Mosquito host-seeking activity was studied using a custom-designed trap to explore: (1) at which time interval of the night adult mosquito abatement would be most effective, and (2) if there exists an avian-specific host-seeking preference. Overnight trials using traps baited with dry ice showed that Aedes taeniorhynchus (Wiedemann) was most active at dusk and was then captured throughout the night. In contrast, Culex spp. (Cx. pipiens (Linnaeus) and Cx. restuans (Theobald) delayed most activity until about two h after dusk and were then captured through the night. This pattern suggests that management activities directed at adult Culex spp. would be most effective if initiated well after sunset. Mosquito capture rates in traps baited with birds in net bags were significantly greater than those with empty net bags, indicating that mosquitoes were attracted to the birds and not incidentally being sucked in by the custom trap's strong fan motor (Wilcoxon matched-pairs signed-ranks test, n= 24, t= 30, p < 0.05). Regression analysis showed that bird weight influenced mosquito attraction (r2= 0.21, p= 0.02). Trials with paired traps that contained different native bird species showed that Gray Catbirds, Dumatella carolinensis, attracted more mosquitoes than the heavier Northern Cardinals, Cardinalis cardinalis (paired samples t-test, t= 2.58, df= 7, p= 0.04). However, attractiveness did not differ substantially among bird species, and Gray Catbirds did not attract more mosquitoes than all other birds combined as a group. American Robins, Turdus migratorius (n = 4) were comparable in attractiveness to other bird species, but not enough American Robins were captured for a comprehensive study of mosquito avian preference.
Knowledge of mosquito feeding patterns on birds is essential to understanding West Nile virus (WNV) transmission patterns (Hayes et al. 2005). Patterns of host feeding by vector mosquitoes depend on such things as temporal and spatial patterns of host-seeking in relation to bird activity, and possibly on innate preferences of mosquitoes for different bird species. These factors should be considered in the implementation of a mosquito control program for surveillance and abatement purposes.
Mosquito host-seeking activity generally falls into one of four categories: nocturnal, crepuscular/nocturnal, crepuscular/diurnal, or diurnal (Kawada et al. 2005). Bird species that are implicated in WNV transmission tend to be active during daylight and to rest at night (Aschoff 1966, Gauthreaux 1996, McMillan et al. 1970), whereby nocturnal mosquitoes can find the relatively inactive bird hosts. Between May and December in Belek, Turkey, the sun sets between 18:00 and 20:00, and it appears to be the most active time for many species of mosquitoes, including Aedes cretinus (Edwards), Ae. caspius (Pallus), Ae. dorsalis (Meigen), and Culex pipiens (Linnaeus). It was noted that Cx. tritaeniorhynchus (Giles) was most active about two h after sunset (Caglar et al. 2003). In the late summer to early fall, the sun sets between 18:30–20:00 in Massachusetts and Culex spp. questing activity was also observed to be greatest around two h after sunset (Reddy et al. 2007).
Recent studies suggest that mosquitoes important for West Nile virus transmission preferentially feed on the American Robin, Turdus migratorius (Kilpatrick et al. 2006, Molaei et al. 2006a). This suggestion was based on blood meal analyses from mosquitoes captured primarily in traps set in suburban areas near human dwellings, where robins are described as highly abundant (Blair 1996). Therefore, trap placement may have biased the results. However, Kilpatrick's showing that 51% of Cx. pipiens blood meals came from the American Robin (that comprised only 4.5% of the avian community at each site) warrants further investigation on the role of avian-specific mosquito host-seeking activity.
Much of our knowledge of host feeding by mosquitoes that transmit WNV has been obtained by collecting mosquitoes and performing PCR and DNA sequencing of their blood meals (Apperson et al. 2002, Lee et al. 2002, Hassan et al. 2003, Kilpatrick et al. 2006, Molaei and Andreadis 2006, Molaei et al. 2006, Richards et al. 2006). While this provides insight on blood meals of mosquitoes in a given area, it is not clear whether this is a result of actual mosquito feeding preferences or another cause (e.g., bird behavior, trap placement bias, or different abundances of local available hosts). Animal-baited traps such as lard can and stable traps can help determine mosquito feeding preferences by quantifying approaches to different vertebrate species (Dow et al. 1957, Main et al. 1966, Service 1993, Balenghien et al. 2006, Darbro and Harrington 2006, Reddy et al. 2007). However, individual birds do not attract large numbers of mosquitoes, so the traps often need to be set for long periods of time, which can be stressful for the birds and possibly modify bird behavior, such as making them more defensive and affecting the success of a mosquito blood meal. In addition, lard can traps and the netting on some stable traps block the mosquito's view of the bird, which can interfere with cues that are important in host-seeking behavior (Kawada et al. 2005).
