Use of Scented Sugar Bait Stations to Track Mosquito-Borne Arbovirus Transmission in California

Abstract

Laboratory and field research was conducted to determine if Culex tarsalis Coquillett expectorated West Nile virus (WNV) during sugar feeding and if a lure or bait station could be developed to exploit this behavior for WNV surveillance. Experimentally infected Cx. tarsalis repeatedly expectorated WNV onto filter paper strips and into vials with wicks containing sucrose that was readily detectable by a quantitative reverse transcriptase-polymerase chain reaction assay. Few females (33%, n = 27) became infected by imbibing sugar solutions spiked with high concentrations (10⁷ plaque forming units/ml) of WNV, indicating sugar feeding stations probably would not be a source of WNV infection. In nature, sugar bait stations scented with the floral attractant phenyl acetaldehyde tracked WNV transmission activity in desert but not urban or agricultural landscapes in California. When deployed in areas of the Coachella Valley with WNV activity during the summer of 2011, 27 of 400 weekly sugar samples (6.8%) tested positive for WNV RNA by reverse transcriptase-polymerase chain reaction. Prevalence of positives varied spatially, but positive sugar stations were detected before concurrent surveillance measures of infection (mosquito pools) or transmission (sentinel chicken seroconversions). In contrast, sugar bait stations deployed in urban settings in Los Angeles or agricultural habits near Bakersfield in Kern County supporting WNV activity produced 1 of 90 and 0 of 60 positive weekly sugar samples, respectively. These results with sugar bait stations will require additional research to enhance bait attractancy and to understand the relationship between positive sugar stations and standard metrics of arbovirus surveillance.

In North America the mosquito-borne encephalitides, including West Nile virus (WNV), are maintained and amplified in enzootic transmission cycles involving vector mosquitoes and a variety of passeriform birds (Komar 2003, Kramer et al. 2008). Surveillance programs track the intensity of enzootic amplification to provide antecedent measures of risk useful in implementing timely intervention to prevent tangential transmission to humans and equines that may suffer neuroinvasive disease and death. Testing groups of mosquitoes (pools) for the presence of virus provides an estimate of infection incidence but not transmission, although virus titer in positive pools may be linked to transmission potential (Armstrong and Andreadis 2010). To monitor the intensity of transmission, groups of sentinel animals such as chickens, are deployed at strategic sites and monitored over time for evidence of infection. However, the bulky nature of chicken coops and the cost of maintenance typically preclude the deployment of large numbers of flocks thereby limiting geographic coverage. In addition, during the on-going WNV epidemic, sentinel chickens were reputedly less sensitive than other measures of virus activity, especially for mosquito species such as CulexpipiensL. that hunt avian bloodmeals within tree canopy (Anderson et al. 2004).

Recently, FTA cards (Whatman International Ltd., Maidstone, United Kingdom) impregnated with honey were used to measure Barmah Forest and Ross River virus transmission by host-seeking mosquitoes collected using traps baited with CO₂ gas (Hall–Mendelin et al. 2010), thereby exploiting previous laboratory reports that flaviviruses were expectorated during sugar feeding (Doggett et al. 2001, van den Hurk et al. 2007). Although technically this method provides data similar to avian baits that measure infection and transmission rates (Bellamy and Reeves 1952, Reeves et al. 1961, Rutledge et al. 2003), the use of honey impregnated cards to collect viral RNA was novel, precluded the use of avian hosts, and allowed trap deployment for longer than a single night. However, the use CO₂ gas to attract host-seeking females failed to take advantage of the fact that most mosquitoes feed repeatedly on fructose throughout the gonotrophic cycle as well as during seasons when gonotrophic activity is suppressed (Reisen et al. 1986). Sugar baits seemed especially useful in arid environments (Schlein and Muller 2008) and have recently been used for Culex control (Muller et al. 2010a,b).

