Jamestown Canyon virus (Bunyavirales: Peribunyaviridae) vector ecology in a focus of human transmission in New Hampshire, USA

Abstract

Jamestown Canyon virus disease (JCVD) is a potentially neuroinvasive condition caused by the arbovirus Jamestown Canyon virus (JCV). Human cases of JCVD have increased in New Hampshire (NH) over the past decade, but vector surveillance is limited by funding and person power. We conducted mosquito surveillance with a focus on human JCVD cases south central NH during 2021. Routine surveillance with CDC miniature traps baited with CO₂ (lights removed) was supplemented by a paired trapping design to test the collection efficiency of octenol, and New Jersey light traps. We performed virus testing, blood meal analysis, and compared morphological identification with DNA barcoding. Over 50,000 mosquitoes were collected representing 28 species. Twelve JCV-positive pools were derived from 6 species of more than 1,600 pools tested. Of those, Aedes excrucians/stimulans (MLE 4.95, Diptera: Culicidae, Walker, 1856, 1848), and Aedes sticticus (MLE 2.02, Meigen, 1838) had the highest JCV infection rates, and Aedes canadensis (MLE 0.13, Theobold, 1901) and Coquillettidia perturbans (0.10, Diptera: Culicidae, Walker, 1856) had the lowest infection rates. One hundred and fifty-one blood meals were matched to a vertebrate host. All putative vectors fed on the amplifying host of JCV, white-tailed deer (36–100% of bloodmeals). Putative vectors that fed on human hosts included Aedes excrucians (8%), Anopheles punctipennis (25%, Diptera: Culicidae, Say, 1823), and Coquillettidia perturbans (51%). CDC traps baited with CO₂ were effective for collecting putative vectors. DNA barcoding enhanced morphological identifications of damaged specimens. We present the first ecological overview of JCV vectors in NH.

Introduction

Jamestown Canyon virus (JCV) is a re-emerging arbovirus (Peribunyavirales: Orthobunyavirus), which impacts human health. JCV was first documented from Culiseta (Cs.) inornata (Diptera: Culicidae: Williston 1893) mosquitoes from Colorado in 1961, and was subsequently connected with neuroinvasive disease across the United States (CDC Arbocat 1984, Grimstad et al. 1986). Over the past 10–20 yr, reported human cases of Jamestown Canyon virus disease (JCVD) have been increasing. Mortality is a rare outcome of JCV infections (CDC Arbonet 2022), although hospitalization was necessary for 48% of reported cases from 2000 to 2013 (Pastula et al. 2015). The enzootic cycle of JCV is poorly understood. The current understanding is that JCV is maintained in a seasonal cycle. In this, Aedes (Ae.) provocans (Diptera: Culicidae, Walker, 1848) overwinters with virus via vertical transmission (Boromisa and Grayson 1990, Farquhar et al. 2022). Upon hatching in spring, these mosquitoes transmit virus to white-tailed deer (Odocoileus virginianus) Artiodactyla: Cervidae Zimmermann, 1780, the only identified amplifying reservoir to date (Issel 1972, Watts et al. 1979), throughout spring and early summer months (Grimstad et al. 1987). Once introduced, other species may play a role as vectors after feeding on infectious deer; to this point, viral RNA has been detected via PRC and viral isolation (Andreadis et al. 2008) from over 20 species.

The endemic range of JCV extends across the United States, and Canada (CA) (Kulkarni et al. 2015). Over 22 states have reported human cases, presumed to be locally acquired (CDC ArboNET 2022). Reports of JCV antibody prevalence in humans range from 4% to >20% across North America from southwestern US National Parks to New Brunswick, CA (Mayo et al. 2001, Kosoy et al. 2016, Mincer et al. 2021). However, human case reporting and mosquito vector surveillance are not uniformly conducted across the United States. States with the highest JCVD case counts include Wisconsin (WI), Minnesota (MN), Michigan, Maine, Connecticut, New Jersey, and New York. From 2011 to 2020, in the Midwestern United States, 67 total cases of JCVD were reported from MN, and 116 total cases were reported from WI. During this same period, 22 total cases were reported from the Northeastern United States. Differences in case reporting and surveillance capacity between the Midwest and other US regions complicate our ability to understand regions of greatest risk for JCVD.

Research conducted over 3 decades suggests that risk of JCVD may be high in the Northeastern United States (Walker et al. 1993, Mayo et al. 2001, Kinsella et al. 2020). Specifically in New Hampshire (NH), 19 human cases of JCVD have been identified since the initial pathogen detection through clinical testing in 2013. In 2021 alone, 5 cases were reported to CDC. Current mosquito surveillance in NH is limited to the seacoast and is intended to target West Nile virus (WNV) and Eastern equine encephalitis virus (EEEv). However, JCV has recently replaced EEEv as the most commonly reported arbovirus in the state, emphasizing the need to establish robust surveillance.

