Enhancing Eastern Equine Encephalitis Virus Surveillance in New Jersey: Optimized Collection of Culiseta melanura
Eastern equine encephalitis virus (EEEV) causes the most clinically severe neuroinvasive arboviral disease in the United States. The virus is endemic in eastern and Gulf Coast states and the Great Lakes region, causing cases annually. To detect EEEV circulation in its enzootic cycle before the virus infects humans and other mammals, mosquito control agencies in New Jersey have conducted mosquito surveillance using a series of permanent wooden resting box sites since 1975. We conducted 2 field studies, 1 evaluating resting traps and 1 evaluating efficacy of CO2 lures, to optimize collection of Culiseta melanura, the primary enzootic vector of EEEV. Resulting mosquito samples were subjected to molecular analysis to determine EEEV infection rates. Corrugated plastic boxes trapped more bloodfed Cs. melanura than other resting trap types (resting boxes, Centers for Disease Control and Prevention [CDC] resting traps, or fiber pots) and were similar to resting boxes in total number of female Cs. melanura caught. Further, non-baited CDC light traps were more successful in trapping host-seeking Cs. melanura than those baited with dry ice, a CO2 lure. The EEEV RNA was identified in Cs. melanura, Aedes vexans, Anopheles quadrimaculatus, and Uranotaenia sapphirina. Our findings indicate that corrugated plastic boxes and non–CO2 baited traps could improve detection of Cs. melanura. Mosquito control agencies are encouraged to periodically assess their surveillance strategy for EEEV.ABSTRACT
INTRODUCTION
Eastern equine encephalitis virus (EEEV) causes the most clinically severe domestic arboviral encephalitis, with a case fatality rate of approximately 30% for those with neurological disease (Feemster 1938, Goldfield and Sussman 1968, Przelomski et al. 1988, Lindsey et al. 2018). Most EEEV infections in humans are asymptomatic with less than 5% of seropositive individuals developing clinical disease (Goldfield and Sussman 1968). Annually, an average of 11 cases are reported in the United States, with most cases occurring in Northeastern or Gulf Coast states (Lindsey et al. 2018, Brown et al. 2021). However, 38 human cases were reported from 10 states, resulting in 19 deaths (a case fatality rate of 50%) during 2019 (Brown et al. 2021, Vahey et al. 2021).
In the northeastern USA, EEEV circulates in an enzootic cycle between passeriform birds and Culiseta melanura (Coquillett) in freshwater swamps. Other mosquito species serve as bridge vectors, which acquire infection from an avian reservoir and, after adequate incubation, transmit the virus to humans and other mammalian dead-end hosts (Crans et al. 1994, Molaei et al. 2015). The EEEV is endemic in New Jersey, with the state’s first outbreak of 32 human cases occurring over an 8 wk period in 1959 (Goldfield and Sussman 1968). Since then, sporadic EEEV cases and outbreaks in susceptible vertebrate hosts (primarily humans and horses) have been observed in New Jersey. In the coastal regions of New Jersey, enzootic circulation is driven by Cs. melanura in Atlantic white-cedar (Chamaecyparis thyoides [L.]) swamps. Inland regions have the highest numbers of Cs. melanura in red maple (Acer rubrum [L.]) swamps (Holden et al. 1954, Crans et al. 1994). Historically, monitoring of Cs. melanura populations early in the season has provided valuable information about enzootic circulation of EEEV, allowing mosquito control agencies to implement enhanced surveillance and vector control strategies.
Less well-established, however, are the bridge vector species driving spillover transmission from the enzootic cycle in swamps into residential or agricultural areas. These species tend to exhibit more catholic feeding behavior, allowing for efficient transmission between avian amplification hosts and incidental hosts (i.e., horses and humans). While the relative importance of bridge vector species varies geographically and seasonally, several species from genera Aedes, Culex, and Anopheles have been implicated throughout the northeastern USA, in addition to Coquillettidia perturbans (Walker) (Armstrong and Andreadis 2010). In New Jersey, Ae. sollicitans (Walker) and Cq. perturbans have been cited as the species with greatest bridge vector transmission potential, based on field-collected mosquitoes screened during epizootic and epidemic periods (Crans et al. 1986).
