-dnnals of Tropical Medicine & Parasitology, Vol. 102, No. 2, 161-171 (2008)

Assessment of a new strategy, based on Aedes aegypti (L.)pupal productivity, for the surveillance and control ofdengue transmission in Thailand

P. BARBAZAN*, W. TUNTAPRASART+, M. SOURIS*, F. DEMORAES*,N. NITATPATTANA++, W. BOONYUAN* and J.-P. GONZALEZ**IRD-UR 178, Centre for Vectors and Vector-Borne Diseases, Faculty of Science, A/IahidolUniversity at Salaya, 25/25 Phutthamonthon 4, Nakhon Pathom 73170, Thailand*Faculty of Public Health, Makidol University, 420/1 Rajavithi Road, Rachathewi, Bangkok10400, Thailand:Center for Vaccine Development, Mahidol University at Salaya, 25/25 Phutthamonthon 4,Nakhon Pathom 73170, Thailand

Received 17 November 2006, Revised 25 June 2007,Accepted 28 June 2007

In the countries where the disease is endemic, control of dengue is mainly based the elimination or treatment ofthe water-filled containers where the main vector, Aedes aegypti, breeds, in interventions usually reliant on

community participation. Although such control activities must be continuous, since vector eradication appearsimpossible, it should be possible to reduce the incidence of dengue significantly, in a cost-effective manner, bytargeting only those types of containers in which large numbers of-de, aegypti are produced. This strategy is nowrecommended by the World Health Organization, although it depends on the most productive types of containerbeing carefully identified, in each endemic region. In Thailand, exhaustive surveys of 3125 wet containers in 240houses in either an urban area (100-120 houses) or a rural (120 houses) were conducted during a rainy and a

dry season in 2004 2005. Indices based on the numbers of-de, aegypti pupae observed were found to correlate withthe 'classical' entomological indices that based on all of the immature stages of the vector. Overall, 2.3 and 0.8Ae. aegypti pupae were observed per person in the rural and urban areas, respectively. Although adult female .de.aegypti laid eggs in all 10 types of wet container that were identified, large water-storage containers produced themajority of the pupae, especially at the end of the dry season (when such containers accounted for 90% of the

pupae detected in the rural area and 60% of those in the urban area). Since these containers large, easy to reachand account for <50% of all wet containers, it should be relatively easy and quick to treat them with larvicide to

cover them. If even such targeted treatment is to be sustainable, however, it will have to be integrated, as one ofseveral activities in which the at-risk communities are encouraged to participate.

Dengue fever is the most wide-spreadarboviral disease in the tropics (Gubler,1997; Mackenzie et al., 2004).Transmission of the viruses that cause thedisease, by Aedes mosquitoes, is affected byvarious environmental conditions and thebehaviour and practices of the local humanpopulation. In endemic areas, the incidence

Reprint requests to: P. Barbazan.E-mail: fnpbb@diamond.mahidol.ac.th; fax: +66 24410267

,(:, 2008 The Liverpool School of Tropical MedicineDOI: 10.1179/136485908X252296

of dengue is largely dependent on thedensity of the local vector population (itselflargely determined by climate) and thefrequency of contact between the vectorsand susceptible human hosts (Kuno, 1995).In Thailand, the vectors are Aedes aegypti(L.) and Aedes albopictus Skuse. Whereas Ae.aegypti is a very anthropophilic species,preferentially biting humans and breedingin almost any type of artificial container,such as jars, cans and tyres (Christophers,1960; Tonn et al., 1969; Strickman and

162 BARBAZAN ETAL.

Kittayapong, 2002; Chareonviriyaphapet aL, 2003), Ae. albopictus is a mostly ruralspecies that usually breeds in natural con-

tainers, such as tree holes or coconut shells,although it can also be found in artificialcontainers.As there is, as yet, no effective treatment

or vaccine for dengue, control of the diseaseis based on surveillance and reduction of thevector population (WHO, 1997). Duringepidemics, outdoor aerial spraying of pesti-cide may significantly reduce the number ofinfected female mosquitoes, although thisdoes not have a long-lasting effect. Duringnon-epidemic periods, surveillance, preven-tion and control often depend on commu-

nity-participation strategies promoted by thenational health authorities. In Thailand, theentomological surveillance system is based

on surveys of water-filled or wet containers,each of which is checked for the immature

stages of Ae. aegypti, by teams from theMinistry of Public Health (MOPH) or

trained village health volunteers (VHV).Most positive containers are then treatedwith larvicide, either by their owners or

MOPH personnel. The members of affectedcommunities are also encouraged to cleardiscarded containers and to cover anycontainers used to store water, to reducethe numbers of potential vector-breedingsites.