Few updates on the diversity and distribution of mosquito species in NH have been published (Lowry 1929, Blickle 1952, Burger and Davis 2008). As a result, the potential role of mosquitoes in JCV transmission remains unclear. To address this crucial knowledge gap, we conducted enhanced JCV vector surveillance in 2021 in collaboration with the NH Department of Health and Human Services. Municipalities with reported JCVD cases, clustering in the south-central part of NH, were targeted for adult surveillance beginning in May 2021. Suspected vectors were tested for JCV RNA, and blood hosts were identified from engorged mosquitoes. Cumulatively, our study yields the first insights into JCV vector ecology in NH, describes basic life history of putative vectors, and uses DNA barcoding to complement morphological identification of subtly different species.

Methods

Potential enzootic and bridge vectors of JCV in NH were identified a priori based on detections of virus from previous studies in the Northeastern and Midwestern United States. Primarily, these were snowmelt Aedes (Diptera: Culicidae Meigen 1818) species known to feed on white-tailed deer with reported JCV detections from multiple sources. Species tested for virus included: Ae. abserratus (Diptera: Culicidae, Felt and Young, 1904), Ae. stimulans (Walker, 1848), Ae. excrucians (Walker, 1856), Ae. Fitchii (Diptera: Felt and Young, 1904), Ae. Intrudens (Diptera: Dyar, 1919), Ae. Aurifer (Diptera: Coquillett, 1903), and Ae. provocans (Boromisa and Grimstad 1986, Boromisa and Grayson 1990, Heard et al. 1990, Walker et al. 1993, Andreadis et al. 2008, Molaei et al. 2008). Additionally, we included Ae. Cantator (Diptera: Coquillett, 1903) (maximum likelihood estimate [MLE] = 1.16), Ae. communis (Diptera: de Geer, 1776) (MLE = 1.66), Ae. sticticus (Meigan, 1838, MLE = 0.23), Ae. canadensis (Theobald, 1901, MLE = 0.23), Anopheles (An.) punctipennis (Diptera: Culicidae, Say, 1823, MLE = 0.69), Cs. morsitans (MLE = 0.40), and Coquillettidia perturbans (Walker, 1856, MLE = 0.04) as potential JCV vectors. An. quadrimaculatus (Diptera: Say, 1824) was included based on a single detection of JCV made in Ohio (Berry 1983), and regional interest in vector competence of this species (Dieme et al. 2022).

Site selection took place in April of 2021, with visits to 13 sites to confirm the presence of snowmelt Aedes larvae, collected by submerging dippers in leaf-lined woodland pools and wetlands (O’Malley 1989). Eight field sites were chosen (Fig. 1), each with multiple habitats ranging from vernal pools, ponds, and semi-permanent and permanent wetlands. Multiple subsites were chosen at each location, hereafter “A–E”, with “A” and “B” sites 20 m apart from each other around 1 breeding site, and “C” and “D” 20 m apart from each other around another breeding site and at least 30 m apart from “A” and “B” to facilitate trap comparisons and control for the effect of trap placement. Additionally, game cameras and data loggers were placed at sites, methodologies and results can be found in Supplementary Material S1.

Fig. 1.

(Bottom left) Map of New Hampshire with field sites highlighted (created in QGIS (v3.6)). (Top right) Zoomed in map showing location of field sites within south-central New Hampshire (background map from Bing Satellite imagery downloaded in QGIS). (Bottom right) Map zoomed into site 1, showing subsites “A” through “E,” made in Google Earth Pro. CDC traps and NJ light traps rotated between subsites 1A and 1B. Paired CDC traps for octenol experiment rotated between subsites 1C and 1D.

Routine surveillance was conducted from 12 May to 10 August 2021 using Centers for Disease Control and Prevention miniature light traps (John W. Hock Company, Gainesville FL) baited with ~1.8 kg of CO₂ (dry ice) with the lights removed to minimize by-catch (CDC traps hereafter), set at each “A and B” subsites (20 m apart). CDC traps were run twice weekly overnight, for an average of 19.5 h per trap night. Wooden resting boxes (2’ × 2’ × 3’), with a red interior and black exterior, were set near CDC traps at the “E” subsite from 22 June to 11 August 2021 to capture blood fed mosquitoes. Resting boxes were deployed in our study to enhance capture of blood fed specimens for blood meal analysis rather than species abundance, and they were not uniformly sampled. Resting boxes were examined visually for mosquitoes and those present were captured by an aspirator (John W. Hock Company, Gainesville, FL, Prokopak model 1419).

Paired trapping was conducted to facilitate comparisons between light, octenol (1-Octen-3-ol, Sigma Aldrich W280518), and CO₂ for their ability to enhance collections of suspect JCV vectors. Octenol is a volatile host cue compound that could enhance collection of mammalophagic mosquitoes (Poldy 2020). For testing light as an attractant, we used New Jersey Light traps, (NJLT hereafter, John Hock Company model 1112.5) powered by a 12 V deep cycle battery (SLI24D24XD 12 V 80 ah) and inverter (Jupiter 56496). NJLT were set at sites 1, 4, 5 weekly from 7 July to 10 August 2021, paired with CDC traps baited with CO₂ (lights removed) at subsites “A” and “B” (20 m apart). For comparing octenol plus CO₂ and CO₂ only, a CDC trap supplemented with ~1.8 kg of CO₂ and 1.5 ml octenol was deployed in tandem with a CDC trap supplemented with ~1.8 kg of CO₂ only. In June and August at sites 1, 3, 6, these traps were set at “C” and “D” subsites on 2 consecutive nights, swapped between subsites on the second night. Octenol was deployed in a 1.7-ml microcentrifuge tube taped upright to the trap with parafilm and a short wick, following the methods of Kline et al. (1990).