For this study, we developed a custom trap that allows access to visual, olfactory, and other host-seeking cues to mosquitoes approaching a potential host. We then assessed overnight host-seeking activity of mosquitoes to optimize timing of trap placement, due to a five-h hold limit on our captive birds. To test for mosquito avian-feeding preferences we set paired traps with different bird species to assess relative attractiveness of different native bird species to mosquitoes.
MATERIALS AND METHODS
The trap consisted of a broad funnel, a black chimney duct (∼20 by ∼15 cm) reducer sourced from a Lowes home improvement store, with a 120 mm high flow fan (114.7 CFM, Panaflo model #FBA12G) mounted inside the top of the ∼15 cm transition to draw mosquitoes into a collecting cup (sourced from our CDC light traps) attached to the ∼15cm bottom opening. The funnel was attached to a round, black plastic lid (also sourced from our CDC light traps; ∼30 cm in diameter) with wires. The bait was suspended between the lid and the funnel by way of an eye hook, which was screwed onto the bottom of the lid (Figure 1).
Mosquito diurnal/noctural activity
In hourly trials, four of the custom traps were suspended ∼1.5 m above the ground and placed ∼45 m apart along the forest edge at East Farm, University of Rhode Island. Each trap was baited with 2.5 kg of dry ice which was wrapped in newspaper and suspended above the trap using twine. Traps were randomly assigned to one of four treatments (one-h, two-h, three-h, and overnight) and each treatment was rotated among the four trap sites. Traps were set out between 18:00–19:00 and picked up after one h, two h, three h, or between 09:00–10:00 the next morning. Traps were set on nine evenings from 18–27 July 2006.
In night vs morning trials, four of the custom traps were set in the same locations as in the hourly trials. Each trap was also baited with 2.5 kg of dry ice. All four traps were set at 22:00 and picked up at 03:30 the next morning for the night collection. All four traps were set up again at 04:00 and picked up at about 09:00–10:00 for the morning collection. Traps were set for five evening-morning cycles from 2–10 August 2006. Mosquitoes were stored frozen for later identification using standard keys. Culex spp. were identified using morphological characters (Andreadis et al. 2005).
Attractiveness to birds
The effectiveness of the custom traps for collecting mosquitoes attracted to live hosts was tested by paired comparisons of the aforementioned traps baited with birds in net bags adjacent to traps with empty net bags. This test was conducted to make sure that mosquitoes collected actually approached the traps and were not suctioned by the force of the fan as they traveled by the trap. All bird trials were conducted at the North Forty area of Floyd Bennett Field, Gateway National Recreation Area (GATE, Jamaica Bay). Birds were collected in mist nets and banded the evening of each trap night. Birds recaptured within seven days were not reused in the captive bird experiments. Birds were handled according to the University of Rhode Island Institutional Animal Care and Use Committee (IACUC) guidelines (Protocol # AN05–02–018) and with permission from GATE services.
Each pair of traps, one bird-baited and one control, was suspended about 1.5 m from the ground and placed at locations at least 91 m apart from any other pair. The captive birds were suspended in net bags made from modified mosquito head nets (Coghlan's Ltd., Winnipeg, MB, Canada, model #8941), each of which was attached to an eye hook on the lid of the trap. Traps were set out between 19:00 and 21:00 and recovered approximately four h later, and the captive birds were immediately released. Trap cups were collected and immediately placed into a cooler with dry ice. Specimens were stored frozen for later identification.
The mosquito feeding preference study was conducted from 17 August through 11 September 2006 and from 26 June through 10 September 2007 (27 total trap nights for functional paired traps) in the same forested habitat as described. All birds used in this experiment were also captured during that evening's mist-netting effort. Two to three pairs of bird-baited traps were set each sample night. Each pair of traps was suspended ∼1.5 m from the ground and placed at one of three locations (trap sites were separated by ∼91 m). The paired traps were situated 1.5 m apart. Care was taken to use birds of similar weights (difference ranged from 0.1 g to 5.9 g). The birds were generally held in the traps for about four (not more than five) h. Traps were set by sunset, recovered by midnight, and the captive birds were immediately released. Trap cups were collected and immediately placed into a cooler with dry ice. Specimens were stored frozen for later identification.