Unfortunately, as there does not appear to

be a simple relationship between the num-

ber ofAe. aegypti-positive containers and thelevel of dengue transmission, it is difficult to

set a valid goal for the reduction in thenumber of containers (Chan, 1985; Kuno,1995; Reiter and Gubler, 1997). Even ifentire communities participate in theremoval or treatment of containers in whichthe vector might breed, new containers (e.g.discarded cans, bottles and old tyres) are

rapidly produced and levels of larvicide intreated water-storage containers rapidly fall

as the water is used and replenished(Nathan, 1993; Okanurak et aL, 1997).The exhaustive elimination of potentialbreeding sites is therefore improbable.

Over the last decade it has become clearthat a more realistic goal is the identificationof those types of container that are most

productive, in terms of the number of vectormosquitoes emerging from them, followedby interventions targeted at just these types.Permanent, large rainwater-collection tanks,holding hundred of litres of water, are, forexample, potentially much more productivethan small containers such as cans or tyres(Tun et aL, 1995).Although container productivity has often

been assessed by counting all the mosquitoor vector larvae and pupae in containers,many mosquito larvae never mature intoadults (many, for example, fail to findsufficient food) and levels of larval mortalityprobably vary with container type.Compared with a count of all the immaturestages in a container, a count of just the

pupae, which do not feed and take just a fewdays to develop into adults, will more closelymatch the number of adult mosquitoesproduced in that container. Interventionstargeted at the most pupa-productive typesof containers should therefore be reasonablyeffective and be quicker and cheaper thanattempts to remove or treat all the contain-

ers in which the vectors oviposit (Focks et al.,2OOO).The World Health Organization (WHO)

recently launched a multi-country study to

test a strategy of dengue surveillance andcontrol based on the identification andremoval/treatment of the most pupa-productive breeding sites (Focks andAlexander, 2006; Kroeger et al., 2006;Nathan et al., 2006). The aim of the presentstudy, part of this larger investigation, was

to evaluate the potential of pupal indices to

characterise the entomological componentof dengue transmission in Thailand, Thenumbers of suitable and productive types ofcontainers and numbers ofAe. aegypti pupaewere evaluated in both urban and ruralenvironments, over one rainy and one dryseason. The implications of the results, in

terms of the national strategies for denguecontrol, are discussed.

DENGUE AND PUPAL PRODUCTIVITY IN THAILAND 163

MATERIAL AND METHODS

Study SitesThe study sites were the rural village ofKhon Kaen, which lies, 100 m above sea

level, 6 km from Khon Kaen city, and a

busy street in the densely populated andvery urban suburbs of Chiang Mai city, at

300 m above sea level. Most ofthe houses inKhon Kaen are detached and follow a

traditional pattern, with a common room,sleeping rooms, a kitchen (outside the mainstructure but under the same roof) and, inthe more recent constructions, one or more

bathrooms. In most cases the bathroom/toilet is in a small shelter, 5-10 m from thehouse. In the Chiang Mai study area therewere a few houses of this type, most peopleliving in a more modern terraced house, ofone or two storeys with no kitchen but a

small shop occupying part of the groundfloor. In both study areas, most houses were

connected to a piped water supply.

Study Households and PeriodsIn Khon Kaen, 120 households, togetherholding 450-500 people, were surveyedthree times during a rainy season

(September 2004) and again at the begin-ning (December 2004) and end (March2005) of the following dry season. Theamounts of rainfall that fell in the study area

in the 2 weeks preceding each survey were

98, 0 and 0 mm, respectively.Similarly, in Chiang Mai, 100-120 house-

holds, together holding 330-400 people,were surveyed during the rainy season of2004 (in August 2004) and at the beginning(November 2004) and end (February 2005)of the dry season of 2004-2005. Theamounts of rainfall that fell in this urbanstudy area in the 2 weeks preceding each

survey were 146, 11 and 0 ram, respectively.

Survey MethodologyDuring each survey, teams (of two or threepeople each) searched each study household(including any yard) for 'wet' containers

(containing any water) and checked eachsuch container detected for mosquito larvaeand pupae. Large stagnant ponds, runningwater (rivers, canal, wells and sewagesystems), septic tanks and paddy fields wereignored. The water in the smaller containersfound positive for the immature stages ofmosquitoes was filtered (by emptying outthe water into a sieve or pumping the waterthrough a sieve) so that all the active larvaeand pupae of mosquitoes of the familyCulicidae could be collected and counted.The larger containers (i.e. those of >200litres) were difficult to empty (practicallyand because householders often wished toretain the water) and so all the visible pupaeand larvae in these were collected with netsand pipettes. On the day of their collection,the larvae were preserved in vials of 70%ethanol (for subsequent morphologicalidentification to species) and the pupaewere kept in glasses of water covered withnets until the adults emerged (with subse-quent morphological identification of theadults to species).The wet containers were classified,

according to their nature and use by thehouseholders (see below), and the source ofthe water in them (tap or rainfall) was

recorded.