All mosquito collections were recorded along with trap type, date, location, set, and check time. Collection bags were transported on dry ice to the NH Public Health Laboratory for storage at −20 °C or 80 °C in filter paper lined petri dishes (VWR Cat No. 25384-088). Mosquitoes were identified morphologically to species using published dichotomous keys (Darsie and Ward 1981, Andreadis et al. 2005) and damaged specimens were reserved for DNA barcoding. Suspected female vectors were pooled for virus testing.

JCV RNA Detection Methods

Non-blood fed mosquitoes of suspect vector species were pooled in groups of 1–50 after identification, transferred to sterile 2-ml tubes, and stored at −80 °C. RNA was extracted from homogenized pooled samples using an automated KingFisher Flex instrument (ThermoFisher Scientific, Waltham, MA) and MagMAX Viral nucleic acid kit (ThermoFisher Scientific Waltham, MA). Quantitative PCR (qPCR) was prepared using a qScript One-Step qRT-CR Kit, Low ROX, and ToughMix (Quanta Biosciences, Beverly, MA). Each qPCR reaction included JCV174 (5ʹ-CAG TCT GTC AGC CGT TAG GA-3ʹ) (forward) + JCV269c (3ʹ-AAT TTC CAC CTG CCA CTC TC-5ʹ) (reverse) primers, and a JCV231c probe (5ʹ-FAM-TCC GCT CCG GTT TAC GAG CG-BHQ1-3ʹ). PCR was performed on the ABI 7500 Fast instrument (ThermoFisher Scientific, Waltham, MA). PCR cycling conditions were 50 °C for 30 min, 95 °C for 10 min, and 45 cycles of 95 °C for 15 s, 60 °C for 60 s. Positive pools were confirmed by repeating the qPCR assay, and homogenates of positive pools were saved at −80 °C. Cycle threshold (CT) values were recorded from qPCRs, with a cutoff of less than 38 indicating a positive result.

Blood Meal Analysis

Mosquitoes with fully engorged abdomens or smaller red blood meals were selected for bloodmeal analysis. Abdomens were removed with flame sterilized forceps and stored individually in sterile microcentrifuge tubes at −20 °C after identification. Heads and thoraces were reserved for DNA barcoding (see below). DNA was extracted using a PureGene kit (Gentra Systems, Minneapolis, MN) following the manufacturer’s instructions. DNA pellets were rehydrated in 50 μl of nuclease-free water (Thermo Fisher AM9930).

The cytochrome c oxidase I (COI) region of vertebrate mitochondrial DNA was targeted for DNA amplification using VertCOI_9174/ModRepCOI primers described by Reeves et al. (2018), and COI long and short primers described by Townzen et al. (2008). VertCOI_9174/ModRepCOI primers were 5ʹ-CGM ATR AAY AAY ATR AGC TTC TGA Y-3ʹ (forward) and 3ʹ-TTC DGG RTG NCC RAA RAA TCA-5ʹ (reverse) with an amplified product size of 324 bp. Thermal cycling conditions for VertCOI_9174/ModRepCOI primers were: 94 °C for 3 min, 40 cycles of 94 °C for 40 s, 51.2 °C for 30 s, 72 °C for 1 min, followed by extension at 72 °C for 7 min. COI long primers were 5ʹ-AAC CAC AAA GAC ATT GGC AC-3ʹ (forward) and 3ʹ-AAG AAT CAG AAT ARG TGT TG-5ʹ (reverse) with an amplified product of 663 bp. COI short primers (nested in COI long) were 5ʹ-GCA GGA ACA GGW TGA ACC G-3ʹ (forward) and 3ʹ-AAT CAG AAY AGG TGT TGG TAT AG-5ʹ (reverse) with an amplified product of 324 bp. For both COI long and short primers, thermal cycling conditions were 95 °C 5 min, 40 cycles of 95 °C for 30 s, 50 °C for 50 s, 72 °C for 1 min, followed by extension at 72 °C for 5 min. Each COI Long or short PCR reaction consisted of a 20 μl total volume of 7 μl H₂0, 10 μl TaqRed Master Mix, 0.75 μl of each primer (10 μM), and 1.5 μl of DNA template from DNA extraction. If the COI_short or COI_long primer pair did not amplify enough DNA, then we performed nested PCR with DNA product from the COI_long primer in a COI_short reaction (Townzen 2008).