To monitor EEEV activity, vector control programs use a variety of mosquito resting trap types, which are used as diurnal resting places by mosquitoes (primarily those in genera Culiseta and Anopheles) (Burbutis and Jobbins 1958, Edman et al. 1968, Crans 1995, Howard et al. 2011). Resting trap types include traditional wooden resting boxes, fiber pots, corrugated plastic boxes (FlexCon), and Centers for Disease Control and Prevention (CDC) resting traps (BioQuip Products). The traditional wooden resting boxes used to trap Cs. melanura by many mosquito surveillance programs are 1 cubic foot, open on 1 end, and painted flat black on the outside with a red interior (Crans 1995). Fiber pots are readily available, inexpensive, lightweight, and effective at trapping Cs. melanura (Komar et al. 1995). Corrugated plastic boxes, structurally similar to traditional wooden resting boxes but 18″ deep (compared to the 12″ deep resting box), offer advantages in being both lightweight and collapsible for transport. The CDC resting traps resemble fiber pots but contain a battery-powered fan that prevents mosquitoes from escaping after they enter the trap.
To detect EEEV circulation in its enzootic cycle prior to viral infections occurring in humans and other mammals, New Jersey has conducted a surveillance program using a series of sites in the southern half of the state since 1975. Each site contains either 25 or 50 wooden resting boxes visited by county vector control staff on a weekly basis from the beginning of June through mid-November. Mosquitoes are aspirated from boxes and female Cs. melanura are pooled and tested for EEEV. Positive test results inform surveillance and control measures in surrounding communities.
In addition to monitoring resting mosquito populations, New Jersey conducts supplemental Cs. melanura and bridge vector surveillance using several types of host-seeking mosquito traps. One of these is the CDC light trap, often baited with dry ice to emulate CO2 emitted from a mammalian host. Because Cs. melanura is an ornithophilic species and birds emit lower quantities of CO2 than mammals, use of dry ice may hinder collection of Cs. melanura (Hachiya et al. 2007).
Here, we implement 2 trap-type comparison studies and present data to aid in optimization of EEEV surveillance methods. We assessed the performance of alternative resting trap types in addition to the wooden resting boxes used in the state-run surveillance program, as well as CDC light traps baited with and without dry ice. We also conducted molecular analysis on mosquitoes collected during site characterization and trap comparison studies to estimate EEEV infection rates.
MATERIALS AND METHODS
Mosquito trapping was conducted at an existing resting box surveillance site in Camden County, NJ, over a 9-day period (September 8 to 16, 2021) corresponding to CDC w 36–37. Mosquitoes were trapped during 2 trap-type comparison studies (September 13 to 16, 2021) and during preliminary characterization of the site (September 8 to 9, 2021). The site is ∼100 feet (∼30.5 m) in elevation and composed of coniferous and deciduous woodlands, including Atlantic white cedar swamp.
Site characterization
During initial site characterization (September 8 to 9, 2021), mosquitoes were collected using a combination of resting boxes (10 trap nights), a CO2-baited CDC light trap (1 trap night), and a CDC resting trap (1 trap night).
Resting trap comparison
For the resting trap-type comparison (September 13 to 16, 2021), we compared 4 trap types, including 1) traditional black wooden resting boxes (12″(30 cm) H 12″(30 cm) W x 12″(30 cm) deep), 2) CDC resting traps from BioQuip Products (Panella et al. 2011), 3) black corrugated plastic boxes from Flexcon (12″(30 cm) H x 12″(30 cm) W x 18″(45 cm) deep) and 4) fiber pots painted black on the inside (Komar et al. 1995) (Fig. 1). Over 4 consecutive nights, we implemented a four-treatment, four-period crossover design (Hedayat and Stufken 2003), placing 12 traps of each type (48 total traps per night) in a grid approximately 5 m apart using a predetermined random sequence (Fig. 1).