Container Categories'Permanent' containers were generally large(100-3000 litres), used throughout the yearfor the storage of water, and filled with rainwater (if the water was only to be used forwashing, laundry or flushing a toilet) or tapwater (if the water was to be used fordrinking and cooking). They took the formof large clay jars (holding 100-200 litres),even larger, cement jars (holding 500-3000litres) and toilet tanks (holding 50-500litres).

'Useful' containers were those used dailyto store or carry water buckets used in thekitchen or garden, fish bowls, flower pots(each generally stood on a plate containingwater), vases and ant traps. All of these

] 64 BARBAZAN ETAL.

containers except the ant traps are generallyleft outdoors and fill naturally with rain

water, but tap water may also be added.'Discarded' containers bottles, broken

objects, cans and tyres (UT) containedonly rain water.

Data AnalysisEach 'wet' container was considered 'posi-tive' if it was found to contain at least one

larva or pupa ofAe. aegypti and 'productive'if it contained one or more Ae. aegyptipupae. The data were used to calculate the'classical' entomological indices that havebeen used for decades in routine surveys bypublic-health authorities and that are based

on the presence/absence of any immature

stages of the vector, not just the pupae.These indices were the Breteau index (thenumber of positive containers found/100surveyed houses), the container index (thenumber of positive containers/100 wet con-

tainers), and the house index (the number ofhouses with at least one positive container/100 surveyed houses). Similar pupal indices,based only on the presence/absence of Ae.aegypti pupae, were also calculated. Theproductivity of each type of container was

calculated as the mean number of Ae.aegypti pupae per wet container of that typeand also as a percentage of all the Ae. aegyptipupae detected during each survey at eachsite.

RESULTS

Overall, 3125 wet containers were surveyed1667 in Khon Kaen and 1458 in Chiang

Mai (see Table). Of these, 685 were positiveforAe, aegypti and 354 contained Ae. aegyptipupae. A total of 4294 Ae. aegypti pupae wascollected. On average, each surveyed househad about eight wet containers during therainy season and four during the dry season.

The larvae and/or pupae ofAe. aegypti werefound in 10%-35% of the surveyed houses.Very few wet containers (<1% of those

found positive for Ae. aegypn) were found> 10 m from a house.Of the containers found harbouring culi-

cine larvae and/or pupae, 58% of those seen

in the rainy season and 81% of those seen at

the end of the dry season were positive forAe. aegypti. The other mosquito speciescollected were Ae. albopictus, Culex quinque-fasciatus, Cx. gelidus, Cx. tritaeniorhynchus,Cx. (Culiciomyia) sp., Cx. (Lutzia) sp. andunidentified species of Toxorhynchites,Anopheles, Armigeres and Orthopodomyia.Although immature stages of Ae. albopictus(a secondary vector of dengue viruses inThailand) were collected from 176 contain-

ers during the rainy season (90 in KhonKaen and 86 in Chiang Mai), none was seen

at the end of the dry season. Culex gelidus,Cx. tritaeniorhynchus and Cx. quinquefascia-tus are potential vectors of Japanese ence-

phalitis virus. Culex (Lutzia) andToxorhynchites larvae are predators of mos-quito larvae but their numbers were very lowduring the rainy season (when they were

detected in < 1% of the containers found to

harbour immature stages) and no immaturestages of these taxa were seen during the dryseason. The most abundant positive con-

tainers were clay jars in Khon Kaen andbuckets in Chiang Mai (see Table).Classical Breteau indices were very high in

every survey and >100 during the rainyseason. The level of positive correlationbetween the number of wet containers/house and the number of positive contain-ers/house decreased over the study period,with r-values in the rainy season, at thebeginning of dry season and at the end ofdry season of 0.53, 0.35 and 0.31, respec-tively, in Khon Kaen, and 0.59, 0.49 and0.38, respectively, in Chiang Mai.From the first survey (in the rainy season)

to the third (at the end of the dry season),there was a gradual decrease in the numbersof productive containers, from 126 to 45 inKhon Kaen and from 61 to 18 in ChiangMai (see Table). Although the numbers ofAe. aegypti pupae collected generally fol-lowed the same trend, two highly productive

