Detection of Cell-Fusing Agent virus across ecologically diverse populations of Aedes aegypti on the Caribbean island of Saint Lucia

Background. Outbreaks of mosquito-borne arboviral diseases including dengue virus (DENV), Zika virus (ZIKV), yellow fever virus (YFV) and chikungunya virus (CHIKV) have recently occurred in the Caribbean. The geographical range of the principal vectors responsible for transmission, Aedes (Ae.) aegypti and Ae. albopictus are increasing and greater mosquito surveillance is needed in the Caribbean given international tourism is so prominent. The island of Saint Lucia has seen outbreaks of DENV and CHIKV in the past five years but vector surveillance has been limited with the last studies dating back to the late 1970s. Natural disasters have changed the landscape of Saint Lucia and the island has gone through significant urbanisation. Methods. In this study, we conducted an entomological survey of Ae. aegypti and Ae. albopictus distribution across the island and analysed environmental parameters associated with the presence of these species in addition to screening for medically important arboviruses and other flaviviruses. Results. Although we collected Ae. aegypti across a range of sites across the island, no Ae. albopictus were collected despite traps being placed in diverse ecological settings. The number of Ae. aegypti collected was significantly associated with higher elevation, and semi-urban settings yielded female mosquito counts per trap-day that were five-fold lower than urban settings. Screening for arboviruses revealed a high prevalence of cell-fusing agent virus (CFAV). Conclusions. Outbreaks of arboviruses transmitted by Ae. aegypti and Ae. albopictus have a history of occurring in small tropical islands and Saint Lucia is particularly vulnerable given the limited resources available to undertake vector control and manage outbreaks. Surveillance strategies can identify risk areas for predicting future outbreaks. Further research is needed to determine the diversity of current mosquito species, investigate insect-specific viruses, as well as pathogenic arboviruses, and this should also be extended to the neighbouring smaller Caribbean islands.


Introduction
Medically important arboviruses that cause human morbidity and mortality are predominantly transmitted by mosquitoes.There are more than 600 known arboviruses and related zoonotic viruses with more than 80 known to be human pathogens.Outbreaks of dengue virus (DENV), Zika virus (ZIKV), yellow fever virus (YFV) -all from the Flavivirus genus (Family: Flaviviridae); and chikungunya virus (CHIKV) -from the Alphavirus genus (Family: Togaviridae) are increasing 1 and there is potential for zoonotic viruses to spill-over into human populations.Arboviral disease transmission mostly occurs in tropical countries of Southeast Asia and South America and has a significant impact on developing countries 2 .Annual DENV infections are estimated at 100-390 million per year 3 and dengue is 're-emerging' mostly due to the expansion of the geographical range of the principal mosquito vector, Aedes (Ae.) aegypti, through globalization and climate change 2,4 .ZIKV is historically thought to be transmitted by Ae. aegypti.Local transmission in the Americas was first reported in early 2014 5 and the latest update from the World Health Organisation (July 2019) reports that 87 countries and territories have had evidence of autochthonous ZIKV transmission 6 .YFV is also transmitted by Ae. aegypti and can result in large urban outbreaks and rapid spread to distant locations 7 and is now endemic in Central American countries and several Caribbean Islands 8 .CHIKV is transmitted by Ae. albopictus (and to a lesser extent by Ae. aegypti) and has spread globally, with outbreaks in the mid 2000s in the Indian Ocean and India, and even in Europe in 2007 9 .Transmission of CHIKV has also been seen recently in the Americas and this rapid geographical expansion (in a similar way to DENV) is likely due to the expanding habitat of the mosquito vectors 10 .
Outbreaks of arboviral diseases including DENV 10 , YFV 8 , CHIKV 11 and ZIKV 12 have recently occurred in the Caribbean.The possibility of additional recent arbovirus transmission in the Caribbean must be considered given some infections result in nearly indistinguishable clinical symptoms.For example, Mayaro virus (MAYV) is an Alphavirus closely related to CHIKV and has resulted in sporadic outbreaks in South America 1,13 .MAYV transmission is restricted to South and Central America where it is thought that non-human primates act as reservoir hosts and Haemogogus mosquitoes (eg.H. janthinomys) found in sylvatic jungle environments are responsible for human cases.Although human cases are strongly correlated with exposure to forest environments, urban transmission of MAYV must be considered given the association of cases and major cities infested with Ae. aegypti 13 .As the Caribbean is a destination for many international tourists, surveillance is needed for individual Caribbean islands to determine the risk of facilitating the spread of arboviral diseases.In particular, arboviruses transmitted by Ae. aegypti are considered important given that prevention predominantly relies on mosquito vector control.
Aedes aegypti was first identified in the Caribbean Islands in 1864 14 and has remained present despite the Pan American Health Organization (PAHO) mosquito control campaign in the 1940s-1960s that was launched to eliminate urban yellow fever.Aedes aegypti was successfully eradicated in many countries including Brazil, Mexico and Guatemala 15 but eradication was not achieved in other countries such as the USA, Suriname, Guyana, French Guyana, Venezuela and the Caribbean Islands.As the eradication campaign deteriorated in the early 1970s and 1980s, many countries became re-infested with Ae. aegypti 16,17 and the geographical expansion of Ae. aegypti with urbanization resulted in the introduction of DENV to many countries 18 .With the exception of YFV, there are no currently available treatments or vaccines for arboviral diseases transmitted by Ae. aegypti and Ae.albopictus.Disease control is currently limited to traditional vector control strategies that rely on insecticides or destruction of larval breeding sites.In most DENV-endemic countries, ultra-low volume space spraying is recommended only during dengue outbreaks.However, widespread insecticide resistance has developed in Ae. aegypti, including high pyrethroid resistance rates in South America [19][20][21][22] and further north in the Caribbean 23 .
The volcanic island of Saint Lucia is located midway down the Eastern Caribbean Chain between Martinique and Saint Vincent and north of Barbados (Figure 1).The first cases of dengue in Saint Lucia were recorded in the 1980s and following Hurricane Thomas in 2011 another outbreak occurred.CHIKV was first introduced to Saint Lucia in 2014 24 but despite these outbreaks of major mosquito-borne arboviruses, vector surveillance has been limited and the last documented studies were carried out in 1976 14 .The landscape of Saint Lucia in many areas has changed over the past 40 years due to natural disasters and urbanisation, which has likely changed the distribution of arbovirus vectors.As the density and habitats of Ae. aegypti have expanded both in urban and rural areas of many tropical countries, we conducted an initial survey of Ae. aegypti and Ae.albopictus distribution and analysed any environmental parameters that were associated with the presence of these species.Female mosquitoes were screened for medically

