The 2000 Epidemic of Rift Valley Fever in Saudi Arabia: Mosquito Vector Studies

Home , Aedes vexans, Culex tritaeniorhynchus

Medical and Veterinary Entomology (2002) 16, 245±252

The 2000 epidemic of Rift Valley fever in Saudi Arabia: mosquito vector studies

P. G. JUPP* ,A.KEMP* , A. GROBBELAAR* ,P.LEMAN* ,F.J.BURT* , A. M. ALAHMEDy,D.ALMUJALLIz,M.ALKHAMEESz and R. SWANEPOEL* *Special Pathogens Unit, National Institute for Communicable Diseases and Department of Virology, University of the Witwatersrand, South Africa, yDepartment of Plant Protection, College of Agriculture, Riyadh, Saudi Arabia and zMinistry of Agriculture and Water, Riyadh, Saudi Arabia

Abstract. In mid-September 2000, Rift Valley fever (RVF) virus was diagnosed as the cause of infection in humans and livestock in Jizan Region, Saudi Arabia. This is the first time that this arbovirus has been found outside Africa and Madagascar. Collections of mosquitoes (Diptera: Culicidae) were therefore undertaken (from 25 September to 10 October) at eight sites during the epidemic to obtain mosquitoes for attempted RVF virus isolation. Among 23 699 mosquito females tested, isolations of RVF virus were made from six of 15 428 Culex (Culex) tritaeniorhynchus Giles and from seven of 8091 Aedes vexans arabiensis Patton. Minimum mosquito infection rates per 1000 at sites with infected mosquitoes were 0.3±13.8 Cx. tritaeniorhynchus and 1.94±9.03 Ae. v. arabiensis. Viral activity moved northwards as collecting was in progress and collectors `caught up' with the virus at the two most northerly sites on the last two trapping evenings. Other species occurred in small numbers and were identified but not tested. Both Cx. tritaeniorhynchus and Ae. v. arabiensis were susceptible to RVF virus and transmitted between hamsters, and an additional quantitative test with Cx. tritaeniorhynchus showed

that 71±73% of mosquitoes became infected after ingesting 6.9±7.9 log10 FFU/mL of virus; transmission rates were 10% (post-infection day 14) and 26% (post- infection day 20). It was concluded that both species were vectors on grounds of abundance, distribution, preference for humans and sheep, the virus isolations and vector competence tests.

Key words. Aedes vexans arabiensis, Culex tritaeniorhynchus, arbovirus vectors, mosquitoes, light-suction traps, Rift Valley fever virus, Jizan, Saudi Arabia.

Introduction Yemen, and to have continued until the end of November. Although the epidemic was centred in Jizan Region and In mid-September 2000, it was confirmed that Rift Valley northern Yemen, it subsequently extended northwards fever (RVF) infection had occurred in both humans and into the Aseer and Al Quenfadah health regions. This is livestock in the Kingdom of Saudi Arabia (CDC, 2000a, b). the first time that RVF virus has been reported outside The epidemic appears to have started simultaneously in the Africa and Madagascar. Jizan Region in the south-west, as well as in neighbouring On 24 September, a team of four scientists from the Special Pathogens Unit of the National Institute for Vir- ology in Johannesburg, South Africa, arrived in Jizan at the Correspondence: Alan Kemp, National Institute for Communic- invitation of the Saudi Arabian Ministry of Agriculture and able Diseases, Private Bag X4, Sandringham 2131, South Africa. Water to help investigate the epidemic. The South African E-mail: [email protected] team members stayed in Saudi Arabia for up to 4 weeks.

