CIESIN Reproduced, with permission, from: Reiter, P. 1988. Weather, vector biology, and arboviral recrudescence. In The arboviruses: Epidemiology and ecology. Vol. 1, ed. T. P. Monath. Florida: CRC Press.

Chapter 9

WEATHER, VECTOR BIOLOGY, AND ARBOVIRAL RECRUDESCENCE

Paul Reiter

I. BASIC PRINCIPLES

A. Recrudescence

Many arboviral diseases are characterized by long periods of invisibility, when little or no evidence of their existence can be detected. At erratic intervals, sometimes separated by several decades, there is a sudden recrudescence, often developing into an explosive epidemic. Retrospective studies of such epidemics frequently suggest, by association, weather-related factors which could have been responsible for triggering this recrudescence, and these associations offer fertile ground for speculative explanations. Nevertheless, the timing of recrudescence remains enigmatic and notoriously unpredictable. In practical terms this is a serious problem; all too frequently, health or veterinary authorities are unaware of the existence of an epidemic until many weeks after its commencement, and may be unable to implement countermeasures until after the majority of infections have already been contracted.

Recrudescence is by no means unique to arboviral infections. In the aftermath of widespread transmission of all viral diseases, the immune system acts within populations as a common defense factor, the herd immunity, which restrains the recurrence of high rates of transmission until a sufficiently large number of nonimmune individuals are recruited, either by natural increase or by immigration from other populations. Recurrent epidemics of measles and influenza viruses are classic examples of this mechanism. However, in the case of the arboviruses, the obligatory involvement of one or more species of arthropod introduces a wide range of additional factors which greatly increase the complexity of transmission. In this chapter we are concerned with weather-related aspects of the natural history of the vector which may influence recrudescence.

B. Climate, Weather, and Arthropods

Climate is the long-term summation of the atmospheric elements---radiation, temperature, precipitation, humidity, and wind---and their variations. The global interactions of the components of climate are highly complex and resolve into a number of long-term meteorological cycles. Best known among these are the southern high-pressure oscillation (the "el nino" phenomenon) and the ice ages. Description of climate is therefore time-dependent, and climatic indices vary from decade to decade and century to century.

Short-term variation within climate is termed weather. The changeability of weather varies widely at different latitudes. It is greatest in the mid-latitudes, where a continuous succession of high and low pressure centers results in a constantly shifting weather pattern, and least in the tropics, where day-to-day and month-to-month changes are so small as to be almost synonymous with climate.

Climate is a major component in the environment of all arthropods, and indeed all living organisms. All species live within defined climatic limits, although the actual limits of their distribution are only partly determined by climate. Weather also exerts a profound effect on arthropods. Within their climatic limits, all the atmospheric elements constantly effect every aspect of behavior, development, and dispersal, while at the boundaries of their climatic limits, relatively minor deviations from the ambient norm can be catastrophic.

It is important to realize the immediacy of these effects. It is possible to measure temperature to a degree by counting the frequency with which crickets chirp: the organism operates in direct linkage to its atmospheric environment, and has only limited physiological mechanisms for responding to short-term fluctuations. Indeed, for most arthropods, even in the humid tropics, daily cycles of temperature and humidity range far beyond the span of their optimal environment, and mean daily conditions are far less important than specific conditions at a select portion of the day. Strict circadian cycles, generally regulated by time cues based on the light regime, dominate their behavior to ensure that crucial activities coincide with the regular occurrence of optimal conditions. If these conditions are not met, behavioral strategies, usually involving evasion and inactivity. are the dominant response; the organism defers normal behavior until the next occasion that its activity cycle coincides with acceptable conditions.

C. Climatic Seasonality and Arthropod Biology

Few arthropods exhibit continuous, year-round activity, and there is a wide variety of seasonal arrangements of active and dormant phases in their life cycles. In temperate zones, winter presents a conspicuous obstacle, and there is a wealth of studies describing dormancy in this season. Far less well known, but widespread and common even at high latitudes, are dormancy phases in other seasons. For example, many insects restrict their active phases to a short fraction of the summer and spend the remainder in a phase of inactivity analogous to overwintering.[1] In the tropics, activity may be restricted to a portion of the wet or the dry season, or both.[2] In addition to these preprogrammed periods of inactivity, transient events, such as periods of low temperature or heavy rainfall may interrupt normal feeding and reproductive behavior during an active season, or may interrupt or terminate inactivity during a dormant season.[3]

In most species, programmed dormancy is expressed in one stage of the life cycle. Thus, Aedes mosquitoes survive winter or drought seasons in the egg stage whereas Culex and Anopheles species survive as adults. Exceptions, such as ticks, generally require 2 or more years to complete a life cycle (see Chapters 7 and 8).