All PCR reactions were performed on a Biorad C1000 touch thermal cycler with a lid temperature of 105 °C. PCR products were visualized via 1.5% agarose gel electrophoresis with GelRed (Biotum inc., #41003) on a BioRad Gel Doc XRS system (Hercules, California). PCR products that were the correct amplicon size were cleaned with ExoSap-IT enzymes (ThermoFisher Cat. No. 78201.1.ML), following the manufacturer’s recommendations and submitted for Sanger sequencing at Cornell Biotechnology Resources Center (https://www.biotech.cornell.edu/core-facilities-brc/services). All bloodmeal sequences were trimmed in Geneious Prime (Dotmatics, Boston, MA, V. 2022.2) and searched in GenBank (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed February–September 2022). Only those samples that matched with greater than 95% nucleotide similarity with e-values near 0, from a logical host (DeGraaf and Yamasaki 2001) were considered successful.

DNA Barcoding for Mosquito Specimen Identification

To complement morphological identifications, DNA barcoding of the COI region was performed to identify damaged specimens and species that are difficult to differentiate morphologically (such as Ae. abserratus/punctor) (Diptera: Kirby, 1837). The COI region was selected based on previous work (Ratnasingham and Hebert 2007). DNA was extracted from whole bodies or heads and thoraces as described above for blood meals. DNA was amplified using published universal invertebrate primers (Kumar et al. 2007, Chan et al. 2014) 5ʹ-GGA TTT GGA AAT TGA TTA GTT CCT T-3ʹ (forward) 3ʹ-AAA AAT TTT AAT TCC AGT TGG AAC AGC-5ʹ (reverse) yielding a product size of ~738 bp. Each PCR 20 μl reaction included 10 μl TaqRed Master Mix, 8.8 μl nuclease free water, 0.4 μl each forward and reverse primer (0.2 μM), and 0.4 μl of DNA template.

A reference library was compiled of DNA barcodes from confirmed voucher specimens as well as high confidence sequences published in the public Barcode of Life Database (BOLD, Ratnasingham and Hebert 2007). Sequences from 30 Aedes (Och.) species, 5 Culex (Diptera: Linnaeus, 1758) species, 5 Culiseta (Diptera: Felt, 1904) species, and Cq. perturbans were downloaded from BOLD (n = 271, accessed September–October 2022). Our morphological voucher specimens (2–27 per species) were pristine samples with all key taxonomic characters which we preserved from our own collections. We sequenced these voucher specimens for Ae. provocans, Ae. fitchii, Ae. stimulans, Ae. excrucians, Cs. morsitans, Cs. Melanura (Diptera: Coquillett, 1902), Ae. abserratus/punctor, Ae. Taeniorhynchus (Diptera: Wiedemann, 1821), Ae. Cinereus (Diptera: Meigen, 1818), and Ae. sticticus. A breakdown of the reference library materials is included in Supplementary Table S4.

Sequences (5ʹ–3ʹ) were trimmed and aligned using Geneious Prime software (v. 2022.2.2). Trees were constructed for each genus (Cs. and Cq. combined) using a Tamura-Nei genetic distance model (Tamura and Nei 1993), with both library and query sequences. Final trees were resampled via bootstrapping (n = 100), and intraspecific distances were calculated using the species delimitation plugin (Masters et al. 2011). Identifications of query sequences were classified as “high confidence”, “unclear”, and “ambiguous” using a variation of the “revised tree-based identifications” method of (Meier et al. 2006), taking into account bootstrap support of each branch. Briefly, sequences that grouped within a strongly supported branch of the tree (Bootstrap >90%) of conspecifics are high confidence identifications. Ambiguous identifications are sequences that grouped within a moderately supported branch (Bootstrap 70–89%), within branches of the same species, or with a mixture of species. Unclear identifications are sequences placed within a weakly supported branch (Bootstrap < 70%), or without a conspecific in the dataset.

Statistical and Analytical Methods

All statistical and descriptive analyses were conducted with MS excel & R Studio, version 3.6 (R Core Team 2021). Bias-corrected maximum likelihood estimates (MLE) of infection rates were determined, using an MS Excel add-in (Biggerstaff 2008). We assessed the effect of octenol supplementation for enhancing the collection of putative JCV vectors using generalized linear models (GLM) created using R package: GLMTMMB (Brooks et al. 2017). Briefly, species specific GLMs were fit for each month (June or August) with site, an offset to account for trap run time, and assessed for octenol significance. We analyzed species specific models to account for large biological variance in abundance and phenology among putative vectors. Model selection was based on AIC values, leading to the selection of negative binomial 2 distributions, which had the lowest AIC. Final models were assessed with ANOVA, a dispersion test, a histogram of residuals, and a Q-Q plot using simulated residuals in R package DHARMa (Hartig 2022). Separately, we also assessed the effectiveness of NJLT (light) versus CDC traps baited with CO₂ with lights removed (CO₂) at enhancing the collection of 12 mosquito species and species groups. For each test, abundance was compared across 25 paired trap nights across 3 sites (1, 4, 5) from 7 July to 10 August. Shapiro–Wilk tests of normality were used to test the normality of abundance, leading to the selection of paired t-test or Wilcoxon ranked sign test. All P-values were adjusted using the Bonferroni adjustment for multiple comparisons. All model forms are provided in Supplementary Materials (see Supplementary Material S2).

