|Home | About | Journals | Submit | Contact Us | Français|
The selection of vertebrate hosts by Culex mosquitoes relative to West Nile virus (WNV) transmission in neotropical countries such as Guatemala is not described. This study determined the feeding patterns of Cx. quinquefasciatus and Cx. nigripalpus and estimated the relative contribution of two common and frequently infected wild bird species, Turdus grayi and Quiscalus mexicanus, to WNV transmission. Engorged mosquitoes were collected from rural and urban habitats after the dry and wet seasons in the Department of Izabal in 2007. Host selection by Cx. nigripalpus varied significantly between urban and rural habitats. Both Cx. quinquefasciatus and Cx. nigripalpus fed predominantly on chickens and other domestic animals. Blood meals from wild birds were rare, accounting for 1.1% of blood meals identified from Cx. quinquefasciatus and 6.5% of blood meals from Cx. nigripalpus. Transmission of WNV by these two mosquito species may be dampened by extensive feeding on reservoir-incompetent hosts.
Serologic evidence of West Nile virus (WNV) was first described in Guatemala in 2003 with the confirmation of WNV-seropositive horses in several departments throughout the country.1 Intensive ecological studies of WNV were initiated within a 80-km2 transmission focus that incorporated the city of Puerto Barrios and the rural village of Machacas del Mar in the Department of Izabal. The clay-colored thrush (Turdus grayi), great-tailed grackle (Quiscalus mexicanus), and domestic chicken (Gallus gallus) have been identified as common and frequently infected resident avian hosts within the transmission focus.2 In 2007 and 2008, T. grayi and Q. mexicanus were among the more abundant wild bird species with the highest WNV seroprevalence rates in the transmission focus.2 However, the relative importance of these and other tropical vertebrate species to enzootic transmission of WNV is unknown. In Puerto Barrios, WNV infections have been detected in Culex (Culex) quinquefasciatus Say, and Cx. (Cx.) mollis/Cx. (Cx.) inflictus.2 WNV has also been detected in Cx. (Cx.) interrogator Dyar and Knab and Cx. (Cx.) nigripalpus Theobald in nearby Chiapas, Mexico.3 WNV isolates were obtained from Cx. nigripalpus, Cx. (Cx.) bahamensis Dyar and Knab, Cx. quinquefasciatus, and Cx. (Cx.) habilitator Dyar and Knab during an outbreak in Puerto Rico in 2007.4 However, no study has yet linked the blood-feeding behavior of these potential WNV vectors to virus transmission among vertebrate hosts in Mesoamerica.
The extraordinary biodiversity of the American tropics presents an enormously complex and challenging ecosystem for understanding arbovirus transmission cycles. Currently, there are 18 recognized mosquito species in the subgenera Culex (Culex) and Culex (Phenacomyia) in Guatemala5 that could potentially serve as vectors of WNV. Vertebrate host selection of these species is poorly understood, and the interactions between these mosquitoes and the myriad of potential WNV-amplifying hosts could vary in response to host availability across habitats and seasons. Comprehensive DNA barcode coverage is available for North American birds, including neotropical migrant species that winter in Guatemala, providing an excellent DNA reference database with which to match mosquito blood-meal sources.6 However, taxonomic coverage of resident wild birds in Guatemala and other tropical vertebrate species likely fed on by mosquitoes is much less complete for both the DNA barcoding gene (mitochondrial cytochrome c oxidase I) as well as mitochondrial cytochrome b.
Understanding how WNV circulates among mosquito vectors and amplifying hosts in tropical ecosystems is a priority for understanding the ecology of WNV and risk of human disease in this part of the world and implementing targeted surveillance and control measures appropriate for this ecological setting. Therefore, the specific aims of this study were to (1) determine the host range of candidate vector mosquitoes in urban and rural habitats and their blood-feeding patterns on different vertebrate hosts during the dry (July) and wet (December) seasons of 2007 and (2) estimate the relative contribution of potential key avian host species to WNV transmission in urban and rural habitats.