The alternative to seasonal dormancy is to continue reproduction throughout the year, with opportunistic bursts of population increase whenever conditions are suitable. In these circumstances, climatic seasonality dictates abundance by the availability of food or breeding sites; however, caution is required when assuming that all members of a species are continually active all of the time. Polymodal strategies, whereby part of a population continues activity while the remainder is dormant are not uncommon. An example is the "bet-hedging" exhibited in the staggered hatching of Aedes eggs. A less well-known example is dormancy in Cx. quinquefasciatus; in the tropics this species breeds throughout the year, yet there is evidence[4] that during the dry season a large fraction of the adult population becomes inactive. A problem with this type of dormancy is that when all stages of the life cycle occur simultaneously, its presence may be difficult to demonstrate.

D. Vectorial Capacity, Weather, and Activity vs. Inactivity

The entomological features of transmission can be expressed in terms of the vectorial capacity, the daily rate of potentially infective contacts between vector and host:

FIGURE

where m is the number of vectors per host, a is the number of blood meals taken by a vector per host per day, p is the daily survival probability of the vector, and n is the number of days between infection of the vector and the time it becomes capable of infecting a new host.[5,6]

The concept of vectorial capacity was developed for models of vector-borne parasitic disease, but can also be applied to arboviral disease. There is only one crucial difference: the ability of the immune system of the host to react rapidly to purge its system of viruses, and to maintain effective barriers against their reintroduction. In contrast to the chronic nature of parasitic infections, which may persist in a host for several decades, the duration of arboviral viremia is very short, usually a matter of days. In addition, the potential for repeated reinfections by a single species of parasite is often high, whereas infections with a particular arbovirus are generally a once-in-a-lifetime event. For this reason, whenever the blood-feeding activity of an arbovirus vector is interrupted for longer than the sum of the host's incubation period and the duration of viremia, the virus effectively ceases to exist in the host, and its survival is entirely dependent on the fortunes of the vector.

The simplest form of vector borne transmission involves the direct transfer of pathogen from one host to another, without replication in the vector. This mechanical transmission is dependent on a vector which takes a number of blood meals on more than one host in rapid succession. and a pathogen which can survive the interval between feeds on the vector, generally on its mouthparts. In this case, p can be regarded as close to unity and n as a fraction of a day. The vectorial capacity is therefore high, but is dependent on the ability of small amounts of pathogen to infect new hosts. Although mechanical transmission can be demonstrated in the laboratory, it is much more difficult to prove in the field. Weather conditions which favor high vector populations and active feeding behavior will presumably increase vectorial capacity.

In more complex forms of transmission, multiplication of the pathogen in the vector ensures that the infective dose is far larger than the quantity which was originally ingested. This generally involves multiplication and movement of pathogen through the gut wall, via the hemolymph, to other organs. Infection of the salivary glands is particularly important because the injection of infected salivary fluid during feeding is the most common mode of transfer of pathogen to a new host. It is the time (n) required for this obligatory incubation that makes longevity the key factor in transmission of all vector-borne disease, for no matter how favorable all other parameters may be, transmission cannot occur unless the vector survives to feed after this incubation period, and small changes in the daily survival probability p will radically alter the frequency of new infections.

After the incubation period, longevity determines how many infective feeds a vector can make. In arboviral disease, because viremia is of short, fixed duration, this is of greater significance than in vector-borne parasitic disease, for the lower the transmission rate, the more the survival of the virus depends on the longevity of the vector.

Some of the effects of climatic seasonality on m and a are readily identified. Maximum vector/host contact will occur in seasons when vectors are active and their populations are high, and epidemics occur at these times. Conversely, low vector populations or inactive vectors imply low contact and an obligatory seasonal maintenance phase.