Mosquito diurnal/nocturnal activity
A total of 465 mosquitoes was captured during the nine evenings of overnight trials, the most common species being Aedes taeniorhynchus (Wiedemann) (n= 217), Culex spp. (Cx. pipiens (Linnaeus) and Cx. restuans (Theobald), n= 82), and Ae. vexans (Meigen) (n= 53). Data were log (x+1) transformed to homogenize variance and normalize data. There was an overall significant difference in the number of mosquitoes captured per hour among the four trap periods (Three-Way ANOVA (date × trap location × time treatment), F= 4.27, p= 0.02). The one-h and overnight traps differed significantly in capture rates by Duncan's Multiple Range Test (p<0.05). Overall, Ae. taeniorhynchus did not show significant differences among time periods (F= 2.63, p= 0.08), but the number captured per hour was greater for the one-h samples than for the overnight samples by Duncan's Multiple Range Test (p<0.05) (Figure 2A). Culex spp. also did not show significant differences in the number captured per hour at different time periods (F= 2.85, p= 0.07). However, for Culex spp. the one-h treatment had significantly fewer mosquitoes captured per hour than the three-h treatment (p < 0.05) (Figure 2B).
Significantly more mosquitoes were captured at night (n= 259) than in the morning (n= 123) (paired samples t-test, t= 8.58, df= 4, p < 0.05). The most abundant mosquito species were Ae. taeniorhynchus (n= 152) and Culex salinarius (Coquillett) (n= 114). Both Ae. taeniorhynchus (t= 3.33, df= 4, p= 0.03) and Cx. salinarius (t= 3.05, df= 4, p= 0.04) were significantly more active during 22:00–03:30 than during 04:00–10:00.
Attractiveness to birds
Bird-baited traps caught significantly more mosquitoes (mean = 4.06 per night, SE = 0.87) than unbaited control traps (mean = 0.88, SE = 0.27) (Wilcoxon matched-pairs signed-ranks test, n= 24, t= 30, p < 0.05). In paired comparisons (Figure 3), Gray Catbird (Dumatella carolinensis)-baited traps caught significantly more mosquitoes than unbaited control traps (paired samples t-test, t= -3.71, df= 27, p < 0.05).
A total of 52 birds was used for the mosquito feeding preference study: Gray Catbird (26), Northern Cardinal, Cardinalis cardinalis (9), American Robin (4), American Redstart, Setophaga ruticillia (3), Eastern Towhee, Pipilo erythrophthalmus (2), Mourning Dove, Zenaida macroura (1), House Wren, Troglodytes aedon (1), Northern Waterthrush, Seiurus noveboracensis (1), Red-eyed Vireo, Vireo olivaceus (1), Veery, Catharus fuscescens (1), Brown-headed Cowbird, Molothrus ater (1), Cedar Waxwing, Bombycilla cedrorum (1), and Eastern Kingbird, Tyrannus tyrannus (1). Mosquitoes attracted to Gray Catbirds, Northern Cardinals, and American Robins in paired comparisons are shown in Figure 4.
In comparisons between paired traps with Gray Catbirds and Northern Cardinals, significantly more mosquitoes were captured in Gray Catbird-baited traps (mean = 0.50 per hour, SE = 0.20) than in Northern Cardinal-baited traps (mean = 0.16 per hour, SE = 0.08; t= 2.58, df= 7, p= 0.04), even though Northern Cardinals were significantly heavier (mean = 40.6 g) than the Gray Catbirds (mean= 37.6 g) (t= 2.40, df= 7, p= 0.03). Bird weight apparently influenced attractiveness overall because a regression of the ratios of the number of mosquitoes (transformed by log (x+2)) attracted to paired birds (all pairs) on the ratios of the log weights of the paired birds (Figure 5), resulted in a positive relationship (r2= 0.21, p= 0.02). Combining all trials in which Gray Catbirds were paired with all other bird species, an average of 0.34 mosquitoes per hour (SE = 0.09) were captured in Gray Catbird-baited traps, and an average of 0.20 mosquitoes per hour (SE = 0.07) were attracted to traps baited with all other birds (t= 1.45, df= 21, p= 0.16). The two most commonly collected mosquito species in these bird-bird comparisons trials were Ae. sollicitans (Walker) (n= 24) and Cx. pipiens (n= 17). Even though mosquitoes in general were significantly attracted to the Gray Catbird-baited traps, individual mosquito species collected (Ae. taeniorhynchus, Ae. sollicitans, Aedes spp., Cx. pipiens, and Culex spp.) did not significantly prefer Gray Catbirds over Northern Cardinals (Fisher Exact Probability Test, n= 8, p= 0.49).