Results

Mosquito Collections

A total of 50,520 adult female mosquitoes were collected from 8 field sites in south central NH from 6 May to 10 August 2021, representing 28 mosquito species and 6 genera. CDC traps with CO₂ (177 trap nights) were the foremost surveillance tool, followed by NJLT (28 trap nights) and supplemental trapping using CDC traps with CO₂ only or CO₂ with octenol (24 trap nights). A total of 56 successful collections were made from resting boxes, resulting in 128 mosquitoes, 78 (61%) were engorged.

Data on phenology and timing of abundance for each species are presented in Fig. 2. Early season CDC trap collections detected Ae. provocans and An. punctipennis first on 7 May 2021, Ae. abserratus/punctor and Ae. sticticus on 12 May 2021, Ae. excrucians/stimulans on 19 May 2021 followed by Ae. canadensis on 27 May. Cq. perturbans was first observed on 8 June 2021, and quickly dominated collections. In addition, we captured a single specimen of Ae. japonicus (Diptera: Theobald, 1901) on 19 June at site 6. We later detected a second emergence of Ae. sticticus observed on 3 and 4 August 2021 at site 1 from refilled emergent pools under mixed forest canopies. Throughout the field season, An. punctipennis and snowmelt Aedes were collected in low numbers, generally less than 10 individuals per trap night.

Fig. 2.

Average abundance of putative vectors (± SE) collected in CDC traps from 12 May to 10 August. Shaded areas indicate detections of JCV, via PCR assay.

Virus Testing Results

In total, 12 detections of JCV were made from mosquito pools collected in 2021 (Table 1) from 6 mosquito species (CT values ranged from 16.10 to 34.82, average of 24.24). A pool comprised of a single An. punctipennis collected 27 May 2021 tested positive. Four pools of Ae. excrucians/stimulans tested positive, from 8 June, 7 July, 20 July, and 10 August 2021. Two pools of Ae. abserratus/punctor tested positive on 11 June and on 18 June 2021. One pool of Ae. sticticus tested positive collected on 17 June 2021. Three pools of Cq. perturbans tested positive, on 18 June, 20 July, and 10 August 2021. One pool of Ae. canadensis tested positive on 10 August 2021, concurrent with peak densities of this species. Virus was detected at sites 1, 3, 5, 6 with woodland pools covered by a mixed canopy of red maple (Acer rubrum) (Sapindaceae: L.) and eastern hemlock (Tsuga canadensis) (Pinaceae: L.). No detections were made from 54 pools (n = 286) of Ae. provocans, 46 pools (n = 105) of An. quadrimaculatus, 38 pools (n = 172) of Ae. aurifer/cinereus, 15 pools (n = 22) of Cs. inornata, 5 pools (n = 10) of Ae. intrudens, nor 5 pools (n = 6) of Ae. communis s.s.

Table 1.

Virus testing and infection rate results

Mosquito	Pools	Detections	No. tested	MLE (95% CI)	MIR
Ae. abserratus/punctor	119	2	2,026	0.99 (0.18–3.24)	1:1,013
Ae. canadensis	289	1	7,875	0.13 (0.01–0.62)	1:7,875
Ae. excrucians a	123	4	827	4.95 (1.61–11.86)	1:207
Ae. sticticus	74	1	485	2.02 (0.12–9.64)	1:485
An. punctipennis	167	1	871	1.14 (0.07–5.48)	1:871
Cq. perturbans	682	3	28,796	0.1 (0.03–0.28)	1:9,599

Bold lettering indicates high infection rates.

^aAe. stimulans included.

Blood Meal Analysis and Host Availability

The blood host(s) of 151 of 315 (47.9%) engorged mosquitoes were successfully identified (Fig. 3), resulting in 159 identified hosts. Blood meals identified from putative vectors were largely taken from mammals (95%, 89/93). Sample sizes were limited from some species including Ae. sticticus (n = 2), Ae. abserratus/punctor (n = 7), and An. punctipennis (n = 8). White-tailed deer host feeding (55%, 51/93) was common from putative vectors, and human feeding was exhibited by An. punctipennis (25%, 2/8), Ae. excrucians/stimulans (9%, 1/11), and Cq. perturbans (51%, 25/49). Coquillettidia perturbans (n = 49) had the broadest host feeding patterns and 5 double bloodmeals were revealed for this species through nested PCR, containing human/white-tailed deer (n = 2), white-tailed deer/river otter (Carnivora: Mustelidae: Lontra canadensis Schreber, 1777), human/barred owl (Strigiformes: Strigidae: Strix varia Barton, 1799), and human/broad winged hawk (Accipitriformes:Accipitridae Buteo platypterus Vieillot, 1823). A detailed account of identified bloodmeals, including those from nonputative vectors (n = 56) are summarized in the Supplementary Table S5. Notably, 2 blood meals were identified from Cs. melanura, including American robin (Passeriformes: Turdidae Turdus migratorius Linnaeus, 1766)/blue jay (Passeriformes: Corvidae Cyanocitta cristata Linnaeus, 1758), and white-tailed deer/human. From Cs. morsitans, 1 double blood meal was identified as scarlet tanager (Passeriformes: Cardinalidae Piranga olivacea Gmelin, 1789)/human.