It is logical to extend such associations in our search for weather-induced recrudescence following a prolonged maintenance phase, but caution is required. For example, during a period of above normal temperatures given adequate breeding sites and nutrition, the rate of development of mosquito larvae and pupae will be above normal. More will survive to adulthood, and ovarian development and the extrinsic incubation period will be shortest factors m, a, and n will all tend to a higher vectorial capacity (C). However, the key factor (p) will be adversely affected, because an increase in the frequency of hazardous events such as host seeking, feeding, and oviposition will increase the mortality rate. Estimates of p for Cx. pipiens s.l. in Memphis ranged from 0.65 (average life expectancy 2.3 days) at the height of summer to above 0.97 (average life expectancy 32.8 days) in the winter.[7] This extreme case illustrates how the longevity of a vector, inversely related to the frequency of risk, can be reduced by a shorter genotrophic cycle. Higher temperatures can thus increase vector-host contact but reduce vectorial capacity. In this context we must remember that cireadian control, which restricts adult activity to specific times of day, dictates that the gonotrophic period can only change by discrete units of time, generally 24 hr. Thus, a small change of temperature at the threshold of the circadian "gate" can have a major effect on p, and hence vectorial capacity.

The summertime estimate of p mentioned above was based on a standard method, the measurement of the percentage of parous insects in filed collected samples by ovarian dissection. The winter estimate was based on weekly counts of marked mosquitoes resting in a tunnel. Ovarian dissection would have been irrelevant in the latter study, because insects were not feeding, and remained nulliparous throughout the study. This illustrates an important point: a vector population which experiences a delay in its activity, by short-term quiescence due to adverse weather, a shortage of oviposition sites or suitable hosts, or a longer, programmed period of dormancy, may present a low physiological age structure but in reality have a high survival rate. Once again, climate or weather factors which prolong the gonotrophic cycle by delaying activity can enhance vectorial capacity and constitute a potential trigger for recrudescence.

Another factor related to quiescence is synchrony. Recruitment to an inactive phase may be sudden in response to an environmental change, or gradual, as individuals attain a particular point in their life cycle. By contrast, the stimuli which terminate inactive phases act on the population as a whole, thereby inducing synchrony in the activated cohort. If incubation of a virus was completed before or during inactivity, then this synchrony will result in an abrupt onset of transmission by all infected individuals.

The arguments set out in this section imply that the inactivity of vectors may be of equal or greater significance than their activity. This paradox may throw light on some of the enigmas of arboviral recrudescence, but the very nature of the covert behavior involved means that direct evidence from the field is difficult to obtain. The problem is compounded by the unpredictability of recrudescence itself: comprehensive long-term field studies are rarely in progress when renewed transmission occurs, and the lag-time inherent in clinical surveillance systems and their interpretation mean that weeks or even months may pass before the event is recognized. The result is that vector biologists almost invariably initiate intensive studies when epidemics are already history, and we know far more about the age structure, vectorial capacity, behavior, and dynamics of vector populations after epidemics than in the crucial periods when they begin.

The analysis of weather patterns themselves presents additional problems. Attempts to correlate weather and biological events require the choice of a suitable time scale, resulting in descriptors like "warmer than normal" months or "wetter than normal" seasons. While it is clearly useful to attempt to resolve hourly or daily data into such units, they frequently give no indication of important detail such as the daily range of temperatures involved, the number of times that rainfall occurred, or the number of short-term anomalies which interrupted the overall pattern observed. Conversely, long-term, year-to-year events escape attention if analysis is focused on shorter units or on calendar years: the springtime population of a vector may be greatly influenced by breeding success in the previous autumn, and a disastrous year may have consequences which extend for many years afterward. In the next section, all these points will be illustrated by examining recrudescence in a single disease, St. Louis encephalitis (SLE).

II. AN EXAMPLE: WEATHER AND URBAN ST. LOUIS ENCEPHALITIS

A. The St. Louis Epidemic of 1933

The largest outbreak of SLE ever recorded in a single city was also the first to be recognized, that of St. Louis, Mo., in 1933. The earliest case was noted on July 7 and the last in late October, with the peak on August 27. The epidemic received great publicity around the world, and was treated as a major emergency.