# 2002 The Royal Entomological Society 245 246 P. G. Jupp etal.

One aspect of the study was a serological survey in livestock inundated by rain but some fields are irrigated with under- throughout Jizan Region and the concomitant vaccination ground water in the absence of rain. of these sheep and goats and a smaller number of cattle and camels. That work will be the subject of a separate paper, while the present paper deals with the entomological aspect Mosquito collecting of the study. In order to determine the vectors of the virus, mosquitoes were collected in the vicinity of human and/or Four methods were used for collection of adults: net traps livestock infections as soon as possible after such infections baited with solid carbon dioxide (CO2) (Jupp & McIntosh, had occurred. The identities of the prevalent mosquito 1967), battery-operated CO2-baited suction traps (ABC species were established, these were tested for the presence trap: Clarke Environmental Mosquito Management Inc., of RVF virus, their host preferences were determined and Roselle, Illinois, U.S.A.) run with or without a light source some observations were made on their breeding places (6 volt Phillips 150 ma bulb), and landing-biting catches on (main larval sites). Additionally, limited laboratory vector human bait. The suction traps were suspended inside and out- competence tests were undertaken. side on the walls of a sheep sleeping hut with a control trap sited 60 m away from the hut. The human-baited catches were also conducted in the manner described by Jupp & McIntosh (1967) by one or two catchers (immunized team Materials and methods members). All the adult collections were undertaken from 18.00 to 20.30 hours, sunset was at about 18.45 hours. Study area Owing to logistical problems and other difficulties, mosquito trapping was limited to 11 evenings. During the daytime Saudi Arabia has a `hotdesert'climate(Fig. 1). The Jizan a few larval collections were made at some of the sites. region is in the south-west, with its western boundary the Larvae were reared in the field to obtain both male and Red Sea, the Yemen Republic to the south and east and the female adults as well as their associated larval exuviae. Aseer region to the north. Jizan Region has been described Adult males were also collected resting in vegetation. briefly by (Ageel & Amin, 1997). There are three distinct zones: `mountain' (the Sarawat range) (2000±2500 m a.s.l., >300 mm rain/year); `hill' (400±600 m a.s.l., <300 mm/year) Mosquito identifications and `coastal plain' (<400 m a.s.l., little if any rain each year). Collection sites were selected after surveillance of the Culicine mosquitoes were identified according to the Jizan Region by helicopter in two phases. First, a south± keys and/or descriptions of Edwards (1941), Mattingly & north survey, close to the Yemen border and also in the Knight(1956), Hopkins (1952), Sirivanakarn (1976) and coastal plane was undertaken. This revealed that the only Jupp (1996). The designation of the Aedes (Aedimorphus) suitable habitat for mosquito breeding existed in the wadis, mosquito arabiensis as a subspecies of vexans followed which run from the Sarawat Mountain range down to the White (1975). Anophelines were identified from the keys Red Sea. Wadi Jizan was then chosen for a second survey of Gillies & De Meillon (1968). Mounts of larval exuviae because it had been associated with the RVF outbreak and and male genitalia were made for confirmation of species also because it was typical of the wadis of the region. Five of identification. our eight collection sites represented various stages of the Wadi Jizan and its tributaries, all within an 8 km radius of Jizan dam (90±300 m a.s.l.): near-Falas westof thedam Virus isolation (coastal plain), a site each on the south and east banks of the dam and two sites in the foothills of the Sarawat Moun- Most mosquito pools contained 200 mosquitoes, which tains, near the Yemen border. One other site in the coastal were processed to obtain a supernatant fluid as described by plain was situated further north near Baysh (50 m a.s.l.). Jupp et al. (2000). Each supernatant was tested by reverse The remaining two sites were situated in the hill zone: transcription-polymerase chain reaction (RT-PCR) to Al Khawbah to the south near the Yemen border identify the presence of RVF viral RNA. If the result of this (400 m a.s.l.) and Harub in the foothills of the mountains test was positive or possibly positive, the supernatant was to the north (500 m a.s.l.). Rainfall at all sites was considerably inoculated into suckling mice on the same day. If the mice higher than usual during 2000, for example, 262 mm fell died in the expected death time for RFV virus, mouse brain from June to September at Al Aridah, midway between was passaged into Vero cells, where the presence of RVF Jizan Dam and the Sarawat Mountains (Fig. 1), so that viral antigen was confirmed by the indirect immunofluor- sites were much wetter than usual, particularly those in the escent antibody test (IFAT) using the procedure described catchment of the Jizan Dam (the Sadd Wadi Jizan) and by Shope & Sather (1979). For those collection sites with Wadi Baysh (with Wadi Harub as one of its tributaries, viral activity, mosquito infection rates were calculated: the via Wadi Wusa and Wadi Shahdan), providing good `minimum infection' rate is the number of infected mosqui- mosquito larval habitats. Such habitats are described in toes per 1000 calculated arithmetically, whereas the `statis- Table 1 for each site. Agricultural fields may become tical infection rate' is the number per 1000 calculated by the

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 Rift Valley fever in Saudi Arabia 247

Fig. 1. Map of Jizan Region of Saudi Arabia, showing mosquito collection sites: 1, Al Khawbah; 2, near Falas; 3, Jizan Dam east bank; 4, Jizan Dam south bank; 5, Al Refoud; 6, near Al Humayrah; 7, near Baysh and 8, Harub.

