| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Prior to 2002, avian hosts of WNV had not been studied in Louisiana but were evaluated in New York City in 19994 and 20005 and in Florida in 2001.6 In these studies, resident passerine birds (pertaining to the order Passeriformes) were implicated as important amplifying hosts for WNV. Because WNV was new to Louisiana, no knowledge existed regarding the identity of local avian amplifying hosts or which bird species were most appropriate sentinels for surveillance programs in Louisiana.
Summer-resident passerine bird species were hypothesized to be the most important amplifying hosts of WNV in St. Tammany Parish. To evaluate this hypothesis, we sampled a variety of birds in early August, several weeks after the peak of epidemic transmission, to determine seroprevalence rates. Because human and equine cases of West Nile neuroinvasive disease (WNND) were clustered in suburban neighborhoods of Slidell, the largest city in St. Tammany Parish, we compared infection rates of birds sampled within a suburban transmission focus around the case clusters to those of birds sampled in rural areas where human population density was low and cases of human WNND were less prevalent. We also evaluated whether WNV activity continued within the suburban transmission focus after the incidence of human WNND cases had subsided by sampling birds in October and looking for evidence of recent WNV transmission. To interpret the significance of our observed seroprevalence rates in relation to the transmission cycle, we generated host competence data for three candidate amplifying hosts, the northern cardinal (Cardinalis cardinalis), northern mockingbird (Mimus polyglottos) and house sparrow (Passer domesticus). Finally, we derived mathematical models using our measured parameters to predict actual WNV infection rates, WNV-attributed population reductions, and relative number of infected mosquitoes deriving from these three species of birds and blue jay (Cyanocitta cristata).
|
MATERIALS AND METHODS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Other collection sites in August included sites B (30°16.2' N, 89°49.8' W) at a suburban residence in southwest Slidell, C (30°33.4' N, 90°02.5' W) about 40 km northwest of Slidell at a rural horse ranch (near mosquito collection site B),7 and D (30°13.5' N, 89°41.2' W) about 9 km southeast of Slidell at a bird-feeding station located on a rural residential property surrounded by dense humid forest and bog. Sites B, C, and D were not associated with WNV cases.
Bird capture and classification. Bird sampling occurred August 3–11, and October 22–29, 2002. A convenience sampling was used, with the objective of sampling as many birds as possible at each study site. Wild birds were trapped in mist nets (Avinet, Inc., Dryden, NY) and captive birds were captured within their enclosures by hand. Wild birds were marked with uniquely numbered aluminum leg bands. Each bird was assigned a residence category as either "breeder" or "nonbreeder". "Breeder" is defined as a bird belonging to a population known to nest locally at or near the study site, and included all the species sampled in early August prior to most landbird migration in Louisiana. "Nonbreeder" is defined as a bird belonging to a population known to breed remotely from the study site, and would include transient birds or northern breeders that had arrived in the study site for the winter season. Some members of a "breeder" population may not have bred locally, and may have either entered the study site as part of postbreeding dispersal (e.g., brown-headed cowbird, Molothrus ater) or migration (e.g., summer tanager, Piranga rubra).
Bird sampling. Blood was obtained (maximum 0.6 mL) by jugular or brachial venipuncture and collected in Microtainer serum separators (Becton, Dickinson and Company, Franklin Lakes, NJ) or diluted 1:2 in BA-1 diluent (Hanks M-199 salts, 0.05 M Tris pH 7.6, 1% bovine serum albumin, 0.35 g/L sodium bicarbonate, 100 U/mL penicillin, 100 µg/mL streptomycin, 1 µg/mL Fungizone) in cryovials for a field serum dilution of approximately 1:5. Blood samples were left at ambient temperature for up to 30 minutes and then incubated on ice until centrifuged for separation of serum or, if diluted in the field, they were frozen on dry ice for transport to the laboratory and stored at –70°C.
Testing procedures. Separated serum samples were frozen at –20°C until tested for neutralizing antibodies (at a dilution of 1:10) by plaque reduction neutralization test (PRNT) using challenge doses of approximately 100 plaque-forming units (PFU) of WNV strain NY99-4132 and Saint Louis encephalitis virus (SLEV, family Flaviviridae, genus Flavivirus) strain TBH-28 in 6-well plates of Vero cells, overlaid with 0.5% agarose in M199 medium containing antibiotics.9
Samples with 
80% reduction in Vero cell plaque forming units of either virus were further titrated in duplicate serial twofold dilutions to determine end-point titers (through 1: 320) for WNV and SLEV. A fourfold or greater 90% neutralization titer (of at least 1:10) for WNV, relative to the titer for SLEV, was considered positive for neutralizing antibodies for WNV. Two specimens had weak SLEV-neutralizing antibody titers, but due to fourfold higher titers against WNV, none were scored positive for SLEV-neutralizing antibodies.
