Field-Based Estimates of Avian Mortality from West Nile Virus Infection

Introduction

S oon after West Nile virus (WNV) arrived in North America, researchers began documenting bird mortality related to WNV infection (McLean 2006, Kilpatrick et al. 2007). Mortality has been most obvious in a few species, which resulted in many public health agencies using dead birds, particularly corvids, as sentinels for WNV activity in an area (Eidson et al. 2001). In areas with high WNV activity, several researchers have documented population declines in several bird species (Yaremych et al. 2004, Caffrey et al. 2003, LaDeau et al. 2007). Although it is known that WNV causes mortality in numerous bird species, field estimates of mortality rates are difficult to measure (Yaremych et al. 2004). Several laboratory studies have estimated mortality in a variety of species experimentally infected with WNV (Komar et al. 2003, 2005, Reisen et al. 2005). However, laboratory studies may not reflect mortality due to WNV in free-ranging birds. In the laboratory, sick individuals can recover from infection in the absence of stressors such as predators, social competition, and feeding young, and they have ad libitum access to food. Conversely, stress associated with captivity may interfere with recovery from experimental infection and increase mortality. We used mark-recapture techniques to estimate the annual apparent survival of a free-ranging resident population of Northern Cardinals (Cardinalis cardinalis) in an area with endemic WNV activity. We used an information theoretic approach to examine differences in the apparent survival of cardinals with and without WNV antibodies and during the period of the year when WNV is active.

Individuals that survive WNV infection develop antibodies, and are likely immune to future WNV infection (Fang and Reisen 2006, Komar et al. 2001, 2005, Nemeth et al. 2007). We assumed that survival of seropositive individuals reflects natural, population-wide survival rates in the absence of WNV. By contrast, the survival rate of birds with no previous WNV exposure (no antibodies) should be equal to natural baseline survival plus additional mortality due to WNV. Thus, by comparing survival of immune individuals (antibody positive) with vulnerable individuals (antibody negative), we can estimate mortality associated with WNV infection. However, an additional confounding factor may be that chronic effects lower the survival of seropositive individuals (Kilpatrick et al. 2006). We refer to chronic effects as an adverse effect stemming from WNV exposure that, while not directly responsible for the individual's death, contributes to its death. Some examples may be slower reflexes and therefore a higher chance of predation, or decreased foraging success and therefore a greater chance of starvation. Further, if chronic effects influence survival, individuals with WNV antibodies would have a lower survival rate than individuals without antibodies during the non-WNV season.

Materials and Methods

We captured, banded, and recorded recaptures of male and female Northern Cardinals every 2 weeks from August 2005 until June 2007 in three woodlots (8, 12, and 25 ha in size) on the South Farms of the University of Illinois, Urbana-Champaign. Cardinals were captured in seven walk-in potter traps and five mist nets positioned in the same locations throughout the woodlots, and only adult birds (i.e., after hatch year individuals) (Halkin and Linville 1999) were used in the analysis. In this location WNV has been detected since 2002, and the overall infection rate in Culex mosquitoes (the species of mosquitoes responsible for WNV transmission in the area) from the beginning of July to the end of September (the peak of WNV transmission) in the Champaign-Urbana area (which includes the study area) was 7.5 per 1000 mosquitoes in 2005 and 21.9 per 1000 mosquitoes in 2006 (Lampman et al. unpublished data).

We had 49 sampling periods with approximately equal sampling effort (3–4 days, ∼4 h in the morning per day, each 2-week period) in which Northern Cardinals were captured. Each time a bird was captured, we collected 0.1–0.2 mL of blood from the jugular vein using a 27-gauge and ½-inch needle, or from the brachial vein by making a small prick in the vein and collecting the blood via capillary tubes. An epitope-blocking enzyme-linked immunosorbent assay was performed, as previously described in Beveroth et al. (2006), to detect the presence of WNV antibodies. We used three monoclonal antibodies, 3.1112G, 2B2, and 6B6C-1. Monoclonal antibody 3.1112G is specific to WNV, whereas 2B2 reacts with both WNV and Koutango Virus and 6B6C-1 can react with all flaviviruses, including WNV.

Results

Of the 204 Northern Cardinals captured, 21 went from being WNV antibody negative to antibody positive (these were not included in the analysis), 24 remained antibody positive throughout, and 159 never tested positive for antibodies. There was an adequate fit of the model to the data (p = 0.23). The model with the most support suggests differences in the biweekly survival rate of Northern Cardinals between individuals with and without WNV, in both the WNV season and the rest of the year, and that recapture probability of the cardinals varied over time (Table 1). In our study population, there was an estimated 34.6% (i.e., 94.9% [95% CI 78.5–100.0%] antibody positive and 60.3% [95% CI 40.8–82.7%] antibody negative) reduction in survival between individuals with and without WNV antibodies during the WNV season (Fig. 1). Conversely, the trend in the non-WNV season (all months except July, August, and September) suggests that individuals without WNV antibodies had a slightly greater (9.0%) survival rate (41.1% [95% CI 24.8–58.6%] antibody positive and 50.1% [95% CI 37.7–62.5%] antibody negative). There was no evidence for sex-specific differences in Northern Cardinal survival. The apparent annual survival rate of antibody-positive individuals was 45.5% (95% CI 59.6–34.4%).