TABLE 1. The numbers and types of wet and Aedes aegypti-positive containers, and the numbers of Ae. aegypti pupae, observed in the two study areas

Type of container:

Permanent Useful Discarded

Clay Cement Toilet Ant Bottles,Area and jars jars tanks Buckets Fishbowls Vases traps broken objects Tyres

KHON KAN

Rainy season

No. of wet containers 217 98 111 121 54 11 49 100 41

No. of positive containers 57 6 48 30 29 25 10No. of pupae 180 77 530 103 0 0 58 55 41

Beginning of dry season

No. of wet containers 157 71 103 39 21 0 30 5 0

No. of positive containers 46 3 66 15 0 21 0

No. of pupae 337 9 1097 96 5 0 31 8 0End of dry season

No. of wet containers 132 41 73 38 28 3 17

No. of positive containers 36 5 26 5 7 0

No. of pupae 220 66 370 20 0 0 25 0 0

CHIANG MAI

Rainy season

No. of wet containers 49 2 24 172 46 102 32 79 24

No. of positive containers 7 0 6 23 6 24 9 19 5

No. of pupae 34 0 24 40 56 29 16 13 48

Beginning of dry season

No. of wet containers 45 6 25 146 68 33 9 21 5No. of positive containers 6 0 5 9 0 8 0 5

No. of pupae 4 0 35 11 0 10 0 3 2

End of dry season

No. of wet containers 35 8 22 130 77 59 16 13 0

No. of positive containers 0 10 11 5 6 2 0

No. of pupae 0 0 293 200 13 12 0 0 0

Other

87

2015

17

625

2

13335

71

70810

114

0

All

889227

1059

4431591608

33583703

663

134

331

42842

75

36740

518

166 BARBAZAN ETAL.

25 60

20 f 50

o 4015

30

z 20

510

0I1

m 0

Container type

FIG. 1. The production ofAedes aegypi pupae in wet containers of each of 10 types, shown as the numbers ofpupae/wet container (), the numbers ofpupae/pupa-positive container (3), and the percentages of all the pupaecollected (0). The data shown combine the results of all three surveys in rural Khon Kaen and urban Chiang Mai.

containers discovered in Chiang Mai at theend of the dry season (a toilet tank with 268Ae. aegypti pupae and a bucket with 127such pupae) led to numbers that went

against the general downward trend.Together, these highly productive contain-

ers held >76% of all the .de. aegypti pupaecollected in Chiang Mai at the end of thedry season. The number of Ae. aegyptipupae/person was always higher in KhonKaen (2.2 in the first survey, 3.3 in thesecond and 1.5 in the third) than in ChiangMai (with corresponding values of 1.0, 0.2and 1.3, respectively).

In each survey, wet containers of almostall the categories identified were foundpositive and productive, the only exceptionbeing the end-of-dry-season survey inChiang Mai, when only four categories ofcontainer (toilet tanks, buckets, vases andfish bowls) were found to be producing .de.aegypti pupae. Only three types of containereach produced > 10% of the pupae collected

over the study period (Fig. 1): toilet tanks(54.7%), clay jars (18.0%) and buckets(10.9%). Together, toilet tanks and clay

jars produced 81.1% of the Ae. aegyptipupae collected in Khon Kaen and 42.2%of those collected in Chiang Mai. The mean

number of Ae. aegypti pupae in each 'wet'clay jar varied from 0 to 2.1 according to thestudy site and survey (with overall means of1.5 in Khon Kaen and 0.3 in Chiang Mai).The corresponding values for 'wet' toilettanks were 1.0 to 10.5 (with overall means

of 7.0 in Khon Kaen and 5.0 in ChiangMai),At the beginning of the dry season in

Khon Kaen, 50% of the productive clay jarsand 39% of the productive toilet tanks onlyharboured one to four Ae. aegypti pupaeeach. In each study area, the contributionmade by toilet tanks to the number of Ae.aegypti pupae produced was lower in therainy season than in the dry season. Clay jarscontributed a similar percentage of the Ae.aegypti pupae collected in both seasons

whereas the contribution made by buckets

was higher in the rainy season than in thedry.Although each productive cement jar,

fishbowl or tyre tended to hold a relatively

DENGUE AND PUPAL PRODUCTIVITY IN THAILAND 167

120

100

80

60 o

O20

40 60 80 100 120 140 160 180 200

Classical Breteau index

FIG. 2. The relationship between the classical Breteauindices and the 'pupal' Breteau indices estimated ineach of the three surveys in Khon Kaen (0) and ChiangMai ((C)). The line represents the linear regression[pupal index=(0.66 classical index)-20.1; Re=0.93].