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 248 P. G. Jupp etal.

Table 1. Site descriptions and numbers of mosquitoes (per trap hour) collected in net and light (suction) traps baited with CO2, ateight different sites in Jizan Region, with proportions of the two principal species.

Mosquitoes/trap hour (total collected) Species ratio Cx. tritaeniorhynchus : Collection site Site description Net trap Light trap Ae. v. arabiensis 1. Al Khawbah Ground pools in stream bed 12 (31) NT 9 : 1 16470 N43130 E 2. Near Falas Temporary lake in Wadi, Jizan 113 (167) 42 (416) 1.4 : 1 17010 N42550 E 3. Jizan Dam, eastside Sedge bed at periphery of large dam 18 600 (37 200)2 NT 1 : 01 17030 N42580 E 4. Jizan Dam, southside Sedge bed at periphery of large dam 500 (2500) 140 (1400) 9 : 1 17020 N42580 E 5. Al Refoud Small dams with emergent vegetation in 813 (4875) NT 4 : 1 17070 N43000 E Wadi Sayal 6. Near Al Humayrah Flooded agricultural fields 483 (15 100)2 800 (10 000)2 1:1 17050 N43020 E 7. Near Baysh Marshy reed beds and flooded agricultural 292 (1458) 12 (119) 1 : 501 17200 N42340 E fields in Wadi Baysh 8. Harub Dry agricultural fields with crops 260 (1300) NT 1 : 3 17260 N42520 E

1These ratios may have been distorted because of insecticidal spraying at these sites prior to mosquito trapping. 2These are approximate numbers. NT no trapping.

method of Chiang & Reeves (1962). The RT-PCR pro- curves, fluorescence vs. temperature (data not shown), cedure had the same RNA extraction as previously described were converted to melting peaks by the LightCycler instru- (Jupp et al. 2000) but because a LightCycler was used, the ment's software, allowing easy interpretation of results. remainder of the procedure differed. LightCycler RT-PCR A melting point of 84C was registered for RVF virus. After and melting curve determinations were performed in 20 mL each LightCycler run, agarose gel electrophoresis with volumes in glass capillaries in a high speed thermal cycler a Tris-borate 1.2% agarose gel, followed by DNA staining (LightCyclerTM, Roche Diagnostics GmbH, Mannheim, with ethidium bromide, was performed to have an inde- Germany) that uses hot air for heating (Wittwer et al., pendent validation check of the presence of the amplicon. 1997). Master mixes were based on a ready-to-use LightCycler-RNA Amplification SYBR Green kit (Roche Diagnostics). The reaction mixture was supplemented Mosquito rearing and vector competence experiments with a final concentration of 5 mM MgCl2 and 0.15 pmol each of primer RVF no. 1 (771/5 0CCA AAT GAC TAC A limited number of live adult field mosquitoes of CAG TCA GC30) and primer RVF no. 2 (1139/50GAC unknown age, collected at Harub, were available for vector AAA TGA GTC TGG TAG AC30). The primers were competence experiments. During their feeds on hamsters to selected to amplify a target region of 369 bp encoding a por- obtain eggs, transmission of virus occurred to two hamsters. tion of the G2 glycoprotein gene of the medium (M) segment A cage of F1 Culex tritaeniorhynchus mosquitoes were of RVF virus. The position of the primers correspond with reared successfully from the field mosquitoes using a the viral complementary DNA sequence of the M segment method similar to that of Jupp & Brown (1967), except of the Zagazig human strain of RVF (ZH501) (Collett hamsters were used for bloodmeals and the cages were et al., 1985). 35 Â 35 Â 35 cm3. Furthermore, the larval water tempera- Reverse transcription was performed at 50C for 15 min ture was increased to 28C (air temperature 30C) and followed by an initial denaturation at 94C for 1 min. sterile sand was added to each larval rearing tray to simu- Cycling was performed for 35 cycles of denaturation late Saudi Arabian conditions more closely. There were too (94C for <1 s), annealing (50C for 10 s) and extension few live Aedes vexans arabiensis adults to permit the rearing (72 C for 20 s). The ramp rates were programmed at of sufficientF 1 mosquitoes for a vector competence test, but 20C/s unless otherwise stated. At the end of the amplification the available females were placed into individual tubes for cycles, melting curves were generated by denaturation of oviposition. the reaction at 95C for <1 s, holding the samples at 65C The Cx. tritaeniorhynchus used for the vector competence for 10 s, heating the samples to 95C for <1 s (with a ramp tests were 20±28 days old. These were infected by allowing rate of 0.2C/s) and finally cooling to 40C. Melting them to feed on hamsters that were viraemic 30 h after