Amendments from Version 1
Our manuscript now includes the addition of an analysis of our trapping data using a negative binomial mixed effects model but significant associations remain the same.We have included trapping nights in Table 1 and an explanation of why a large amount of mosquitoes were collected at Soufriere Hospital.Additional CFAV sequences which became publicly available after the original analysis have now enabled a re-analysis to be carried out which has revealed our sequences are identical or very closely related to CFAV sequences from Guadeloupe potentially representing a Caribbean lineage of CFAV.We have adjusted our manuscript in all relevant areas to reflect this and replaced the previous Figure 3 and Figure 4 phylogenetic trees with a single updated phylogenetic tree (now Figure 3) from the re-analysis which incorporates additional more recently available sequences of CFAV for a more comprehensive comparison.We have now also included our % nucleotide homologies following re-analysis (Underlying data) and modified our conclusions to include some sentences on insect-specific flaviviruses.important arboviruses and other flaviviruses to investigate whether there was any evidence of infection.

Study sites and mosquito collection
Mosquito collections were carried out on the island of Saint Lucia (N 13˚ 54' 33.9984, W 60˚ 58' 44.0148) in July 2015.Saint Lucia has a population of ~166,000 people and is 27 miles long and 14 miles wide, with forest covering 77% of the island.The tropical climate includes a dry season (December to June) and a wet season (July to November).Biogents (BG) Sentinel 2.0 mosquito traps baited with BG lure® were used at various sites across the island (Supplementary Figure 1, Extended data) 25 during the beginning of the wet season.Site selection was undertaken based on geographical and environmental variation in urban and semi-urban areas across the island and factors based on island topology including forested areas, brackish water bodies, freshwater bodies and mangrove habitats in communities with previously high mosquito numbers recorded, gathered from local knowledge.In some locations traps were placed inside houses.Four longer-term permanent traps were connected to power supplies at Canaries Wellness Centre, Soufriere Hospital, Etangs Health Centre and the Riviere Doree Anglican Primary School (Figure 1) and traps were run for a period of 22-24 days with mosquitoes collected at 24-hour intervals.Four temporary traps powered by Power King®12-volt batteries were deployed at various locations across the island (Supplementary Figure 1, Extended data) 25 to collect mosquitoes over a 24-hour period.Due to logistical difficulties accessing all sites across the island, not all trapping days had four temporary traps running synchronously with the four longer-term permanent traps.EasyLog USB data loggers (EL-USB-2, LASCAR electronics Ltd.) were placed in permanent traps to record relative humidity and temperature at hourly intervals.Garmin eTrex® 20 hand-held GPS units (Garmin Ltd.) were used to determine co-ordinates of both permanent and temporary traps.Trapped mosquitoes were collected, killed on ice for genera morphological identification to identify individuals belonging to the Aedes genus using basic defining characteristics including patterns of light and dark scales on the abdomen and thorax and alternating light and dark bands on the legs.Larval dipping was undertaken at Soufriere Town, Choiseul Village, Marisule and Gros Islet to sample immature stages (larvae/pupae) from domestic containers (e.g.tanks and drums, discarded containers and tyres).Immature stages were reared and allowed to emerge in mosquito cages.Individual mosquitoes that were identified by morphology to be Ae.aegypti were placed in RNAlater and stored at -20°C to preserve RNA for molecular analysis.