The disease was new and transmission theories abounded. It was noted that because the summer was the driest on record, there was very little flow in the open drainage ditches used for sewage disposal; these developed a particularly offensive stench and were breeding unusually large numbers of mosquitoes. Because there were no obvious indications of contagion between cases or links to water, food, or milk supply, and because mosquitoes were abundant. many people felt that mosquito transmission was the answer. The investigators themselves did not think this likely, but decided on a thorough investigation of the possibility, if only to "obtain...information...in the negative sense". Exhaustive attempts were made to demonstrate transmission by Ae. aegypti, Anopheles quadrimaculatus, and Cx. pipiens s.l., the latter chosen because it was the most common species, but no transmission was observed. It was concluded that the results spoke against mosquito transmission. The diffuse distribution of cases was cited as supportive evidence because this was considered analogous to that of poliomyelitis and quite unlike that of previous epidemics of yellow fever.

It is fascinating to read the official report of the epidemic[8] in the light of another document, commissioned by the Surgeon General in September 1933 but not published until 25 years later.[9] Its author, L.L. Lumsden, conducted a classic epidemiological study from which he produced abundant evidence that SLE was indeed mosquito-borne and that the vector was Cx. pipiens s.l. breeding in sewage-polluted water. He hypothesized that the virus entered the mosquito from the sewage, and that the exceptionally dry weather had precipitated the epidemic by boosting the mosquito population.

B. Patterns of Weather and Outbreaks of Urban SLE

Many authors have commented on weather factors in their descriptions of subsequent SLE outbreaks. In a review of these, Monath[10] compared epidemic to nonepidemic years in terms of deviations from normal monthly temperatures and precipitation. A pattern emerged for epidemic years in most localities: above-normal temperatures in January, February, and May through August, and below-normal temperatures in April, January and February were abnormally wet, but July was abnormally dry.

The epidemic in St. Louis was not included in Monath's analysis, but the weather of 1933 fits his pattern precisely. Indeed, the extremes of that year were so remarkable that they deserve further mention; for this purpose I draw liberally from contemporary weather descriptions, published weekly by the U.S. Department of Agriculture.[11] The winter of 1932--1933 was the second warmest on record, with temperatures in December and January approximating conditions appropriate for April. Abnormally high temperatures persisted until late March, but farming activities were severely delayed because of incessant rainfall. April was cool, May was slightly warmer than average, but the rain continued at twice the normal rate so that by the beginning of June conditions were critical---continued wetness had made field operations impossible. At this point a dramatic change occurred: rainfall for the next 3 months was the lowest ever recorded, June was the hottest on record, and monthly average temperatures for July and August were also well above normal.

Lumsden and his contemporaries drew attention to the coincidence of SLE and an abnormal lack of rainfall, but made no mention of conditions in the preceding 5 months, which were also very unusual. My purpose in giving so detailed a description has not been to highlight their omission---without knowledge of the natural history of the mosquito and the complex zoonosis which is involved, this information could not have seemed relevant---but to underline some of the pitfalls of summarizing and drawing conclusions from meteorological data in studies of arboviral recrudescence. However, I first review some of the speculations which have been offered to explain the association between SLE outbreaks and weather patterns.

C. Speculations on Entomological Reasons for Associations between Weather Patterns and Urban SLE Outbreaks

Serologic studies on birds indicate that in nonepidemic years, enzootic transmission of urban SLE occurs in spring and autumn, but in epidemic years there is a sudden recrudescence in summer, generally in late June or July. Human cases become apparent some weeks later; this is the basis of SLE surveillance by sentinel flocks.[12]

During urban epidemics Cx. pipiens s.l. appears to be the most abundant potential vector, and more isolations of SLE virus have been made from this than any other species. Field and laboratory evidence indicate that two other mosquitoes. Cx. restuans Theobald and Cx. salinarius Coquillett are also involved in transmission. but these are not commonly collected after recrudescence. In Florida, all three species are present, but another vector, Cx. nigripalpus Theobald, is more abundant during epidemics.[13]

Cx. pipiens s.l. overwinter as adults. It has been suggested that a warm, wet winter could be conducive to high survival during quiescence, resulting in a rapid buildup of populations in the spring. In addition. if SLE overwinters in the mosquito, high survival rates of the vector would favor the virus as well. However. no difference in the mortality rate of overwintering Cx. pipiens s.l. was observed in Memphis in the winters of 1981--1982 and 1982--1983,[7] although the first was colder than normal and the latter was the fourth warmest on record. Indeed, since depletion of fat reserves, predation, and fungal disease are the principal causes of winter death in this species,[14] it would seem that mortality should be more likely to increase rather than decrease in abnormally warm, wet conditions. Moreover, despite many attempts to prove the contrary, there is very little evidence that SLE virus overwinter in Cx. pipiens s.l.

