West Nile Virus–infected Mosquitoes, Louisiana, 2002

Culex quinquefasciatus was identified as probable vector.

S ince the first appearance of West Nile virus (WNV) (family Flaviviridae: genus Flavivirus) in the Western Hemisphere in 1999 (1), the virus has spread rapidly south and west from its initial focus in the New York City metropolitan area. By the end of 2001, WNV-infected mosquitoes, birds, horses, or humans had been reported from 27 states, and human cases of WNV disease occurred as far south as southern Florida and as far west as Arkansas and Louisiana (2,3).
In the northeastern United States, the primary epizootic/epidemic vector of WNV is Culex pipiens, a species that feeds primarily on birds (4)(5)(6). Other potentially important vector species, based on frequency of isolations of WNV or laboratory vector competence studies, include Cx. restuans and Cx. salinarius (7,8). WNV has been isolated from an additional 57 species, but their status as vectors is unknown (Centers for Disease Control and Prevention [CDC], http://www.cdc.gov/ncidod/dvbid/westnile/Mos-quitoSpecies.htm). In the southern United States, WNV was isolated from Cx. quinquefasciatus, Cx. salinarius, and Cx. nigripalpus in Florida and Georgia (9), Cx. nigripalpus in northern Florida (10), and from Anopheles atropos, Deinocerites cancer, and Aedes taeniorhynchus in the Florida Keys (11). However, the role these species play in epidemics of WNV disease in the southern states has not been determined. Ae. albopictus is common in urban, suburban, and rural residential settings throughout the southern states and is a competent laboratory vector of WNV (12,13). Although the virus has been isolated from Ae. albopictus in the Northeast (14), this species' importance in transmission of WNV to humans is unknown.
During May and June 2002, WNV infection was identified in chickens, horses, dead wild birds, and in pools of Cx. quinquefasciatus mosquitoes from St. Tammany Parish, on the north shore of Lake Pontchartrain in southeastern Louisiana (15). Human cases of WNV neuroinvasive disease began to appear in late June, and 27 cases were reported by the end of July. Intense local WNV transmission was indicated by the St. Tammany Parish Mosquito Abatement District's surveillance program, which detected WNV immunoglobulin (Ig)M antibody in 17% of their sentinel chickens and WNV antigen from 11 mosquito pools by the end of July (15). The human cases tended to cluster in 2 areas of St. Tammany Parish, Slidell and the Covington-Mandeville area. In neighboring Tangipahoa Parish, human cases were also being reported, with most clustering in the Hammond-Pontchatula area (Louisiana Department of Health and Hospitals, unpub. data).
The recognition of a growing outbreak of WNV disease in humans provided an opportunity to describe the transmission dynamics of WNV in locally occurring mosquitoes during epidemic transmission and to compare these dynamics to patterns seen in the northeastern states (4)(5)(6). Accordingly, we conducted an entomologic survey in St. Tammany and Tangipahoa Parishes during August 2002. The specific aims of the survey were to document species composition, relative abundance, and WNV infection rates in mosquitoes at residences of patients with confirmed cases and at residences of patients with suspected cases of WNV fever, the most likely locations where transmission to humans occurred. We were particularly interested in attempting to ascertain the importance of Cx. quinquefasciatus and Ae. albopictus as vectors of WNV in this epidemic.

Study Sites and Specimen Collection
Mosquitoes were collected in St. Tammany and Tangipahoa Parishes from August 3 to August 15, 2002. Two study sites were selected in each parish (denoted as St. Tammany A and B, and Tangipahoa A and B). These sites were located at or near residences of patients with confirmed cases of WNV neuroinvasive disease. As suspected cases of WNV fever (persons reporting as outpatients with undifferentiated febrile illness with headache) were identified, collections were made at the residences of these patients.
Mosquitoes were collected primarily by using CDC miniature light traps baited with dry ice to collect hostseeking females, Reiter gravid traps (16) to collect females seeking a location to deposit eggs, and ovitraps to collect eggs from container-breeding mosquitoes. Both light and gravid traps at the 4 initial study sites were operated for 24 h/day in an attempt to maximize the collection of Ae. albopictus, a daytime feeder. Some additional collections were made by using Fay-Prince traps and duplex cone traps and by aspirating resting adult mosquitoes from the outside of residences or other structures. Collections were transferred to 2.0-mL cryovials and frozen on dry ice until returned to the CDC laboratory in Fort Collins, Colorado, where they were stored at -80°C. Mosquito eggs collected in ovitraps were hatched in the insectary, reared to adulthood, held for 48 h at 27°C and 80% relative humidity, then identified and processed for virus testing as described below.