RNA extraction and PCR analysis
A sub-sample of Aedes adult female mosquitoes were selected to maximise diversity of geographical trapping locations and pooled according to trap location and date of collection (1-3 females/pool) and RNA was extracted using Qiagen 96 RNeasy Kits (cat no.74182) according to manufacturer's instructions after using a Qiagen Tissue Lyser II (Hilden, Germany) with 3mm stainless steel beads to homogenise mosquitoes.RNA was eluted in 45 μl of RNase-free water and stored at -70˚C.A Qiagen QuantiTect Reverse transcription Kit (cat no.205314) was first used to remove any genomic DNA co-purified during the RNA extraction protocol and then reverse transcription was performed with random primers to generate cDNA from all RNA extracts according to manufacturer's instructions.Confirmation of species identification was undertaken using an internal transcribed spacer 1 (ITS1) realtime PCR assay that discriminates between Ae. aegypti and Ae.albopictus 26 .Cycling conditions were as follows: 95°C for 15 minutes followed by 40 cycles of 95°C for 10 seconds, 55°C for 30 seconds, 72°C for 20 seconds.Amplification was followed by a dissociation curve (95°C for 10 seconds, 65°C for 60 seconds and 97°C for 1 second) to ensure the correct target sequence was being amplified.
Arbovirus screening included the major arboviruses of public health importance, suspected or having the potential of being transmitted by Ae. aegypti / Ae. albopictus in the Caribbean: DENV, CHIKV, ZIKV, YFV and MAYV (Supplementary Table 1, Extended data) 25 .In addition, Pan-Flavivirus PCR screening was undertaken that allows simultaneous detection of numerous flaviviruses using a conserved region of the NS5 gene 27 .PCR reactions for all assays except ZIKV were prepared using 5 μl of Qiagen QuantiTect SYBR Green Master mix (cat no.204145), a final concentration of 1 μM of each primer, 1 μl of PCR grade water and 2 μl template cDNA, to a final reaction volume of 10 μl.Prepared reactions were run on a Roche LightCycler® 96 System (product no.05815916001) and PCR primer sequences are described in Supplementary Table 1 (see Extended data) 25 .Cycling conditions were as follows: DENV, CHIKV, YFV and MAYV -95°C for 15 minutes followed by 40 cycles of 95°C for 10 seconds, 55°C for 30 seconds, 72°C for 30 seconds; Pan-Flavivirus -95°C for 15 minutes followed by 50 cycles of 95°C for 10 seconds, 60°C for 10 seconds, 72°C for 10 seconds.Amplification was followed by a dissociation curve (95°C for 10 seconds, 65°C for 60 seconds and 97°C for 1 second) to ensure the correct target sequence was being amplified.ZIKV screening was undertaken using a probe-based assay 28 that used 5 μl of Qiagen QuantiTect Probe PCR Kit (cat no.204345), a final concentration of 1 μM of each primer, 1 μl of PCR grade water and 2 μl template cDNA, to a final reaction volume of 10 μl.Cycling conditions for ZIKV were 95°C for 15 minutes followed by 40 cycles of 95°C for 10 seconds, 55°C for 30 seconds, 72°C for 30 seconds.PCR results were analysed using the LightCycler® 96 software (Roche Diagnostics).Synthetic long oligonucleotide standards of PCR products were generated in the absence of biological virus cDNA positive controls and each assay included negative (no template) controls.

Sanger sequencing and phylogenetic analysis
Pan-Flavivirus PCR products were submitted to Source Bio-Science (Source BioScience Plc, Nottingham, UK) for PCR reaction clean-up, followed by Sanger sequencing to generate both forward and reverse reads.Sequencing analysis was carried out in MEGA7 29 as follows.Both chromatograms (forward and reverse traces) from each sample were manually checked, edited, and trimmed as required, followed by alignment by ClustalW and checking to produce consensus sequences.Consensus sequences were used to perform nucleotide BLAST (NCBI) database queries.Maximum Likelihood phylogenetic trees were constructed from Sanger sequences as follows.The evolutionary history was inferred by using the Maximum Likelihood method based on the Tamura-Nei model 30 .The tree with the highest log likelihood is shown.The percentage of trees in which the associated taxa clustered together is shown next to the branches.Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Joining and BioNJ algorithms to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach, and then selecting the topology with superior log likelihood value.The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.Codon positions included were 1st+2nd+3rd+Noncoding.All positions containing gaps and missing data were eliminated.The phylogeny test was by Bootstrap method with 1000 replications.Evolutionary analyses were conducted in MEGA7 29 .

Statistical analysis
Count data analysis was conducted using a generalized linear model because the response variable (mosquito counts) had a non-normal error distribution.Models were run using Stata MP (version 14, Stata Corp, College Station, TX, USA).Both Poisson and negative binomial link functions were used in analysis, with the superior model identified from visual inspection of fits (Supplementary Figure 2, Extended data) 25 .A univariate analysis included elevation, humidity and temperature as continuous explanatory variables, and urbanisation level (urban or semi-urban) as a factor.Incident rate ratios (IRRs) and corresponding 95% confidence intervals were calculated.

Ethical approval
The study protocol was reviewed and approved by the Research Governance & Integrity Office of the London School of Hygiene and Tropical Medicine (LSHTM ethics no.9308).

Results
A total of 3,701 adult mosquitoes were collected across the island of Saint Lucia over a four-week period using BG Sentinel 2 traps (Table 1).Culex was the dominant genus, comprising 78.7% of the total mosquitoes collected and the remaining 21.3% were morphologically identified as species within the Aedes genus.No Ae.albopictus females were collected in any of the locations despite traps being placed in diverse ecological settings.Ae. aegypti adults were collected  1; Supplementary Figure 1, Extended data) 25 .In contrast, low numbers of Ae. aegypti were collected using the permanent trap at Riviere Doree Anglican Primary School, with Ae. aegypti comprising 0.5% (n=5) of the collection and an average of 0.13 female mosquitoes per 24 hours of trapping.
A generalized linear model (GLM) was used to analyse the associations between the counts of female Ae.aegypti (combining counts from both the temporary traps and permanent traps) and four independent variables: peak daily temperature, peak daily humidity, trap elevation and ecological zone (semi-urban or urban).Plotting count frequencies against alternative, competing models, assuming either a Poisson or a negative binomial distribution, clearly demonstrated the superiority of a negative binomial model in fitting the data distribution (Supplementary Figure 2, Extended data) 25 .Exponentiation of the coefficients resulting from a GLM (negative binomial family) produced the IRR associated with the independent variables.Here, IRR can be interpreted as the ratio of counts per trap-day associated with the tested variable.These are described along with 95% confidence intervals in Table 2.No significant association was found with temperature.Because previous studies have shown a non-monotonic association between Ae. aegypti and temperature (i.e.Ae. aegypti thrive at a non-trivial optimal temperature) 31 , we subsequently attempted to fit a more complex (quadratic) function between these variables but this did  not improve model fit (p=0.850).Higher counts were significantly associated with higher elevation although the effect size was small; and semi-urban settings yielded female mosquito counts per trap-day that were five-fold lower than urban settings.We tested for interactions between all covariates but none were found to be significant.We also used a negative binomial mixed effects model to account for the potential additional random effect of trap site and found the same significant associations (Supplementary Table 2, Extended data) 25 .
A sub-sample of adult female Ae.aegypti mosquitoes (n=381) collected from BG traps were screened for arboviruses.No evidence was seen for infection of the major medically important arboviruses that have historically been transmitted by Ae. aegypti.However, the presence of a Flavivirus closely related to cell-fusing agent virus (CFAV) was detected in  3 and Supplementary Table 3, Extended data) 25 .We also detected this Flavivirus in adult females that had been reared from larval collections (Supplementary Table 3, Extended data) 25 .Phylogenetic analysis indicated the Flavivirus sequences obtained were identical to, or closely matched to, CFAV NS5 sequences, particularly sequence obtained from Ae. aegypti specimens collected in Guadeloupe in 2016 (GenBank accession number: LR694081), revealing they are potentially forming a Caribbean strain of CFAV (Figure 3 and Jeffries et al. qPCR and sequencing data file, Underlying data) 25 .This phylogenetic analysis also confirmed these Flavivirus sequences cluster with other insectspecific flaviviruses (ISFs) and are separate from pathogenic flaviviruses such as DENV, ZIKV and YFV (Figure 3).