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 Rift Valley fever in Saudi Arabia 249 intraperitoneal inoculation of the AN1830 strain of RVF small numbers, which were identified but not tested. These virus. The virus had been passed five times in infant mice. were Aedes (Aedimorphus) vittatus (Bigot) (57), Ochlerotatus The concentration of virus in the blood of each hamster was caspius (Pallas) (6), Culex (Culex) pipiens complex (1) and determined in a sample taken just before feeding. For this, Anopheles (Cellia) azaniae Bailly Choumara (2). The last 0.1 mL of blood was removed from the heart and diluted species is a new distribution record for Saudi Arabia and with 0.9 mL of 10% Leibovitz medium. The titration was the two individuals were collected in a net-trap at Al Refoud done in Vero cells as described elsewhere (Jupp et al. 2000) (17070 N43000 E). The identification of the two predom- and titres were expressed as log10 fluorescentfocus forming inantspecies was confirmed by examinationof link-reared units (FFU) per mL. Infected insects were held at 25±26C specimens, i.e. male and female adults with corresponding and 75±80% relative humidity for 14 days (experiment 1) larval exuviae. Aedes v. arabiensis females removed from the and 20 days (experiment 2) until single mosquitoes were traps showed variation in regard to the extent of pale apical tested by the in vitro `capillary' method to determine if banding on the abdominal tergites. Apical bands occurred they could transmit virus. On day 25, another group of 13 either on segments 6 and 7 or on segments 5, 6 and 7, or mosquitoes, which were infected in experiments 1 and 2, occasionally on segments 2±7. Previously, narrow pale were pooled together and allowed to feed on a susceptible bands have only been reported on segments 6 and 7 hamster. (Edwards, 1941). Transmission of virus to hamsters was established by making an extract of the liver of the dead animal and passaging this in infant mice. The brains of some of the Ecological observations dead mice were then themselves passaged into Vero cells, which were tested by IFAT. The capillary method for show- The number of mosquitoes collected in net and light ing in vitro transmission by individual mosquitoes was traps, both baited with CO2, are shown in Table 1, as well based on the method of Aitken (1977). With this technique, as the ratio of Culex tritaeniorhynchus to Ae. v. arabiensis in a fine capillary tube containing a mixture of equal parts of each collection. Considering the short period over which the foetal calf serum and 10% sucrose solution was inserted traps were set each evening, the catches were very high over the proboscis of a mosquito, which was then allowed except at Al Khawbah and Falas. Thus, both types of trap to feed on the mixture. The mosquito injects saliva and also functioned well in the environment of Jizan Region. The virus, if present in its salivary glands, into the capillary tube ratios of Cx. tritaeniorhynchus to Ae. v. arabiensis indicated while feeding. Before feeding, the tube contains about that Cx. tritaeniorhynchus was more prevalent than the 6.8 mL of the mixture, which after feeding decreased to other species near the permanent or semi-permanent aqua- about5.0 mL. This was diluted 10-fold by expression into tic habitats (sites 2, 3, 4 and 5), whereas Ae. v. arabiensis was 25 mLof10% Leibovitz's tissue culture medium and the more common near temporarily flooded agricultural fields same day passaged into Vero cells. The Vero cells were (crops or pasture, sites 6, 7 and 8). Limited larval collecting subsequently tested by the IFAT to show presence of at three of the sites showed the presence of Cx. tritaenio- virus. After feeding, each mosquito was given the same rhynchus (site 3), Cx. tritaeniorhynchus and Ae. v. arabiensis number as its matching capillary fluid and was stored at (site 6) and Ae. v. arabiensis (site 7). At Harub (site 8), the À70C. flooded pasture had just dried up and we appear to have The infection rate (the proportion of fed mosquitoes sampled mosquitoes resting in nearby stands of maize. becoming infected) was subsequently determined by intra- Four mosquito species were collected in the landing-biting cerebral inoculation of a suspension of each individual catches made on human bait in 5.25 man-hours at three of mosquito into infant mice. The transmission rate, i.e. the the collecting sites. The numbers of Cx. tritaeniorhynchus percentage of infected mosquitoes that successfully trans- and Ae. v. arabiensis taken were 137 (26/man h) and 84 mitted virus into capillary tubes, could then be calculated. (16/man h), respectively, and a few Ae. caspius and Ae. vittatus A vector competence index (VCI) was also calculated, were also collected. The numbers of Cx. tritaeniorhynchus which was the product if the infection and transmission and Ae. v. arabiensis entering suction traps set inside and on rates. The maximum possible VCI is 1.0. the outside walls of a sheep hut were 42 and 151, respectively (ratio 1 : 3.6), with no mosquitoes entering the control traps. No mosquito activity started until after dark. Results