Fig. 3.

Mammal blood meal proportions derived from 6 putative vector species and Aedes aurifer/cinereus.

Images from game cameras deployed at 7 of the 8 sites showed 10 species of mammals and 3 bird species, as well as the presence of white-tailed deer at each site where cameras were deployed. Game cameras also revealed a high level of human activity at sites 1 and 5, where people were geocaching. More detailed findings in Supplementary Material S1.

Trap Type Comparisons

Across paired trapping events, octenol supplementation did not have a statistically significant effect on the abundance of putative vectors in paired collections of CDC traps baited with CO₂ with and without octenol, in June or August. In paired collections of CDC traps and NJLT (Fig. 4, n = 25 each), CDC traps captured more mosquitoes (n = 6,310) than NJLT (n = 1,365). In July and August, CDC traps caught significantly more snowmelt Aedes of interest (Ae. excrucians/stimulans, Ae. sticticus and Ae. abserratus/punctor) than NJLT (Wilcoxon test, P < 0.05). In addition, significantly more Cq. perturbans, Ae. canadensis, Ae. trivittatus (Diptera: Coquillett, 1902) and Psorophora ferox (Diptera: Humboldt, 1819) were captured in CDC light traps than NJLT (Wilcoxon test, P < 0.05). NJLT collections had Culex (Cx.) territans (Diptera: Walker, 1856) and Uranotaenia (Ur.) sappharina (Diptera: Osten Sacken, 1868), which were not found in CDC traps. Significantly more Cs. melanura, Cs. morsitans, Ae. aurifer/cinereus were captured with NJLT than CDC traps (Wilcoxon test, P < 0.05). Collections of An. punctipennis (Paired t-test, P = 0.541), Ae. vexans (Diptera: Meigen 1830) (Wilcoxon test, P = 0.104), Cx. Pipiens (Diptera: Linnaeus, 1758)/restuans (Diptera: Theobald, 1901)/salinarius (Diptera: Coquillett, 1904) (Wilcoxon test, P = 0.343) were not significantly influenced by trap type. A summary table of NJLT and CDC trap comparisons can be found in Supplementary Material S2.

Fig. 4.

Stacked bars representing the percent of each species in total paired collections of NJLT and CDC traps (n = 50). Differences by trap type for each species are indicated (Wilcoxon rank tests or paired t-test, **P < 0.05).

DNA Barcoding of Damaged Mosquito Specimens

In the Aedes (Och.) tree (Fig. 5), 54.6% (53/97) of query specimens were identified with high confidence, 40.2% (39/97) were ambiguous, and (5.2%) (5/97) were unidentifiable. Ambiguous identifications could only be made for some sister taxa (Ae punctor/implicates (Diptera: Vockeroth, 1954), Ae intrudens/Ae diantaeus (Diptera: Howard, Dyar and Knab, 1913), Ae aurifer/decticus (Diptera: Howard, Dyar and Knab, 1917), Ae. excrucians/stimulans, Ae. communis s.l.) due to a lack of branch support indicating nondistinct groupings. However, voucher specimens were correctly placed for Ae. excrucians, Ae. stimulans, Ae. fitchii, Ae. cinereus, Ae. taeniorhynchus, and Ae. sticticus. Aedes communis s.l. (churchilliensis, hexodontus, communis s.s.) were not well defined due to a lack of specificity in BOLD sequences. Reexamination of an Ae. provocans voucher grouped within Ae. communis s.l. confirmed that it was not Ae. communis s.l., via the presence of a hypostigmal area scale patch (Darsie and Ward 1981). Morphological reexamination of 2 Ae. abserratus/punctor vouchers (grouped within Ae. punctor/Ae. implicatus) revealed that both specimens were not Ae. implicatus via the presence of scales on the mesokatepisternum reaching an anterior angle, and an elongate foreclaw (1 specimen missing foreclaw) (Darsie and Ward 1981).

Fig. 5.

Aedes (Och.) Collapsed tree used for specimen matching (DNA barcoding) created in Geneious Prime (v.2.2022.2). Brackets indicate [# BOLD vouchers, #Morphological vouchers | #Queried specimens (morphological identifications of this species), # high confidence matches, # ambiguous matches] for the species indicated. Branch support values indicate bootstrap resampling, values < 50% not shown, subtrees were trimmed at a distance of 0.029 for fit.

Damaged specimens that were missing morphological features were also identified with high confidence using the approach described above. This includes: Ae. communis s.l. (n = 5), Ae. cinereus (11), Ae. canadensis (1), Ae. sticticus (4), and Ae. trivittatus (4). Additional specimens were identified to 1 of 2 species including Ae. diantaeus or Ae. intrudens (n = 11), Ae. aurifer or Ae. decticus (5), Ae. punctor or Ae. implicatus (14), and Ae. excrucians or Ae. stimulans (23). Intraspecific distances for well-defined groups ranged from 0.01% (Ae. abserratus) to 2.6% (Ae. sticticus). Detailed accounts of the identification success of Culex spp. and Culiseta spp., as well as intraspecific genetic distances can be found in Supplementary Material S3. A breakdown of the total mosquito collections made as well as the use of morphological and BOLD vouchers can be found in Supplementary Table S5.