Discussion
Entomological indices including the abundance of adult mosquitoes are often used to assess the risk of disease transmission and this, in turn, influences vector control strategies.The lack of surveillance studies, to our knowledge, for major vectors of arboviruses of public health importance on the island of Saint Lucia needed to be addressed, given the recent outbreaks of arboviruses such as DENV, CHIKV and ZIKV in the Caribbean and surrounding regions.The principal vector of these arboviral diseases, Ae. aegypti, is highly invasive and is now present in much of the Americas including the USA 5 .In this study, we collected adult mosquitoes to determine the geographical distribution of Ae. aegypti across the island of Saint Lucia and to determine any correlation with environmental variables.BG Sentinel 2 traps were selected as these traps were redesigned to provide increased durability in field conditions and were recently shown to be effective at trapping Aedes species 32 .The durability of traps was particularly important for the four permanent traps that were used for approximately 24 days.Adult collections indicate that Ae. aegypti is present throughout the island of Saint Lucia and population densities are significantly higher in urban areas compared to semi-urban or rural settings.We also demonstrated that adult counts were positively correlated to elevation.
The comparatively larger number of mosquitoes collected at Soufriere hospital is likely due to this trapping location having several optimal characteristics for Ae.aegypti.Firstly, the trap was placed in the hospital garden which is located in the centre of an urban area with a high abundance of available human hosts for blood feeding.Secondly, breeding sites resulting from the accumulation of rainwater in animal troughs and deep gutters lining the edges of roads were in close proximity.
The trapping of a high number of Ae. aegypti (47 females and 10 males) during a 24-hour period from Dugard using a trap placed indoors in a semi-urban area suggests an interesting behavioural observation.The biology and behaviour of Ae. aegypti in the Caribbean has not been extensively studied but previous work on a Trinidad strain using human landing catches revealed the periodicity consisted of 90% arriving during daylight and twilight and 10% during the night 33 .This study included both urban and rural sites and a consistently larger number of mosquitoes were collected outside vs. inside houses.Light intensity was also significantly correlated with mosquito landing patterns 33 .The trapping of Ae. aegypti inside houses in Saint Lucia could indicate a change in behavior, with mosquitoes biting indoors during the night in houses with lights on (an anecdotal observation that occurred during our study).Indoor resting of Ae. aegypti has recently been documented in Mexico 34 , which has implications for control methods.Other studies in the Caribbean have also shown that high temperatures in open environments can result in Ae. aegypti breeding in underground sites 35 and indoor oviposition has been demonstrated 36,37 .
Confirmation of the presence of Ae. aegypti on the island of Saint Lucia is not particularly surprising given this species is widespread throughout the Caribbean and is now widespread in the Americas 4,5 .The association with urban environments in Caribbean Islands is seen with the most common breeding sites being drums/barrels, uncovered tanks and cisterns, brick holes, flower pots, used tyres and utility manholes 38 .Saint Lucia now provides the ideal environment for Ae.aegypti due to recent changes in the climate.The El Niño period in 2009 -2010 introduced dry hot periods and provided an environment that was not conducive for mosquito production.Water conservation has become a critical issue for Saint Lucia and the majority of the water supply comes from surface runoff collected in rivers, streams and dams.Rain water is collected and stored haphazardly and inappropriately in various containers such as water tanks, drums, and buckets, creating ideal breeding grounds for this species 1 .This study was undertaken during the commencement of the wet season with the average rainfall in June and July 2015 being 37.1mm and 175.8mm, respectively.This indicates that greater mosquito abundance is likely throughout later stages of the wet season and follow-up studies should be undertaken to determine this.
Clearly climatic patterns resulting in unpredictable rainfall will provide ideal breeding grounds (unpolluted water in artificial and natural containers) for Ae.aegypti in Saint Lucia 35 .
Aedes albopictus was not identified in the mosquitoes collected in our study despite many traps being set in or near forested areas.The range of Ae. albopictus has expanded to Europe, USA and many South American countries 4 .In the Caribbean, this species was first reported in the Dominican Republic 39 and has also been recently found in Jamaica 40 and implicated in CHIKV transmission in Haiti in the eastern Caribbean 41 .Therefore, recent outbreaks of CHIKV on Saint Lucia and neighbouring Caribbean islands suggest that there might be a possibility that Ae. albopictus may also play a role in the spread of the disease.With Ae. albopictus present in the US to the north and the Cayman Islands to the south, Saint Lucia is clearly considered at risk for establishment of Ae. albopictus 42 .
Aedes albopictus has also been shown to harbour Eastern equine encephalitis virus (EEEV) 43 , highlighting the potential transmission risk of additional arboviruses.The traditional ways of importing Ae. albopictus through the trade of tyres is also a possible source of introduction for this species.
Although Saint Lucia has signed onto the International Health Regulations (IHR) 2005, to prevent and control the international spread of disease, port surveillance systems are not fully implemented and might not be sufficient to monitor containers present on ships and ensure that they are fumigated before they arrive in port.Saint Lucia is also faced with the problem of tyre disposal where there is no functional shredding equipment, which is of great concern, particularly so because tyres in landfills are in close proximity to urban communities.
The detection of Flavivirus sequences identical, or closely related to CFAV in diverse ecological populations of Ae. aegypti across the island of Saint Lucia suggests the potential for undiscovered and widespread viruses in the Caribbean which may have implications for vector potential and transmission of pathogenic arboviruses.A large study was undertaken in Trinidad, screening more than 185,000 mosquitoes representing 46 species, and 85 different viruses were isolated 44 .The isolation of Mucambo virus (MUCV), (a Venezuelan equine encephalitis complex subtype IIIA), follows a history of isolating Alphaviruses from mosquitoes in Trinidad 45 .More recently, high prevalence of a Phasi Charoen-like virus was reported in Ae. aegypti in Grenada 46 , suggesting a high diversity of unknown viruses are present in Caribbean populations of Ae. aegypti.A potentially novel strain of CFAV was discovered in Ae. aegypti populations from Mexico 47 and CFAV was detected in Ae. aegypti populations from Kenya 48 .The year after our specimens were collected, Ae. aegypti specimens from Guadeloupe were collected which resulted in CFAV sequences being identified in the population from this nearby Caribbean island 49 .Several studies have shown the potential for mosquitospecific viruses to interfere with arboviruses [50][51][52] .Interestingly, CFAV infection significantly enhanced replication of DENV (and vice versa) in Ae. aegypti Aag2 mosquito cells 53 .Furthermore, a study looking at ZIKV replication in Ae. aegypti and Ae.albopictus cell lines suggested that insect-specific viruses (including CFAV) may interfere with ZIKV, DENV and La Crosse virus replication 54 .Further studies are needed to determine if this occurs in field settings by determining the prevalence of mosquito-specific viruses in mosquito populations.Newly described viruses/viral sequences allow a more comprehensive understanding of virus evolution and virus-host interactions and could also contribute to efforts to target both insects and pathogens 55 .
CFAV has been shown to be vertically transmitted in Ae. aegypti lab colonies, suggesting the possibility of using CFAVs and closely related ISFs for control of medically important arboviruses 56 .The presence of insect-specific viruses in Ae. aegypti might be underestimated given a recent study suggested up to 27 insect-specific viruses (23 currently uncharacterized) in populations from Cairns (Australia) and Bangkok (Thailand) 57 .The question remains as to whether insect-specific viruses like CFAV have not yet gained the ability to infect vertebrates and therefore become arboviruses or whether they have lost this ability 58 .Phylogenetic studies focused on the Flavivirus E gene would suggest CFAV is a basal lineage that diverged prior to the separation of mosquito and tick-borne flaviviruses 59 .Our results indicate the presence of CFAV but it has been shown that some Flavivirus genome-integrated sequences can be transcribed and therefore caution must be taken to assume the presence of an active Flavivirus infection 60 .Furthermore, Flavivirus-related integrated DNA sequences were detected in wild Ae.aegypti mosquitoes most likely resulting from preceding infection by the corresponding RNA viruses 61 .As we removed genomic DNA during our reverse transcription step then this is less likely to have been the source of the Flavivirus sequences detected in this study.Caution must be taken as our Flavivirus sequences were short partial NS5 sequences from Pan-Flavivirus PCR products, limiting the phylogenetic analysis possible.However, the matching of our sequences to CFAV NS5 sequence from Guadeloupe (where genome sequences were obtained) 49 adds weight to the potential for a Caribbean lineage of CFAV being present in these Ae.aegypti populations.Full genome sequencing of these new isolates from Saint Lucia would provide much greater confidence on the phylogeny and level of genetic diversity and would allow further comparisons to CFAV infections in geographically diverse Ae. aegypti populations 61 .This is particularly important given insect-specific viruses are now being considered for mosquito control strategies including CFAVs in Ae. aegypti 62 .