This study was conducted in a previously identified WNV transmission focus comprising the city of Puerto Barrios and the rural town of Machacas del Mar, located in the Department of Izabal along the Caribbean coast of Guatemala (15°50′ N, 88°28′ W) (Figure 1). A complete description of the transmission focus is published elsewhere.2 Within the WNV transmission focus, 10 1-km2 quadrants were selected, each containing a sentinel house at which WNV surveillance activities, including engorged mosquito collections, were conducted. Four of these quadrants were classified as rural, and six sites were urban because of the prevalence of urban facilities (houses, roads, bridges, etc.) and vegetation. Urban quadrants were defined as quadrants having > 30% road and human dwelling infrastructure. In addition to the sentinel houses, six rural microhabitats were also identified for mosquito collections: road, secondary forest, pasture, shaded pasture, river, and hedge.
Mosquitoes were aspirated from either the vegetation or 9-in diameter circular plastic flower pots placed horizontally on the ground using a Modified Centers for Disease Control and Prevention (CDC) Backpack Aspirator Model 1412 (John W. Hock Company, Gainesville, FL). For the pot collections, 24 pots were placed in the yard surrounding each of four rural and six urban sentinel houses, and 50 pots were placed in each of the six rural microhabitats. Mosquitoes were aspirated from pots during 1 week each month (3–5 collection days per site per month) between July 1–6, 2007 and November 29 to December 7, 2007. Pots were left in place for the entire collection period, and missing pots were replaced as necessary. After aspiration of mosquitoes from pots, approximately 10 minutes were spent at each site aspirating resting mosquitoes from the surrounding vegetation. Collections from pots and vegetation were kept separate. Aspiration cups of mosquitoes were stored on dry ice until sorting. Female mosquitoes were morphologically identified7; male mosquitoes were discarded. The identity of a subset of known species (Cx. quinquefasciatus, Cx. (Cx.) lactator, Cx. interrogator, Cx. nigripalpus, Cx. (Cx.) chidesteri, and Cx. (Cx.) coronator) and engorged specimens that could only be classified as Cx. spp. were confirmed by polymerase chain reaction (PCR).8 Mosquitoes were classified as unfed, fully engorged, half-gravid, or gravid, and they were stored individually at ambient temperature in 0.5-mL tubes containing silica gel desiccant and cotton sorted by date and location.9 Additional engorged Cx. quinquefasciatus collected in 2007 were obtained from CO2-baited CDC light traps10 (John W. Hock Company, Gainesville, FL) and gravid traps11 from the same sites within the municipality of Puerto Barrios. Cx. quinquefasciatus blood-meal identification data from all collection methods were combined after it was determined that feeding patterns were not significant by trap type.12
To generate density and relative abundance data for wild birds and chickens, avian point counts were performed.2,13 Point counts were conducted along a single transect in each of the 10 1-km2 quadrants. For each quadrant, four 4-minute counts of bird species seen or heard within 50 m were conducted within the first 3 hours of daylight. Mosquito sampling sites were located within 50 m of each transect. Bird surveys were conducted during the same weeks as mosquito collections. Domestic animals, chickens, and humans were counted at each sentinel house while mosquitoes were being collected.
Abdomens of engorged mosquitoes in CDC light trap and gravid trap collections were removed over clean microscope slides using clean insect pins and forceps and placed into separate labeled tubes on ice. Engorged abdomens were manually triturated in phosphate buffered saline. DNA was extracted from these homogenates using the QIAamp DNA mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. When the QIAamp DNA kit was not available, DNA was extracted with DNAzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Pellets from DNAzol extractions were incubated overnight at 4°C to maximize pellet rehydration. All DNA samples were stored at −20°C until tested by PCR assays.