Mosquito Processing and Testing
Mosquitoes were identified to species on a refrigerated chill table. Pools of <50 specimens sorted by species and collection site and date were triturated in 1.75 mL of diluent by using a Mixer Mill apparatus (Qiagen Inc., Valencia, CA, USA) and centrifuged (17). Supernatants from the mosquito suspensions were tested for the presence of WNV RNA by TaqMan reverse transcription-polymerase chain reaction (RT-PCR), and positive pools were retested by using a different primer set to confirm the presence of WNV RNA (18). Mosquito infection rates were determined by calculating the maximum likelihood estimate (MLE) with 95% confidence intervals (19).

Mosquito Collections
Collections were made at 14 sites, 12 in St. Tammany Parish and 2 in Tangipahoa  Trapping effort at each site and elapsed time between onset of illness and mosquito collection are shown in Table 1. Although traps were run for 24 h/day at some sites, only mosquitoes collected overnight are used to calculate mosquitoes per trap night. The earliest date of onset was June 21, and the latest date of onset was August 4. Mosquito collection dates ranged from 8 to 50 days after onset of illness. Trapping effort per site ranged from 2 to 60 trap nights for light trap collections, and from 2 to 59 trap nights for gravid trap collections. No notable changes in the weather occurred during the collection period that might bias comparisons of mosquito abundance.
A total of 31,215 mosquitoes were collected during the trapping period of August 3 to August 15 ( No relationship was shown between the population densities of the species examined and whether the site was a case-patient or non-case-patient residence, except for Cx. quinquefasciatus, for which much higher densities were found at sites of non-case-patients. Cx. quinquefasciatus gravid trap counts per trap night ranged from 0.4 to 44.1 for confirmed WNV disease case-patient residence sites, and 59.6 to 142.8 for non-case-patient sites (p<0.001, Wilcoxon rank sum test).

WNV Detection
WNV RNA was detected in 9 mosquito pools by TaqMan RT-PCR (Table 4) Table 4). The highest infection rate was seen in Cx. salinarius and the lowest in Cq. perturbans. Infection rates in Cx. quinquefasciatus were similar among sites (2.31/1,000-5.64/1,000).
No relationship was found between the relative densities of mosquitoes collected and the finding of WNVinfected mosquitoes (Tables 3 and 4). Three infected pools of Cx. quinquefasciatus were collected from Tangipahoa site B, with 15.1 mosquitoes per gravid trap night, whereas no infected pools were collected from St. Tammany site m, which had the highest Cx. quinquefasciatus count per gravid trap night (142.8). Likewise, the only WNV-infected Cx. salinarius pool was from St. Tammany site B, which had 1.6 mosquitoes per light trap night, 1 of the lower density sites for that species. Eight other sites had higher light trap counts but no WNV-positive mosquitoes were detected. Cq. perturbans was found in high densities at only Tangipahoa sites A and B, and the densities at these sites were similar at 17.7 and 16.2 per light trap night, respectively. Infected Cq. perturbans were found only at Tangipahoa site A.
Detection of WNV-infected mosquitoes was not influenced by elapsed time between dates of onset of illness (a surrogate for date of infection) and mosquito collection dates. We obtained 3 isolates from Tangipahoa site B, where the date of onset was 47-50 days before mosquito collection (Tables 1 and 4).