Conclusions
The impact of arboviral diseases is increasing due to the expanding geographical range of many mosquito species, particularly Ae. aegypti and Ae.albopictus.As most arboviral diseases occur in sporadic epidemics, vector control options are often limited to the use of insecticides that are becoming less effective due to insecticide resistance.As re-emerging arboviral diseases such as DENV and ZIKV continue to spread geographically, the fight to eradicate or reduce the transmission potential of Ae. aegypti is increasing in importance.Although we found no evidence of human arboviruses through screening, the detection of sequences which appear to be CFAV, an insect-specific Flavivirus, warrants further investigation, particularly given that these viruses can affiliate phylogenetically with flaviviruses that infect vertebrates and they may influence the transmission potential of pathogenic flaviviruses 60 .Outbreaks of arboviral diseases, including DENV, CHIKV and ZIKV, have a history of occurring in small tropical islands.ZIKV emerged for the first time outside of Africa and Asia in Yap State in Micronesia and then a large outbreak in French Polynesia was followed by transmission in other Pacific islands 63 .Small Islands Developing States (SIDS) and territories such as Saint Lucia are particularly vulnerable to arboviral disease outbreaks for several reasons 64 .Natural disasters are more frequent and these change the geographical landscape, allowing rapid mosquito proliferation.SIDS often lack safe water supplies and sanitation and local governments have limited resources to undertake vector control and manage outbreaks.An increasing ability for travel between SIDS and continental regions facilitates the spread of arboviruses to previously unexposed populations.For these reasons, surveillance strategies need to be monitored, risk areas need to be mapped out and epidemic trends recorded for predicting future outbreaks.For the Caribbean island of Saint Lucia, further research is needed to determine the diversity of current mosquito species and investigate the presence of viruses within these species, and this should also be extended to the neighbouring smaller Caribbean islands.-Supplementary Figure 2 (The observed proportions along with the Poisson and negative binomial probabilities for the count type variable using 'nbvargr' function in Stata).

