Over 20 years since its introduction, the West Nile virus (WNV) continues to be the leading cause of arboviral disease in the USA. In Panama City Beach (Bay County, FL), WNV transmission is monitored using sentinel chickens and testing mosquito pools for presence of viral RNA. In the current work, we monitored WNV transmission from 2014 to 2020 through weekly serology sampling of sentinel chickens; mosquito populations through biweekly mosquito collections by suction traps (1 m and 9 m) and weekly gravid trap collections; and mosquito infection rates using a reverse transcriptase–polymerase chain reaction (RT-PCR) assay. Samples were sent to the Bureau of Public Health Laboratories (Tampa, FL) for testing presence/absence of WNV via RT-PCR assay. Our results indicated that canopy surveillance could augment ground collections, providing greater proportions of Culex mosquitoes with less bycatch compared with ground collections. Serology indicated 94 seroconversions to WNV in the study area from 2014 to 2020. The most active year was 2016, which accounted for 32% (n = 30) of all seroconversions reported during the study period. We detected 20 WNV-positive mosquito pools from Culex quinquefasciatus during 2014–17; mosquito infection rates ranged from 2.02 to 23.81 per thousand (95% CI). Climate data indicated anomalously high precipitation in 2014–19 preceding WNV transmission. Data analyzed herein indicate utility in year-round continuous and diversified surveillance methodologies. This information is needed to properly calibrate future models that could assist with predicting transmission events of WNV in Panama City Beach, FL.
West Nile virus (WNV; Flaviviridae, Flavivirus) is the most prominent arthropod-borne virus in the continental USA and has been endemic in Florida for more than 20 years. Outbreaks, sporadic and focal (Shaman et al. 2005), resulted in over 400 recorded human cases from 2001 to 2020 (FDOH 2022a). In Florida, WNV was first isolated in corvids from postmortem tissue samples in 2001 in Jefferson County, located in the northeastern Panhandle of Florida (Fig. 1; Blackmore et al. 2003). The Florida Panhandle region has repeatedly observed outbreaks in humans (37% of all human cases in Florida are assumed to have occurred in this region) and is a focus of residual transmission during nonepidemic years (FDOH 2022a). West Nile virus transmission in Florida is monitored using a statewide sentinel chicken surveillance program (approximately 26 participating counties) and testing mosquito pools for presence of WNV RNA. Sentinel chicken flocks have been used to reliably detect arbovirus activity (antibody testing) in Florida since the 1970s (Day et al. 1991, Rutledge et al. 2003, Shaman et al. 2005, Sallam et al. 2016, Beeman et al. 2021). Mosquito pool testing (reverse transcriptase–polymerase chain reaction [RT-PCR]) is also a reliable method to survey virus populations in host species (Shroyer 2001, Sutherland and Nasci 2007, Gu et al. 2008), but trap bias and bycatch hinder mosquito collection and processing efforts, resulting in increased financial costs and low infection rates. Therefore, knowledge of trapping methodologies that target Culex species are essential to optimize mosquito pool testing protocols.
Like the type of traps and attractants used, height of trap placement (e.g., at ground level, in the canopy) can result in significant differences in estimates of mosquito community composition, physiological status, and abundance (Anderson et al. 2004, Andreadis and Armstrong 2007). Moreover, mosquito vertical patterns are plastic and controlled by fine-scale habitat use (i.e., resource availability) and climatic variables (temperature, humidity, precipitation), which influence abundance, movement, and interaction with potential hosts (Burkett-Cadena et al. 2013, Janousek et al. 2014, Wardhaugh 2014), demonstrating a need for collecting local and regional canopy surveillance data of WNV vector species (Giordana 2021). Unlike other orders within Diptera that have continuity in studies performed across vertical strata (Axtell et al. 1975, Leprince et al. 1994, McGregor et al. 2018), regionally specific canopy surveillance data are considered sporadic and are not continuous in reporting for Culicidae, constituting a substantial gap toward understanding WNV transmission and missed opportunities for optimizing control efforts.
