Rapid Surveillance for Vector Presence (RSVP): Development of a novel system for detecting Aedes aegypti and Aedes albopictus

Background The globally important Zika, dengue and chikungunya viruses are primarily transmitted by the invasive mosquitoes, Aedes aegypti and Aedes albopictus. In Australia, there is an increasing risk that these species may invade highly urbanized regions and trigger outbreaks. We describe the development of a Rapid Surveillance for Vector Presence (RSVP) system to expedite presence- absence surveys for both species. Methodology/Principal findings We developed a methodology that uses molecular assays to efficiently screen pooled ovitrap (egg trap) samples for traces of target species ribosomal RNA. Firstly, specific real-time reverse transcription-polymerase chain reaction (RT-PCR) assays were developed which detect a single Ae. aegypti or Ae. albopictus first instar larva in samples containing 4,999 and 999 non-target mosquitoes, respectively. ImageJ software was evaluated as an automated egg counting tool using ovitrap collections obtained from Brisbane, Australia. Qualitative assessment of ovistrips was required prior to automation because ImageJ did not differentiate between Aedes eggs and other objects or contaminants on 44.5% of ovistrips assessed, thus compromising the accuracy of egg counts. As a proof of concept, the RSVP was evaluated in Brisbane, Rockhampton and Goomeri, locations where Ae. aegypti is considered absent, present, and at the margin of its range, respectively. In Brisbane, Ae. aegypti was not detected in 25 pools formed from 477 ovitraps, comprising ≈ 54,300 eggs. In Rockhampton, Ae. aegypti was detected in 4/6 pools derived from 45 ovitraps, comprising ≈ 1,700 eggs. In Goomeri, Ae. aegypti was detected in 5/8 pools derived from 62 ovitraps, comprising ≈ 4,200 eggs. Conclusions/Significance RSVP can rapidly detect nucleic acids from low numbers of target species within large samples of endemic species aggregated from multiple ovitraps. This screening capability facilitates deployment of ovitrap configurations of varying spatial scales, from a single residential block to entire suburbs or towns. RSVP is a powerful tool for surveillance of invasive Aedes spp., validation of species eradication and quality assurance for vector control operations implemented during disease outbreaks.


Methodology/Principal findings
We developed a methodology that uses molecular assays to efficiently screen pooled ovitrap (egg trap) samples for traces of target species ribosomal RNA. Firstly, specific real-time reverse transcription-polymerase chain reaction (RT-PCR) assays were developed which detect a single Ae. aegypti or Ae. albopictus first instar larva in samples containing 4,999 and 999 non-target mosquitoes, respectively. ImageJ software was evaluated as an automated egg counting tool using ovitrap collections obtained from Brisbane, Australia. Qualitative assessment of ovistrips was required prior to automation because ImageJ did not differentiate between Aedes eggs and other objects or contaminants on 44.5% of ovistrips assessed, thus compromising the accuracy of egg counts. As a proof of concept, the RSVP was evaluated in Brisbane, Rockhampton and Goomeri, locations where Ae. aegypti is considered absent, present, and at the margin of its range, respectively. In Brisbane, Ae. aegypti was not detected in 25 pools formed from 477 ovitraps, comprising % 54,300 eggs. In Rockhampton, Ae. aegypti was detected in 4/6 pools derived from 45 ovitraps, comprising % 1,700 eggs. In Goomeri, Ae. aegypti was detected in 5/8 pools derived from 62 ovitraps, comprising % 4,200 eggs. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
In Australia, quarantine authorities intercept both species at international seaports and airports [16,25], with the frequency of detections increasing dramatically since 2012. However, there is a concurrent threat of range expansion from endemic Queensland populations. Aedes aegypti occurs throughout most of Queensland, although dengue outbreaks only occur in north Queensland [26][27][28][29]. Aedes albopictus has not colonized the Australian mainland following the rapid invasion of island communities of the Torres Strait, north Queensland [16,30], largely due to suppression programs at transport hubs [31].