The current laboratory and field research was conducted to determine if Cx. tarsalis frequently expectorated WNV during sugar feeding and if a lure or bait station could be developed to exploit this behavior for use in arbovirus surveillance programs. Herein, we confirmed the frequent expectoration of flaviviruses during sugar feeding by experimentally infected Culex females and verified our ability to detect WNV RNA deposited on dental wicks baited with sugar and the floral attractant phenyl acetaldehyde, a component of some flower scents (Jhumur et al. 2006, 2008). Also presented are promising field results from California, where WNV RNA positive sugar baits were detected before seroconversions by sentinel chickens.

Methods and Materials

Laboratory Studies

Before the start of field trials, laboratory experiments evaluated 1) whether WNV RNA was detectable on sugar wicks fed upon by WNV-infected mosquitoes, 2) the limits of RNA detection in wicks with 66% sucrose and phenyl acetaldehyde, and 3) if WNV deposited on sugar wicks may serve as a source of virus for infecting mosquitoes. The latter experiment addressed the important potential issue of contaminant infection during cohabitation within holding cartons used for vector competence experiments.

Experiment 1 used two colonies of Culex tarsalis Coquillett for experimentation including BFS, a colony started in 1956 from mosquitoes collected in Bakersfield, Kern County, CA. (Bellamy and Kardos 1958) and YOLO, a colony founded in 2004 from mosquitoes collected at the Yolo Bypass Wildlife Area, Yolo County, CA. Mosquitoes were infected by feeding on heparinized chicken blood mixed with 10⁸ plaque forming units (PFU) per ml of the CA04 strain of WNV (isolated from a dead Yellow-billed Magpie in Sacramento County, CA, during 2004 and passaged three times in Vero cells; GenBank accession number DQ080059) using a Hemotek feeding system and membranes provided by the manufacturer (Discovery Workshops, Accrington, Lancashire, United Kingdom). Postinfection blood-fed mosquitoes were transferred to clean 0.67 liters (1 pint) paper cartons and held for 13 d postinfection (dpi) at 26°C. During this holding period mosquitoes were offered either sugar wicks or sugar papers that were initially positioned on day 1, and then replaced at 6, 11, and 13 dpi. Sugar wicks were 3 cm long cotton dental rolls soaked in a 66% honey solution and tinted blue with food coloring. The sugar papers were 1 × 2 cm pieces of no. 2 filter paper treated as above, but allowed to air dry before use. Sugar wicks were inserted and protruded through the side of the paper carton, whereas sugar papers rested on the screened lid of the carton.

At 13 dpi, mosquitoes were anesthetized with tri-ethylamine and expectorant collected using the capillary tube method (Aitken 1977) as described previously (Reisen et al. 2006). Mosquito legs were removed and placed in separate vials. Sugar wicks, papers, mosquito bodies, legs, and expectorant were stored at −80°C. Mosquito bodies and legs were homogenized in 1.0 ml of Dulbecco’s Modified Eagle Medium (DMEM; Life Technologies, Carlsbad, CA) containing 500 U/ml penicillin, 50 μg/ml streptomycin, and 20% fetal bovine serum. RNA was extracted from a 0.1 ml supernatant aliquot by a MagMax system (Life Technologies) using the viral isolation kit and manufacturer protocols. Samples were tested for WNV RNA by a quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) assay using an ABI7900 platform (Life Technologies) and primers/probe specific for the envelope region of the viral genome (Lanciotti et al. 2000). Sugar wicks and papers were tested for WNV RNA by qRT-PCR as described above, except before extraction 1.0 and 2.0 ml MagMax lysis/binding solution concentrate was added to the wicks and papers, respectively. Samples were incubated at 4°C overnight before RNA extraction. Diagnostic procedures and laboratories complied with Biological Use Authorization protocol 0873 approved by the University of California Davis Internal Biosafety Committee.