In the Florida Panhandle, Cx. p. quinquefasciatus (hereafter Cx. quinquefasciatus) and Cx. nigripalpus Theobald are the primary vectors for WNV when considering amplification and epizootic and epidemic transmission cycles (Rutledge et al. 2003, Vitek et al. 2008, Day et al. 2015). Culex quinquefasciatus is opportunistic in its host seeking and abundant in urban areas (Wilke et al. 2019), implicating this species as a primary enzootic vector in anthropogenic landscapes (Rochlin et al. 2019). Culex nigripalpus along with Cx. salinarius Coquillett are considered bridge vectors (Godsey et al. 2015) and, generally, but not always, are more abundant in rural and sylvatic habitats (Rey et al. 2006), although populations can occur throughout suburban and urban terrain (Schwartz 2014). Due to a paucity of studies for host seeking for these 2 vector species, vertical stratification host-seeking behaviors need to be investigated further, where mosquito control agencies should be implementing canopy trap surveillance in endemic areas for WNV.
The FDOH response to WNV transmission is strategic communication to public health, mosquito control, and media agencies (FDOH 2022b, see Chapter 11). Advisories, alerts, and threats are guided by animal surveillance activity (e.g., avian, equine, sentinel chicken), weather conditions, time of year, vector species surveillance, arbovirus epidemiology models, historic arbovirus distribution data, and human cases of disease in the targeted county and also adjacent counties. Robust estimates of mosquito population density and infection/transmission rates are essential. Nevertheless, limitations remain with regard to sampling bias, the lag time associated with antibody production, and differential diagnostic testing algorithms. This exacerbates lag times from sample collection to reporting, in some cases exceeding 3 wk (FDOH 2022a).
Despite over 2 decades of mosquito and virus surveillance, we are still unable to accurately identify when and where WNV epidemics will emerge in Florida. In the current work, we analyzed 7 years of mosquito and WNV surveillance data from Beach Mosquito Control District (BMCD) located in Panama City Beach, FL (Bay County; Fig. 1). The BMCD monitors mosquito and arbovirus abundance and diversity through a variety of sampling methodologies, including ground and canopy space collections of wild-type mosquitoes and sentinel chicken surveillance to infer estimates of arbovirus activity. The objectives of the current work are to summarize current surveillance methodologies employed in Panama City Beach, FL, describe seasonal and vertical distributions of mosquito abundance and sentinel chicken seroconversion, describe the sampling effort for WNV vector species, describe mosquito community composition, and elucidate patterns between sentinel chicken seroconversion, mosquito abundance, mosquito infection data, and environmental data.
MATERIALS AND METHODS
Study area and collection sites
Panama City Beach, FL (Bay County), is located in the Florida Panhandle along the Gulf Coast and has a population of 36,640. Panama City Beach habitat is characterized by hydrophytic plants and seasonally saturated hydric soils. Suburban areas are dominated by wetlands and semiaquatic land saturated by seasonal precipitation. Wetlands cover approximately 30% of the landscape and are estuarine marshes, mixed hardwoods, scrub shrub, wet pine flatwoods, pine savannah, and cypress–gum tree swamps (USFWS 2022). Ditch systems for drainage of these habitats have been installed since 1952 and have been gradually replaced with cyclic retention and catch basin systems for stormwater management with increasing human settlement and development. The southern suburban area is set against sand dunes, dune lakes, and coastal strand habitats that can be observed southernly from the east side to the west side of the region. Brackish salt marshes are located on the northern east side and are encompassed by bay waters.
Mosquito, arbovirus surveillance, and control operations are managed by BMCD, originally formed as The Gulf Mosquito Control District on October 14, 1952, when the citizens of the Panama City Beach area voted to create a mosquito control district. The district's name was changed in early 1997. The BMCD is an independent special taxing district that is governed by a board of 3 elected 4-year-term commissioners. Funding is received to operate the district through the levying of local ad valorum taxes and state matching funds. The district operates under Chapter 388 of the Florida Statutes and Rule 5E-13 under the Florida Department of Agriculture and Consumer Services. Surveillance has been conducted at 16 statically placed collection sites from 1998 to 2020 (Fig. 1).