Identification and taxonomy Virus isolations

The total number of mosquitoes collected was more All those mosquito pools that were positive for RVF than 74 500. Of these, 23 699 were identified and tested for virus by LightCycler analysis showed a single band at RFV virus: there were three species, namely Culex (Culex) 369 bp on their confirmation by gel electrophoresis; this tritaeniorhynchus Giles (15 428), Aedes (Aedimorphus) band represented the RVF virus-specific amplicon. Nearly vexans arabiensis Patton (8091) and Ochlerotatus caballus all the mosquitoes collected were tested for RVF virus (Theobald) 180. There were four other species collected in except in the case of sites 3, 6 and 8. At the first two of

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 250 P. G. Jupp etal.

Table 2. Number of mosquitoes tested for RVF virus and number of isolations made in Jizan Province.

Culex tritaeniorhynchus Aedes v. arabiensis Aedes caballus

No. mosquitoes No. pools No. mosquitoes No. pools No. mosquitoes No. pools Collection site tested (No. positive) tested (No. positive) tested (No. positive) 1. Al Khawbah 28 1 2. Near Falas 361 2 262 2 3. Jizan Dam, eastside 3575 19 4. Jizan Dam, southside 3333 17 (1) 359 2 5. Al Refoud 3829 20 (3) 934 5 105 3 6. Near Al Humayrah 4126 23 4548 25 75 4 7. Near Baysh 31 1 1545 8 (3) 8. Harub 145 4 (2)1 443 5 (4)1 Totals 15 428 87 (6) 8091 47 (7) 180 7

1Some of the mosquitoes in the catch from this site were kept alive before being killed for virus assay so that they could feed on hamsters. This resulted in one transmission by Ae. v. arabiensis and another by Ae. v. arabiensis and/or Cx. tritaeniorhynchus.

Table 3. Mosquito infection rates with RVF virus at the four sites 0.3 to 13.8 and the highest rates for both mosquito species where infected mosquitoes were collected. occurred atsite8, atHarub.

MIR1 (IR2) Collection site Cx. tritaeniorhynchus Ae. v. arabiensis Vector competence experiments 4. Jizan Dam, 0.30 (0.30) south bank 5. Al Refoud 0.78 (0.81) When the live field mosquitoes from Harub were fed on 7. Near Baysh 1.94 (2.35) a hamster in the insectary, there was transmission of RVF 8. Harub 13.803 9.033 virus to the hamster. The 15 Ae. v. arabiensis that had engorged were subsequently tested in one pool and found 1Minimum infection rate no. infected mosquitoes/1000. to be positive (isolation included in Table 2), indicating 2Statistical infection rate/1000 calculated by method of Chiang that this species had transmitted. There was also a second & Reeves (1962). similar transmission but in this case the hamster had been 3 The IR was not calculated for this site because the pool size was bitten by a mixture of mosquitoes, so that transmission too variable. was due to Ae. v. arabiensis and/or Cx. tritaeniorhynchus, as a pool of seven Ae. v. arabiensis and a pool of 10 these sites, after a large sample had tested negative no Cx. tritaeniorhynchus each subsequently tested positive further testing was considered worthwhile. At site 8 a por- (isolation included in Table 2). tion of the catch was kept alive but some of these mosqui- Because a number of Cx. tritaeniorhynchus remained toes died in transit and had to be discarded. Thirteen alive for several weeks it was possible to obtain F1 egg isolations of RVF were made from mosquitoes collected at rafts by placing mosquitoes singly into `laying tubes' con- four of the eight sites (Table 2). Six isolations came from taining grass infusion (Jupp & Brown, 1967) after hamster Cx. tritaeniorhynchus and seven from Ae. v. arabiensis. The feeds and to rear the resulting larvae into adults. Although mosquito infection rate for the species infected at each of there was a high larval mortality, the surviving F1 adults the four sites is shown in Table 3. Where two different were healthy. These mosquitoes were used for vector com- methods were used to calculate the infection rate the petence tests (Table 4). Attempts were also made to obtain results agreed well; the minimum infection rates varied from an F2 generation but the F2 rafts, although they were