Field diluted serum samples collected in August were tested in duplicate on 6-well plates by Vero plaque assay for evidence of viremia.9 Any plaques were harvested by standard techniques into 1 mL of BA-1 containing 20% fetal bovine serum. The harvested suspension was then identified using WNV-specific RT-PCR techniques previously published.10
Host competence studies. Northern cardinals (N = 13) and northern mockingbirds (N = 4) were captured using mist nets in Mississippi and Alabama and transferred to animal facilities at the University of Southern Mississippi, Hattiesburg. House sparrows (N = 6) were captured in baited potter traps in northern Colorado and transferred to animal facilities at CDC, Fort Collins. Birds seronegative for both WNV and SLEV (all but 4 cardinals) were inoculated subcutaneously with approximately 1,000 PFU (cardinals and mockingbirds) or 600 PFU (sparrows) of low-passage WNV-NY99-4132 and blood-sampled daily for 6 (cardinals and mockingbirds) or 5 (sparrows) days post-inoculation. Whole blood (0.05 mL) was frozen at –70°C until tested by plaque assay to determine viremia profiles. For plaque assay, the whole blood was diluted with 225 µL BA-1 for a 1:10 dilution of serum, and titrated by serial 10-fold dilution. Competence index, Ci, values were calculated as the product of three parameters: susceptibility, infectiousness and duration of infectiousness.11,12 These values represent the relative number of infectious Culex quinquefasciatus (or Cx. pipiens) mosquitoes that may result from feeding on infected vertebrate hosts, assuming that all vertebrates are equally attractive to vector mosquitoes. To convert these needle-derived Ci values to mosquito bite-derived Ci values available for other species, a conversion factor of 3.24 was derived for house sparrows, using a published mosquito bite-derived Ci value. 12 This conversion factor was applied to the other species. The four WNV-seropositive cardinals were treated identically as the infected cardinals as a control for the effects of handling.
Model construction. Simple mathematic models were constructed to predict actual WNV infection rates and WNV-attributed population reductions, and a mosquito inoculation index was developed to estimate the relative contributions of various bird species to WNV transmission in Slidell. The models were applied to four bird species (for which sufficient data were available), including blue jay, northern cardinal, northern mockingbird and house sparrow.
A species-specific WNV infection rate, ir, was predicted using the following equation:
 |
However, none of these terms were measurable in our study system. Number of survivors is the product of the measured seroprevalence rate, s, and the estimated post-epizootic population sampled, P. Number of deaths is the product of the pre-epizootic population, P0, infection rate (ir), and the mortality rate, m. P0 can be expressed in terms of P, ir and m by the expression P/(1 – ir*m). Thus ir can be solved for as follows:
 |
This equation simplifies to:
 |
The WNV-attributed population reduction, p
–, was predicted using the following equation:
 |
This equation can be expressed in terms of P, ir, and m, as follows:
 |
This equation simplifies to:
 |
For calculating a vertebrate host species-specific mosquito inoculation index, MIi, that predicts the species’ relative contribution of infected vector mosquitoes, the following equation was used:
 |
However, P' used here can be either P0 or P, depending on whether one is interested in estimating the contributions of each species for the epizootic, or potential for future contributions (which would use the current, or post-epizootic, population), respectively. We are evaluating the former situation (using P0), and therefore we modified the equation as follows:
 |
Statistical analyses.
We calculated 95% confidence intervals (CI) for seroprevalence proportions using the Wilson score method (S-PLUS 6.1 Professional software, Insightful, Inc., Seattle, WA). Seroprevalence proportions were compared using the Fisher exact test or the Pearson 
2 test. For multiple comparisons, Bonferroni adjustments were applied. Bird population sizes were estimated using the Lincoln-Petersen estimator for mark-release-recapture data,13 with variances and CI calculated by the methods of Williams and others.14
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), and detected 41 (15.5%, CI 11.7–20.4%) with WNV-neutralizing antibodies, ranging in ninety-percent neutralization titers (PRNT90) from 1:40 to 1:320. To determine whether the seroprevalence rates between rural and suburban sites were significantly different, we restricted analysis to site A (suburban) and site C (rural) because of the larger sample collections at these sites. This difference (25% versus 7%) was statistically significant (Pearson 2, P = 0.0003). Because northern cardinals were so abundant, and easily captured in both suburban and rural settings, we compared the seroprevalence rates between rural cardinals (2.1%, N = 47) and suburban cardinals (52.6%, N = 38), and also found this difference to be significant (Fisher exact test, P < 0.0001).
|
Within Site A, the seroprevalence rates of birds varied among species from 0 to 100%, but only chickens (67% positive, including 100% of six adults > 7 months old and 0% of three chicks < 1 month old) and northern cardinals differed significantly (P < 0.004, 
= 0.05, 11 comparisons including chickens and emu) from all other bird species combined.