high number of Ae. aegyptz pupae, these

types of container were rare or rarelypositive and so their contribution to thetotal number of pupae was low (Fig. 1).No correlation was observed between the

number of wet containers detected in eachhouse and the number of Ae. aegypti pupaecollected/house. Each of the 'classical' ento-

mological indices, calculated from theabsence/presence of any immature stages,was about twice as high as the correspond-ing pupal index. There was, for example, a

good positive correlation between the clas-sical Breteau indices and the 'pupal' Breteauindices (Fig. 2). There was also a positivecorrelation between the number of positivewet containers (in each survey at each site)and an index created by multiplying thenumber of wet containers by the number of

pupae collected (r=0.85).

DISCUSSION

In the present study, the values recorded forthe classical entomological indices indicatedhigh Ae. aegypti densities in both study

areas, especially during the rainy season.

The World Health Organization considers

areas with classical Breteau indices of >5and house indices of >5 to be at significantrisk of arbovirus transmission (WHO,1997). All the entomological indices andthe numbers of pupae/person were alwaysfound to be higher in rural Khon Kaen thanin urban Chiang Mai. In Chiang Mai most

houses are very close to each other and thereis simply not enough space for large storagecontainers, leading to the more frequent use

of buckets. Recent economic developmentin Khon Kaen, like that occurring in manyrural areas of Thailand, has, however,brought with it the water-supply systems,electricity, consumerism and shopping areas

that were once restricted to urban environ-

ments. It was, therefore, no surprise to find

as many 'useful' and 'discarded' containersin Khon Kaen as in Chiang Mai. In general,the most productive containers were of the

same types at both sites.During the dry season, the total numbers

of water-containing 'useful' and 'discarded'containers fell, increasing the importance ofthe permanent storage containers (in terms

of the numbers of wet or positive containersand pupal productivity). Jars, toilet tanksand buckets were found to be the most

productive containers in the present studyand in previous investigations in Thailand(Tonn et al., 1969; Strickman andKittayapong, 2002; Chareonviriyaphapet al., 2003) and there is little in the presentstudy to indicate that dengue-control stra-

tegies in Thailand need to be modifiedaccording to land use (Focks andAlexander, 2006).

In India, an unexpected effect of theconnection of households to a piped supplyof water was increased pupal productionduring the dry season, in permanent con-

tainers filled with tap water (Mahadev andGokhale, 1988). The recent connection ofKhon Kaen to a piped water supply hasprobably had the same effect.The characteristics of storage containers

favourable to the development ofAe. aegypti

168 BARBAZAN ETAL.

are a large volume (generally >100 litres),an outdoor situation in the sun, frequentfilling by rainfall or household owners

(which helps to induce egg hatching) andthe presence in the water of organic materi-als on which the larvae can feed(Christophers, 1960). In the present studyareas, there appeared to be negligible pre-dation of the immature stages ofAe. aegyptiby the larvae of other mosquitoes.On a community scale, the vector pro-

ductivity of an area is dependant not only onthe presence of suitable containers but also

on the seasonal dynamics of the Ae. aegyptipopulation. At the end of the rainy season,in the present study areas, the small contain-

ers that had been kept filled by rainfall beginto dry up and the large permanent contain-

ers used for water storage represent an

increasing proportion of the wet containersavailable to mosquitoes. In Khon Kaen, forexample, the permanent containers repre-sented 49% of the wet containers in therainy season but had already become 74% ofthe wet containers by the beginning of thedry season. Early in the dry season, whilethe moderate temperatures and high humid-ity needed by the adult mosquitoes persist,there may be at least as many gravid femaleAe. aegypti seeking oviposition sites, and justas many containers being colonized by thesemosquitoes, as during the rainy season

(Sheppard et al., 1969). This may explainwhy, with many large containers perma-nently filled with water, pupal productivityin Khon Kaen appeared higher at thebeginning of the dry season than in therainy season, despite the absence of rain and

an overall decrease in the total number of

wet containers. By the end of the dry season,however, several months of low humidityand relatively low (<20C in January) andthen high temperatures (>35C) havereduced the numbers of gravid Ae. aegyptiso that not only is pupal production low butthe 'positivity' of the wet containers thatremain available is also low. The numberof pupae observed is an indicator of thedensity of gravid females seeking favourable

containers in which to oviposit. The highlevel of correlation observed between thenumber of positive containers and (numberof wet containers x number of pupae) illus-trates the impact of environmental charac-teristics (climate, number and types of wet

container) and Ae. aegypti populationdynamics on pupal production (Edmanet al., 1998; Getis et al., 2003).