Discussion

We detected 12 JCV RNA positive mosquito pools during the course of our study. This represents the first detection of JCV in mosquitoes in NH since the initial 2013 report of JCV human illness (NH Arbo Bulletin 2021). We sought to understand what mosquitoes may serve as bridge vectors in the region as an initial step to understanding risk and JCV prevention strategies. We identified 6 species as potential JCV vectors. These mosquitoes can harbor JCV viral RNA but may not be competent to transmit virus to new hosts. We detected white-tailed deer (the presumed JCV reservoir host) and human feeding by Ae. excrucians, An. punctipennis and Cq. perturbans. In addition to gaining a better understanding of JCV vector potential in New Hampshire, we also evaluated surveillance methods to determine optimal approaches for JCV vector surveillance. CO₂ baited CDC traps were more effective in capturing snowmelt Aedes putative vectors than NJLT. CDC traps supplemented with octenol may be useful in enhancing collections of An. punctipennis.

Our results support prior research demonstrating white-tailed deer as an important blood host for An. punctipennis, Ae. abserratus/punctor, Ae. excrucians/stimulans and Ae. canadensis (Boromisa and Grimstad 1986, Molaei et al. 2008, Murdock et al. 2010, Shepard et al. 2016). The presence of deer at each field site was confirmed by game camera surveys, suggesting that blood meals were taken locally. Viral detections were made from Ae. abserratus/punctor in mid-June. An. punctipennis, Ae. excrucians/stimulans and Ae. sticticus had the highest detection rates of viral RNA and exhibited human feeding in less than 20% of identified bloodmeals. JCV viral RNA was rarely detected from the most abundantly collected species: Cq. perturbans and Ae. canadensis. However, humans were identified as the blood host of 47% of Cq. perturbans bloodmeals. In contrast to other published studies (Boromisa and Grimstad 1986, Andreadis et al. 2008) we found higher JCV infection rates for Ae. excrucians/stimulans and Ae. sticticus. Two additional detections of JCV RNA were made by NH DHHS apart from this study. These included a pool of Ae. excrucians collected in Derry, NH on 24 June 2021, as well as a pool of Ae. canadensis collected in Kingston, NH on 30 June 2021 (NH DHHS Arbo Bulletin 2021). No JCV was detected from Ae. provocans, an incriminated vector of JCV from studies conducted in New York, Wisconsin (Boromisa and Grayson 1990, Farquhar et al. 2022), and Connecticut (Andreadis et al. 2008).

The purpose of vector surveillance is to infer potential human disease risk before human cases occur. For this study, we targeted sites with known populations of putative JCV vector mosquito larvae, based on habitats and detection of larvae in April 2021. By starting adult surveillance in May, we were able to characterize the enzootic and bridge vector populations and began detecting JCV as early as 27 May 2021 at site 5 in An. punctipennis. Site 5 had continued virus presence across the field season with 1 detection on 11 June 2021 from Ae. abserratus/punctor, and 2 on 20 July 2021, Ae. excrucians/stimulans and Cq. perturbans. Later, in September 2021, 1 human case occurred in Loudon NH, a town adjacent to field sites 4 and 5. Confirming the presence of virus detected from vector surveillance.

Future JCV surveillance efforts in NH and the NE region should prioritize mid-May to early June, before snowmelt Aedes populations peak. This will enable assessment of disease transmission risk through species abundance and infection rates. For example, if surveillance efforts in NH had begun on the first of July in 2021, surveillance would have missed 50% (6/12) of virus detections and captured only 1.5% (32/2026) of Ae. abserratus/punctor and 39% (321/827) of Ae. excrucians/stimulans. These early season vectors may be both enzootic bridge vectors for JCV given several JCVD cases in NH with onset in May (NH Arbo Bulletin 2021), and detection of human blood meals from Ae excrucians and An. punctipennis. Surveillance efforts targeting An. punctipennis, Ae. abserratus/punctor, Ae. excrucians/stimulans as well as Ae. sticticus, beginning at least by the first of June, will provide useful entomological data in advance of human cases.

In nature, vertical transmission of JCV has been documented in Ae. triseriatus in Ohio (Berry et al. 1977), Ae. provocans and Ae. stimulans s.s. (Boromisa and Grimstad 1986, Boromisa and Grayson 1990, Farquhar et al. 2022) in New York. Vertical transmission is known to occur in other California Serogroup (CSG) viruses and is a likely overwintering mechanism (Rosen 1987), influenced by extrinsic factors such as temperature. Given viral RNA detections coinciding with the adult emergence of Ae. abserratus/punctor, Ae. excrucians/stimulans, Cq. perturbans, and Ae. canadensis, vertical transmission may be occurring as a form of viral maintenance. Rigorous work on vertical JCV transmission is limited by an inability to rear snowmelt mosquitoes in the laboratory, but larval surveillance could be employed to overcome this. In our study, we detected JCV in An. punctipennis for the first time early in the mosquito season (27 May). Given their laboratory demonstrated vector competence for JCV (Heard et al. 1991) and adult overwintering biology, future research should target their potential for JCV virus maintenance in NH.