Experiment 2 assessed the sensitivity of our qRT-PCR assay for detecting WNV RNA on our field sugar bait (66% sucrose scented with phenyl acetaldehyde). One milliliter of bait solution was added to a 1.5 ml cryovial. Dental wicks (1.0 cm) were inserted and the tubes were inverted allowing the wicks to saturate with sugar bait. WNV (CA04) was diluted 10-fold from 10⁸ to 10¹ PFU/ml and 0.01 ml pipetted in triplicate onto saturated sugar wicks. Because field samples are stored at −80°C before testing, spiked sugar wicks also were stored at −80°C for 24 h before testing. RNA was extracted and qRT-PCR performed in triplicate as described above.

Experiment 3 evaluated whether mosquitoes became infected after feeding on cotton pads moistened with a WNV-sucrose mixture to simulate contaminated sugar pads such as those found in experiment 1 and possibly could be encountered at sugar bait stations in the field. WNV was mixed 1:10 in 10% sucrose in a 10-fold dilution series from 10⁸ to 10² PFU of WNV/ml, and then used to wet sugar pads placed on 0.67 liters paper cartons housing 3- to 5-d-old Cx. tarsalis from the YOLO colony. Mosquitoes were allowed access to the virus-sugar mixture for 48 h, after which the pads were replaced with clean pads containing only 10% sucrose. Mosquitoes were held for 10 d at 26°C. All mosquitoes surviving to 10 d, the WNV-infected sucrose solutions, and the sugar pads collected after 48 h of exposure to mosquitoes were tested individually for WNV RNA by qRT-PCR as described for experiment 1 and for infectious WNV by standard plaque assay on Vero cell culture (Kramer et al. 2002).

Field Studies

Five sugar bait stations were placed 3–7 m apart along vegetative ecotones within 30–50 m from flocks of sentinel chickens deployed at duck clubs along the shore of the Salton Sea in Coachella Valley, in urban habitats in Los Angeles, and in agricultural or riparian habitats near Bakersfield in Kern County, CA, during the summer of 2011.

Sugar bait stations were constructed of a one-half inch (3.8 cm) pvc pipe slip cap fitted with a three-fourths inch (1.9 cm) slip plug glued to the inside and drilled to accept a 1.5 ml cryovial (Fig. 1). Vials were filled with a 66% sucrose solution scented with phenyl acetaldehyde and plugged with a 1 cm segment of dental wick (Fig. 1). At collection, the cap of the cryovial was screwed on and the entire vial placed in a cryovial storage box. Boxes were stored at −80°C until shipped on dry ice to the Center for Vectorborne Diseases laboratory where they were tested by qRT-PCR as described above. We attempted to confirm positive samples by retesting using primers/probe from the NS1 region of the WNV genome (Shi et al. 2001) and/or by re-extraction and retesting using the envelope gene primers/probe as described above.

Fig. 1.

Ventral aspect of sugar bait station showing a cryovial loaded with sucrose solution dyed blue with food coloring and the dental wick plug.

Sentinel chickens were bled bi-weekly and sera tested for WNV antibodies using an enzyme immunoassay (EIA) (Patiris et al. 2008) by the California Department of Public Health laboratory in Richmond, CA. In addition, mosquitoes collected by either dry ice-baited Centers for Disease Control and Prevention style (Newhouse et al. 1966) or hay infusion-baited gravid (Cummings 1992) traps were tested for infection by qRT-PCR as described above. In this way both WNV infection (mosquito pools) and transmission (sentinel seroconversions) were monitored concurrently in areas where the sugar bait stations were deployed. During the summer of 2011, stations were deployed continuously, with wicks replaced at 3–7 d intervals, at five sites in the Coachella Valley from 22 June to 7 October, at two sites in Los Angeles (Encino, 19 September to 14 October and Rowland Heights, 2 September to 24 October) and at three sites in Kern County from 5 to 19 September. The protocol for the maintenance and sampling of sentinel chickens was approved by the University of California Internal Animal Care and Use Committee.