Blood meals from specimens collected by CDC light traps and gravid traps were initially screened in Guatemala for common domestic hosts: cow (Bos taurus), human (Homo sapiens), pig (Sus scrofa), goat (Capra hircus), dog (Canis familiaris), and chicken (G. gallus) were known to be present at the study sites.9,14,15 Amplification of chicken DNA was carried out in a separate tube using specific primers UNFOR1029 and Chick1123R under the same reaction conditions.9 DNA extracted from whole blood of each vertebrate species tested was used as positive control. For blood meals not identified by multiplexed PCR, vertebrate cytochrome b (cytb) amplicons generated by either avian16 or mammalian17 primers and/or vertebrate mitochondrial cytochrome c oxidase 1 (COI) barcoding primers18 were sent to the CDC in Fort Collins, CO for sequencing. Blood meals from aspirated specimens were first screened using chicken primers as above. Blood meals not containing chicken DNA were identified by PCR amplification and sequencing of a fragment of either the COI or cytb gene.16–19 DNA amplifications and sequencing reactions were completed on a DNA Engine PTC-200 thermal cycler (Bio-Rad Laboratories, Inc., Hercules, CA) and analyzed on an ABI 3130 genetic analyzer. Sequences were assembled using Lasergene 9 software (DNA STAR, Inc, Madison, WI). Two databases were used to identify mosquito blood-meal sequences. The DNA Barcode database was used to identify COI sequences (www.barcodinglife.org) and GenBank was searched to identify cytb sequences. Greater than 98% sequence homology was considered a positive match. To aid in blood-meal identification, novel COI and cytb16 reference sequences were generated for 40 species of resident birds in Guatemala (GenBank accession nos. EU442290–EU442363).
The blood-meal host selection of Cx. quinquefasciatus and Cx. nigripalpus was analyzed by χ2 to test the null hypothesis that there was no difference in mosquito host selection across habitats or seasons. Preferential host selection favoring one host over another was determined by calculation of the feeding index.20 The feeding index (FI) is expressed as FI = (Ne/Ne′)/(Ef/Ef′), where Ne is the number of feeds on host I, Ne′ is the number of feeds on host II, and Ef/Ef′ is the expected proportion of feeds on host I/host II based on their relative abundance. A feeding index of 1.0 indicates equivalent host selection for the two hosts being compared. Values less than and greater than 1.0 indicate lesser and greater host selection, respectively, in feeding on the first host relative the second host. For estimation of the term Ef/Ef′, the relative abundance of dogs, humans, cattle, chickens, and wild birds was directly interpreted as the expected proportion of feeds on each species. Raw counts of the domestic animals and chickens and total numbers of wild birds seen or heard along urban or rural point count transects during a given sampling period were, therefore, used in calculation of the feeding indices using the methods in the work by Kay and others.20 All feeding index calculations were set compared with chickens, where chickens were host I.
The relative number of WNV-infectious mosquitoes resulting from feeding on T. grayi and Q. mexicanus was calculated as described previously.14,21 For Cx. quinquefasciatus and Cx. nigripalpus feeding on T. grayi and Q. mexicanus, the relative number of infectious mosquitoes (F) resulting from feeding on each avian species (i) was estimated by multiplying reservoir competence, C, by the square of the proportion of blood meals (b) from species i each month: Fi = Ci × bi2.14 Avian reservoir competence index values used for T. grayi (0.1), Q. mexicanus (1.8), and G. gallus (0.0) were derived from experimental infection with the Tabasco strain of WNV.22 The relative number of infectious mosquitoes derived from feeding on Q. mexicanus compared with T. grayi was determined by dividing FQuiscalus by FTurdus.
In total, 686 blood meals from Culex mosquitoes were identified during 2007, including blood meals from several potential vectors of WNV in the subgenus Culex (Culex) (Table 1). Of these blood meals, 486 blood meals were from the aspiration collections in July and November/December of 2007, and 200 blood meals were from CDC light trap and gravid trap collections performed throughout 2007. Regardless of collection method, most blood meals were from domestic animals. Domestic birds fed on in addition to chicken (G. gallus) were turkey (Meleagris gallopavo), Muscovy duck (Cairina moschata), and one white-winged dove (Zenaida asiatica). Mammal blood meals included human (H. sapiens), cattle (Bos spp.), horse (Equus caballus), dog (C. familaris), pig (S. scrofa), cat (Felis catus), black rat (Rattus rattus), Norway rat (R. norvegicus), and gray four-eyed opossum (Philander opossum).