A cool April has been explained as delaying the nesting activity of birds, thereby inducing synchrony between high mosquito populations and the availability of nonimmune nestlings. However, low temperatures will also delay the multiplication of vectors, and in many areas synchrony between mosquito production and nesting activity occurs in a normal April. A normal or warm April followed by a warm May would therefore seem more advantageous to continued virus amplification.

Warm weather in May might favor transmission by accelerating the life cycle of the mosquito and the multiplication of the virus, but here we are faced with an anachronism: in much of the area where urban SLE occurs there is little or no Cx. pipiens s.l. activity in May. The first major population peak occurs some 2 months later, from early to mid-July in Memphis[15] and mid- to late July in the northern states and Canada.[16] For this reason it is not easy to implicate Cx. pipiens s.l. in the pre-epidemic recrudescence of SLE, nor to account for weather-associated factors which might precipitate this.

By contrast, Cx. restuans is abundant in springtime but rarely evident after recrudescence. In the adult stage Cx. pipiens s.l. and Cx. restuans are so alike that many authors consider field-caught specimens to be indistinguishable. For these reasons there is little precise information on the population dynamics of either species. However, in the larval stages they are readily separated, and recent studies have therefore used quantitative egg-raft collections to monitor adult activity.[17] They share a preference for breeding in foul-smelling water and in many ways can be regarded as ecological homologs, except for the temporal separation of their activity. It is this last detail that may be of crucial importance to the recrudescence of SLE: when the adult population of Cx. restuans is at its annual peak, there is a sudden cessation of activity and the species enters a state of dormancy. This disappearance of adult Cx. restuans is dramatic, and few egg-rafts appear at ovitraps during the summer. Resumption of activity in the autumn can be equally dramatic, and may coincide with the return of cooler weather.[15 ]If temperatures permit, the species continues breeding into the winter months, although there is some evidence of a winter diapause mechanism in its northern range.[18]

An exception to this pattern was noted in Memphis in 1983: a major peak of Cx. restuans activity occurred suddenly in July, 8 weeks after the normal May peak, followed by a smaller but well-defined peak in mid-August. The remarkable feature of these peaks was that they coincided with frontal systems which brought unseasonably cold weather during an abnormally hot, dry summer.[15] Studies in Vero Beach, Fla. have indicated a similar pattern: both Cx. restuans and Cx. salinarius suddenly cease to oviposit in late April or early May, but resume oviposition at any time that temperatures drop below 20deg.C.[26]

The coincidence of Cx. restuans activity and SLE transmission in spring and autumn suggests that this species may be a major vector and that in nonepidemic years quiescent adults function as a bridge over the summer period. Warm weather in May would boost populations before quiescence, and reactivation by periods of unseasonably cool summer weather would cause the sudden, synchronous release of virus at a time when it would be relayed to the burgeoning Cx. pipiens population. A high survival rate in the inactive Cx. restuans would ensure a high transmission rate, and an oscillating summer weather pattern would be more conducive to recrudescence than one of consistent high temperature.