In field surveys, unfortunately, generaltrends in pupal production can be maskedby sampling bias, particularly whether or not

rare containers that are extraordinarilyproductive are detected. When Chiang Mai

was surveyed at the end of the dry season,for example, >76% of all the pupaecollected came from just two containers.The impact of such containers onAe. aegyptipopulation at the village scale is probablyminor, and such highly productive contain-

ers need not form part of any controlstrategy if they are exceptional (Focks andAlexander, 2006).Those involved in vector surveillance also

need to balance the rarity of each type ofcontainer against the pupal production ineach container of that type. In the presentstudy, the number of pupae per positivecontainer was relatively high (6.2) for thefishbowls but only 12 of the 294 wet

fishbowls surveyed were found positive (all12 being fish-free) and so the number of

pupae produced/wet fishbowl was relativelylow (0.25).

Focusing vector-surveillance activities on

pupae (rather than larvae and pupae) and onthe types of container that contribute most

to pupal production would facilitate theroutine work of MOPH teams, who couldignore the less productive types of container,which, as luck would have it, are often thosethat are hardest to investigate (Strickmanand Kittayapong, 2003). During the rainy-season survey, almost half of the containersfound to contain the immature stages ofmosquitoes were negative for Ae. aegypti. Ifsurveillance were restricted to pupae thenthe species could be identified, as in the

present study, by keeping the pupae until

DENGUE AND PUPAL PRODUCTIVITY IN THAILAND 169

the adults emerge. Although this is a

relatively simple procedure, it is alsotime-consuming.The possibility that the positive correla-

tion seen, in the present study, between theclassical Breteau indices and the pupalBreteau indices (Fig. 2) is not only fixedbut also widely applicable is an interestingone. It indicates, for example, that historicalvalues for pupal Breteau indices could beestimated from the values for the classicalBreteau indices that have been regularlycalculated, for many years, by public-healthauthorities in dozens of countries withendemic dengue.

In Thailand, a control strategy that istargeted only at the most productive types ofcontainer would be simplified by the easyidentification of those types, although, as

the most productive types are kept and filledwith water by household members, anycontrol strategy needs to be based on

community participation (Gubler, 1989;Perez-Guerra et al., 2005). To be sustain-able, those planning such interventions alsoneed to provide health education (in terms

of the methods that are likely to be not onlyeffective but also feasible) and a clearlydefined goal (significant risk reduction).Control methods based only on the elimina-tion of the most productive containers(which is not a practical option for most

storage containers) or on covering them, to

prevent oviposition in them, could simplyreorientate gravid female mosquitoestowards other types of containers and/orincrease the dispersal of the females (Edmanet a/.,1998). The present results indicatethat female Ae. aegypti looking for an

oviposition site are extremely opportunistand flexible in their choice, laying eggs in

wet containers of all types. Since breedingsuccess in the types of wet containerconsidered relatively unproductive in thepresent study is, however, generally low(e.g. because such containers are smalland more susceptible to drying out or

running out of food for the larvae), thenthe diversion of gravid females to such

containers may still cause a marked reduc-tion in pupal (and thus adult) productivity.The use of a cover seems the only acceptablemosquito-control intervention for contain-

ers in which drinking water is stored, as

most people do not want to drink water thathas been treated with a larvicide. Slow-release formulations of larvicides (Bacillusthuringiensis israeliensis, temephos, pyriprox-ifen etc) or copepods could be used in otherjars and in toilet tanks. Buckets pose a

particular problem as they generally becomepositive when they are left forgotten out-

doors and fill with rainwater. Their treat-

ment with a larvicide appears impractical,since any larvicide would be quickly elimi-nated as the buckets overflow with water or

are emptied by hand, and water frombuckets is often used in food preparation.Communities need to be told of the threatposed by buckets and be encouraged to

keep buckets empty or covered if they are

needed or to eliminate them if they are not

required.Compared with a wide-spread untargeted

intervention against all wet containers, a

targeted 'assault' against just clay storagejars and toilet tanks (which represented<50% of all the wet containers in KhonKaen and <14% of those in Chiang Mai)should be relatively easy, quick and cheap. Ifsustained, such an assault is likely to have a

cumulative negative effect on Ae. aegyptidensities and may even decrease the numberof potential vectors to a level that is too low

to support the stable transmission of dengueviruses (Focks et al., 2000). As Khon Kaenand Chiang Mai are, respectively, fairlyrepresentative of rural and urban areas ofThailand, the most productive types ofcontainer identified in the present studymay be those that are most productivethroughout Thailand. They could there-fore be the main targets of a nationalsurveillance and control strategy. In othercountries, other types of container needto be targeted. In the Americas, for example