Only 17 of 686 (2.5%) blood meals identified from Culex mosquitoes were obtained from wild birds. Blood meals from Cx. nigripalpus included one Strix spp. owl, one T. grayi, one Q. mexicanus, one bare-throated tiger heron (Tigrisoma mexicanum), one Northern oriole (Icterus galbula), one white-eyed vireo (Vireo griseus), one yellow-crowned night heron (Nyctanassa violacea), and one unidentified raptor in the family Accipitridae. One T. grayi and three Q. mexicanus blood meals were identified from Cx. quinquefasciatus. One spot-breasted oriole (I. pectoralis), one unidentified passerine in the family Thamnophilidae (antshrike spp.), and one T. grayi were identified from Cx. spp. along with one red-winged blackbird (Agelaius phoeniceus) from Cx. lactator Dyar and Knab and one T. grayi from Cx. interrogator. Of these species, I. galbula and V. griseus are neotropical migrant species, whereas the remainder of species are resident in Mesoamerica all year. Eleven blood meals were obtained from reptile and amphibian species. These blood meals included green iguana (Iguana iguana) and a Gambelia spp. lizard from Cx. nigripalpus, I. iguana from Cx. taeniopus, and an unidentified skink, unidentified amphibian, unidentified reptile, and three frogs in the family Ranidae from Cx. spp. Most of the blood meals obtained from Cx. spp. were from Cx. mollis Dyar and Knab, Cx. declarator Dyar and Knab, or Cx. inflictus Theobald. Adult females of these species are difficult to differentiate morphologically, and these three species are not included in the multiplexed PCR assay for Culex identification.8
The two mosquito species from which the largest numbers of blood meals were obtained were Cx. nigripalpus and Cx. quinquefasciatus. Cx. nigripalpus was collected from both rural and urban habitats, and it had a relatively opportunistic blood-feeding behavior (Table 1). Overall, 36.5% of Cx. nigripalpus blood meals came from mammals, 61.5% came from birds, and 1.9% came from reptiles. The host selection of Cx. nigripalpus on livestock, humans, dogs, domestic birds, wild birds, and reptiles differed significantly by habitat (χ2 = 18.7, degrees of freedom (df) = 5, P < 0.0025, α = 0.004) (Figure 2). No seasonal shift in Cx. nigripalpus blood-feeding among humans, dogs, domestic birds, and wild birds was observed in urban areas (χ2 = 6.8, df = 3, P > 0.05, α = 0.006) or on livestock, humans, dogs, domestic birds, wild birds, and reptiles in rural areas (χ2 = 7.9, df = 5, P > 0.15, α = 0.004). Overall, 52% of identified Cx. nigripalpus blood meals came from chickens (Table 1).
In contrast, Cx. quinquefasciatus was very common in urban areas but more rarely collected in rural areas. Within urban areas, Cx. quinquefasciatus blood meals were predominately from chickens (85%) (Table 1). Cx. quinquefasciatus blood-meal composition did fluctuate significantly in some months (χ2 = 124.5, df = 32, P < 0.001, α = 0.001) (Figure 3). However, because of the low number of observations (e.g., only four observations from January to April), temporal fluctuations should be studied more thoroughly. The most common mammalian hosts fed on by Cx. quinquefasciatus were dogs (N = 27) (Table 1). Other mammalian blood meals included human (N = 1), cattle (N = 2), cat (N = 4), and rat (N = 1).
Feeding indices were calculated from the numbers of blood meals from host I compared with host II and the raw counts of these hosts at the collection locations as determined from avian point counts and domestic animal censuses in July and December of 2007 (Table 2). Feeding indices indicated that Cx. nigripalpus selected both cattle and dogs over chickens when these hosts were available. Chickens were used over both humans and wild birds in urban and rural habitats (Table 2). Feeding index analysis also suggested that Cx. quinquefasciatus selected chickens over humans, dogs, cattle, and wild birds (Table 2).
The most common wild bird species fed on collectively by Culex mosquitoes were T. grayi and Q. mexicanus. Of the 17 total Culex blood meals from wild birds, 4 each (23.5%) were from T. grayi and Q. mexicanus (Table 1). We estimated that, for every 1 WNV-infectious Cx. quinquefasciatus derived from feeding on T. grayi, 162 WNV-infectious Cx. quinquefasciatus were derived from feeding on Q. mexicanus, and for every 1 WNV-infectious Cx. nigripalpus derived from feeding on T. grayi, we estimated that 18 WNV-infectious Cx. nigripalpus were derived from feeding on Q. mexicanus. However, only 0.003 (0.3%) and 0.008 (0.8%) of Cx. quinquefasciatus blood meals originated from T. grayi and Q. mexicanus, respectively. The proportion of Cx. nigripalpus blood meals from each T. grayi and Q. mexicanus was 0.009 (0.9%). A high proportion of blood meals from each of these mosquito species came from chickens. Chickens are not known to be competent amplifying hosts for WNV,22 and estimates for the relative number of infectious mosquitoes derived from feeding on them were, therefore, zero.