We collected mosquitoes at 3 sentinel chicken monitoring stations that have been active from 1998 to present time (Fig. 1). Site A (14th Street, Florida Department of Health [FDOH] ID 0008; 30.24777, −85.93150) is located in a suburban neighborhood and positioned on a county-owned drainage easement set in between 2 homesteads on either side and adjacent to a deciduous swamp to the east; this site has been active since 2006. Site B (Treatment Plant; FDOH ID 0001; 30.21764, −85.85190) is located in a suburban government commercial property wastewater treatment facility, positioned on the north-median side of the property line backed up against a deciduous swamp; this site has been active since 1998. Site C (Ed's Sheds; FDOH ID 0005; 30.19035, −85.77700) is located in a suburban commercial storage facility on the northeast property line corner set against a cedar swamp to the east; this site has been active since 2003.
Mosquito collection, identification, and viral testing
At each location, one Centers for Disease Control and Prevention (CDC) updraft suction trap (Model 1312 with the ultraviolet light removed; John W. Hock, Gainesville, FL), baited with pressurized carbon dioxide (250 ml/min) and 1-octen-3-ol (Nosquito: NS16; Kaz, Marlborough, MA), was hoisted 9 m into surrounding canopy and understory layers using a pulley (hereafter referred to as canopy trap); one CDC updraft suction trap (as previously described) was hung at shoulder height (1 m); and one CDC gravid trap (Model 1712; John W. Hock) was placed on the ground (0 m). The CDC updraft suction trap (9 m and 1 m) collections occurred twice per calendar week from February through November each season. Gravid traps were baited with a hay infusion (50 gal of tap water, 10 g of brewer's yeast, and 2 lb of hay fermented for 7 days; prepared fresh biweekly) and set once per calendar week from February through November. Each trap was set in the field for 24 h. Thereafter, trap contents were transported to BMCD on ice packs and immediately placed in a −20°C freezer. A −40°C ultrafreezer was purchased in 2020 and used for the duration of the study period. Mosquitoes were identified to species using the keys of Darsie and Morris (2003), Burkett-Cadena (2013), and Harrison et al. (2016). Culex nigripalpus and Cx. quinquefasciatus were separated into species pools of no more than 50 individuals and shipped on dry ice to the Bureau of Public Health Laboratories (BPHL; Tampa, FL) to be tested for presence of WNV by RT-PCR assay. We only pooled and tested gravid females of the aforementioned species due to budgetary and staffing constraints. Since mosquito pool size varied throughout the study, infection rates were calculated using bias-corrected maximum likelihood estimation (MLE) in Microsoft Excel 2017 using the PooledInfRate v.4 statistical software package (Biggerstaff 2006). Infection rates represent the number of infected mosquitoes per 1,000 mosquitoes.
Proportions and seasonal distributions
Mosquito trap data (9- and 1-m updraft traps and 0-m gravid traps) for 2014 through 2020 were recorded in MapVision Generation 1 (Leading Edge Incorporated), a web-based geospatial database management software, and exported into Microsoft Excel 2010 by BMCD personnel. To standardize mosquito abundance, we calculated the total number of mosquitoes collected per trap-night for each taxon and for all trap types. We plotted the proportion of each taxon using the “ggplot2” package in R (Wickham 2016). We calculated weekly totals of mosquitoes per trap-night and averaged across all years to generate seasonal distributions of taxa. The weekly binned data set was visualized using the “ggplot2” package in R (Wickham 2016).
Diversity and rarefaction extrapolation curves
To examine differences in the diversity of mosquito species collected from each trap, measures of Hill numbers (q = 0, richness; q = 1, Shannon diversity; and q = 2, Simpson diversity) were calculated with an individual-based rarefaction and extrapolation procedure, using the “iNEXT” package in R (Chao et al. 2014, Hsieh et al. 2020). Abundance-based Morisita–Horn index was computed to assess pairwise similarities between mosquito communities across each trap, using the “SpadeR” package in R with a bootstrap of 1,000 replications (Chao et al. 2006, 2016). The Morisita–Horn index is not biased to sample size (Chao et al. 2006) and takes into account richness and abundance. Results range from 0 (no community overlap) to 1 (complete community overlap).