OPEN IN VIEWER

FIG. 1.

Apparent biweekly survival rates of Northern Cardinals with (Anti+) and without (Anti−) West Nile virus (WNV) antibodies during the WNV season and the non-WNV season. These estimates were generated from the most supported model [ϕ(season*anti) p(t)]. 95% Confidence intervals are shown.

Our estimate of mortality attributed to WNV infection (34.6%) was not significantly different than that reported from the experimental infections of this species by Komar et al. (2005) of 22.2% (p = 0.39). We also did not detect any decline in the local Northern Cardinal population with the advent (circa 2002) or continued WNV activity in the area (slope of 0.01 for the CBC and 0.02 for the SBC; Fig. 2).

OPEN IN VIEWER

FIG. 2.

The number of Northern Cardinals per party hour during the Champaign County Spring Bird Count (SBC) and the National Audubon's Christmas Bird Count (CBC) conducted in Champaign County from 1996 to 2008. The line represents the arrival of WNV in the area.

Discussion

Our results suggest that WNV affects the survival of free-ranging Northern Cardinals. Support for the season*anti model indicates differences in survival between birds with and without WNV antibodies in both the WNV season and during the remainder of the year. Although the model suggests differences in both time periods, difference in apparent survival was much greater during the WNV season. Individuals that have been infected with WNV and survive, develop antibodies and have a relatively high rate of survival during the WNV season, whereas susceptible individuals have a relatively low survival rate. Although the survival rate of Northern Cardinals without WNV antibodies was low in the WNV season, it was still not as low (though the 95% CI do overlap) as it was in the winter. This suggests that survival is lowest in the non-WNV season, probably due to winter mortality (Newton 1998). We also did not detect any difference in WNV-related mortality between this study and the experimental inoculations by Komar et al. (2005), suggesting that the experimental inoculations may produce mortality estimates similar to what would be observed in the field.

The difference in apparent survival between the antibody-positive and antibody-negative individuals may reflect the percentage of the population that succumb to WNV. This is assuming there is no difference in dispersal rates and non-WNV-related mortality. The recapture probability varied over time, likely because in winter we had a higher recapture rate as birds were moving between feeding locations, which resulted in the concentration of birds near our nets and traps. During the breeding season the cardinals were dispersed across the area and we generally caught fewer individuals.

A comparison of the apparent survival rate of antibody-positive individuals (45.5%) in this study to other studies suggest antibody-positive individuals may have a slightly lower survival rate when compared with what has been found in other studies. A similar study in Maryland before the arrival of WNV estimated annual survival for cardinals at 60% (SE 6.0%) (Karr et al. 1990), and a study in southern Ontario found that 65% of breeding birds returned between years (Scott 1967). This finding of slightly lower survival than other species before the arrival of WNV, and the result of the survival analysis suggest the possibility of chronic effects from WNV. Although the magnitude of differences in the apparent survival rates for the non-WNV season is much smaller than during the WNV season, the difference is consistent with chronic effects. These results are also consistent with Kilpatrick (2003), who found that individuals that survive infection of malaria had lower survival rates, and Reisen et al. (2006) have found that birds can be chronically infected with WNV. A longer-term study with a larger sample size would be needed to adequately address the chronic effects of WNV in birds. If chronic effects persisted after the WNV season, given that winter mortality may be the bottleneck for resident bird populations (Newton 1998), it is likely that these lower conditioned birds would quickly be removed from the population, so the effect may be most pronounced in early winter.

Although substantial WNV-related mortality is occurring in our study area, WNV does not appear to be impacting the local population. This is in agreement with LaDeau et al. (2007), who estimated population trajectories of Northern Cardinals before and after the arrival of WNV in Illinois using the Breeding Bird Survey data. They found that the Northern Cardinals were exhibiting no decline, and there were no clear impacts from WNV in this population. This was true for Northern Cardinals not only in Illinois, but also in all areas they analyzed. The finding that Northern Cardinal populations appear to be unaffected by WNV, while WNV has an impact on survival rates, suggests that Northern Cardinal populations are able to compensate for the WNV-related morality. Komar et al. (2005) estimated that WNV may have had a small impact (12% decline based on formulas associated with WNV-related mortality, population size, and seroprevalence rate) on a local Northern Cardinal population in Slidell, Louisiana. Although there are likely many reasons why some species are impacted while others are not, this study suggests that, while mortality in Northern Cardinals may increase with the presence of WNV, the population appears to be resilient to this disease.