Results

Laboratory Studies

In experiment 1, 78% (21 of 27 samples) of the sugar papers and all of the sugar wicks tested positive for WNV RNA by qRT-PCR (Table 1). There was significantly less (two-way analysis of variance (ANOVA); F = 62.2; df = 1, 48; P < 0.001) WNV RNA on paper samples (mean = 32.8 Ct, cycle threshold) than on wicks (mean = 24.3 Ct), but the quantity of RNA at the three time points did not change with mosquito age, because the overall means of 29.0, 27.9, and 28.8 Ct on 6, 11, and 13 dpi, respectively, were statistically similar (F = 0.35; df = 2, 48; P > 0.05). Because the sugar was lightly tinted with blue food coloring, mosquito feces were visible throughout the paper carton, on the screened lid and on the sugar wicks. Some of the wicks may have become contaminated by defecated viral RNA, because all nine wicks were positive on day 6, before most mosquitoes should be able to transmit WNV at 26°C (Reisen et al. 2006). In addition, wicks in carton three were positive, even though expectorant samples from all surviving mosquitoes were negative. Although sugar papers placed on the cage lid should have had less fecal contact, as they were positioned above the enclosure area, six of nine sugar papers collected at 6 dpi also were positive for WNV RNA and two additional cartons had sugar papers positive for WNV RNA even though all expectorant samples were negative. However, each carton with a positive sugar paper or wick and negative expectorant samples contained mosquitoes that had bodies and legs positive for WNV RNA. The high proportion of WNV RNA positive wicks and papers demonstrated that WNV RNA could be detected on artificial sugar sources fed upon by WNV-infected mosquitoes. In addition, these data indicated that WNV RNA was sufficiently stable over the 5 d exposure period to be detected using qRT-PCR.

Table 1.

Detection of WNV RNA in sugar wicks and papers at 6, 11, and 13 dpi after being fed upon by experimentally infected Cx. tarsalis mosquitoes

Wick/paper	Carton	Colony	nb	Mosquitoes			Sugar wicks and papers
				% WNV RNA Positive			qRT-PCR Cta
				Body	Expect.c	Legs	6 dpid	11 dpi	13 dpi
Wick	1	BFS	9	100	44	100	23.8	24.1	25.0
Wick	2	BFS	11	100	55	100	25.1	23.0	24.6
Wick	3	BFS	9	78	0	78	22.9	25.4	26.5
Wick	4	BFS	4	75	50	50	21.7	23.1	28.7
Wick	5	BFS	5	100	80	100	25.9	23.0	23.6
Wick	6	BFS	22	100	NT	100	21.8	22.3	21.8
Wick	13	YOLO	12	92	92	92	27.1	26.2	26.0
Wick	14	YOLO	10	100	60	90	25.6	21.5	23.2
Wick	17	YOLO	22	59	41	59	23.9	24.1	27.4
Paper	7	BFS	8	63	38	38	26.2	UD	35.8
Paper	8	BFS	7	43	14	43	25.9	UD	UD
Paper	9	BFS	9	100	11	89	31.3	32.6	27.6
Paper	10	BFS	5	40	20	40	UD	31.3	35.9
Paper	11	BFS	3	33	0	33	30.5	21.8	32.2
Paper	12	BFS	11	82	0	82	UD	32.9	32.8
Paper	15	YOLO	8	75	63	75	34.2	33.4	29.2
Paper	16	YOLO	10	90	80	90	UD	29.6	26.5
Paper	18	YOLO	14	93	65	86	35.7	28.6	31.4

^aCt = cycle threshold, generated by qRT-PCR, a relative measure of WNV RNA; UD = undetermined with Ct > 40.

^bn = sample size or no. of mosquitoes.

^cExpectorant.

^dDays postinfection.