Three sentinel flocks (6 chickens in each) were active from January 1, 2014, to December 9, 2014; May 27, 2015, to December 14, 2015; February 1, 2016, to December 5, 2016; May 22, 2017, to December 4, 2017; March 5, 2018, to December 10, 2018; February 11, 2019, to December 23, 2019; and March 1, 2020 to December 6, 2020. An additional susceptible flock was maintained each year in a mosquito-proof (screened-in) enclosure located at BMCD. All chickens were confirmed to be negative for anti-WNV antibodies prior to being placed in the field. Sentinel chicken blood was extracted by trained phlebotomists from the brachial wing vein weekly and shipped to the BPHL (Tampa, FL) to screen for presence of anti-WNV antibodies (FDOH 2022b). Sera were first screened with a flavivirus antibody group hemagglutination inhibition assay (HAI). In December 2017 the antigen in the HAI assay was switched out, which led to an increased sensitivity for WNV. After parsing for flavivirus infections, samples with titers >1:10 were tested for presence of West Nile–specific IgM antibodies by enzyme-linked immunosorbent assay (ELISA). Negative, equivocal, or inconclusive samples were rebled and assayed in a WNV plaque reduction neutralization test (PRNT). All 6 chickens in each coop were sampled weekly. Seropositive or deceased chickens were removed from the flock and replaced with a susceptible chicken.
Weekly counts of seropositive sentinel chickens for 2014 through 2020 were recorded in Microsoft Excel 2010 by BMCD. We calculated weekly seroconversion rates by dividing the number of seropositive chickens (ELISA or PRNT) by the total number of chickens tested each week. Chickens produce detectable levels of WNV-specific antibodies 5 days postexposure (Senne et al. 2000). Therefore, we lagged sentinel chicken data by 1 week to account for this delay.
We obtained daily total precipitation (mm), minimum temperature (°F), and maximum temperature (°F) from 2 weather stations located in Panama City Beach, FL, from 2014 to 2020: 1) NOAA climate station USC00086842 (30.2491, −85.6605); and 2) Florida State University Panama 5N station (30.1899, −85.7234). We computed weekly average minimum temperature, maximum temperature, and precipitation to obtain a single value representing the sampling location. We used the anomalize() function in the “anomalize” package (Dancho and Vaughan 2020) to detect anomalies in precipitation and temperature time series using the interquartile method. Weekly average minimum temperature, maximum temperature, and precipitation and anomalies were plotted using the “ggplot2” package in R (Fig. 2) (Wickham 2016, R Core Team 2020).
Mosquito abundance and seasonal distribution
We collected a total 131,442 mosquitoes comprising 8 genera and 41 species. We identified 13 species of medical and veterinary importance (pestiferous or nuisance) in our traps, including Aedes albopictus (Skuse) (2.5%), Ae. atlanticus Dyar and Knab (4.9%), Ae. canadensis canadensis (Theobald) (4.9%), Anopheles crucians Wiedemann complex (16.1%), An. quadrimaculatus Say complex (0.3%), Culiseta inornata (Williston) (0.1%), Cs. melanura (Coq.) (1.1%), Cx. coronator Dyar and Knab (0.1%), Cx. erraticus (Dyar and Knab) (4.1%), Cx. nigripalpus (18.9%), Cx. quinquefasciatus (19.2%), Cx. restuans Theobald (0.9%), and Cx. salinarius (15.5%). We classified all other species as “other mosquito species” (11.4%). The proportions and seasonal distributions of 13 mosquito species we deemed to be medically relevant in northwestern Florida are presented in Fig. 3 and Supplementary Table 1. We present all mosquito collection data in minimum information standard for reporting arthropod abundance data (MIReAD) format (Supplementary Table 2; Rund et al. 2019).
Culex was the most abundant genus across all trap types, totaling 62.8% of 9-m collections, 48.4% of 1-m collections, and 92.5% of gravid collections. Culex nigripalpus was the most abundant vector species in 9-m and 1-m updraft traps, which made up 30.0% and 21.1% of collections, respectively. Culex quinquefasciatus showed a clear affinity for gravid traps (87.6% of gravid trap collections) (Fig. 3 and Supplementary Table 1). Culex nigripalpus and Cx. quinquefasciatus were abundant throughout the sampling period. We consider Cx. coronator to be a rare occurrence at 9 m and in gravid traps (<1% of total collections). In 1-m updraft collections, Cx. coronator was occasionally collected throughout the summer and early fall. Culex salinarius was the 2nd-most abundant vector species captured from the canopy, totaling 17.1% of collections but was a rare occurrence in gravid trap collections (<1%). Collections from the 9-m and 1-m traps indicate an early springtime peak in abundance (Fig. 3). An additional 3 Culex species were identified, including Cx. interrogator Dyar and Knab (<0.1% of total collections), Cx. pilosus (Dyar and Knab) (0.1%), and Cx. territans Walker (<0.1%).