Discussion
The results of our survey indicate that the natural history of WNV in the southern United States is similar to that seen in the northern states, where Cx. mosquitoes, especially Cx. pipiens, Cx. restuans, and Cx. salinarius, are thought to be the species primarily involved in enzootic, epizootic, and epidemic transmission (3)(4)(5)(6). Seven of 9 (78%) WNV-infected mosquito pools were Cx. quinquefasciatus. Both Cx. pipiens and Cx. quinquefasciatus are primarily ornithophilic, although some studies indicate that Cx. quinquefasciatus feeds more readily on mammals (20)(21)(22). One of the 2 other positive pools was of Cx. salinarius, which feeds primarily on mammals (20)(21)(22). WNV has been isolated frequently from this species (5,6,23), and laboratory studies indicate that it is a competent vector (8).
Cx. salinarius has been associated with an outbreak of human WNV illness in New York City (6) and appears likely to be important in transmitting WNV to humans and domestic mammals in the southern United States as well. The other positive pool was of Cq. perturbans. WNV isolates previously have been obtained from this species, but it is an inefficient vector in the laboratory (8).
Eight mosquito pools containing WNV RNA were collected at 5 (45%) of 11 confirmed WNV case-patient residences, while the remaining pool was from 1 (33%) of 3 non-case-patient sites. This finding suggests that many, perhaps most, human infections are acquired near their residences.
Although substantial numbers of Ae. albopictus were tested, no virus was detected in this competent laboratory vector of WNV. This finding was perhaps due to the bloodfeeding habits of this species. Two studies of engorged specimens wild caught in the continental United States found that 1% and 17% of blood meals were taken from birds (24,25). The remaining meals were from a variety of mammals, including humans. In our study area, relatively few blood meals may have been taken from birds, thus reducing the exposure of host-seeking Ae. albopictus to the high-titered levels of WNV viremia seen in many species of birds. Little data have been published on WNV viremia levels in mammals, but in horses, dogs, and cats, viremia levels are transient, of low titers, or both (12,26). If this condition is also the case for other mammalian species, then most blood meals taken by Ae. albopictus from WNVinfected hosts would be below the threshold titer necessary to initiate infection.
In our study, gravid traps were clearly preferable to light traps as an effective surveillance tool for detecting WNV RNA in mosquitoes. All the positive Cx. quinquefasciatus pools and 91% of total Cx. quinquefasciatus were from gravid traps. The other 2 WNV-positive pools were from mosquitoes collected in light traps. Gravid traps were a more effective means of collecting Ae. albopictus than were light traps. Unlike Cx. quinquefasciatus, most female Ae. albopictus collected in gravid traps were not gravid, and numerous males were also collected. Ae. albopictus were also readily collected by aspiration and ovitrapping.
Although active transmission of WNV was still occurring at the time of our collection efforts during the first half of August, most human patients had dates of onset between late June and late July. Thus, the relative numbers and species composition we observed may not have been representative of the situation when most human infections were occurring. Mosquito control activities intensified in St. Tammany Parish in response to the high level of WNV activity (15). Mosquito surveillance by the parish showed large reductions in total mosquito counts and in Cx. quin-quefasciatus counts in CDC light traps and in New Jersey light traps from May to August. Eleven WNV antigen-positive mosquito pools were detected, all in June and July. Ten of these positive pools were of Cx. quinquefasciatus, and 1 was of Cx. salinarius, similar to our findings in August. Notably, the number of sentinel chickens developing WNV IgM antibody peaked during the third week of July, declined during early August, then rose again during late August (15). This finding suggests that exposure of sentinel chickens to infected mosquitoes was ongoing, and perhaps increasing, during the period of our study. Serologic conversions in sentinel chickens continued to be detected into November. Serologic studies of wild birds caught in mist nets in St. Tammany Parish were conducted in August, and again in October (27). These data indicated that enzootic WNV transmission continued to occur in the parish, although likely at a reduced level, after human cases were no longer being reported. Long-term studies are needed to monitor the transmission dynamics of WNV in mosquito populations during epidemic and nonepidemic years.