One WNV virus isolate was made from a Carolina wren (Thryothorus ludovicianus) sampled from site A. The titer was 104.1 PFU/mL serum. The same specimen was negative for WNV-neutralizing antibodies.
In October, we resampled free-ranging wild birds at site A (Table 2). Of 166 resident breeder birds sampled, 40 (24.1%, CI 18.2–31.1%) tested positive, with 90% neutralization titers ranging from 1:40 to 1:320. The seroprevalence rates for different species of free-ranging breeders varied from 0 to 100%, but only the northern mockingbird and the northern cardinal differed significantly (P < 0.006, = 0.05, 9 comparisons) from all other breeder bird species combined. No positives were detected from among 7 nonbreeder birds (CI 0–35.4%).
|
), we noticed that the seroprevalence rate remained stable at 24–25%. A statistically significant increase in seroprevalence rates within a species was observed only for the northern mockingbird (P = 0.01, right-tailed Fisher exact test). No statistically significant decreases in seroprevalence rates within a species were observed. Because seroprevalence rates did not change significantly between August and October (except for the mockingbird), we combined these data (excluding the mockingbird) to look for any species-specific difference from the larger, more robust data set (Table 2). No additional species-specific differences were observed.
In October we recaptured a small number of birds that had been sampled in August at site A. Recaptured birds that remained seronegative included a chickadee, 3 titmice, 4 house sparrows, a cardinal, a red-bellied woodpecker, and a captive cockatiel. Two cardinals that tested positive in August remained positive in October, and both showed declines in 90% neutralization titers from 1:320 in August to 1:160 or 1:40 in October. One blue jay that tested negative in August (< 1:10) seroconverted to positive in October ( 1:320).
Mark-recapture data from site A were used to generate crude estimates of bird population sizes (Table 3). For birds with recaptures, cardinals and house sparrows were estimated to be the most abundant, with several hundred birds each utilizing the study site. Red-bellied woodpeckers, blue jays and Carolina chickadees were also common, but their populations were about an order of magnitude less than cardinals and house sparrows. Carolina wrens and northern mockingbirds were probably intermediate in abundance but no recaptures were made. Titmice were relatively uncommon.
|
). All inoculated birds survived infection except for two cardinals, suggesting a crude mortality rate of 25% for cardinals. Four cardinals serving as handling controls all survived. The viremia profile for fatally infected cardinals was dramatically different than for surviving cardinals (Figure 2). The fatally infected cardinals developed a mean peak viremia titer of 109.4 PFU/mL serum (range, 107.1–109.7) compared with 105.6 PFU/mL serum (range, 104.7–106.3) for the survivors. Susceptibility, Culex quinquefasciatus infectiousness, and duration of infectiousness parameters were estimated from the viremia profiles, and competence index values were derived from these (Table 4).
|
|
|
). These simple mathematical equations predicted the greatest infection rate (> 60%) and population reduction (about 50%) for the blue jay. The northern cardinal, the northern mockingbird and the house sparrow were infected at moderate rates (30–60%) but suffered low population impacts (0–20%).