Importantly, both species are considered to be absent from southeast (SE) Queensland (population 3.4 million), where % 70% of the Queensland population reside. This status is not based on systematic entomological baseline monitoring, but rather on the lack of local disease transmission following the importation of ZIKV, DENV and CHIKV in viremic travellers, coupled with negative results from ad hoc larval surveys and trapping of peridomestic mosquitoes. Southeast Queensland is considered to be receptive to invasion by both species. Indeed, Ae. aegypti was present in this region until the mid-1950s [29], whilst predictive models indicate the eastern seaboard of Australia could be colonized by Ae. albopictus [25,32]. Receptivity is considered to be increasing [33,34], partly due to the recent proliferation of water-storage containers, such as rainwater tanks [35]. In the future, ineffective mosquito-proofing of rainwater tanks or rainwater harvesting structures may increase the number of available larval habitats [36][37][38].
Contemporaneous baseline monitoring is essential to increase certainty that vector species are absent in geographies that are vulnerable to invasion [39] to minimize the risk of cryptic disease outbreaks. However, existing surveillance options for peridomestic mosquito species have operational weaknesses. Larval and pupal surveys are labor-intensive [40-42] and can be compromised by inaccessible premises or larval habitats, cryptic containers [43, 44], or timing a survey when mosquito abundance is low [16]. In terms of adult traps, some designs (e.g., Biogents BG-Sentinel trap) are highly sensitive [45] but require electricity and incur significant procurement, servicing, and maintenance costs [46,47]. Novel adult traps (BG-Gravid Aedes Trap) are cheaper but less sensitive [48][49][50] and may be expensive to deploy in extensive arrays over a large spatial scale [6,47,51,52]. Ovitraps provide a cheap and sensitive tool for presence-absence surveys of peridomestic species [17,53,54] but require an investment in time and laboratory resources to count eggs, rear and identify larvae [54,55].
We report the development of a sensitive, easy-to-use, and cost-effective system, which we call Rapid Surveillance for Vector Presence (RSVP), to expedite presence-absence surveys of invasive Aedes species. The RSVP provides a powerful validation tool to confirm a species is absent at various spatial scales (property, suburban block, town or region), within an eradication program, and potentially as a quality assurance measure for vector control strategies implemented in response to ZIKV, DENV or CHIKV outbreaks.

Methods
Development and optimization of molecular assays to detect Ae. aegypti and Ae. albopictus in ovitrap samples Eggs were hatched overnight by submerging ovistrips in 200 mL of Milli-Q water in rectangular 750 mL polyethylene containers, to which % 10 mg of brain-heart powder was added to stimulate hatching. Ovistrips were removed and larvae extracted from the water via a vacuum filtration protocol. Specifically, the membrane and support pad of a MicroFunnel 300 filter funnel (Pall Life Sciences, Ann Arbor, MI, USA) was replaced with an FTA card (GE Healthcare Life Sciences, PA, USA) trimmed to fit the base of the funnel. The filter funnel was then placed on top of a vacuum flask. The water containing the larvae was gently agitated before it was poured into the cylinder of the filter funnel and a vacuum applied. The filter funnel and hatching container were rinsed with Milli-Q water until there were no larvae visible. The cards were then removed from the base of the funnel and air-dried overnight at room temperature.
The protocol of Hall-Mendelin et al. [56] was used to prepare the FTA cards and dried larvae for nucleic acid extraction. Briefly, the cards were cut into 4-5 strips and placed in a 5 mL vial containing 1 mL of Milli-Q water. Vials were placed on wet ice and vortexed every 5 min for 15 sec for a total of 20 min. The cards were then placed in a 5 mL syringe from which the plunger had been removed. The plunger was used to squeeze the liquid from the cards into a 2 mL vial. Nucleic acids were extracted from 140 μL of the eluate using the Bio Robot Universal System (Qiagen, Hilden, Germany) and the QIAamp Virus BioRobot MDx Kit (Qiagen, Clifton Hill, Australia) according to the manufacturer's instructions.