Our results from the mosquito diurnal/nocturnal activity trials conducted in a forest-edged/rural habitat conform to previously published results that host-seeking activity of Aedes species such as Ae. taeniorhynchus is substantial at dusk. However, Culex spp. were not active until well after dusk, as reported by Reddy et al. (2007) in a suburban/residential study and Anderson et al. (2007) in a wooded study site, who observed Culex spp. activity to be the greatest two h after sunset. This pattern suggests that management activities directed at adult Culex spp. would be most effective if initiated well after sunset. Our results indicate that the risk of encounter with mosquitoes likely to transmit WNV is greatest at dusk and in the first half of the night. In the night vs morning study, significantly more mosquitoes were captured at night (22:00–03:30) than in the morning (04:00–10:00). Perhaps the cooler temperatures in the morning resulted in lower mosquito activity.
Dry ice-baited CDC traps are effective at capturing many mosquitoes partly because of the high volume of CO2 expelled when dry ice sublimates and partly because of the length of time the trap is set. The bird-baited traps reported here are limited by the volume of CO2 expelled by a single bird and by a short four to five h trap time (to avoid negative effects on the bird from long-term confinement). All birds appeared in good health when released from these traps, indicating no apparent negative effects on the birds. The Ae. aegypti (Linnaeus) mosquito has been observed to have dichromatic vision which provides it with good contrast sensitivity (Allan 1994), allowing the mosquito to see the host (unobstructed) and use other cues to finalize the seeking process. Therefore, an advantage of our trap design is that mosquitoes are not only attracted to the CO2 expelled by the birds but also to other bird odor and visual cues.
The size of the host has been suggested as an important factor in mosquito attraction (Dow et al. 1957), since larger hosts would expel more CO2 and other kairomones than smaller individuals. This was not the case for Gray Catbirds, which attracted more mosquitoes than Northern Cardinals, even though Gray Catbirds were significantly lighter. The attractiveness of a bird species might include more than olfactory cues. For example, probing and biting activity might depend on textural factors, and differences in defensive behaviors among bird species might influence mosquito feeding success. However, an overall analysis of all bird pairs in our study showed a positive relationship between weight and attractiveness (Figure 5). These results suggest that mosquitoes are attracted to heavier birds, but individual preferences also exist that are unrelated to bird weight as shown by the Gray Catbird-Northern Cardinal-baited trap trials. This pattern contrasts with the results of Darbro and Harrington (2006), who found no differences in attractiveness between chickens and house sparrows. The difference may be that we used native wild birds in this study, instead of domestic or introduced species, so that supposed evolved feeding preferences might account for the different results. However, the lack of a difference in attractiveness between Gray Catbirds and all other captured bird species combined implies that although there were some individual differences among species, most birds were similarly attractive to host-seeking mosquitoes. These results suggest that the strong preponderance of robin blood in engorged mosquitoes reported by Kilpatrick et al. (2006) and Molaei et al. (2006a) may result from ecological, spatial, and avian behavioral factors that influenced successful blood meals rather than from feeding preferences by mosquitoes. This interpretation is supported by recent results suggesting important roles of blue jays and house sparrows, in addition to robins, in Cx. pipiens infection with WNV in Chicago (Hamer et al. 2009). In hindsight, the paired bird-baited traps should have been situated in the canopy as this is more likely where birds roost. Doing so may have yielded more ornithophilic, WNV-implicated mosquito species.
Lastly, our studies on the attractiveness to birds were conducted in a forested habitat in Brooklyn, NY, which might have a different distribution of WNV-implicated bird species than in other published studies. The most abundant passerine bird species captured during mist-netting were Gray Catbirds (42.9%), American Robins (7.2%), and Traill's Flycatchers (5.3%). The most abundant mosquito species captured during CDC light trap collections were Ae. vexans (38.6%), Culex spp. (Cx. pipiens and Cx. restuans; 28.9%), and Ae. sollicitans (19.1%).
We had difficulty capturing sufficient American Robins to conduct a comprehensive study of mosquito host preference. The large preponderance of robin blood reported by Kilpatrick et al. (2006) and Molaei et al. (2006a) in mosquito blood meals warrants further research to determine whether a feeding bias actually exists, and if so, its mechanism in nature.
We thank the staff at Gateway National Recreation Area, notably K. Tripp and J. Zuzworski, for logistical support. C. Hahn and J. Day provided constructive comments on early drafts of the manuscript. Use of trade or product names does not imply endorsement by the U.S. Government. This work was funded by the Natural Resource Preservation and Protection (NRPP) program, U.S. National Park Service.