Aedes albopictus was present throughout the study period, albeit in low abundance (<5% of total collections) and more persistent during the summer months in 1-m and gravid trap collections. Aedes atlanticus distribution began early in the season and persisted into the fall; 9-m trap collections were sporadic. In contrast to the other species, Ae. canadensis distribution began early in the season during weeks 15 to 20, and, thereafter, rapidly declined. Anopheles crucians was the 3rd-most abundant species collected during the sampling period, accounting for 21.9% of 1-m trap collections, and was abundant throughout the study period. Anopheles quadrimaculatus and Cs. inornata were rarely collected across all collection methods (<1%). In general, Ae. atlanticus, Ae. canadensis, Cs. melanura, Cx. erraticus, Cx. restuans, and Cx. salinarius distributions were similar across all collection methods and had lower abundances in gravid collections (Fig. 3).
Trapping effort and similarities
Rarefaction/extrapolation curves (q = 0) of the 3 sampling methodologies are presented in Fig. 4. The extrapolation curves of the 9-m and 1-m traps were closest to the total observed richness in this survey (n = 41). The 1-m trap curve indicated this method captured the greatest number of mosquitoes and was the most complete of the 3 approaches. Nevertheless, a greater sampling effort is required to provide an improved diversity estimation, specifically for the 9-m traps, as indicated by truncation of the extrapolation curves prior to reaching the asymptote. Neither the interpolated nor extrapolated 9-m trap curve approached a defined asymptote, indicating that more sampling is needed to observe greater species diversity estimates. As expected, the extrapolated gravid trap curve approached an asymptote at a lower richness observed across all 3 traps in the current study.
The Morisita–Horn similarity indices and number of shared taxa are presented in Table 1. Morisita–Horn similarity indices ranged from 0.184 (9 m/gravid) to 0.873 (9 m/1 m) (Table 2). The 9-m and 1-m traps were the most similar assemblages (32 shared species, Morisita–Horn = 0.873) and the 9-m and gravid traps were least similar (22 shared species, Morisita–Horn = 0.184). In general, traps placed at 9-m and 1-m heights exhibited greater estimated species richness and diversity compared with gravid traps (Table 1). The Shannon diversity estimates for the updraft traps (9 m and 1 m) each indicated 10 common species, and the Simpson diversity estimates indicated 6 and 7 dominant species, respectively (Table 2). As expected, gravid traps yielded the lowest richness, and the alpha-diversity metrics indicated 2 common and 1 dominant species from this assemblage (Table 2).
Sentinel chicken seroconversion
A total of 5,153 chicken serum samples were extracted and tested during the study period. Two percent (94/5,123) of serum samples tested showed evidence of seroconversion to WNV. The proportion of seroconverting chickens (no. seropositive birds/total no. susceptible chickens) was low (≤0.04) across all years and sampling locations (Table 3). Weekly seroconversion rates across all sampling sites are presented in Fig. 4. We observed July through August in Panama City Beach to be the most active for mosquito transmission of WNV. The most active weeks are summarized as follows: in 2014, 5 of 18 chickens seroconverted during week 33; in 2016, 8 of 18 chickens seroconverted during week 25, 5 of 18 during week 32, and 6 of 13 chickens during week 33 (5 chickens were recorded as dead during this week); in 2019, 6 of 18 chickens seroconverted during week 25; and in 2020, 6 of 18 chickens seroconverted during week 32. In 2014 and 2018, chickens seroconverted to WNV for 4 consecutive weeks and continued for an additional 3 wk thereafter, indicating sustained and optimal environmental conditions to support mosquito transmission events.