In experiment 2, 0.01 ml of each viral aliquot from a 10-fold dilution series containing 10⁸ to 10¹ PFU of WNV was added into 1.0 ml of sugar bait solution, reducing the final titer two-fold or by 10² as shown on the x-axis of Fig. 2. Overall, WNV RNA was detected by qRT-PCR in all sugar baits containing from 10⁶ to 10¹ PFU/ml. When the Ct scores of the WNV RNA standards from viral culture were compared with samples collected from sugar baits, the sugar bait samples were on average 3.2 (SD = 1.0) Ct higher (Fig. 2), also indicating that the titer of WNV in sugar baits was decreased. WNV RNA was detected in all three replicate samples spiked with 10⁸ to 10⁴ PFU (effectively 10⁶ to 10² PFU/ml; Fig. 2); RNA was detected in two out of three replicates spiked with 10³ PFU. The mean Ct score for 10⁴ of 35.8 (SD = 0.15) was not different than the mean Ct score for 10³ of 35.8 Ct (SD = 0.9). Samples spiked with 10² or 10¹ PFU or with an effective concentration of ≤10¹ PFU/ml tested negative by qRT-PCR.

Fig. 2.

Cycle threshold scores (Ct) for WNV RNA from sugar baits and standards plotted as a function of effective WNV titer in plaque forming units per milliliter. (Online figure in color.)

In experiment 3, six groups of mosquitoes exposed for 48 h to sugar pads containing 10¹ to 10⁶ PFU of WNV/ml tested negative for WNV RNA on 10 dpi. In contrast, 9 (33%) of 27 mosquitoes offered 10⁷ PFU of WNV/ml tested positive (Table 2). These data indicated that it was possible to infect some sugar feeding mosquitoes with WNV, but only when ≥10⁷ PFU WNV/ml was present in the sugar meal. WNV RNA and presumably infectious virus decreased over the 48 h exposure period on the sugar wicks, with Ct scores increasing significantly (paired t-test = 2.3; df = 6; P= 0.03) from a mean of 21.5 Ct at 0 h to 25.5 Ct at 48 h. Furthermore, WNV did not retain infectivity on sugar pads. By 48 h, infectious WNV was only detected on one of two sugar pads spiked with 10⁸ PFU WNV and the titer here was reduced to 10^2.3 PFU WNV/ml.

Table 2.

Detection of WNV RNA in Culex tarsalis mosquitoes that were allowed to feed on sugar

Titera	Ct sugar T0b	Ct sugar T48c mean ± SD		nd	% WNV RNA pose
107	9.5	15.9	1.3	27	33
106	13.3	16.2	0.3	17	0
105	16.5	22.1	2.5	20	0
104	20.3	24.8	0.4	19	0
103	23.7	27.4	0.4	24	0
102	27.3	32.1	0.3	20	0
101	30.6	UD		15	0

Pads spiked with increasing titers of WNV.

^aTiter of WNV in plaque forming units per milliliter added to each sugar pad.

^bCt = cycle threshold, generated by qRT-PCR; 10% sugar solution spiked at time T₀ with given quantities of WNV.

^cMean ± SD Ct scores at time T₄₈ when WNV/sucrose solutions were collected from sugar pads after a 48 h feeding period.

^dn = no. of mosquitoes tested.

^ePercentage of mosquitoes positive for WNV RNA by qRT-PCR.

Field Studies

During the summer of 2011, 27 of 400 sugar samples (6.8%) from Coachella Valley tested positive for WNV RNA by qRT-PCR (mean Ct = 34.6; range = 27.4–37.1). Of these, 18 (67%) were confirmed: 8 by using primers/probe from the NS1 region and 10 by re-extraction and retesting using the envelope gene primers/probe. Twenty-six of the 27 positive sugar samples were among 275 samples collected during Aug–Sep when WNV activity was detected by sugar wicks or concurrent surveillance measures (Fig. 3). Although positives were found at all five sites, the percent positive ranged from 2 to 18% (Table 3). By comparison, 3 (5%) of 70 mosquito pools comprising 2,056 female Cx. tarsalis collected from the same locations tested positive for WNV RNA (Table 3). In addition, during the sugar baiting period 10 chickens seroconverted at Adohr Duck Club (nine detected on 19 September and one on 3 October) and six at Gordon’s Ranch (one on 6 September, three on 19 September, and two on 3 October). None of the other sites had positive mosquito pools or seroconversions. Assuming a lag time of a week before detection, there was one concurrent sugar positive at Gordon’s Ranch and none at Adohr Duck Club. These data have been placed on a time-line to illustrate the sequence of detection and showed that WNV positive sugar baits were detected earlier than the other surveillance metrics (Fig. 3).