Table 4. Results of vector competence tests with RVF virus in the F1 progeny of field-collected Cx. tritaeniorhynchus infected orally. Titre of virus in serum of infective Days after Infection Transmission 1 2 3 4 Experimentno. hamster in log10 FFU/mL infective meal rate rate VCI 1 7.4, 7.9 14 10/14 (71%) 1/10 (10%) 0.07 2 6.9 20 19/26 (73%) 5/19 (26%) 0.19

1Also the number of days before transmission was attempted. 2Numerator no. of mosquitoes infected, denominator no. of infected mosquitoes feeding. 3Numerator no. of mosquitoes transmitting, denominator no. of infected mosquitoes feeding. 4VCI vector competence index, see text.

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 Rift Valley fever in Saudi Arabia 251 deposited fairly readily in the `laying tubes', failed to hatch, sampled greater numbers of Ae. v. arabiensis (3.6 : 1) even indicating that no mating had occurred because of the eury- though the CO2 baited traps showed that equal numbers of gamous nature of the mosquito. Table 4 shows that after the two species were present at the site (Al Humayrah). feeding on hamsters with a high titre viraemia, Cx. tritae- About the same number of isolations were obtained niorhynchus had a high infection rate after two different from the two species and both species transmitted the virus periods of incubation. The in vitro transmission rates, how- in the laboratory, although we were only able to under- ever, were low À only 10% after 14 days but increasing to take quantitative vector competence experiments with 26% after 20 days. Although these rates are not significantly Cx. tritaeniorhynchus, owing to insufficient live specimens different by the Chi-squared test (P 0.3), they probably of Ae. v. arabiensis. The tests with Cx. tritaeniorhynchus would have been if the sample size had been larger. A group gave high infection rates but the transmission rate increased of 13 mosquitoes that fed on a hamster on day 25 success- with a longer interval after the infecting bloodmeal to a mod- fully transmitted, showing that Cx. tritaeniorhynchus was erate rate of 26%, indicating a rather lengthy extrinsic incu- also capable of in vivo transmission of the virus. bation period. Interestingly, the same trend was shown in tests on Culex zombaensis in South Africa (McIntosh et al., 1983). Quantitative tests are still needed on Ae. v. arabiensis Discussion to see whether it has a vector capability superior or not to Cx. tritaeniorhynchus. This is the first time that RVF has occurred outside Africa Culex tritaeniorhynchus has not been implicated in the and Madagascar and in Saudi Arabia. Itseems clear transmission of RVF virus in Africa but this may be that the virus came from East Africa, according to the because it does not occur at high densities in countries on sequencing of six of the mosquito isolates, five from that continent. It certainly does not occur at high density in Ae. v. arabiensis and one from Cx. tritaeniorhynchus. South Africa. It is, however, an important vector of the These all had the same genomic sequence, a sequence very Flavivirus Japanese encephalitis in India and the Far East close to that of the isolates from the 1977±1978 epidemic (Burke & Leake, 1988). A floodwater mosquito referred to in EastAfrica (Grobbelaar et al., unpublished). as Aedes (Aedimorphus) vexans was implicated in the trans- Five of the collecting sites that were used are grouped mission of RVF in Senegal when 10 viral isolations were fairly close together in three neighbouring districts, made from itthere(Zeller et al., 1997). According to White Abu Arish, Al Aridah and Al Humayrah, and there was (1975), this mosquito is the same subspecies as occurred in one remaining site to the south at Al Khawbah where only the Jizan region ± Ae. vexans arabiensis. As the type form a few mosquitoes were collected. The other two sites that Ae. vexans vexans is widely distributed in Europe and parts were productive for mosquitoes, those near Baysh and at of the Middle East, comparative vector competence experi- Harub, are much further north (Fig. 1). The viral infection ments are urgently needed on this taxon and subspecies rates were either zero or less than one infected mosquito per arabiensis. Although Ochlerotatus caballus has been impli- 1000 at the five central sites, whereas at the other two northerly cated as a possible minor vector in South Africa (McIntosh sites they were much higher (Table 3). It would appear that we & Jupp, 1981), only 180 individuals were collected among started our mosquito collecting in the Abu Arish ± Al Aridah ± the 23 699 specimens identified and tested for virus in the Al Humayrah districts after this area had been the centre presentcollections. of intense viral activity: that there had been such activity is It may be concluded that both the most prevalent mos- clear from the number of human and livestock infections that quito species collected were evidently active vectors of RVF had occurred there. On the other hand, we appear to have virus during the epidemic. The higher immune rates in live- sampled the populations at Baysh and Harub just after or stock resident in the foothills of the Sarawat mountains that during the occurrence of human and livestock infections. were shown in the companion study may be related to the Therefore, itis probable thatvirus activitymoved north- presence of the many agricultural fields in the terraces wards during the 3 weeks from 25 September to 10 October between the wadis in the foothills of the mountain range. when we collected mosquitoes, and that we caught up The temporary flooding of these crop and pasture fields with the virus on our last two trapping evenings near Baysh provided an ideal habitat for oviposition and subsequent (9 October) and at Harub (10 October). larval development of both species, but particularly the Determining the identity of the mosquito vectors respon- floodwater mosquito Ae. v. arabiensis. The latter would sible for transmission of RVF virus during the epidemic was have been the first to appear in the area when the aedine simplified by the small variety of species present and eggs beneath the soil of these fields became inundated by the because only Cx. tritaeniorhynchus and Ae. v. arabiensis first rains. There is the possibility that Ae. v. arabiensis will were prevalent species. Both these species showed ecological transmit the virus transovarially to the next generation characteristics that fitted them for vectorship. Apart from when infected eggs hatch once the next season's rains their abundance, both were plentiful in the vicinity of farm- arrive, in a manner similar to that shown in Kenya for ers and their livestock and we showed that they fed readily Ae. (Neomelaniconion) mcintoshi (Linthicum et al., 1985). on humans and sheep. Aedes v. arabiensis may, however, Significantpopulationsof Cx. tritaeniorhynchus probably have a higher preference for sheep than Cx. tritaenio- remained in the Jizan Region after the 2000 rains ceased rhynchus, as our suction trap collections at sheep huts and the epidemic waned, as this mosquito breeds in more