Temperature and precipitation anomalies are presented in Fig. 4 and total counts summarized in Table 4. The results of the anomalize analysis revealed a total of 36 precipitation, 19 minimum temperature, and 16 maximum temperature anomalies were observed during the study period. The hottest summers (average maximum temperature from June to September) were 2016 (33.3°C ∓ 0.3 SD) and 2019 (33.2°C ∓ 1.0) (Supplementary Table 3).
Generally, chickens seroconverted during the warmest weeks of each season, coinciding with minimum and maximum average temperature peaks. Very little transmission occurred at the troughs, exceptions being early-season transmission events recorded in 2016 (week 14) and 2019 (week 16) and late-season in 2014 (week 46), 2015 (week 44), 2016 (week 45), and 2018 (week 48) (Fig. 4).
Mosquito infection rates
Mosquito infection data presented as bias-corrected MLE are summarized in Table 5 and Supplementary Table 4. The BPHL tested a total of 5,678 mosquitoes grouped into 297 pools. Fifteen percent (20/297) of Cx. quinquefasciatus pools tested positive for presence of WNV. Ninety percent (18/20) of positive pools were collected from July to September. West Nile virus infection rates were the highest in 2014 (11.93 per thousand [5.37–23.81]) and lowest in 2017 (1.74 per thousand [0.31–5.75]). In 2017, serology testing indicated no presence of WNV in sentinels, although 3 positive pools of Cx. quinquefasciatus were confirmed on July 26, August 3, and August 22. None of the 164 mosquito pools collected in 2018, 2019, and 2020 amplified WNV genetic material. Infection rates by month are presented in Supplementary Table 5.
Mosquito advisories, alerts, and control measures for Bay County
Confirmed reports of WNV sentinel chicken seroconversion, mosquito pools, and human cases resulted in mosquito-borne disease advisories and alerts being issued in Bay County in 2016 (n = 2) and 2018 (n = 2). During the week of June 27, 2016, 8 chickens seroconverted and 2 Cx. quinquefasciatus mosquito pools tested positive for presence of WNV. An additional 17 chickens seroconverted July 18 through August 22, and confirmation of a human case on September 7 resulted in a county-wide alert. In 2018, an advisory was issued on June 29 after confirmation of 1 human case. The advisory was updated to an alert after a 2nd human case was confirmed on June 30. Eight additional chicken seroconversions were reported from July 9 to July 30. Thereafter, 3 human cases were reported from August 23 to September 24, and 11 chickens seroconverted from August 6 through October 10.
During the 2016 outbreak, aerial ultra-low volume applications of Dibrom (AMVAC Chemical Corporation, Axis, AL) were performed using a single boom Micronair spray system (Model AU6539; Bromyard, United Kingdom) with a 1972 Bell OH-58C helicopter navigated with a Wingman GX management system (ADAPCO, Lake Mary, FL). From July through September of 2016, 46,077 acres were treated with Dibrom, spanning 3 separate treatments (243.1 gal/application rate; 0.67 oz per acre/swath; 1,065 feet/airspeed; 86 knots). Additionally, in 2018, 5 aerial applications treated 57,869 acres with Dibrom (300.0 gal; 0.67 oz per acre/swath; 1,065 feet/airspeed; 86 knot) from June through September.
In Florida, mosquito community composition across vertical strata is not well studied. Here we demonstrate that WNV vector species occupy heights >1 m and in the canopy and understory layers year-round in Panama City Beach. Though more mosquitoes were collected at 1 m compared with 9 m, a lower sampling effort is required to detect the same diversity of Culex species. In the 9-m collections, Culex mosquitoes were collected in greater abundance when compared with other genera collected. All Culex species were more abundant in 1-m collections versus 9-m collections except for Cx. quinquefasciatus, which was evenly distributed across suction trap collections. Our results conflict with previous findings throughout the eastern USA and indicate the need for region-wide local surveillance data. For instance, in Indian River County, Cx. nigripalpus showed no affinity for height (Giordano et al. 2021). In the current work, Cx, nigripalpus was collected at 1 m in abundances 4 times that of 9-m collections (Supplementary Table 1). And in Louisiana, Culex salinarius were collected in significantly greater abundances at 3 m compared with 1.5 m (Mackay et al. 2008); ornithophilic Culex were collected at abundances up to 6 times that of ground-level collections in Connecticut (Anderson et al. 2006); in North Carolina, Culex mosquitoes were more abundant in 7.6-m collections than ground-level collections (Savage et al. 2014); but no significant difference in abundance was observed in New York (Darbro and Harrington 2006). Nevertheless, the presence of unfed females in 9-m collections baited with host emanations indicated host-seeking behavior across vertical strata in Bay County. Together, our results indicate capacity for WNV surveillance vertically in Panama City Beach, and where hosts and vectors are likely to come into contact, although species-specific attraction is predicated by positions through the gonotrophic cycle.