Extended data
-Supplementary Figure 3 (qRT PCR fluorescence and melting temperatures of Pan-Flavi NS5 positive samples)

Lícia Natal Fernandes
Institute of Tropical Medicine, University of São Paulo, São Paulo, Brazil The article is about an entomologycal survey conducted in Sant Lucia, a Caribbean island.Association between some environmental parameters and the counts of female Aedes aegypti mosquitoes was performed and a screening of arbovirus was also carried out.Higher counts of Aedes aegypti were significantly associated with elevation and mosquito counts in semi-urban settings were lower than urban settings.Nucleotide sequences of cell fusing agent virus, a insect specific flavivirus, was detected with a high prevalence in Aedes agypti pools.It was very important to perform such study considering the island where the study took place has the potential for occurrence of arbovirus outbreaks.This is a well written article and only minor changes are suggested, as follows.
In the first paragraph authors say: "YFV is also transmitted by Ae. aegypti and can result in large urban outbreaks and rapid spread to distant locations".In order to avoid confusion I would say something like: "YFV is also transmitted by Ae. aegypti in the urban circle of transmission, what can result in large urban outbreaks and rapid spread to distant locations where population has not been vaccinated yet".
In my opinion, citation of Figure 1 should be removed from the last paragraph of the introduction chapter because its title mentions the traps, which are only described in the methods.Therefore, the citation should be placed only in methods.
It would have been more appropriate to use a taxonomic key to separate mosquitoes Aedes from other mosquitoes.However I believe results are correct as the authors confirmed the morphology results by molecular biology based on the ITS1 gene.
In the Results, authors say "No Aedes albopictus females were collected", what gives the impression that males might have been collected.So I think it would be more appropriate to say that no Aedes albopictus were collected.
It's not clear how many adults from collected larvae were analysed, if they were processed in pools or individually.Also, it would be very interesting to add information on the larvae collection (frequency of collection and kinds of breeding sites sampled -natural or artificial or both).
In the last paragraph of the results, authors begin saying the number of mosquitoes analysed (381), but when they talk about the positive results for PCR, it seems that information is given in number of mosquito pools.It would be important to have the number of mosquito pools analysed in the text, together with the number of positive pools.Also, it should be informed that positivity is given in number of pools and not in number of individual mosquitoes, if that really is the case.As mosquitoes were processed in pools, it's not possible to know the real number of positive mosquitoes, therefore authors should consider to calculate and inform the minimum infection rate.
One limitation of the study is the fact that very small nucleotide sequences were analysed, nonetheless this is very well discussed by the authors, that even comment on the importance of sequencing the entire genome of the virus detected.
Authors found significantly association between counts of Aedes aegypti and elevation although do not discuss it.If any explanation for that is available, it would be very important to be in the text.
-Figure 1 -extended material says: "yellow pins represent semi-permanent traps".Is it correct or would it be "permanent traps" instead of "semi-permanent traps"?

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?I cannot comment.A qualified statistician is required.

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Entomology / Molecular biology applied to detection of protozooans and viruses in mosquitos.

I confirm that I have read this submission and believe that I have an appropriate level of
expertise to confirm that it is of an acceptable scientific standard.
Reviewer Report 05 January 2021 https://doi.org/10.21956/wellcomeopenres.18175.r41856 © 2021 Villinger J.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Jandouwe Villinger
International Centre of Insect Physiology and Ecology (icipe), Nairobi, Kenya The issues of the original version of the manuscript have now been sufficiently addressed.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Molecular entomology of mosquito-and tick-borne diseases.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Version 1
Reviewer Report 23 October 2020 https://doi.org/10.21956/wellcomeopenres.17585.r40686 © 2020 Neira M. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Marco Neira
The Cyprus Institute, Nicosia, Cyprus Major Revisions: By comparing Table 1 with the mosquito trapping data available on the supplementary files, it would seem that most locations sampled with temporary traps were only sampled for a single day, with a single trap each, while locations sampled with permanent traps had a much higher number of trapping nights.If this was the case, the authors should address whether this style of sampling is appropriate for performing the type of statistical analysis presented in the paper.In any case, I suggest they include in Table 1 the number of trapping nights for each location.
The data shown in Figure 2 and Table 1 suggest that one of the permanent trapping sites (Soufriere Hospital) consistently produced much higher yields than any of the other sites, including those sampled with temporary traps.However, the authors don't discuss potential explanations for this phenomenon, or the ways in which this fact could (or not) create bias on their data analysis.I suggest they revise the discussion addressing these observations.
As the authors correctly point out, the detection of flavivirus-like RNA in a mosquito could very well be the product of a transcribed cell silent agents (CSA) integrated in the insect's genome sometime in the past.While it is certainly possible that a novel, extant virus could be the source of such CSA, it is also possible that the sequences obtained by the authors represent a CSA originating from a known virus and subsequently mutated after insertion in the mosquito genome.Alternatively, the sequences could correspond to a CSA derived from an extinct virus.In any case, I think that in the absence of a full viral genome, the authors should be more careful about their claim of having identified a novel insect-specific virus.
The 'Conclusions' section discusses aspects such as the spread of arboviral diseases, the limited options available for vector control, the history of Zika outbreaks (among other), which are at best tangentially related to the paper's original results.I recommend that the authors re-write this section, presenting conclusions that are actually related to their original work.