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252 252 P. G. Jupp etal. permanent water and is not dependent on the flooding of Edwards, F.W. (1941) Mosquitoes of the Ethiopian Region. III. agricultural fields. In this way Cx. tritaeniorhynchus could Culicine Adults and Pupae. British Museum (Natural History), have continued to transmit virus after Ae. v. arabiensis had London. disappeared. Gillies, M.T. & De Meillon, B. (1968) The Anophelinae of Africa As future outbreaks may be anticipated in Saudi Arabia, South of the Sahara (Ethiopian Zoogeographical Region), 2nd edn. Publications of the South African Institute for Medical ways to reduce the exposure of humans and livestock to Research no. 54. South African Institute for Medical Research, mosquito bites will become important. Insecticidal mosquito Johannesburg. control has already been undertaken but the knowledge Hopkins, G.H.E. (1952) Mosquitoes of the Ethiopian Region, that both the vectors bite exclusively in the dark should I Larval Bionomics of Mosquitoes and Taxonomy of Culicine lead to better protection measures at night, for example, Larvae. British Museum (Natural History), London. wider use of bed-nets, especially for farmers and their families, Jupp,P.G.(1996).Mosquitoes of Southern Africa Culicinae and screening of the huts where sheep and goats often sleep. and Toxorhynchitinae. Ekogilde Publishers, Hartebeespoort, Furthermore, the monitoring of viral activity in both livestock South Africa. and mosquitoes should continue. Jupp, P.G. & Brown, R.G. (1967) The laboratory colonization of Culex (Culex) univittatus Theobald (Diptera: Culicidae) from material collected in the Highveld region of South Africa. Journal of the Entomological Society of Southern Africa, 30, Acknowledgements 34±39. Jupp, P.G., Grobbelaar, A.A., Leman, P.A., Kemp, A., The South African authors wish to thank the Minister of Dunton, R.F., Burkot, T.R., Ksiazek, T.G. & Swanepoel, R. (2000) Experimental detection of Rift Valley Fever virus by Agriculture and Water of the Kingdom of Saudi Arabia, reverse transcription-polymerase chain reaction assay in large Dr Abdullah Bin Abdul Aziz Bin Muammar, for inviting samples of mosquitoes. Journal of Medical Entomology, 37, them to investigate the epidemic and for providing the field 467±471. team with accommodation and necessary facilities. Various Jupp, P.G. & McIntosh, B.M. (1967) Ecological studies on Sindbis personnel from his Departmentweremosthelpful. and West Nile Viruses in South Africa II, Mosquito bionomics. Mr Siraj Omah and colleagues of the Ministry of Health South African Journal of Medical Sciences, 32, 15±33. in Jizan Province are thanked for directing the entomologists Linthicum, J.K., Davies, F.G., Kairo, A. & Bailey, C.L. (1985) Rift to strategic sites for mosquito collection. We are indebted Valley fever virus (family Bunyaviridae, genus Phlebovirus): to Drs Tom Ksiazek and Kent Wagoner of CDC, Atlanta, Isolations from Diptera collected during an inter-epizootic for providing meteorological data and to