Citation: Journal of the American Mosquito Control Association 40, 2; 10.2987/23-7148
CDC light trap with and without CO2 comparison
To determine if using dry ice as a CO2 source impacts trapping efficacy of Cs. melanura using CDC light traps (John W Hock, Gainesville, FL; Model 512), a 2-treatment, 2-period crossover design was employed (Hedayat and Stufken 2003). On the first night, 4 CO2-baited CDC light traps and 4 non-baited CDC light traps (8 traps in total) were placed in a row approximately 50 m apart, using a predetermined random sequence. A two-day washout period (nights 2 and 3) was implemented to prevent residual carryover (i.e., “trap-out”) effects from the first night. On the fourth night, the opposite trap type was used at each location within the sampling site (8 traps in total) to control for position-specific effects. On each night, traps were operated from dusk to dawn.
Mosquito collection and processing
For both field studies, mosquitoes were collected from traps each morning within a four-hour period, transported on dry ice, and stored at –80°C. Mosquitoes collected from wooden boxes, corrugated plastic boxes, and fiber pots were anesthetized with triethylamine (TEA) and aspirated using a handheld aspirator. Mosquitoes were morphologically identified using the criteria of Darsie and Ward (2005) and sorted by species, sex, and bloodfed status on a chill table. Unfed Cs. melanura, in addition to all other mosquito species (regardless of engorgement status), were processed at New Jersey Public Health and Environmental Laboratories (NJPHEL). Bloodfed Cs. melanura were sent to Division of Vector-Borne Diseases (DVBD), Centers for Disease Control and Prevention, in Fort Collins, Colorado, for further molecular analyses.
Statistical analysis
For the resting trap study, trap efficacy was assessed by comparing total number of female Cs. melanura (both bloodfed and unfed) and bloodfed-only Cs. melanura, referred to as “all” and “bloodfed,” respectively. For the CO2-baited trap study, total number of (host-seeking) female mosquitoes were counted and sorted by genus for analysis. Poisson regression was used to compare trap types using these counts, and we evaluated period and trap location as potential adjustment factors. Pairwise comparisons were calculated as mean ratios (MR) (95% confidence intervals [CI]), computed by exponentiating estimates of model parameters. Model selection was based on likelihood ratio chi-squared tests, with significance level 0.05. Firth’s adjustment maximum likelihood (Firth 1993) was used to estimate model parameters, and final estimates (95% equal-tailed, percentile CI) were computed by bootstrap (Davison and Hinkley 1997) with 1,000 replicates. Poisson models were tested for overdispersion (Dean 1992), and Tukey’s method was used for multiple comparisons adjustment as required. Analysis was conducted using R version 4.0.3 (The R Project for Statistical Computing [r-project.org]). Infection rate estimates were calculated by species, reported as maximum likelihood estimate using Firth’s correction with 95% skew-corrected score CI (PooledInfRate R Package, version 1.5, CDC DVBD [see www.cdc.gov/westnile/resourcepages/mosqsurvsoft.html]).
Molecular testing of mosquitoes for EEEV RNA
All mosquitoes collected during site characterization activities and trap comparison studies were subjected to molecular testing. Abdomens of bloodfed Cs. melanura were homogenized individually and RNA extracted using QIAamp Viral RNA MiniKit (QIAGEN, Valencia, CA) at CDC DVBD. Extracted nucleic acid was subjected to real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) targeting EEEV, and those testing positive were subjected to confirmatory qRT-PCR (Lambert et al. 2003). To test for viral dissemination, the legs of abdomen-positive mosquitoes were tested using the same extraction process and qRT-PCR assay.
All other mosquito species, in addition to unfed Cs. melanura, were tested for EEEV RNA at NJPHEL. Briefly, mosquitoes were pooled by trap night, trap type, and species before homogenization by adding 500µl of TE buffer and bead beating with a copper bb in a Qiagen TissueLyser at a frequency of 300 1/sec for 5 min followed by centrifugation. RNA was extracted using QIAamp DNA Blood BioRobot MDx kit (Qiagen, Catalogue #965152) and target detection performed using a quadruplex real-time RT-PCR assay developed in NJPHEL. Primers and probes were based off the CDC singleplex assay for EEEV (Lambert et al. 2003) and performed in a multiplex along with primer/probes for Jamestown Canyon virus, West Nile virus, and St. Louis encephalitis virus.