Real-time TaqMan reverse transcription-polymerase chain reaction (RT-PCR) assays were used to detect target Ae. aegypti and Ae. albopictus against a background of endemic species. The Ae. aegypti ribosomal RNA was detected using real-time TaqMan RT-PCR with forward GCAGTCAGATGTTTGATACCGC and reverse GGTTGACGTATTATCAGGTCACACTA primers at 500 nM, and probe FAM-TGGGCGCCTCGGTGTCCCG-TAMRA at 300 nM. The Ae. albopictus ribosomal RNA was detected using forward CCGACAAGGCAATATGTC and reverse ACGCGTACGGACATTG primers at 300 nM, and probe FAM-TTCCCTCCGATCA GCGAACTC-TAMRA at 200 nM. The cycling conditions consisted of one cycle at 50˚C for 5 min, one cycle at 95˚C for 2 min, and 40 cycles at 95˚C for 3 sec and 60˚C for 30 sec. A positive result, indicating the presence of target species RNA, corresponded to cycle threshold values of 40 cycles. Reaction controls always included a no-template and synthetically generated template samples. The specificity of the 2 TaqMan RT-PCR assays was tested using 4 th instar larval samples of Ae. aegypti and Ae. albopictus, as well as other peridomestic species, including Aedes katherinensis, Aedes notoscriptus, Aedes palmarum, Aedes scutellaris, Aedes tremulus and Culex quinquefasciatus.
Pools of varying sizes were produced to test the sensitivity of the TaqMan RT-PCR assays in detecting Ae. aegypti and Ae. albopictus. The Ae. aegypti eggs used to optimize the sensitivity of the assay were obtained from the F 2-4 generations of a colony originating from Townsville, Australia, while F o Ae. notoscriptus were collected in ovitraps deployed in Brisbane, Australia. The Ae. albopictus were from a colony established from eggs collected from Hammond Island (Torres Strait, Australia) and had been in colony for < 10 generations. Following hatching as described above, single 1 st instar Ae. aegypti or Ae. albopictus larvae were added to batches of 1 st instar Ae. notoscriptus or Ae. aegypti larvae, respectively, to produce pool sizes of 10, 100, 1,000 and 5,000 larvae. In addition, a pool comprised of a single Ae. aegypti or Ae. albopictus 1 st instar larvae, and pools containing only Ae. notoscriptus or Ae. aegypti 1 st instar larvae were produced. The pools were processed using the vacuum filtration method described above and Ae. aegypti and Ae. albopictus detected using the species-specific TaqMan RT-PCR assays.
Field validation of the RSVP as a tool for detection of Ae. aegypti Study sites. Three sites were chosen to represent the trichotomy of Ae. aegypti distribution in Queensland. Brisbane (27˚28'15.6"S, 153˚1'24.4"E) is the state capital (> 1 million population; > 380,000 private dwellings) [57] within SE Queensland with no recent detections of Ae. aegypti. The city provided a large geographical area (> 1,367 km 2 ) to test RSVP in a species absence context. Rockhampton (-23˚22'49"S, 150˚30'21"E) is a small regional city 636 km NW of Brisbane (80,000 population; > 32,000 private dwellings). Aedes aegypti is locally abundant and therefore provided a suitable site for field validation of species presence. Goomeri is a small town (< 500 population, > 200 private dwellings), 235 km NW of Brisbane and represents the ill-defined margin of Ae. aegypti distribution nearest SE Queensland. Aedes aegypti has previously been detected during infrequent larval surveys.
Trapping strategy. Large ovitraps were assembled to maximize visual and olfactory cues [58,59]. Black plastic buckets (9.3L) were provided with an overflow hole and exposed to the outdoor environment for 2 weeks prior to deployment, to remove volatile chemicals that may repel ovipositing females. An 'ovistrip' [60] (380 x 50 mm) made from white Tork Cleaning Cloth (SCA, Stockholm, Sweden) was affixed to the inside wall of each bucket with a fold-back clip. A pellet of compacted alfalfa was added to 3-L of tap water to produce an organic infusion [61]. Black plastic mesh covers (20 mm aperture) were fitted to prevent access by children or domestic animals.
