West Nile virus infection in birds and mosquitoes, New York State, 2000.

West Nile (WN) virus was found throughout New York State in 2000, with the epicenter in New York City and surrounding counties. We tested 3,403 dead birds and 9,954 mosquito pools for WN virus during the transmission season. Sixty-three avian species, representing 30 families and 14 orders, tested positive for WN virus. The highest proportion of dead birds that tested positive for WN virus was in American Crows in the epicenter (67% positive, n=907). Eight mosquito species, representing four genera, were positive for WN virus. The minimum infection rate per 1,000 mosquitoes (MIR) was highest for Culex pipiens in the epicenter: 3.53 for the entire season and 7.49 for the peak week of August 13. Staten Island had the highest MIR (11.42 for Cx. pipiens), which was associated with the highest proportion of dead American Crows that tested positive for WN virus (92%, n=48) and the highest number of human cases (n=10).

The percentage of WN virus-positive birds was analyzed for the epicenter and non-epicenter regions. Data are included for avian species for which at least 10 birds were tested in one of the regions ( Table 2). For all submitted birds, 51% and 23% WN virus-positive birds were found in the epicenter and nonepicenter regions, respectively. WN virus infection in dead birds was highest for American Crows (67%) in the epicenter. In the non-epicenter, WN virus infection for crows was lower, similar to infection in all birds in this region. High numbers of crows were tested in both regions, and the percentage positive differed by almost threefold.
WN virus infection in dead birds was examined over time for American Crows and all other birds in the epicenter and four radial regions in New York State ( Figure 2). Comparison of American Crows in the five regions ( Figure 2A) revealed the highest peak in the epicenter during September. In addition, the peak for American Crows in the epicenter was much broader than for the other four regions. For September, the peaks for American Crows in R1 and R2 were greater than those in the more distant regions, R3 and R4, suggesting a minor extension of the epicenter during this month. Comparison of all birds except American Crows in the five regions ( Figure 2B) revealed little difference between the regions, even for the epicenter. These data support the hypothesis that the susceptibility to WN disease was greatest in crows in the epicenter.
We tested 9,954 mosquito pools with 317,668 mosquitoes, representing 28 species and eight genera (Table 3). Of eight positive species representing four genera, most positive pools were Culex species (n=341), compared with only 22 positive pools in the other three genera. All but five of the positive pools were collected in the epicenter. The MIR was calculated for each species in the epicenter for which at least 1,000 mosquitoes were tested (Table 3). In the epicenter, the MIR ranged from 0.47 to 3.55. The MIR of Cx. pipiens was the highest for an individual species. All the pure Cx. restuans pools were negative, and the MIR for Cx. pipiens-restuans was almost half that of the pure Cx. pipiens pools; therefore, the positive mosquitoes in the mixed pools of Cx. pipiens-restuans

West Nile Virus
were most likely Cx. pipiens. The MIRs for Cx. pipiens and Cx. pipiens-restuans were compared with the percentage of positive American Crows in the epicenter for each week of the transmission season ( Figure 3). The MIRs for Cx. pipiens and Cx. pipiens-restuans peaked 3 and 2 weeks, respectively, before the peak for positive crows. These data support the hypothesis that Cx. pipiens is an important enzootic vector of WN virus in New York. The epicenter was examined as individual boroughs and counties to compare vertebrate and invertebrate WN virus infections ( Table 4). The percentage of positive American Crows was calculated, and human and equine cases were noted for each county over the entire season. The MIRs of mosquitoes from each county were calculated for species with at least 1000 mosquitoes tested. Six mosquito species or groups met this criterion.