RESULTS
Mosquito identifications and sorting
A total of 2,175 mosquitoes were collected during site characterization (n = 279) and the trap comparison studies (n = 1896). Overall, 2,020 were identified to the species level, with 20 species represented. The remaining 155 specimens were identified to the level of genus (Aedes, Anopheles, Culex) or species complex (Cx. pipiens/restuans/salinarius). Culiseta melanura comprised 20.14% (n = 434/2,155) of these mosquitoes, and 103 were engorged.
Analysis of resting trap-type comparison
Model fitting for the resting trap-type comparison indicated that neither period (P = 1.0) nor trap location (P = 0.36) contributed significantly to model fit, and the test for overdispersion was not significant (P = 0.94), indicating the Poisson model sufficiently characterized variation. Of the 4 resting trap types under evaluation, corrugated plastic boxes caught the highest number of Cs. melanura mosquitoes (mean per trap night: 3.15; 95% CI: 2.63–3.74) (Table 1). Corrugated plastic boxes and resting boxes were equally effective at trapping female Cs. melanura (MR: 0.87; 95% CI: 0.65–1.16; see Supplemental Table 1.1), and both were more effective at trapping Cs. melanura than fiber pots or resting traps (Fig. 2). Corrugated plastic boxes trapped the highest mean number of bloodfed Cs. melanura compared to all other trap types (mean per trap night: 0.95; 95% CI: 0.68–1.30) (Table 2), including the wooden resting boxes (mean per trap night: 0.58; 95% CI: 0.38–0.86) currently in use by NJDEP (Fig. 3).
Citation: Journal of the American Mosquito Control Association 40, 2; 10.2987/23-7148
Citation: Journal of the American Mosquito Control Association 40, 2; 10.2987/23-7148
Analysis of CDC light traps baited with and without CO2
For the CO2-baited trap comparison, model analysis indicated that neither period (P = 1.0) nor trap location (P = 0.45) significantly contributed to the fit, and the test for overdispersion was not significant (P = 0.79). Of the 498 mosquitoes collected during the CO2 analysis, CDC light traps baited with CO2 caught more mosquitoes, regardless of species (n = 307, 62%) than CDC light traps without CO2 (n = 191, 38%). However, the mean number of Cs. melanura collected from non-baited CDC light traps per trap night was 2.19 (95% CI: 1.09–4.40) times higher than the mean number of Cs. melanura collected from CO2-baited CDC light traps (P = 0.03). Furthermore, the non-baited CDC light traps were more favorable for Cs. melanura, which comprised 12.6% (24/191) of mosquitoes collected without CO2 bait compared to 3.3% (10/307) when CO2 was present (P < 0.01) (Fig. 4).
Citation: Journal of the American Mosquito Control Association 40, 2; 10.2987/23-7148
Viral testing
The 331 unengorged Cs. melanura collected during site characterization activities and trap-type comparison studies were split into 133 pools, 2 of which tested positive for EEEV, resulting in an estimated infection rate (IR) of 6.1 (95% CI: 1.1–19.6) per 1,000 mosquitoes. The abdomens and legs of 2 of the 103 engorged Cs. melanura tested positive for EEEV RNA (IR: 19.42 (95% CI: 5.34–68.05) per 1,000 mosquitoes).
Other mosquito species from which EEEV RNA was amplified included Ae. vexans (Meigen) (IR: 15.3 (95% CI: 0.9–75.7) per 1,000 mosquitoes), An. quadrimaculatus Say (IR: 21.4 (95% CI: 5.7–56.4) per 1,000 mosquitoes) and Uranotaenia sapphirina (Osten Sacken) (IR: 107.2 (95% CI: 6.6–409.3) per 1,000 mosquitoes) (Table 3).
DISCUSSION
In evaluating mosquito traps for EEEV surveillance in New Jersey, we determined that corrugated plastic boxes are a viable alternative to wooden resting boxes for the collection of resting Cs. melanura. We also showed that refraining from use of CO2 as a bait in CDC light traps resulted in collection of a higher number of host-seeking Cs. melanura and reduced collection of non-target species.