|
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The finding that all seropositive birds in the October sampling of site A were breeders (rather than "nonbreeders," i.e., transient or winter residents) is consistent with observations following the Queens 1999 and the Staten Island 2000 outbreaks.4,5 However, insufficient numbers of non-breeders were sampled to determine the seroprevalences in these groups of free-ranging birds. Some breeders, such as cowbirds and doves, sampled in October, may in fact have been immigrants or transients, having arrived into the transmission focus after the period of intense transmission. However, one mourning dove was recaptured at the same site 1 and 4 days after sampling in October, suggesting that this bird was not migrating at the time of sampling. As with other WNV out-breaks in North America, summer-resident birds appear to be important amplifying hosts for WNV.4–6
Of the birds sampled, most were resident breeders that were common or abundant, and therefore candidate amplifying hosts for WNV.15 The array of species sampled was biased towards those species likely to be trapped in mist nets placed at ground level, so infection rates could not be assessed in other species, such as fish crows and waterfowl. However, our sample identified certain land-dwelling species such as the northern mockingbird, the northern cardinal, the Carolina wren, the house sparrow, the blue jay, and the red-bellied woodpecker that were infected at high frequencies. The high seroprevalence rates observed for these species (ranging from 24% to 78% of captured birds) indicated that they may be useful target species in surveillance programs that use free-ranging birds as sentinels.16 In particular, cardinals and mockingbirds were more frequently seropositive than the average bird, and thus should be targeted by free-ranging bird surveillance programs. Higher seroprevalences in adult chickens suggest that this domestic species may be even more sensitive than wild species for detecting transmission to birds in ecosystems similar to that of site A. Captive chickens have been used historically to monitor arbovirus infections, particularly SLEV and eastern equine encephalomyelitis virus, in St. Tammany Parish, and they were useful for tracking WNV activity during the 2002 outbreak in Slidell.17
The seroprevalence rates reported herein may underestimate species-specific infection rates, because of WNV-attributed mortality in birds. The candidate amplifying host species named above have been reported dead with confirmed WNV infections through avian mortality surveillance.1 Thus, a proportion of infected individual birds of each species may disappear from the sampled population due to mortality, and the result is a skewing of the measured seroprevalence in a negative direction. For example, experimental infection data for the blue jay suggests that approximately 75% of infected blue jays die of WNV infection.12 If only 25% of infected birds survive, then our observed seroprevalence rate of 30% for blue jays grossly underestimates the true proportion of the blue jay population infected. Our model for predicting actual WNV infection rates concluded that 62% of the pre-epizootic blue jay population was infected.
High WNV mortality rates reported in laboratory studies do not necessarily portend large numbers of reported WNV-positive carcasses. Although the possible reasons for this are many, the bottom line is that few data are available on the actual reductions of bird populations due to natural transmission of WNV.18 We used the measured seroprevalence rates in Slidell together with experimentally derived mortality rates to predict the actual effects of WNV infections on populations of four bird species in Slidell. We predicted a large reduction for the blue jay (about 50%), smaller reductions for the house sparrow and northern cardinal (10–20%), and none for the northern mockingbird. These predictions have significant implications for avian mortality surveillance, confirming the utility of blue jays in Slidell (and possibly elsewhere in the southeastern United States), but casting doubt on the utility of the other three species. We assume that a 10–20% population reduction for a common bird species like the house sparrow or the northern cardinal may go unnoticed by most people. The blue jay represented 91.5% of all WNV-positive dead birds tested in Harris County, Texas, in 2002.19 Our derivation of a mathematical equation for predicting population reductions for birds due to WNV may have great utility beyond our small study in Slidell. The equation depends only on laboratory-derived mortality rates and field-derived sero-prevalence rates. Thus, site-specific population reductions can be predicted for many species. For example, the reported WNV seroprevalence of 60% for house sparrows in Queens, New York City in 19994 results in a prediction of approximately 40% population reduction there. Moreover, this calculation would apply to any infection and any host population.
With the exception of the woodpecker, all of the sampled candidate amplifying hosts were passerine birds and thus were likely to be competent for infecting mosquitoes. Published competence data are available for the blue jay and the house sparrow.12 Of 25 species evaluated in the laboratory, the blue jay was the most infectious to mosquitoes, and was expected to infect about 60% more Culex quinquefasciatus mosquitoes than the house sparrow. Cx. quinquefasciatus is suspected to be the vector responsible for the Slidell outbreak7 and is susceptible to WNV infection at increasing rates with orally ingested doses above 105 PFU/mL.20 Based on the competence data presented herein, the house sparrow and the cardinal would infect 156% and 40%, respectively, more mosquitoes than the mockingbird. By inference, the blue jay would infect 250% more mosquitoes than the mockingbird. Using the predicted actual WNV infection rates from our mathematical model together with the competence data and species-specific population estimates, we predicted the relative importance of the blue jay, the house sparrow, the northern cardinal and the northern mockingbird as amplifying hosts in the transmission cycle by calculating a relative mosquito inoculation index value, MIi. Because field conditions are not standardized, species-specific MIi values are expected to vary for each transmission focus, and will not always correlate with laboratory-derived host competence Ci values. Thus, whereas the blue jay was the most competent of the four bird species evaluated, it inoculated many fewer mosquitoes than either the house sparrow or the cardinal in Slidell.