As already mentioned. June 1933 was the hottest on record, but closer scrutiny reveals a remarkable anomaly. After 2 weeks of abnormally hot weather, temperatures plunged. A minimum reading of 10deg.C was recorded in St. Louis, contrasting with minima as high as 24deg.C the previous week. Frost damage was reported over considerable areas including the lake region and the northern Ohio River Valley. In the midwest, oat crops were abandoned and mowed for hay, and in the southern and eastern states the growth of sugar, cotton, and tobacco was retarded. Abnormally hot weather returned in the last week of June, with minimum temperatures as high as 29deg.C, but was interrupted by another unusually cold week around the 4th of July, with minima of 18 to 20deg.C. For the remainder of the summer there were at least three more cycles of cold weather, with minima as low as 13deg.C.[19] Thus, the requirement for an oscillating pattern of summer temperatures predicted by the Cx. restuans hypothesis is fulfilled by the St. Louis data. Moreover, the large geographic region affected by these transient weather phenomena accounts for the concurrence of widely dispersed SLE outbreaks which is frequently observed in epidemic years. Similar spells of unseasonably cold weather in warmer than normal summers can also be identified in years when SLE transmission was high. Conversely, in hot summers when such cool spells were absent transmission has never been high.

The St. Louis data clearly illustrates the limitations of summarizing weather data in seasonal or monthly terms, and this applies equally to other times of the year. Arthropods can exploit warm periods in an otherwise "cold" spring. High total rainfall may suggest a wet summer but be too intermittent to provide suitable groundwater breeding sites. A brief killing frost can obviate the effects of a warm autumn.

Finally, the St. Louis example also encompasses long-term factors which may have influenced vector populations and SLE recrudescence. The winter of 1932--1933 was the 12th abnormally warm winter in the preceding 13. The hot, dry summer was one of a series in a period which came to be known as the "dust bowl years", and the drought of 1930 was even more severe than that of 1933, as were those of 1934 and 1936.

III. OTHER EXAMPLES OF ASSOCIATIONS BETWEEN WEATHER AND ARBOVIRAL RECRUDESCENCE

SLE has served to illustrate the complexity of weather/vector/arbovirus interactions. Throughout this book, examples of weather-associated recrudescence in many other arboviral diseases can be found. It is unnecessary to review these in detail, but brief mention of some is justified.

A. Rainfall and Temperature

References to rainfall and temperature are common in the literature. Japanese encephalitis (JE) is perhaps the most studied, and appears to have similar weather associations to SLE. Other mosquito-borne viruses include eastern equine encephalomyelitis (EEE), western equine encephalitis (WEE), Murray Valley encephalitis (MVE), Rift Valley fever (RVF), and dengue. Tick-borne fevers include Crimean-Congo hemorrhagic fever (CCHF) and Omsk, hemorrhagic fever (OHF). In these and many other cases, a direct influence of rainfall and temperature on vector abundance and vector survival has often been suggested.

B. Humidity

Oviposition by Cx. nigripalpus is inhibited when relative humidity at sunset is below 90%. Provost[20] suggested that a prolonged delay of oviposition by drought could trigger an SLE epidemic by increasing the probability of completion of extrinsic incubation. In principle, this was the forerunner of the Cx. restuans hypothesis of urban SLE epidemics described in the previous section.

A seasonal shift in host selection by Cx. nigripalpus from birds in winter and spring to mammals in summer and autumn has also been attributed to humidity: high humidities apparently enable mosquitoes to leave their woodland habitat (where birds are abundant) to feed on mammals in open fields. "Clearly, this shift may be important as a coupling mechanism between avian and mammalian cycles of SLE transmission. A similar pattern of host preference has been observed in Cx. tarsalis Coquillet, the principal vector of rural SLE in the western U.S.[22]

In many regions of the hot, dry African savannah there is no surface water for up to 8 months of the year. During part of this dry season, mean monthly temperatures exceed 40deg.C and mean monthly relative humidities drop below 10%. A study in one such area in the Sudan[23] revealed that adult An. gambiae s.l. can survive in these conditions for 9 months by sheltering in human dwelling places, dry wells, and animal burrows and taking an occasional blood meal. This adaptation to extremely dry conditions results in year-round malaria transmission, and is terminated by the advent of moister conditions. I mention it here as an extreme example of the quiescence/transmission phenomenon in relation to humidity.

C. Wind

Wind plays an important role in insect dispersal and migration. and there is a large literature on the possible role of infected, wind-borne vectors in arboviral recrudescence.[25,26] The difficulty of explaining the "disappearance" of arboviruses for long periods of time makes wind-borne reintroduction from endemic areas an attractive possibility. In addition, there is firm evidence for the spread of plant viruses by plant-sucking insects, and the appearance of adult Simulium damnosum sl. infected with Onchocerca volvulus in the Sahel at the start of the West African rainy season appears to substantiate migration of a human pathogen on the Intertropical Convergence Front.