Minor Revisions:
Throughout the paper, please italicize scientific names.

Introduction:
The authors state that "22 countries and territories have now been identified to have autochthonous transmission", which could be interpreted as being a global tally of countries reporting Zika transmission.However, this number (a) corresponds exclusively to the Americas, and (b) represents an outdated figure.The latest figures available from the Pan-American Health Organization (available here) indicate that just within the Americas, 48 countries and territories have reported mosquito-borne Zika transmission.Considering that earlier on the paper the authors provide global data for the epidemiology of dengue, I would recommend revising this figure to reflect updated global data on Zika distribution.

Are sufficient details of methods and analysis provided to allow replication by others? Partly
If applicable, is the statistical analysis and its interpretation appropriate?I cannot comment.A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? No
Competing Interests: No competing interests were disclosed.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
statistically increased mosquito count for higher elevations and for urban (vs semiurban) areas but no statistical association between counts and either humidity or temperature.As these significant associations are the same, the results of this analysis are now presented in Supplementary Table 2, extended data).We have also now included the trapping nights for each location in Table 1.
The data shown in Figure 2 and Table 1 suggest that one of the permanent trapping sites (Soufriere Hospital) consistently produced much higher yields than any of the other sites, including those sampled with temporary traps.However, the authors don't discuss potential explanations for this phenomenon, or the ways in which this fact could (or not) create bias on their data analysis.I suggest they revise the discussion addressing these observations.
Soufriere Hospital did indeed produce higher yields and we have added three sentences at the start of the second paragraph in the discussion to show how the trapping location had several optimal characteristics for Ae.aegypti.
As the authors correctly point out, the detection of flavivirus-like RNA in a mosquito could very well be the product of a transcribed cell silent agents (CSA) integrated in the insect's genome sometime in the past.While it is certainly possible that a novel, extant virus could be the source of such CSA, it is also possible that the sequences obtained by the authors represent a CSA originating from a known virus and subsequently mutated after insertion in the mosquito genome.Alternatively, the sequences could correspond to a CSA derived from an extinct virus.In any case, I think that in the absence of a full viral genome, the authors should be more careful about their claim of having identified a novel insect-specific virus.
We agree and have modified the end of our discussion to provide a more balanced view.Additional CFAV sequences which became publicly available after the original analysis have now enabled a re-analysis to be carried out which has been incorporated into the latest version of the manuscript, with relevant areas updated.
The 'Conclusions' section discusses aspects such as the spread of arboviral diseases, the limited options available for vector control, the history of Zika outbreaks (among other), which are at best tangentially related to the paper's original results.I recommend that the authors re-write this section, presenting conclusions that are actually related to their original work.
We agree and have modified our conclusions to include some sentences on insectspecific flaviviruses.However, confirmation of a wide distribution of Ae. aegypti on the island of St Lucia as part of our entomology survey and screening for human arboviruses undertaken does (in our opinion) relate to the possibility for future arboviral outbreaks.

Minor Revisions:
Throughout the paper, please italicize scientific names.
We have done this.

Introduction:
The authors state that "22 countries and territories have now been identified to have autochthonous transmission", which could be interpreted as being a global tally of countries reporting Zika transmission.However, this number (a) corresponds exclusively to the Americas, and (b) represents an outdated figure.

Methods:
The authors should provide a clearer summary of their trapping strategy.In the current text, they state that trapping was performed using four permanent traps and four temporary traps, for what can be understood as a total of eight traps per each 24-hour period.However, mosquito trapping data available on the supplementary files suggests that on most dates only six or less traps were used.

○
Yes that is correct and this was due to logistical difficulties accessing all sites across the island so not all trapping days had four temporary traps running synchronous with the four longer-term permanent traps.Our methods have now been modified for clarity.
Please specify which taxonomic key was used to separate mosquito genera.If no specific key was used, please provide a complete list of characters used to sort specimens by genus.Please specify what kind of primers were used for reverse-transcription reaction.

○
We have added in the methods that random primers are used with the QuantiTect Reverse transcription Kit Discussion: In the first paragraph of the discussion, the sentence "The principle vector of these arboviral diseases..." should be changed to "The principal vector of these arboviral diseases...".We agree that although we included all relevant sequences that we were able to obtain at the time of original analysis, on which we based our original results and conclusions, there have since been further CFAV sequences made publicly available.We have therefore undertaken a reanalysis which now indicates our sequences are identical or very closely related to the CFAV sequence from Guadeloupe (From Baidaliuk et al. 2020) 1 and alongside this sequence may potentially form a Caribbean lineage of CFAV.We have adjusted our manuscript in all relevant areas to reflect this and replaced the previous Figure 3 and 4 phylogenetic trees with a single updated phylogenetic tree (now Figure 3) from the re-analysis which incorporates additional more recently available sequences of CFAV for a more comprehensive comparison.
We have now also included our % nucleotide homologies following re-analysis, in addition to further sequencing analysis information, within the Jeffries et al. qPCR and sequencing analysis file in the Underlying data.
As we sequenced products from a Pan-flavivirus PCR assay which detects a wide range of viruses 2 , we obtained sequences less than 200bp and therefore Genbank unfortunately will not accept these.However, as stated in the underlying data section, raw sequence data is available at https://doi.org/10.17605/OSF.IO/DACKS.