The analysis above assumes accurate population data derived from the mark-recapture study. However, these data are crude, and may be influenced by numerous factors related to the bird species studied, including differing probabilities of capture and recapture, different probability for survival after blood sampling (possibly an issue for small birds such as Carolina wren and Carolina chickadee), and others. An inherent limitation of mark-release-recapture data is the assumption of a closed population, which would require an island situation to achieve. Although no birds were likely to have been born between early August and late October, it is likely that some individual birds entered the population as a consequence of postbreeding dispersal or even migration. For example, both cardinals and blue jays are partially migratory in some parts of their ranges. Furthermore, some birds may have emigrated from this population during this period. Certainly some birds would have died due to natural causes, although natural mortality probably would not have exceeded 20% for most passerine birds during the study period.21 If WNV was still circulating in the study population, then mortality may have been even higher for some species. Mortality, permanent emigration or immigration would result in a negative bias of detection probability and a positive bias in population size.
By sampling at site A in both August and October, we were able to obtain evidence for ongoing transmission between these sampling periods. Although the overall seroprevalence in birds remained constant, one of 12 recaptured birds sero-converted between the two sampling periods indicating that transmission did occur during the interim. Also, WNV was isolated from mosquitoes collected at other locations near case residences in Slidell in mid-August.7 However, most of the transmission had probably occurred prior to the August sampling, a finding consistent with the human epidemic study in Slidell (Bunning M, pers. comm.). Transmission may have been curtailed by intensive mosquito control efforts used during July and August.17
In summary, this study evaluated the avian hosts of WNV within a suburban transmission focus in Slidell, Louisiana. Primarily the northern cardinal and the house sparrow, but also the blue jay and the northern mockingbird, have been implicated by our model as important amplifying hosts due to high levels of exposure, moderate to high abundances, and experimentally demonstrated host competence. The northern cardinal and the northern mockingbird are likely to serve as effective target species for a surveillance program that utilizes free-ranging birds as sentinels in Slidell. The blue jay was implicated as an important target for avian mortality surveillance due to a very high predicted population reduction in Slidell. Chickens were frequently infected and will likely serve as effective captive sentinels in Slidell. This study supports the conclusion that certain summer-resident passerine birds are important amplifying hosts for MNV Slidell.
Received July 22, 2004. Accepted for publication January 6, 2005.
Acknowledgments: Brad Biggerstaff assisted with statistical analyses. Paul Doherty provided advice on estimating bird populations. Discussions with Andrew Spielman led to the concept of "mosquito inoculation index" as used in this paper. Field assistance was provided by Leah Colton, Sarah Lasater, Gabrielle Dietrich, Michel Bunning, Kevin Caillouet, Dawn Wesson, Megan Hughes, Rachel Bru, James Bru, and Emily Moore. The authors thank property owners for access to study sites. The St. Tammany Parish Mosquito Abatement District (Director, Charles Palmisano), Slidell Memorial Hospital (Associate Administrator, Bruce Clement), Grand Bay National Wildlife Refuge (Manager, Pon Dixson), Grand Bay National Estuarine Research Reserve (Manager, David Ruple), Bon Secour National Wildlife Refuge (Manager, Jereme Phillips), University of Southern Mississippi Animal Research Facility (Manager, Juliann Rich), and Michel Bunning provided logistical support. The authors thank Oliver Komar for his critique of the manuscript. Funding was provided by the Centers for Disease Control and Prevention.
* Address correspondence to Nicholas Komar, CDC, P.O. Box 2087, Fort Collins, CO 80522. E-mail: nkomar{at}cdc.gov 
Authors’ addresses: Nicholas Komar, Nicholas A. Panella, Stanley A. Langevin, Aaron C. Brault, and Eric Edwards, Centers for Disease Control and Prevention, P.O. Box 2087, Fort Collins, CO 80521, Telephone: 970-221-6400, Fax: 970-221-6476. Manuel Amador, Centers for Disease Control and Prevention, 1324 Calle Cañada, San Juan, Puerto Rico 00920, Telephone: 787-706-2399, Fax: 787-706-2496. Jennifer C. Owen, Department of Biological Sciences, University of Southern Mississippi, 118 College Drive, Box 5018, Hattiesburg, MS 39406, Telephone: 601-266-6215, Fax: 601-266-5797.
Reprint requests: Nicholas Komar, CDC, P.O. Box 2087, Fort Collins, CO 80522, Telephone: Office (970) 221-6496, Mobile (970) 567-4970, Fax: (970) 221-6476, E-mail: nkomar{at}cdc.gov.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. M. Nemeth, P. T. Oesterle, and R. A. Bowen Passive Immunity to West Nile Virus Provides Limited Protection in a Common Passerine Species Am J Trop Med Hyg, August 1, 2008; 79(2): 283 - 290. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Darbro and L. C. Harrington Avian defensive behavior and blood-feeding success of the West Nile vector mosquito, Culex pipiens Behav. Ecol., July 1, 2007; 18(4): 750 - 757. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