In our resting trap comparison conducted shortly after the peak of EEEV activity in New Jersey, corrugated plastic boxes outperformed the historically used wooden resting boxes in trapping bloodfed Cs. melanura. The black interior of the corrugated plastic boxes may have provided visual cues to attract mosquitoes, as previous studies have demonstrated the color preference of Aedes mosquitoes to be black, followed by red (Brett 1938; Brown 1954; Browne and Bennett 1981). Further, the greater depth of the corrugated plastic boxes (18″ (45 cm) deep compared to the 12″ (30 cm) depth of wooden resting boxes) might have resulted in a more substantial microclimatic gradient, providing a more favorable ambient temperature to resting Cs. melanura. However, without direct measurements of temperature and RH within each trap, differentiation of these effects from other abiotic factors (e.g., light) remains difficult. An important advantage of the corrugated plastic boxes is their maneuverability in the field—the material is more lightweight than wooden resting boxes, and the boxes easily collapse down when not in use. Fiber pots and CDC resting traps were not as effective as wooden resting boxes or corrugated plastic boxes in trapping Cs. melanura. Interestingly, battery-powered CDC resting trap demonstrated the lowest success rate in trapping Cs. melanura. This contrasts with our previous findings demonstrating that fiber pots caught as many Cs. melanura as wooden resting boxes when operated during July–October (Komar et al. 1995), and that addition of a battery-powered fan to CDC resting traps increased efficacy in trapping Culex spp. vectors of West Nile virus in Colorado (Panella et al. 2011). Geographic and seasonal variations in humidity and temperature might influence the relative efficiency of different trap types for different mosquito genera because temperature has been shown to impact mosquito phenology and behavior (Mushegian et al. 2021). Therefore, trap-type comparisons should be performed at different locations and at different times of the transmission season to determine optimal trap types for Cs. melanura.
We also characterized whether presence of dry ice as a CO2 source influenced the relative efficacy of CDC light traps in collection of Cs. melanura because different volatiles emulating those emitted by their preferred host could impact trap efficacy and species diversity (Kline and Mann 1998, Bernier et al. 2008, Verhulst et al. 2010, Busula et al. 2017, Díez‐Fernández et al. 2020, Giordano et al. 2021). Congruent with previous studies, our study demonstrated that non-CO2 baited CDC light traps caught more Cs. melanura per trap night than those baited with dry ice as a CO2 source (Hachiya et al. 2007). This may be due to this mosquito species’ ornithophilic feeding preferences, as birds emit smaller quantities of CO2 than mammals (Cooperband and Cardé 2006). Further, Cs. melanura comprised a greater proportion of overall mosquitoes in non-CO2 baited traps, resulting in reduced trapping of ‘non-target’ species. Thus, programs or agencies targeting primarily Cs. melanura (e.g., by EEEV vector surveillance programs) might consider foregoing use of CO2 as a means of reducing non-target species and expediting the sorting process. Not using dry ice in EEEV surveillance programs also has implications for cost, effort, and convenience. Dry ice is costly, sublimates more drastically under extreme ambient conditions, and can present hazards to individuals handling it in the field. However, the lack of CO2 might decrease the yield of potential bridging vectors so trap types used should be based on the specific goal of the surveillance program (e.g., identification of enzootic or bridging vectors for EEEV).
In preparing for and conducting the trap-type comparison study, we had the opportunity to further investigate the potential contribution of Cs. melanura and other mosquito species to EEEV amplification in southern New Jersey. The EEEV was detected not only in the enzootic vector, Cs. melanura, but in 3 other species of mosquitoes (Ae. vexans, An. quadrimaculatus, and Ur. sapphirina). Aedes vexans is a suspected bridge vector of EEEV transmission, having been the first species from which EEEV was isolated in Connecticut (Wallis et al. 1960). While Ae. vexans has ecological and life history characteristics that make it an optimal bridge vector (e.g., aggressively anthropophilic feeding behavior and large broods following heavy rains), laboratory competence studies indicate Ae. vexans might not play a significant role in EEEV transmission (Nasci and Mitchell 1996, Vaidyanathan et al. 1997, Armstrong and Andreadis 2010). Anopheles quadrimaculatus is known to feed on many mammal species, including white-tailed deer (Molaei et al. 2009), a species where evidence of EEEV infection has been described (Tate et al. 2005, Schmitt et al. 2007, Mutebi et al. 2011). Further, An. quadrimaculatus is competent for EEEV in the laboratory and has been suggested as a potential epidemic vector for EEEV (Vaidyanathan et al. 1997, Moncayo et al. 2000). Given this species’ local abundance and shared habitat with Cs. melanura, further monitoring of this species to characterize its potential as a bridge vector of EEEV in New Jersey is warranted.