Wind patterns offer fertile ground for speculation. There is evidence that on a number of occasions, African horse sickness and bluetongue virus may have spread to nonendemic areas in wind-borne Culicoides midges. Study of surface synoptic charts indicates a coincidence between the tracks of tropical storms and outbreaks of Ibaraki and bovine ephemeral fever in Japan, the inference being that infected insects are launched into these weather systems as they pass over Asian countries where the diseases are endemic. Seasonal winds associated with the annual movements of the Intertropical Convergence Zone suggest a regular vehicle for the reintroduction of JE to Japan and Korea from the tropics, and there are several reports of Cx. tritaeniorhynchus (the major vector of JE in that region) being collected at sea. A similar pattern has also been suggested for all the encephalitides in North America, including SLE.

A major problem with these ideas is that wind is not an isolated weather phenomenon; the movement of air masses is a major cause of changes of temperature, precipitation, humidity, and other parameters, and, as we have seen throughout this chapter, a case can be built for the involvement of any or all of these in arboviral recrudescence. Contemporary techniques for identifying geographical strains of viruses will resolve many of these speculations. To date, such evidence does not support the suggestion that SLE virus migrates from the tropics to North America.

D. Host Behavior and Other Factors

Although not strictly of vector origin, a few other factors deserve mention. During unusually hot weather, people may prefer to live and sleep out of doors, thus increasing exposure to mosquitoes and other vectors. On the other hand, when air conditioning is available, people may spend more time inside buildings, protected from vectors by closed doors and windows. In regions without piped water supply, increased domestic storage of water during drought may promote breeding by peridomestic vectors such as Ae. aegypti. In many countries, activities like berry picking, mushroom gathering, recreational camping, or hunting, may be greatly influenced by weather, and attract large numbers of people to areas where ticks and other vectors may be common. Finally, nectar flow in plants. which is an important food source for all mosquitoes, is strongly influenced by weather conditions.

IV. CONCLUSION

There is clearly a wealth of evidence that weather plays a role in arboviral recrudescence, and this can be explained in terms of vector biology. Behind most attempts to rationalize this role is the hope that weather forecasting and weather analysis might eliminate the element of surprise in arboviral epidemics. This is an attractive proposition, but the sobering truth is that arboviral epidemiology is a complex, multifactorial process, and that coincidental events involving some or all variables are the true precipitating factors for recrudescence.

A better understanding of the weather/arthropod portion of the equation is nonetheless worth pursuing, not only because of its potential contribution to epidemic forecasting, but also to improve our insight into the advisability of a vector control response. For example, if epidemics are triggered by brief bursts of activity in vectors which are otherwise hidden in inaccessible places, then a major investment in aerial adulticiding measures will not be justified unless operations can be linked to accurate short-term weather forecasts or effected in a season when the vector is more active.

What is clearly required is a comprehensive, in-depth monitoring of the recrudescence process, a blow-by-blow account which cannot begin when transmission has already accelerated, but must be in place and running during the interepidemic phase, however long this may last. All the factors contributing to vectorial capacity must be accounted for. New methods of age determination are required which indicate chronological rather than physiological age. Because infected insects are rare, even during peak transmission periods, emphasis needs to be placed on adequate sample size, even if this means very large collections. At the same time, the quantitative aspects of sampling must be strictly maintained, so that the dynamics of populations can be accurately portrayed. Improved sampling methods may be indicated, and in many cases need to be devised to target the infected cohorts of the vector population rather than aiming at the maximum number of specimens. Meteorological observations must be directly relevant to study areas, and include adequate appraisal of the microhabitats involved.

The effort required for a comprehensive project of this type may seem too great for the average research team to contemplate, but participation of several interested groups could spread the load of laboratory work involved. In addition, the initial phase could be viewed as archival; material would be routinely collected and stored under suitable conditions until an epidemic had come and gone. At that point, intensive processing would begin, working back from the time of known transmission. Whatever strategy is developed, an aggressive approach is clearly required to convert the wealth of random speculation on the weather/recrudescence relationship into useful information.