This study characterized the seasonal vertebrate host selection of Cx. quinquefaciatus and Cx. nigripalpus in urban and rural habitats in Guatemala during 2007, and it provided blood-meal identification data from numerous additional mosquito species. Both Cx. nigripalpus and Cx. quinquefasciatus are widely reported to have opportunistic host selection throughout the Americas.23–26 The host selection of Cx. nigripalpus in Florida was described as opportunistic, with the majority of blood meals taken from cattle and rabbits, but also, many were from birds, including wading birds.23 Along the southeastern Pacific coast of Guatemala, Cx. nigripalpus fed mostly on birds (61–80% across several years) and occasionally, mammals and reptiles.24 Cx. quinquefasciatus is also known to use a wide variety of avian, mammalian, and to a minor extent, reptilian hosts. Whether the mosquitoes were predominantly ornithophilic27–30 or mammalophilic26,31,32 varied across habitats, geographic locations, and whether collections were performed indoors and/or outdoors.
As opportunists, the host selection behaviors of Cx. quinquefasciatus and Cx. nigripalpus are, in part, determined by the availability of domestic and free-ranging hosts at different collection sites.23,31 In our study, avian blood meals from free-ranging species were rarely detected for both Cx. nigripalpus and Cx. quinquefasciatus. Less than 1% of blood meals from these candidate WNV vectors were derived from T. grayi and Q. mexicanus, providing little support that these two wild bird species are significant sources of WNV-infectious mosquitoes. Still, WNV seroprevalence was approximately 11% in T. grayi (N = 171) in 2007 and 32% in Q. mexicanus (N = 67), and WNV infection rates in Cx. quinquefasciatus were 5.7 per 1,000 in July of 2007 and 15.7 per 1,000 in August of 2007, indicating a relatively high amount of virus activity in these vertebrate and vector species.2
In this study, the majority of blood meals from wild birds was clustered in May and June of 2007 (Figure 3). It is possible that this spike in mosquito blood-feeding activity on wild birds is the result of mosquitoes capitalizing on the presence of defenseless nestlings and brooding adult birds during the nesting season. Investigations into the importance of nestling birds to WNV transmission have resulted in disparate conclusions. WNV infection/seroprevalence rates in nestling birds were very low in several studies,33–35 which concluded that nestlings were not important early-warning sentinel species or amplification hosts for WNV. Mosquito landing rates were also found to be higher on adult birds than on nestlings, and parental brooding reduced landing rates of mosquitoes on nestlings.36 In contrast, mosquito feeding on specific vertebrate hosts in Alabama peaked during the reproductive periods for those hosts.37 The lack of evidence for WNV infection in nestlings during the nesting season may be the result of sampling early in the season, when mosquito and virus activity is relatively low.38 In studies that have highlighted a role for nestling or juvenile birds in WNV amplification, there seems to be strong spatial and seasonal components in the focality of WNV transmission among juvenile birds.14,38 The seasonality of mosquito blood-feeding on wild birds in Guatemala may have contributed to the peak in mid-summer WNV amplification observed in 2007 in the work by Morales-Betoulle and others,2 and it warrants additional study.
The majority of blood meals for Cx. quinquefasciatus and Cx. nigripalpus in both urban and rural habitats originated from domestic animals, particularly chickens. Chickens have been previously reported to be heavily used hosts of both Cx. quinquefasciatus23,27,29,39 and Cx. nigripalpus,23 and this finding likely reflects host composition where mosquito sampling efforts were concentrated. Chickens were very prevalent in both urban and rural peridomestic study sites. Mosquito collections close to the ground in these habitats naturally would yield a large number of blood meals from poultry and other domestic animals from mosquitoes foraging at ground level. Although adult chickens are not considered competent amplifying hosts and may dampen WNV transmission, we recognize that very young chicks (< 1 week old) may be competent amplifying hosts for WNV.40 Future research should include blood-meal identification of mosquitoes in locations where domestic animals are not present to further elucidate which mosquito species are feeding on and infecting wild birds.