in Trinidad, Puerto Rico, Cuba, Peruand E1 Salvador (Moore et al., 1978; Focks

170 BARBAZAN ETAL

and Chadee, 1997; Scott et aL, 2000; Hayeset aL, 2003; Focks and Alexander, 2006)

discarded containers are the most

productive.On their own, entomological indices

cannot provide an absolute value of the riskof dengue transmission, since host density,'herd immunity', the viruses present andenvironmental factors, especially climateand urbanization, all contribute to such risk(Kuno, 1995). In any given endemic area,however, where these other factors are fairlystable, it is the number of potential vectors

that is the main factor triggering changes inthe incidence of dengue. A control strategybased on the treatment of the types ofcontainer that contribute most pupae, withregular monitoring and data dissemination

so that participating communities can

always see how close they are to achievingsome pre-set goal, therefore appears to bepromising. At the same time, at-risk com-

munities could also be encouraged to

eliminate all those less-productive contain-

ers, such as used tyres (Fig. 1), that are

worthless to them but still contribute to

vector populations.

ACKNOWLEDGEMENTS. This study could not

have been done without the support of theMinistry of Public Health in Thailand,especially that of the Dengue FeverControl Division, Drs T. Suwich and S.Seevigaa, and the provincial offices and localvolunteers working with the ResearchCentre for Emerging Viral Disease. Theauthors thank Dr M. B. Nathan and Pro-fessor A. Kroeger, both from the UNICEF/UNDP/World Bank/WHO Special Pro-

gramme for Research and Training inTropical Diseases (TDR), for their con-

stant support during the field work and theanalysis of the results. This investigationreceived financial assistance from TDR, theDepartment of Technical Cooperation inThailand, and the French Institut deRecherche et Ddveloppement's Unit deRecherche 178 (IRD-UR 178).

REFERENCES

Chan, K. L. (1985). Methods and indices used in thesurveillance of dengue vectors. Mosquito BorneDiseases Bulletin, 1, 79-87.

Chareonviriyaphap, T., Akratanakul, P., Nettanomsak,S. & Huntamai, S. (2003). Larval habitats anddistribution patterns of Aedes aegypti (Linnaeus)and Aedes albopictus (Skuse), in Thailand. SoutheastAsian Journal of Tropical Medicine and Public Health,34, 529-535.

Christophers, S. R. (1960). Aedes aegypti (L.). TheYellow Fever Mosquito. London: CambridgeUniversity Press.

Edman, J. 1)., Scott, T. W., Costero, A., Morrison, A.C., Lorentz, L. K., Clark, G. G., Strickman, D. &Kittayapong, P. (1998). Aedes aegypti (Diptera:Culicidae) movement influenced by availability ofoviposition sites. Journal of Medical Entomology, 35,578-583.

Focks, D. A. & Alexander, N. (2006). MulticountryStudy of Aedes aegypti Pupal Productivity SurveyMethodology: Findings and Recommendations.Document: TDPUIRM/DEN/06.1. Geneva: WorldHealth Organization.

Focks, D. A. & Chadee, D. D. (1997). Pupal survey: an

epidemiologically significant surveillance method forAedes aegypti: an example using data from Trinidad.American Journal of Tropical Medicine and Hygiene, 56,159-167.

Focks, D. A., Brenner, R. J., Hayes, J. & Daniels, E.(2000). Transmission thresholds for dengue in

terms of Aedes aegypti pupae per person withdiscussion of their utility in source reduction efforts.American Journal of Tropical Medicine and Hygiene, 62,11-18.

Getis, A., Morrison, A. C., Kenneth, G. & Scott, T. W.(2003). Characteristics of the spatial pattern of thedengue vector, Aedes aegypti, in Iquitos, Peru.American Journal of Tropical Medicine and Hygiene,69, 494-505.

Gubler, D. J. (1989). Aedes aegypti and Aedes aegypti-borne disease control in the 1990's: top down or

bottom up. American Journal of Tropical Medicine andHygiene, 40 571-578.

Gubler, D. J. (1997). Dengue and dengue hemorrhagicfever: its history and resurgence a global publichealth problem. In Dengue and Dengue HemorrhagicFever, eds Gubler D. J. & Kuno, G. pp. 1-22.Walling'ford, U.K.: CAB International.