Minor comments:
The authors should ensure that conventions are followed and that genus and species names are italicized (double check all 'Ae.aegypti' and 'Alphavirus').

○
We have checked and corrected our manuscript for these conventions.
It is not clear to me why Figure 3 is required as the same information is fully represented in Figure 4.I think that Figure 3 can be removed

○
We agree and as mentioned above we have removed both the previous Figure 3 and 4

Figure 1 .
Figure 1.Sampling locations of the longer-term permanent mosquito traps on the island of Saint Lucia used throughout the duration of the study (July 2015).Maps were produced using ArcGIS® software (version 10.4.1) by Esri.

Figure 2 .
Figure 2. Population dynamics of local female Ae.aegypti mosquitoes caught from the four longer- term traps positioned in field sites detailed in Figure 1.

Figure 3 .
Figure 3. Maximum Likelihood molecular phylogenetic analysis of selected Flavivirus NS5 partial sequences showing the Flavivirus sequences from Saint Lucia Ae. aegypti clustering alongside other cell-fusing agent virus (CFAV) sequences, within the insect-specific flaviviruses group.Sequences generated in this study are shown with node markers, circular markers for Flavivirus sequences obtained from adult female Ae.aegypti and diamond markers for those obtained from larval collections.The tree with the highest log likelihood (-2701.53) is shown.The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.The analysis involved 62 nucleotide sequences.There was a total of 166 positions in the final dataset.

○
In our methods we have the following sentences: 'Trapped mosquitoes were collected, killed on ice for genera morphological identification to identify individuals belonging to the Aedes genus using basic defining characteristics including patterns of light and dark scales on the abdomen and thorax and alternating light and dark bands on the legs.'We then used ITS1 real-time PCR assays that discriminates between Ae. aegypti and Ae.albopictus.The legend in Figure1makes reference to an inset showing a "representative BG Sentinel 2 trap placed in the Des Cartier rainforest".However, the actual figure does not contain any inset showing a representative trap.Thank you for pointing this out -this was error from a previous figure version and we have removed this reference to the trap from the figure legend.

○
Thank you for spotting this error and we have amended thisCompeting Interests: No competing interests were disclosed.Reviewer Report 14 July 2020 https://doi.org/10.21956/wellcomeopenres.17585.r39242© 2020 Villinger J.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Jandouwe VillingerInternational Centre of Insect Physiology and Ecology (icipe), Nairobi, Kenya This is a generally well written manuscript on a study surveying mosquito viruses in St Lucia.Although no arboviruses were detected, they found a 19% prevalence of Aedes-specific flaviviruses.There is a general lack in the literature of information on individual prevalence of insect-specific flaviviruses in mosquito populations.Therefore, their infection rate data could be highlighted more in the paper My main concern is that the authors are likely over interpreting their finding of a new Aedesspecific flavivirus (AeFV) sequence as a novel insect-specific flavivirus.Just a few CFAV sequences and only one other AeFV sequence are included in the phylogenetic analyses.CFAV is part of a clade of AeFVs that are broadly distributed and show variation.The Saint Lucia AeFV seem to fall within the expected variation of AeFVs, if not even CFAVs.At the very least the sequence they found simply represents a new strain of AeFV associated with Aedes aegypti in St. Lucia.Percent nucleotide homologies are not given, so I cannot fully judge, but these are likely a St. Lucia lineage of CFAV.It would also be useful if there were GenBank accession numbers associated with the study's ISF sequences so that their phylogenies can be independently assessed and the sequences can be used in future analyses of ISF variation/evolution.Please consider the recent papers: Patterson et al. (2020) 1 .Thomas Walker, London School of Hygiene & Tropical Medicine, Keppel Street, London, UK Response to reviewer 1 Many thanks Jandouwe for your thoughtful review.Our responses are shown below in bold:My main concern is that the authors are likely over interpreting their finding of a new Aedes-specific flavivirus (AeFV) sequence as a novel insect-specific flavivirus.Just a few CFAV sequences and only one other AeFV sequence are included in the phylogenetic analyses.CFAV is part of a clade of AeFVs that are broadly distributed and show variation.The Saint Lucia AeFV seem to fall within the expected variation of AeFVs, if not even CFAVs.At the very least the sequence they found simply represents a new strain of AeFV associated with Aedes aegypti in St. Lucia.Percent nucleotide homologies are not given, so I cannot fully judge, but these are likely a St. Lucia lineage of CFAV.It would also be useful if there were GenBank accession numbers associated with the study's ISF sequences so that their phylogenies can be independently assessed and the sequences can be used in future analyses of ISF variation/evolution.Please consider the recent papers: Patterson et al. (2020), Baidaliuk et al. (2020)

thanks for pointing this out and we have updated our figures based on the last available WHO update (July 2019).
The latest figures available from the Pan-American Health Organization (available here) indicate that just within the Americas, 48 countries and territories have reported mosquito-borne Zika transmission.Considering that earlier on the paper the authors provide global data for the epidemiology of dengue, I would recommend revising this figure to reflect updated global data on Zika distribution.One potential source is the World Health Organization's Zika Epidemiology Update (available here).The authors state that "widespread insecticide resistance has developed in Ae. aegypti, including high pyrethroid resistance rates in South America and further north in the Caribbean".To back up this statement, the authors cite a single paper focused on resistance found on a specific location in Brazil, and one paper focusing on a location in the Caribbean.I would recommend that the authors increase the number of citations to better represent the actual spread of pyrethroid resistance in the region.