Mosquito collections from urban sites with substantial human activity produced surprisingly few human-derived blood meals from Cx. quinquefasciatus. In Mexico, humans were the second most common blood source of Cx. quinquefasciatus, comprising 9.4% of the blood meals from outdoor collections in one site.29 In Brazil, Cx. quinquefasciatus preferred humans to chickens; however, these mosquitoes were collected indoors as well as outdoors.26 Cx. quinquefasciatus in Southern California fed predominantly on doves and passerine birds but took approximately 2% of blood meals from humans, showing the capacity of this species to serve as both an enzootic and bridge vector of WNV.41 Housing in the sampling sites in Guatemala was relatively open, providing mosquitoes free movement in and out of homes. Therefore, the intensive collections performed around houses, yards, and porches using four different collection methods conceivably should have produced human-fed Culex. Alternatively, this finding may reflect a highly ornithophlic Cx. quinquefasciatus population, with chickens and other domestic animals diverting host-seeking Cx. quinquefasciatus away from humans. Zooprophylaxis through the presence of chickens and other domestic animals in urban areas is one hypothesis to explain the lack of reports of human WNV cases in Puerto Barrios. Additional investigation, including paired indoor–outdoor mosquito collections, is needed to further evaluate the anthropophily of Cx. quinquefasciatus in our study sites.
Interestingly, species of Culex (Culex) and Culex (Melanoconion), including Cx. (Culex) nigripalpus and Cx. (Mel.) taeniopus, fed on various species of frogs, skinks, and lizards. These species as well as Cx. quinquefasciatus have been previously shown to feed occasionally on reptiles and amphibians.23–25,28 The role of these vertebrates as potential arbovirus reservoirs in the tropics has not yet been determined. The green iguana (I. iguana) and the North American bullfrog (Rana catesbeiana) have been shown to develop only low-titered WNV viremia after experimental infection (< 4.0 log pfu/mL serum).42 However, lake frogs (R. ridibunda) in Russia43 and juvenile American alligators (Alligator mississippiensis)44 are known to develop high-titered viremias infectious to feeding mosquitoes. Both Cx. quinquefasciatus and Cx. nigripalpus were found to have fed on captive alligators in Louisiana.45 A high rate of feeding on reptiles and amphibians was reported for Cx. quinquefasciatus in Puerto Rico during a period of elevated WNV transmission.46
In North America, seasonal shifts in blood-feeding behavior have been well-documented for several species of Culex (Culex), including Cx. nigripalpus.14,23,47–49 Such shifts can be epidemiologically significant if fluctuations in particular vector–host contacts result in increased enzootic or epizootic virus transmission.23,49,50 In Florida, this shift was attributed to seasonal patterns of relative humidity and rainfall that regulate Cx. nigripalpus movements between resting refugia and open fields, where they encounter primarily mammalian hosts.23 Although Cx. nigripalpus blood meals were analyzed at the end of dry and wet seasons during this study, no significant seasonal shift in blood-feeding behavior was seen for this species in either urban or rural area. More frequent mosquito collections and correlation of blood meals with individual rainfall events may be necessary to detect such a pattern. Similarly, no observable shift in blood-feeding behavior was observed for Cx. quinquefasciatus, although its blood-feeding pattern did differ significantly in some months. This variation could be explained by clusters of blood meals from particular vertebrate species in certain months (e.g., Muscovy ducks clustered around a gravid trap one night in June). Seasonal shifts in host selection have been reported previously for Cx. quinquefasciatus.31,32,51
Another specific aim sought to estimate the relative contribution of different potential avian amplifying hosts to WNV transmission in Puerto Barrios during 2007. The relative abundance of wild birds used in feeding index calculations was determined from the point count data. Although point counts are an accepted, standardized methodology for performing avian surveys, this method is subject to a number of limitations. For example, the detectability of different bird species by sight and sound, weather conditions, skill level of the observer, distance from the observation point, and time of day that the survey was conducted all influence the recorded abundance of each bird species. Although we designed and conducted our surveys to control for these limitations and performed engorged mosquito collections in the same locations as the bird surveys were conducted, the relative abundance of the wild birds compared with the other vertebrate species available to mosquitoes at each site, particularly at the times that mosquitoes were actively feeding, cannot be precisely determined. Fortunately, Q. mexicanus and T. grayi, the two potential WNV-amplifying hosts primarily discussed in this study, were very abundant and easily observable at our study sites.2 Both Cx. nigripalpus and Cx. quinquefasciatus, as well as other species of Culex (Cx. spp. and Cx. interrogator) fed on T. grayi and Q. mexicanus, although apparently at very low rates. Given the high WNV seroprevalence in Q. mexicanus in 2007, future mosquito collection efforts should be focused around communal roosts of Q. mexicanus. Q. mexicanus forms large aggregations in the evening at several locations throughout Puerto Barrios, and it is at these roosting locations where they may be susceptible to mosquito attack rather than at the residential homes where engorged mosquitoes were collected. Communally roosting passerines have previously been hypothesized to be a rich source of blood meals for host-seeking Culex mosquitoes as well as a focus of WNV amplification.14 More work is needed to determine mosquito feeding behavior at communal bird roosts and the contribution of Q. mexicanus communal roosts to enzootic WNV amplification.