Ovitrap sites included well-shaded urban yards, premises near imported dengue cases, commercial precincts and transport hubs (Fig 1). Ovitraps were deployed with occupier consent in outdoor sites that were adjacent to premises and protected from rain and wind [62]. Brisbane ovitraps were primarily deployed in a routine monitoring context from January to July 2015 and were collected after 2 weeks. Rockhampton ovitraps were deployed in May and June 2015, and were collected after 7 days. Goomeri ovitraps were deployed in February and March 2016 for periods ranging from 11-15 days. Ovistrips were transferred to the local office in individual sealable bags and dried overnight at room temperature. Brisbane ovistrips were pooled each fortnight, whilst additional ad hoc pools were submitted from suburbs where an imported dengue case was notified. Rockhampton ovistrips were pooled according to geographical clusters, or the suburb where an imported case of dengue was notified. Goomeri ovistrips were pooled per each trapping event. All ovistrips were forwarded to Forensic and Scientific Services, Department of Health, Queensland Government, Brisbane, for molecular identification.
Egg quantification using image analyses  For the latter method, an average value of 168 pixels per Ae. notoscriptus egg (n = 25) was calculated.
The script was automated upon selection of a directory and dual results (a spread sheet containing quantification for each image and binary images showing the regions selected by the macro as being egg-covered) placed in a subdirectory. The binary images provided a quality control mechanism to check that the macro functions correctly and egg regions are located accurately.

Results
Sensitivity of the novel real-time TaqMan RT-PCR assays for detection of Ae. and Ae. albopictus The Ae. aegypti TaqMan RT-PCR assay was able to detect single 1 st instar larvae in all pools that contained increasing numbers of Ae. notoscriptus up to a pool size of at least 5,000 ( Table 1). The Ae. albopictus assay was not as sensitive, with the limit of detection being a single 1 st instar larva in a pool of 1,000 Ae. aegypti.

Field validation of the RSVP system
A total of 477 ovitraps were deployed in Brisbane of which 400 (83.9%) collected >75,000 Aedes spp. eggs. The processing of 25 pools, representing an estimated % 54,400 eggs, using the Ae. aegypti TaqMan assay did not detect Ae. aegypti.
A total of 45 ovitraps were deployed in Rockhampton, of which 18 (40.0%) collected an estimated total of % 1,700 Aedes spp. eggs. Aedes aegypti was detected in 4 of 6 pools.
A total of 62 ovitraps were deployed in Goomeri, of which 32 (51.6%) collected an estimated % 4,200 eggs. Aedes aegypti was detected in 5 of 8 pools.

Utility of image analysis for quantifying eggs on ovistrips
Due to an absence of visual contaminants, the eggs on 107 of 193 (55.4%) of ovistrips were able to be accurately quantified using ImageJ. Of the remainder, accretion of contaminants rendered 41 of 193 (21.2%) of ovistrips unsuitable for analysis. The eggs on 45 of 193 (23.3%) ovistrips could be quantified with manipulation, such as removing debris or digitally enhancing the image. Table 1. Identification of single 1st instar Ae. aegypti or Ae. albopictus larvae in pools of 1st instar Ae. notoscriptus or Ae. aegypti larvae, respectively using the TaqMan RT-PCR assays.