REFERENCES

1. Masaki, S., Summer diapause. Ann. Rev. Entomol.. 25, 1, 1980.

2. Denlinger. D. L., Dormancy in tropical insects Ann Rev. Entomol., 31, 239, 1986.

3. Beck, S. D., Insect thermoperiodism Ann. Rev. Entomol., 28, 91, 1983.

4. Gillett. J. D.. Mosquitos, Weidenfeld & Nicolson. London. 1971. 77.

5. Macdonald. G., The analysis of equilibrium in malaria. Trop. Dis. Bull.. 49, 813, 1952.

6. Garrett-Jones, C., The human blood index of malaria vectors in relation to epidemiological assessment. Bull. WHO 30, 241, 1964.

7. Reiter, P., Survival rates of overwintering Culex pipiens s.l. in Memphis, Tennessee, U.S.A. . manuscript in preparation.

8. Cumming, H. S. et al., Report on the St. Louis outbreak of encephalitis. Public Health Bull. No. 214. U.S. Treasury Department. Public Health Service. Washington. D.C., 1935.

9. Lumsden, L. L., St. Louis encephalitis in 1933. Observations on epidemiological features. Publ. Health Rep. 73, 340, 1958.

10. Monath, T. P., Epidemiology. in St Louis Encephalitis. Monath. T. P. Ed.. American Public Health Association Washington D C. 1980, 239.

11. U.S. Department of Commerce and U.S. Department of Agriculture. Weekly Weather and Crop Bulletin Ser., 1933.

12. Bowen, G. S. and Francy, D. B., Surveillance. in St. Louis Encephalitis Monath. T. P., Ed.. American Public Health Association. Washington, D.C., 1980. 473.

13. Mitchell, C. J., Francy, D. B., and Monath, T. P., Arthropod vectors, in St Louis Encephalitis. Monath. T. P., Ed.. American Public Health Association. Washington. D.C., 1980. 313.

14. Sulaiman, S. and Service, M. W., Studies on hibernating populations of the mosquito Culex pipiens L. in southern and northern England. J. Nat. Hist., 17, 849, 1983.

15. Reiter, P. and Francy, D. B., Weather-related variations in seasonal activity patterns of urban Culex mosquitoes and their implications in the epidemiology of St. Louis encephalitis manuscript in preparation.

16. Madder, D. J., Surgeoner, G. A., and Helson. B. V., Number of generations, egg production, and developmental time of Culex pipiens and Culex restuans (Diptera: Culicidae) in southern Ontario, J. Med. Entomol., 20, 275, 1983.

17. Reiter, P., A standard procedure for the quantitative surveillance of certain Culex mosquitoes by egg raft collection, J. Am. Mosq. Control Assoc. 219, 1986.

18. Eldridge, B. F., Bailey, C. L., and Johnson, M. D., A preliminary study of the seasonal geographic distribution and overwintering of Culex restuans Theobald and Culex Salinarius (Diptera: Culicidae). J. Med. Entomol., 9. 233, 1972.

19. U.S. Weather Bureau. Climatological Data: Missouri Section. Vol. 37, 1933.

20. Provost, M. W., The Natural History of Culex nigripalpus. Fla. State Bd. Health Monogr. No. 12. 46, 1969.

21. Edman, J. D., Host-feeding patterns of Florida mosquitoes. III. Culex (Culex) and Culex (Neoculex) J. Med. Entomol., 11 95, 1974.

22. Tempelis, C. H., Reeves, W. C., Bellamy, R. E., and Lofy, M. F., A three-year study of the feeding habits of Culex tarsalis in Kern County, California. Am. J. Trop Med. Hyg., 14, 170, 1965.

23. Omer, S. M. and Cloudsley-Thompson, J. L., Survival of female Anopheles gambiae Giles through a 9-month dry season in Sudan. Bull. WHO, 42, 319, 1970.

24 Sellers, R. F., Weather, host and vector---their interplay in the spread of insect-borne animal virus diseases. J. Hyg. (Cambridge). 85, 65, 1980.

25. Pedgley, D. E., Wind-borne spread of insect transmitted diseases of animals and man, Philos. Trans. R. Soc. Lond. Ser. B. 302, 463, 1983.

26. Haeger, J. S., personal communication.