification, insect screens, bednets, biological control, pesticides, repellents, and more effective integration of these approaches. Targeting of pathogens with drugs or vaccines also provides options to reduce transmission and to reduce symptoms. To identify the tools that have the greatest potential for use to address global change, it is necessary to evaluate their future sustainability and robustness under a wide range of conditions (309). Some of the options that are currently available are rated for these attributes in Table 6. Preventative measures based on environmental management rate highly in terms of their robustness, but their potential sustainability is limited by their demands on management. To date, no approach has been successful without social stability and maintenance of an effective public health system.

Responses to changes in the timing and intensity of disease transmission under global change need to be designed a priori to enable them to be proactive rather than reactive, as illustrated by the preventative vaccination against rabies in North America (274, 283). This calls for a thorough understanding of the biogeography and epidemiology of each disease in the local area at risk. High-risk areas and communities need to be identified in order to focus efforts toward adaptation of the existing disease management programs. Some indication of the needs for modification of current management tools, in order to accommodate global change, are summarized in Table 7. All techniques and approaches need to be tuned to make them more flexible and adaptable to a changing environment, as shown by the low success rate with their current application around the world. This demands a more knowledge-based, rather than product-based, approach.

Adaptation to climate change requires a combination of monitoring, interpretation of field observations based on an understanding of the system involved, and adjustment of existing management actions to accommodate the changes in geographical distribution, seasonal numbers, and timing of vectors (307). As discussed above, it is critical that the cause of changed transmission rates is determined to avoid inappropriate responses based on assumptions about the cause of failure (307). Provision of protection for the most sensitive human populations, without acquired immunity to the pathogen concerned, will be a priority. These people will be found in the vicinity of the altitudinal and horizontal boundaries of the geographical distribution of the species when the variation in climate leads to exceptional divergence from average conditions. They are likely to be found among the most underprivileged communities.

In the case of local environmental changes brought about by local disturbances such as introduction of dams and irrigation or deforestation, there are less likely to be qualitatively different risks, and so implementation of current management practices will be appropriate.