Aedes aegypti Aedes albopictus
Species ratio a n b Mean ± SD C t value n Mean ± SD C t value

Discussion
The RSVP system is a powerful and flexible tool for presence-absence surveillance of invasive Aedes (Stegomyia) mosquito species. RSVP utilizes sensitive molecular diagnostics to resolve logistical and time constraints typically associated with ovitrap surveillance. Molecular diagnoses take approx. 48 h, including hatching of eggs, nucleic acid extraction and real-time RT-PCR. This represents an appreciable time saving, considering that 1,000s of larvae from multiple ovitraps can be processed concurrently. In comparison, ovitraps require approx. 7-10 days to rear cohorts to 4 th instar larvae or adults before each specimen is identified by the microscopic examination of key morphological characters. Molecular analyses mitigates the risk of human error in detecting a small number of specimens of an invasive species that are obscured by many 1,000s of larvae from endemic species [64]. Hatching eggs to 1 st instar larvae prior to RT-PCR analysis removes requirements for a specific ovitrap format (design, size, material), ovistrip size [65] or substrate (e.g., cloth, paper, germination paper, wood). Shipment of large numbers of egg samples by postal service is also potentially more logistically viable than couriering water samples for newly developed environmental DNA analysis [66]. When viewed in isolation, this application of molecular diagnostics may seem expensive, with a single test costing approx. AU$50.00. However, compared to morphological identification, the saving in time and labor to perform one molecular diagnosis for 10 pooled samples (or AU$5.00 per ovitrap) more than offsets the cost of performing the molecular assays. Furthermore, the cost to perform each test will decrease as the number of cards tested increases. Molecular diagnostics can detect multiple target species (e.g. Ae. albopictus and Ae. aegypti) once the nucleic acids have been extracted, with only a marginal cost increase to include an additional species. Other real-time TaqMan assays can be developed to identify other invasive species, such as Aedes japonicus and Aedes koreicus, which have recently colonized North America and Europe, respectively [67].
Estimates of egg abundance on each ovistrip will guide the aggregation of ovistrips into pool sizes that are below the TaqMan RT-PCR assay's threshold of detection, and enable comparisons between trap sites or to generate spatial 'heat maps' [7]. Egg counting software has the potential to remove the need for manual counting and improve the accuracy and consistency of subjective rapid visual estimates [68]. We demonstrated that systems developed to count eggs harvested under laboratory conditions [69,70] may not be suitable for rapid analyses of field collections due to greater exposure of ovistrips to contaminants. It should be noted that eggs on any contaminated ovistrips would remain viable for molecular analyses. Pre-trial testing in Brisbane determined that a 2 week ovitrap deployment minimized the accretion of contaminants, such as plant debris and/or bacterial and fungal growth. Whilst the relatively high number of contaminated ovistrips limited its accuracy for automated egg counting in our trials, ImageJ or other image processing software warrants further investigation.
The spatial scale of an RSVP ovitrap network can be adjusted to service early-warning, eradication or quality assurance programs. RSVP was trialled by inexperienced field staff in locations where Ae. aegypti is present (Rockhampton), absent (Brisbane) and near the margin of its geographic range (Goomeri). The simplicity, operational flexibility and the rapidity of diagnosis that RSVP provided was well received by these staff. The RSVP system is currently being translated into an expanded local government program in southeast Queensland. Furthermore, the RSVP is also providing a diagnostic platform for a pilot citizen-science project in southeast Queensland. This pilot will emulate citizen-science programs that monitor invasive mosquitoes in Europe [71] and North America (Invasive Mosquito Project), and may complement community-release programs for novel biological control strategies, such as Wolbachia-based programs [72][73][74].
RSVP may provide a powerful quality assurance tool [75] for ZIKV, DENV or CHIKV responses, either through rapid analyses of eggs from lethal ovitraps [60,76] or non-lethal ovitraps. Vector presence-absence data within small spatial scales (e.g. residential block) [27, 77,78] may facilitate risk-based assessment for persistence of local transmission and direct subsequent control operations toward blocks that are infested. The RSVP has potential to be integrated with the roll-out of Wolbachia-based programs. For example, ovitraps were used to assess the rate of invasion by Ae. aegypti infected with the endosymbiont Wolbachia pipientis in open field releases in north Queensland [78]. RSVP may also provide a tool to facilitate the measurement of the heterogeneity of Ae. aegypti in the assessment of candidate sites before releases [79], or aid the assessment of Wolbachia invasion during and after releases for interventions aimed either at the reduction of vector populations [80][81][82] or for control of ZIKV [83,84], DENV [77,85,86] and CHIKV [87,88]