Fig. 3.

Chronology of WNV detection in Coachella Valley during the summer of 2011, where sugar bait is the percent of weekly sugar bait station samples WNV RNA positive, mosquito pools are the percent of groups of ≤50 host-seeking female Cx. tarsalis positive for WNV RNA, and sentinel chickens are the percent of sera from sentinel chickens positive for antibodies against WNV.

Table 3.

Number of weekly sugar bait samples testing positive by qRT-PCR compared with infection detected by testing mosquito pools by qRT-PCR and transmission determined by seroconversions in sentinel chickens deployed in the vicinity of bait stations in Coachella Valley, CA, during the virus activity period from Sept. to Oct. 2012

Site name	Sugar bait			Mosquito pools			Sentinel chickens
	Samples tested	WNV pos	%	Pools tested	WNV pos	%	Chickens tested	WNV aby pos	%
Adhor duck club	45	8	18%	19	1	5%	10	10	100%
North Shore State Park	55	7	13%	2	0	0%	10	0	0%
Gordons Ranch	65	6	9%	14	2	14%	10	6	60%
Jessup Ranch	55	4	7%	12	0	0%	10	0	0%
Cook St. sewer plant	55	1	2%	23	0	0%	10	0	0%
Total	275	26	9%	70	3	4%	50	16	32%

Sugar baits were less effective in Los Angeles and Kern counties. At Rowland Heights, 1 of 35 weekly sugar samples were positive, whereas one of three pools of Culex quinquefasciatus Say and three of seven sentinel chickens seroconverted (one on 3 October and two on 17 October). Assuming the 1 wk lag, the single positive sugar bait coincided with the later seroconversions. In contrast, none of 17 weekly samples from Encino were positive, whereas one of three pools were WNV RNA positive and five of seven chickens seroconverted 4 d after the sugar bait sampling ended, indicating that infective mosquitoes may have been in the vicinity while sugar baits were deployed. None of 30 weekly samples were positive from Kern County near Bakersfield, although 5 of 15 pools of Cx. tarsalis tested positive from one of the sites; the other two sites had 0 of 9 pools positive. Two of these sample sites had chicken flocks, although only the Amos site was within 100 m range of the sugar baits. There were four serconversions at Amos concurrent with the sugar baiting period.

Discussion

Laboratory experiments demonstrated that WNV RNA was expectorated and readily detectable by qRT-PCR on sugar wicks and papers offered to experimentally infected Cx. tarsalis mosquitoes, even if phenyl acetaldehyde was present in the solution. Overall, the sugar wicks were more often WNV RNA positive, contained more RNA, and retained moisture better than the filter papers, and therefore were selected for field deployment. Interestingly, all of the wicks and six of nine papers were WNV RNA positive as early as 6 dpi at 26°C. According to previously published temperature studies (Reisen et al. 2006), WNV transmission rates were <15% for experimentally infected females held for 6 dpi at 26°C. Fecal material was observed throughout the holding cartons and on the wicks, and therefore may have contributed to this early positivity. When we attempted to infect mosquitoes by feeding them on sugar pads spiked with increasing titers of WNV, only those mosquitoes fed the highest titer (10⁷ PFU of WNV/ml) became infected. In agreement, by 48 h infectious WNV was only detected on one of two sugar pads spiked with 10⁸ PFU WNV and the titer here after the 48 h holding period was reduced to 10^2.3 PFU WNV/ml. Considering the relative volumes involved, this titer greatly exceeds what is normally expectorated by Culex mosquitoes (Colton et al. 2005, Colton and Nasci 2006, Reisen et al. 2006), so deployed sugar bait stations would not likely serve as a source of mosquito infection. Previously, a small number of Cx. tarsalis were infected experimentally by feeding on a bloodmeal containing a low concentration of 10^2.3 PFU WNV/ml (Wheeler et al. 2012), but this was considered a rare event. Therefore, although WNV RNA was relatively stable on the sugar wicks, papers, and cotton pads, infectious virus was not. These data have important implications for experimental infection studies in that uninfected mosquitoes were not likely to become infected by sugar feeding on cotton pads concurrently with infected mosquitoes.