all the staff of the period in Kenya. Journal of Hygiene (London), 95, 197±209. Mattingly, P.F. & Knight, K.L. (1956) The mosquitoes of Arabia I. Special Pathogens Unit in Johannesburg for providing Bulletin of the British Museum (Natural History) Entomology, 4, technical assistance. One of us (P.G.J.) acknowledges a 89±141. grant from the Poliomyelitis Research Foundation towards McIntosh, B.M. & Jupp, P.G. (1981) Epidemiological aspects of the laboratory work. Rift Valley fever in South Africa with reference to vectors. Contributions to Epidemiology and Biostatistics, 3, 92±99. McIntosh, B.M., Jupp, P.G., dos Santos, I. & Rowe, A.C. (1983) Field and laboratory evidence implicating Culex zombaensis and References Aedes circumluteolus as vectors of Rift Valley fever virus in coastal South Africa. South African Journal of Sciences, 79, Ageel, A.R.M. & Amin, M.A. (1997) Integration of schistosomiasis- 61±64. control activities into the primary-health-care system in the Shope, R.E. & Sather, G.E. (1979) The Arboviruses. Diagnostic Gizan region, Saudi Arabia. Annals of Tropical Medicine and Procedures for Viral Rickettsial and Chlamydial Infections,5th Parasitology, 91, 907±915. edn (ed. by E. H. Lennette and N. J. Schmidt), pp. 767±814. Aitken, T.H. (1977) An in vitro feeding technique for artificially American Public Health Association, Washington, D.C. demonstrating virus transmission by mosquitoes. Mosquito Sirivanakarn, S. (1976) Medical Entomology Studies À III, News, 37, 130±133. a revision of the subgenus Culex in the Oriental region (Diptera: Burke, D.S. & Leake, C.J. (1988) Japanese encephalitis. The Arbo- Culicidae). Contributions of the American Entomological Insti- viruses: Epidemiology and Ecology III (ed. by T. P. Monath), tute, 12, 1±269. pp. 63±92. CRC Press Inc, Boca Raton, Florida. White, G.B. (1975) Notes on a catalogue of Culicidae of the CDC (2000a) Outbreak of Rift Valley fever ± Saudi Arabia, Ethiopian region. Mosquito Systematics, 7, 303±344. August±October, 2000. MMWR, 49, 905±908. Wittwer, C., Ririe, K., Andrew, R., David, D., Guntry, R. & CDC (2000b) Update: outbreak of Rift Valley fever ± Saudi Balis, U. (1997) The Lightcycler: a microvolume multisample Arabia, August±November, 2000. MMWR, 49, 982±985. fluorimeter with rapid temperature control. Biotechniques, Chiang, C.L. & Reeves, W.C. (1962) Statistical estimation of virus 22, 176±181. infection rates in mosquito vector populations. American Journal Zeller, H.G., Fontenille, D., Traore-Lamizana, M., Thiongane, Y. of Hygiene, 75, 377±391. & Digoutte, J.P. (1997) Enzootic activity of Rift Valley fever Collett, M.S., Purchio, A.F., Keegan, K., Frazier, S., Hays, W., in Senegal. American Journal of Tropical Medicine and Hygiene, Anderson, D.K., Parker, M.D., Schmaljohn, C., Schmidt, J. & 56, 265±272. Dalrymple, J.M. (1985) Complete nucleotide sequence of the M RNA segmentof theRiftValley fever virus. Virology, 144, 228±245. Accepted 20 March 2002

# 2002 The Royal Entomological Society, Medical and Veterinary Entomology, 16, 245±252