Conclusion
In the 2000 transmission season in New York State, we found 63 bird species infected with WN virus, compared with 14 species in 1999 (6). The percentage of WN virus-positive birds was higher in the epicenter than outside it. This high percentage almost entirely reflects infected crows in the epicenter; no increase in WN virus infection was noted in birds other than crows. In contrast, high WN virus infection of dead crows was not observed outside the epicenter, where the percentage of WN virus positivity was similar in crows and other birds over the entire season. High numbers of dead crows were also observed in 1999 (3,6). The cause of the increased sensitivity of crows to WN disease or infection is unknown, but may be due to virus-host interactions, mosquito-bird interactions, mosquito feeding preferences, crow population density, or crow behavior. The presence of WN virus in dead birds does not indicate a definitive diagnosis of WN virus as the cause of death. Many of the birds did not show gross pathologic lesions of WN disease (12). In addition, the rate of WN virus-positive birds in our surveillance samples is not equivalent to prevalence of infection, since we are sampling only dead birds.
The analysis of percentage positive birds over time revealed differences between various regions in New York State. The percentage of WN virus-positive crows was highest in the epicenter compared with other regions of New York State throughout the season, supporting the importance of crows as indicators in the epizootic. The percentage of WN virus-positive crows was higher than that for all other birds early in the season only in the epicenter. Outside the epicenter, the percentages of WN virus-positive crows and all other birds were similar until the peak month of September. At this time, the two radial regions closest to the epicenter showed higher infection in crows than in other birds, suggesting that the intense level of viral activity may have spread beyond the epicenter. An explanation for this apparent spread may be increased movement of crows during the fall. The surveillance data on avian deaths have implications for future surveillance activities. The similar percentages of positive crows and other birds outside the epicenter indicate the importance of testing all birds, not only crows, outside the epicenter.
Sampling errors are likely with the avian surveillance samples. The samples were from dead birds submitted to the Wildlife Pathology Unit, which relied on the cooperation of the general public, individual county health departments, and other agencies; therefore, surveillance samples do not represent a random sampling of dead birds throughout the state. The size and degree of urbanization of various bird species may have resulted in differences in submission. For example, large urban-dwelling species, such as crows, were more likely to have been submitted than small rural dwellers. In addition, more birds were likely sampled from areas with human cases, greater media coverage, and higher human population. The similarity in WN virus infection in birds other than crows from different regions suggests that the impact of sampling bias was not significant. Additionally, specimens submitted for testing to the Wadsworth Center from the Wildlife Pathology Unit represent a sample of those submitted by the public. Sampling at this level may introduce further bias into our surveillance sample. The similarity in WN virus infection in birds other than crows outside the epicenter, however, suggests that such bias was minimal.
The highest MIR for mosquitoes in the epicenter was for Cx. pipiens. Positive pools of Cx. pipiens also were identified in 1999 in New York (13). Cx. pipiens-pipiens mosquitoes feed almost exclusively on birds (14); thus, they are likely an important enzootic vector in the bird-mosquito cycle in North America. Other Cx. species have been implicated as enzootic vectors worldwide (15). High MIRs of Cx. pipiens and Cx. salinarius were associated with human and equine WN virus cases and high infection rates of crows in counties in the epicenter. Cx. salinarius feeds on both birds and mammals (16); therefore, it is a likely candidate as a "bridge vector," transmitting the virus from bird to mammal. Cx. salinarius has been proposed as a bridge vector for Eastern equine encephalitis virus (17). All MIR data were calculated by using RT-PCR-positive pools and therefore cannot be directly compared with MIRs calculated by using infectious viruspositive pools. No virus was isolated from five RNA-positive pools of Ae. japonicus, even after six serial passages through C6/36 cells. Much attention has been focused on this species because it has been reported to be a highly competent laboratory vector of WN virus (9), but current field data do not support this experimental observation. Different populations of Ae. japonicus have been described in the eastern United States (18), and differences in vector competence between the populations may explain the discrepancy between the field and experimental data.
The possibility of sampling error also exists for the mosquito surveillance samples. Individual counties collected mosquitoes in different numbers and set traps by different criteria. In addition, some counties used mosquito adulticides or larvicides during the season. These sources of bias are unlikely to have been uniformly introduced, and their impact on our analyses is unclear.
The results from the 2000 surveillance season for WN virus leave a number of unanswered questions. Many avian species can become infected with WN virus, but the prevalence of infection for each species is unknown without systematic serosurveys of the wild bird population. It is also unknown which birds have a high enough viremia for efficient transmission to the vector. The apparent mortality rate caused by WN virus is higher for crows than for other birds, but laboratory experiments are required to determine WN virus mortality rates and the pathogenic mechanisms in crows and other avian species. In addition to crows, many other birds, such as raptors and other corvids, also showed significant pathology. Some nonmigrating species (e.g., ruffed grouse) have potential use as an indicator species for WN virus infection. Vector competence, blood meal identification, and transovarial transmission studies for the potential mosquito vectors are also important research areas.