Mosquitoes frequently feed on sugars throughout their lifetime (Foster 1995) and are attracted to plants by a variety of odors and visual cues (Foster and Hancock 1994, Yuval 1992), perhaps providing an alternative to host-related attractants for sampling. However, our previous attempts using a variety of plants, fruits and inflorescence as well as sucrose solutions as trap attractants for Culex were generally unsuccessful, even though both males and females frequently feed on fructose (Reisen et al. 1986). In the current study, the addition of the floral attractant phenyl acetaldehyde (Jhumur et al. 2006, 2008) seemed to enhance the attraction of a 66% sucrose solution in desert habitats in the Coachella Valley, and we repeatedly were able to detect WNV RNA in sugar wicks exposed for a 5 d period. Our data and those of others (Hall–Mendelin et al. 2010, van den Hurk et al. 2007) clearly showed that infectious mosquitoes expectorate flaviviruses while sugar feeding and that WNV RNA is relatively stable on sugar wicks for over 5 d at ambient temperatures, agreeing with previous laboratory experiments (Johansen et al. 2002, Turell et al. 2002).

Our sugar wick bait stations worked well in the Coachella Valley, but not elsewhere. A single sample was positive in Los Angeles and none were positive when exposed for a brief period during active enzootic transmission in Kern County. Although desert habitats in Coachella Valley were hotter and drier than the other sites sampled, bait stations were deployed along Tamarix tree lines where mosquitoes have been observed to sugar feed at floral masses, so there was competition for attraction. Further studies are planned in Los Angeles and Kern counties to improve station deployment strategies and thereby improve sensitivity.

Advanced warning of viral amplification is important in surveillance programs informing decision support systems for intervention. In Coachella Valley, sugar bait samples detected WNV RNA 4–7 wk before samples from adjacent mosquito monitoring sites or sentinel chickens were positive. In addition, sugar bait station samples were positive from three sites where surveillance data were negative. This was difficult to understand, because females obviously must become infected and complete their extrinsic incubation period before being able to expectorate virus. Even at hot summer conditions in Coachella Valley, this takes ≈6 d, and therefore we should have been able to detect WNV infection in pooled mosquitoes collected host-seeking before sugar bait stations. However, dry ice-baited traps were deployed biweekly over a broader geographical area and possibly sample sizes were inadequate to detect the low level of infected females. If these females fed repeatedly at sugar bait stations that were available continually, it may be that the sugar bait system was more sensitive, because it sampled mosquitoes on a daily basis. In addition, early detection may have been enhanced by deposition of WNV-infected fecal material onto the sugar bait stations by infected but not yet infectious mosquitoes.

Unexpectedly, sugar samples were positive much earlier than sentinel chicken seroconversions. However, determining the exact infection date for the sentinel chickens was difficult. Although we initially offset our detection dates by 7 d to accommodate antibody rise after infection (Fig. 3), it may be that this was inadequate. Sentinel chickens take ≈7–10 dpi to become serologically positive by our EIA post infection (Patiris et al. 2008). This lag period combined with our biweekly sampling interval means that some chickens may have been infected as much as three wks earlier than depicted in Fig. 3 and therefore closer to the time of peak sugar bait station positivity. Future research will attempt to resolve this disparity in chronology.

In summary, sugar bait stations may present an attractive alternative to sentinels for monitoring arbovirus transmission. Because RNA seemed to be relatively stable for up to a week, multiple bait stations could be deployed thereby enhancing spatial coverage. Testing the wicks was done in 96 qRT-PCR well format allowing high throughput using the same reagents and methods used for testing mosquito pools. On-going research seeks an alternative attractant better suited to compete with native flora and improve detection in urban and agricultural landscapes.