Hayes, J. M., Garcia-Rivera, E., Flores-Reyna, R.,Suarez-Rangel, G., Rodriguez-Mata, T., Coto-Portillo, R., Baltrons-Orellana, R., Mendoza-Rodriguez, E., de Garay, B. F., Jubis-Estrada, J.,Hernandez-Argueta, R., Biggerstaff, B. ]. & Rigau-Perez, J. G. (2003). Risk factors for infection during a

severe dengue outbreak in E1 Salvador in 2000.

DENGUE AND PUPAL PRODUCTIVITY IN THAILAND 171

AmeriCan Journal of Tropical Medicine and Hygiene, 69,629-633.

Kroeger, A., Nathan, M. B., Hombach, J., Dayal-Drager, R. & Weber, M. W. (2006). Dengue researchand training supported through the World HealthOrganization. Annals of Tropical Medicine andParasitology, 100 (Suppl. 1), $97-S101.

Kuno, G. (1995). Review of the factors modulatingdengue transmission. Epidemiological Reviews, 17,321-335.

Mackenzie, J. S., Gubler, D. J. & Petersen, L. R.(2004). Emerging flaviviruses: the spread and

resurgence of Japanese encephalitis, West Nile anddengue viruses. Nature Medicine, 10 (Suppl. 12), 98109.

Mahadev, P. V. M. & Gokhale, M. D. (1998). Water

storage and Aedes aegypti breeding in multistoreyedflats: a case study in Pune, Maharashtra, India.Entomon, 23, 173-183.

Moore, C. G., Cline, B. L., Ruiz-Tiben, E., Lee, D.,Romney-Joseph, H. & Rivera-Correa, E. (1978).Aedes aegypti in Puerto Rico: environmental determi-

nants of larval abundance and relation to denguevirus transmission. American Journal of TropicalMedicine and Hygiene, 27, 1225-1231.

Nathan, M. B. (1993). Critical review ofAedes aegypticontrol programs in the Caribbean and selectedneighboring countries. Journal of the MosquitoControl Association, 9, 1-7.

Nathan, M. B., Focks, D. A. & Kroeger, A. (2006).Pupal/demographic surveys to inform dengue-vectorcontrol. Annals of Tropical Medicine and Parasitology,100 (Suppl. 1), S1-$3.

Okanurak, K., Sommani, S. & Indaratna, K. (1997).The cost of dengue haemorrhagic fever in Thailand.Southeast Asian Journal of Tropical Medicine and PublicHealth, 28, 711-717.

Perez-Guerra, C. L., Seda, H., Garcia-Rivera, E. J. &Clark, G. G. (2005). Knowledge and attitudes in

Puerto Rico concerning dengue prevention. RevistaPanamericana de Salud Pfblica, 17, 243-253.

Reiter, P. & Gubler, D. J. (1997). Surveillance andtrol of urban dengue vectors. In Dengue and DengueHemorrhagic Fever, eds Gubler, D. J. & Kuno, G.

pp. 425-462. Wallingford, U.K.: CAB International.Scott, T. W., Morrison, A. C., Lorenz, L. H., Clark, G.

G., Strickman, D., Kittayapong, P., Zhou, H. &Edman, J. D. (2000). Longitudinal studies of Aedesaegypti (Diptera: Culicidae) in Thailand and PuertoRico: population dynamics. Journal of MedicalEntomology, 37, 77-88.

Sheppard, P. M., MacDonald, W. W., Tonn, R. J. &Grab, B. (1969). The dynamics of an adult popula-tion of Aedes aegypti in relation to dengue haemor-rhagic fever in Bangkok. Journal of Animal Ecology,38, 661 702.

Strickman, D. & Kittayapong, P. (2002). Dengue and its

vectors in Thailand: introduction to the study andseasonal distribution ofAedes larvae. American Journalof Tropical Medicine and Hygiene, 67, 247-250.

Strickman, D. & Kittayapong, P. (2003). Dengue andits vectors in Thailand: calculated transmission riskfrom total pupal counts of Aedes aegypti and associa-tion of wing-length with aspects of larval habitats.American Journal of Tropical Medicine and Hygiene, 68,209-217.

Tonn, R. J., Sheppard, P. M., MacDonald, W. W. &Bang, Y. H. (1969). Replicate surveys of larvalhabitats of Aedes aegypti in relation to denguehaemorrhagic fever in Bangkok, Thailand. Bulletinof the World Health Organization, 40, 819-829.

Tun, L. W., Kay, B. H. & Barnes, A. (1995).Understanding productivity, key to Aedes aegyprisurveillance. American Journal of Tropical Medicineand Hygiene, 53, 595-601.

World Health Organization (1997). DengueHaemorrhagic Fever." Diagnosis, Treatment, Preventionand Control, 2nd Edn. Geneva: WHO.