In this study we examined patterns of WNV infection of mosquitoes and sentinel chickens in Panama City Beach, FL, from 2014 to 2020. This work demonstrates that in Bay County many sentinel chickens seroconverted to WNV, indicating this region can provide suitable conditions for focal WNV epidemics. In the current work, we focused our arbovirus surveillance efforts on Cx. nigripalpus and Cx. quinquefasciatus given their historical role in WNV transmission in the region. West Nile virus RNA was recovered only from Cx. quinquefasciatus pools. We did not detect WNV RNA in Cx. nigripalpus pools, though, given the low sample size (only 178 individuals were grouped into 25 pools) this was expected. Pools of Cx. nigripalpus were not performed generally for arbovirus surveillance due to the lack of abundance in gravid trap collections where pooling of mosquitoes from canopy collections were not performed over the study period. In Monroe County, FL, Anderson et al. (2022) indicated in 2016–17 a WNV infection rate of 6.04 in Cx. nigripalpus during the same time period as WNV infection rates recorded for Cx. quinquefasciatus in Bay County, FL.
Upon visual inspection of climate data and seroconversion time series, 2014, 2015, 2016, 2018, and 2019 mosquito transmission events followed anomalously high precipitation recordings (Fig. 4). In the Florida Panhandle, the dry season typically runs from October to April. Fifty-three percent (19/36) of anomalously high precipitation days observed occurred during the dry season. For instance, in 2014, 6 of 9 anomalously high precipitation days occurred from February to April, historically dry months in Panama City Beach (Fig. 4 and Table 4). And in 2016, 3 of 5 anomalously high precipitation days occurred in March and April. In 2018, Bay County received a total of 81 inches of rain, breaking the previous record of 66 inches in 2014 (https://www.weather.gov/tae/). Increased precipitation events flush mosquitoes from larval habitat but also have potential to create new ones. Furthermore, anomalously high precipitation events, if followed by increased temperature, can result in substantial increases in relative humidity. Together, this represents optimal conditions for mosquito proliferation: abundance of larval habitat, decreased larval development times, and optimal humidity for questing adults.
For this study, it is important to consider WNV's pattern of sporadic and focal transmission in Florida. During periods of low amplification, i.e., low mosquito and/or host infection rate, fine-sampling resolution is required to detect a signal. Zero or low infection/seroconversion rate does not necessarily mean transmission is not occurring, but rather the sampling effort may be insufficient to detect a signal. In the current work, mosquito trapping was performed from May to December as follows: suction traps were set 2 days/wk and gravid collections once per week. The sentinel program was inactive each year from December to February, but in some years did not resume until March or May. These gaps in surveillance restrict aggregating data into a continuous time series and, therefore, limit our ability to elucidate spatiotemporal drivers of transmission dynamics. From this study, the issue of these gaps in data is being corrected in Panama City Beach by continuous surveillance seasonally, including serology testing for WNV. We recommend programs perform at minimum biweekly sampling during the off-season, to ensure a continuous time series of data collection and to establish baseline estimates of wintertime mosquito and virus populations.