In conclusion, this study characterized the blood-feeding patterns of Cx. quinquefasciatus and Cx. nigripalpus in urban and rural habitats in a WNV transmission focus in Guatemala. Cx. quinquefasciatus was one of the most common mosquito species collected in urban areas by aspirations, and one from which several WNV isolates have been made within the transmission focus.2 This species fed predominantly on chickens, with few blood meals also identified from humans and wild birds, including T. grayi and Q. mexicanus. With a highly ornithophilic Cx. quinquefasciatus population taking approximately 85% of blood meals from chickens, the effect that these domestic birds have on WNV transmission and the potential role of Cx. quinquefasciatus also serving as a bridge vector responsible for transmission of WNV to humans in Puerto Barrios need to be further evaluated. Because of its prevalence in urban and rural areas, previous virus isolations, and propensity to feed on both WNV-amplifying and dead-end hosts, Cx. nigripalpus has the potential to serve as both an enzootic and bridge vector of WNV in Guatemala. This species could be involved in circulating WNV among competent wild bird species, including T. grayi and Q. mexicanus, as well as bridging WNV to chickens, livestock, and humans in both urban and rural settings in Guatemala. However, the high proportion of mosquito blood meals from reservoir-incompetent hosts could function to dampen WNV transmission in Puerto Barrios. Much more work is needed to elucidate vertebrate host selection for the other numerous Culex species and the role that those species play in transmission of WNV in this tropical environment.
For assistance with mosquito collections, the authors thank Alfonso Salam, Bernarda Molina, Danilo A. Alvarez Castillo, Silvia M. Sosa Echeverria, Maria L. Muller Theissen, and Maria de Lourdes Monzon. Assistance with DNA extractions was provided by Claudia Paiz. Mosquitoes collected by Centers for Disease Control and Prevention light trap and gravid trap were identified by Silvia Sosa and Maria Luisa Muller Theissen. Bird counts were conducted by Jean-Luc Betoulle and Cristina Chaluleu. House and land owners are acknowledged for the permission to work on their properties. Sequencing was performed by Rich Tsuchiya in the Centers for Disease Control and Prevention Sequencing Core Facility.
Financial support: This research was funded by the Centers for Disease Control and Prevention, Cooperative Agreement U50/CCU021236-01 between the Centers for Disease Control and Prevention and the Universidad de Valle del Guatemala, and a Robert E. Shope International Fellowship in Infectious Diseases award.
Authors' addresses: Rebekah C. Kading and Nicholas Komar, Centers for Disease Control and Prevention, Division of Vector-Borne and Infectious Diseases, Arbovirus Diseases Branch, Fort Collins, CO, E-mails: vog.cdc@7kxf and vog.cdc@6kcn. Ana Silvia Gonzales Reiche, Department of Veterinary Medicine, University of Maryland, College Park, MD, E-mail: tg.ude.gvu.sec@zelaznoga. Maria Eugenia Morales-Betoulle, Viral and Zoonotic Diseases Research Program, US Naval Medical Research Unit No. 3 (NAMRU-3), Cairo, Egypt, E-mail: firstname.lastname@example.org.