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 host-seeking 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 Parish. Residences of WNV neuroinvasive disease or fever case-patients are denoted by upper case letters. Non-case-patient residences are denoted by italicized lower case letters. Eight St. Tammany sites (A, C, D, E, F, g, I, J) were in or near the city of Slidell in the southeast corner of the parish, St. Tammany site B was located in Abita Springs, east of Covington, and 3 sites (K, l, m) were in Pearl River in the east-central region of the parish. The 2 Tangipahoa parish sites (A, B) were on the northwest and northern outskirts of Ponchatoula.

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 (Table 2). Cx. erraticus was the most commonly collected species, accounting for 28% of the total collected. Cx. quinquefasciatus, Ae. albopictus, Coquillettidia perturbans, and Cx. salinarius were other commonly collected species. Ovitraps yielded 335 Ae. albopictus and 778 Ae. triseriatus/hendersoni reared to adults. Aspirator collections yielded 658 mosquitoes of 16 species, of which 474 were Ae. albopictus. Cone traps collected 33 mosquitoes (9 species) and Fay-Prince traps yielded 214 mosquitoes (15 species). Mosquitoes were sorted into 2,471 pools for processing and virus testing.

Relative population densities (light trap or gravid trap counts per trap night) of the species in which we detected WNV RNA, and of Ae. albopictus, were calculated for case and non-case sites (Table 3). For most species, light trap counts per night greatly exceeded gravid trap counts. For Cx. quinquefasciatus, however, gravid trap counts were 7–58 times greater than were light trap collection counts. Neither gravid traps nor light traps collected large numbers of Ae. albopictus. Light trap counts per trap night for Ae. albopictus were approximately the same as gravid trap counts except at site l where 35.5 mosquitoes were collected per gravid trap night compared to 4.5 per light trap night.

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). Five viral RNA positive pools were from St. Tammany Parish and 4 were from Tangipahoa. Seven of the positive pools contained Cx. quinquefasciatus; 4 of these were from St. Tammany Parish, and 3 were from Tangipahoa. The other 2 positive pools consisted of a pool of Cx. salinarius from St. Tammany and a pool of Cq. perturbans from Tangipahoa. All of the WNV-positive Cx. quinquefasciatus were collected in gravid traps, while the positive Cx. salinarius and Cq. perturbans were collected in light traps. No virus was detected in mosquitoes collected by the other methods. WNV infection rates ranged from 0.81/1,000 to 10.91/1,000 by MLE (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 WNV-infected 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).