Furthermore, the number of mosquitoes sampled is important in determining detection probability for arboviruses. We calculated detection probability using the formula suggested by Gu and Novak (2004) and using an estimated infection rate of 0.1%. Detection probability for WNV was moderate to moderately high across 2014 (49%), 2015 (62%), 2016 (50%), 2019 (74%), and 2020 (88%), but very low sample sizes in 2017 (n = 121) and 2018 (n = 447) resulted in low detection probabilities (11% and 35%, respectively). The preponderance of positive mosquito pools during focal epidemics from 2014 to 2016, despite low sample size (<1,000 mosquitoes were pooled and tested), suggests infection rates may have been greater than 0.1%. This hypothesis is supported by the large number of sentinel chickens seroconverting for WNV during these years. The lack of human case data reported in 2014 and 2015 is perplexing but could be attributed to WNV's clinical presentation and diagnosis, i.e., approximately 80% of human cases are asymptomatic and approximately 20% present acute systemic febrile illnesses, which may not be severe enough to warrant clinical diagnosis (CDC 2022). Furthermore, health officials are challenged with determining approximately when and where each WNV-human infection occurred (often based on personal testimony and health records), a difficult task given the complexity of individual human behavior and movement patterns and ubiquitous mosquito populations. Due to these reasons, human case data can be an unreliable measure of mosquito transmission activity. The serological data we present clearly show WNV transmission to sentinel chickens with a 1-wk lag for antibody production.
In Florida, there is a lack of mosquito pool testing data when compared with other states with organized mosquito and arbovirus surveillance regimes, in large part due to the support of the statewide sentinel chicken surveillance program which has been in place since 1977. A dearth of species-specific information sets limits our understanding of mosquito-borne disease systems and prevents the development of targeted mitigation campaigns. Moreover, several mosquito control districts perform mosquito pool viral assays at their own facilities, but the use of commercial viral detection assays, a failure to maintain cold chain, and inadequate storage can result in RNA degradation of the original sample, demonstrating the importance of proper sample handling and storage and the dangers of relying on ostensibly negative results. Adequate cold chain and cold storage (−40°C to −80°C) are essential to preserve RNA, but ultra-freezers are costly equipment that require maintenance and placement and can quickly incapacitate budget allocations. While we encourage programs to expand their arbovirus surveillance capacity, it is important to be aware of specialized training and financial and infrastructural limitations required to handle RNA safely and responsibly in the laboratory. West Nile virus, though recently downgraded to a Biological Safety Level 2 agent, remains to be a nationally notifiable disease. In Florida, all mosquito pool testing results are verified by the state public health laboratories (Tampa) before inclusion in the state arbovirus surveillance reports curated by the Florida Department of Health (FDOH 2022a, 2022b). This requires an extra step at the cost of the programs, i.e., programs supply collection materials, dry ice/cold packs, waste disposal, and shipping costs.
The present study documents WNV infections in mosquitoes and sentinel chickens at 3 sampling sites over 7 years, 2014–20, in the Florida Panhandle. We include vector species' seasonal and vertical distributions, mosquito infection rates, and reports of sentinel chicken seroconversion. Our results indicate that Panama City Beach is capable of supporting focal and sporadic outbreaks of WNV, but a diverse mosquito and arbovirus surveillance program provides valuable information about transmission risk to the public and a comprehensive mosquito control program to mitigate adult and juvenile populations of mosquitoes through the use of insecticides. We observed sentinel chicken seroconversion and pooling Cx. quinquefasciatus mosquitoes in the gravid physiological state to be an efficient method of detecting WNV activity in the study area. Nevertheless, no one trapping methodology yielded optimal catch for all Culex species. Results of this study demonstrate the utility of diversified and strategic sampling regimes that have potential to reduce sampling effort and improve surveillance of target species. This in turn will improve decision-support systems, while efficient and cost-effective surveillance programs reduce financial and labor burden of mosquito control and public health agencies.
We dedicate this manuscript to the late Dale Martin, BMCD entomologist (1998–2013), for his arduous efforts constructing these sites from scratch including the canopy traps. We thank Daniel R. Hood, lead mosquito technician for arbovirus surveillance, for performing site maintenance and handling collections from 2010 to 2020. We also thank BMCD for their support and the operations crew who assisted in maintaining the integrity of the arbovirus surveillance system from 1998 to 2020. Salary support for BVG was provided by the Florida Medical Entomology Laboratory's Applied Mosquito Research Program (Florida Department of Agriculture and Consumer Services Contract Number 27396). This work is/was supported by the USDA National Institute of Food and Agriculture, Hatch project 1021482. The funders did not influence and had no role in study design, data analysis, and manuscript preparation.
Beach Mosquito Control District, 509 Griffin Boulevard, Panama City Beach, FL 32413.
Florida Medical Entomology Laboratory, University of Florida | IFAS, 200 9th Street, Vero Beach, FL 32962.