Lastly, we detected EEEV RNA in 1 pool of Ur. sapphirina, a species known to prefer feeding on annelids (e.g., worms and leeches), but which has also been shown to feed on amphibians, reptiles, birds, and humans (Molaei et al. 2008, Reeves et al. 2018, Khalil et al. 2023). Its predilection to feed on ectotherms has resulted in its incrimination as a potential overwintering host of EEEV (White et al. 2011, Bingham et al. 2012, Graham et al. 2012). Other reports detecting EEEV RNA in Ur. sapphirina in the southern USA postulate that in addition to the well-characterized enzootic cycle between Cs. melanura and passeriform birds, an additional transmission cycle might exist between Ur. sapphirina and ectotherms (Cupp et al. 2003, Graham et al. 2012).
The trap comparison studies were subjected to limitations inherent to the opportunistic nature of this field survey and its short duration. Intensive trapping in specific locations during a 1-week period limits extrapolation of our results to the rest of the year. However, we conducted both trap-type comparison studies during epidemiologic wk 37, a period during which Cs. melanura populations have remained active (according to historical data available through Rutgers University, https://vectorbio.rutgers.edu/reports/vector/). Additionally, the limited trapping period and optimizing the trapping for Cs melanura might have restricted the identification of potential bridging vectors. Further, these studies took place at 1 study site in Camden County, and caution should be exercised in translating these results to other settings—particularly those with differing ecology. Lastly, small mosquito sample sizes resulted in sometimes relatively wide, and imprecise, confidence intervals. To further corroborate the results of the trap-type comparison studies before implementation in a routine surveillance program, additional trap-type comparisons are needed (preferably at different sites throughout New Jersey or other states in the northeastern USA).
The EEEV infection causes severe neurologic illness in humans and other dead-end hosts, and detection of viral activity early in the enzootic cycle allows for prompt implementation of the vector control measures required to disrupt epizootic transmission into human or domestic animal populations. Evaluation of vector monitoring programs is an important step in assessing overall arboviral surveillance and can aid in obtaining a more granular understanding of local EEEV ecology. Each jurisdiction with EEEV circulation is encouraged to periodically reassess their vector control program to optimize their early detection and prevention of EEEV transmission.
SUPPLEMENTAL MATERIAL
Data tables describing estimated mean ratios of Culiseta melanura trapped during the resting trap-type comparison analysis.
Image of 4 resting trap types used in the resting trap comparison study with an example of the four-treatment, four-period crossover design. Note: images not to scale.
Ratios of estimated mean numbers (95% CIs) of Culiseta melanura females trapped per trap night for each pair of trap types, adjusted for trap location and trapping period and multiple comparisons. A dashed line marks equality of the means. For specific ratio measurements, see Supplemental Table 1.1 (CP = black corrugated plastic boxes; FP, fiber pots; RB, traditional wooden resting boxes; RT, CDC resting traps).
Ratios of estimated mean numbers (95% CIs) of bloodfed Culiseta melanura female mosquitoes trapped per trap night for each pair of trap types, adjusted for trap location and trapping period and multiple comparisons. A dashed line marks equality of the means. For specific ratio measurements, see Supplemental Table 1.2 (CP = black corrugated plastic boxes; FP, fiber pots; RB, traditional wooden resting boxes; RT, CDC resting traps).
Number of female mosquitoes, by genus, trapped using CDC light traps with (n = 307) and without (n = 191) the use of dry ice as a CO2 source (8 trap nights/treatment). Culiseta melanura and Coquillettidia perturbans were analyzed at the species level. * “Other” category includes Psorophora spp. and Uranotaenia spp.
Contributor Notes