We also report the first updated record of mosquito collections in New Hampshire since 1952 (Blickle 1952). Our collection confirms the presence of Ae. japonicus within the state (Burger and Davis 2008). Some Ae. abserratus/punctor specimens were putatively identified as Ae. implicatus via DNA barcoding, and if present, would be a novel detection of this species in NH. However, we were not able to discern the identity of these mosquitoes beyond Ae. punctor/implicatus. Collections from May through August 2021 suggest a possible bivoltine cycle of Ae. canadensis and Ae. sticticus, which is similar to reports from Anderson et al. (2018) in CT. For Ae. sticticus, it has been proposed that some floodwater species undergo installment hatching (Wilson and Horsfall 1970, Hendrichs et al. 2020). In NH in 2021, it is possible that reflooding of vernal pools with heavy rainfall in July stimulated a new hatch of Ae. sticticus eggs.

It is unclear what other animals may serve as JCV reservoirs in the region. Penned white-tailed deer were shown to develop JCV antibodies from May and June (Grimstad et al. 1987) in Michigan. Experimental infections of white-tailed deer show viral amplification for 4–5 days (Issel 1972, Watts et al. 1979), but viral maintenance remains poorly understood. JCV vectors with viral detections later than June such as Ae. canadensis also fed on moose (Alces alces) (Artiodactyla: Cervidae Linneaus 1758), gray squirrel (Sciurus caroliniensis) (Rodentia:Sciuridae Gmelin, 1788), and woodcock (Scolopax minor) (Charadriiformes:Scolopacidae J. F. Gmelin, 1789). In addition to white-tailed deer, Cq. perturbans fed on river otter (Lontra canadensis), moose, red fox (Vulpes vulpes) (Carnivora: Canidae Linnaeus, 1758), barred owl, black-capped chickadee (Poecile atricapillus) (Passeriformes: Paridae Linnaeus, 1766), and broad winged hawk. The presence of CSG and JCV antibodies in moose from Canada (Trainer and Hoff 1971, McFarlane et al. 1981) indicates exposure to JCV is likely occurring. Moose were only at sites 1, 3, and 5 in mosquito blood meals at time points roughly coinciding with detections of JCV at these sites. Moose thermoregulate in wetlands during the summer, overlapping with the presence of mosquito vectors (DeGraaf and Yamasaki 2001). However, moose have yet to be tested for reservoir competence for JCV or other CSG viruses.

Identifying certain snowmelt Aedes exclusively based on morphological features can be unreliable, (Defoliart et al. 1969, Farquhar et al. 2022) creating yet another challenge in incriminating mosquito vectors of JCV. A lack of reliable species level identifications has resulted in the use of informal groupings, such as “Ae. stimulans group” and “Ae. communis group”. While convenient for operators, these groupings lack the rigor needed to distinguish important differences in species level biology, ecology, and behavior. Furthermore, some important taxonomic features required for accurate species level determination, can easily be lost during collection. While not always practical or cost efficient, DNA barcoding can be employed to identify specimens damaged from collection fans and handling. However, we noted incorrect sequences uploaded to BOLD for Cs. Impatiens (Diptera: Walker, 1848), Cx. Erratics (Diptera: Dyar and Knab, 1906), Ae. Grossbecki (Diptera: Dyar and Knab, 1906), Ae. communis s.l., and Ae. implicatus decreasing the accuracy of this tool. To overcome this limitation, we barcoded confirmed specimens from pristine samples and uploaded them to the BOLD database. Moving forward, removal of misidentified sequences and contribution of sequences from voucher specimens identified by a taxonomic expert will improve the ability of the research community to study mosquito vector ecology.

Conclusion

In NH, JCV represents a significant public health threat as indicated by sporadic cases of JCVD. In the absence of treatment or vaccination, surveillance, public education, and outreach represent the only modes of primary prevention for JCVD. Here, we provide the first data on surveillance and blood feeding ecology of likely mosquito vectors of JCV in NH. Snowmelt Aedes species and An. punctipennis represent a focus of virus activity and should be targets for future studies of JCV transmission dynamics and ecology.

Funding

This publication was supported by cooperative agreement 1U01CK000509-01 between Cornell University and the CDC. Contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC, Cornell University, or the Department of Health and Human Services. Supplemental funding was awarded to J. D. Poggi by the NJMCA Jobbins Scholarship in 2020. Additional funding was provided by CDC Epidemiology and Laboratory Capacity for Infectious Diseases (ELC) NU50CK000522-02-00 and Public Health Emergency Preparedness Cooperative Agreement (PHEP): 5NU90TP922018.

Disclaimer

This study was completed as part of the main author’s Master of Science degree at Cornell University and is also published as 1 thesis chapter. The findings and conclusions of this report are those of the authors alone.