Assessment of the Effectiveness of Various Adult Mosquito Sampling Methods in a Thickly Populated Urban Slum Settlement - A study from Besant Nagar, Chennai, India

and light trap) approaches were assessed to nd out the appropriate collection method and time which could yield the maximum number of An. stephensi with the existing resources. The study revealed that dawn collections during the early hours as the most suitable time to collect wild An. stephensi in an urban setting. The present study thus would help in chalking out an operationally feasible vector control strategy with the most appropriate methodology, timeframe useful for effective control of vectors.


Abstract Background
In an urban scenario, it is an arduous task to collect adult Anopheles stephensi, unlike the immature forms due to various reasons such as the complex spatial heterogeneity, intricacies in feeding, and resting preferences. Thus it is necessary to have more speci c and accurate assessments of adult vector density by performing various collection methods and timeframes to achieve appropriate and sustainable vector control strategies.

Methods
The study was undertaken in two phases, (i) resting dawn collections in cattle sheds from Jan 2015 to Dec 2016 to assess the possibility of maximum collection of An. stephensi compared to dusk collections done previously and to nd out the best appropriate collection strategy for vector control and (ii) light trap collections from Jan 2016 to Apr 2017 to determine the e ciency and reliability of CDC light trap in sampling the anophelines including the urban malaria vector, Anopheles stephensi in human dwellings besides, other mosquito species. The man-hour density of An. stephensi in cattle sheds were calculated and its infection rate was analyzed by circum-sporozoite ELISA (CS-ELISA). Binary logistic regression analysis was done to ascertain the variables associated with the collection of Anopheles stephensi using a light trap.

Results
The resting collections in cattle sheds indicated that dawn collections yielded more Anopheles stephensi than at dusk. The resting stage female (fully fed, semi-gravid, and gravid) mosquitoes were more in dawn collections than in dusk collections. The CDC-light trap collections revealed that An. stephensi collected indoors were more than outdoor in human dwellings. Further, vector incrimination results observed that 0.56% of the female An. stephensi from cattle sheds were infected in 2014 (dusk), 0.15%, and 0.09% in 2015 and 2016 respectively in dawn collections. Nevertheless, 2.3% of An. stephensi collected by CDC light trap in human dwellings were positive for Plasmodium vivax (Pv210) infection. Binary logistic regression analysis proved that the presence of An. stephensi in human dwellings was signi cantly in uenced by seasons, the number of rooms in the house, number of household members, and also the use of repellents.

Conclusions
The different collection (resting and light trap) approaches were assessed to nd out the appropriate collection method and time which could yield the maximum number of An. stephensi with the existing resources. The study revealed that dawn collections during the early hours as the most suitable time to collect wild An. stephensi in an urban setting. The present study thus would help in chalking out an operationally feasible vector control strategy with the most appropriate methodology, timeframe useful for effective control of vectors.

Page 3/19
Background Malaria remains a life-threatening disease in many tropical and sub-tropical areas, caused by the protozoan parasite Plasmodium, transmitted by female Anopheles mosquitoes, which bite mainly between dusk and dawn. In 2018, India contributed 3.4% of an estimated 228 million cases and 2% of 4.05 lakh deaths globally. India shared the highest (47%) among the total Plasmodium vivax cases worldwide [1]. Tamil Nadu state recorded 3758 cases in 2018, 75% of which was contributed by Chennai [2]. Nonetheless, urban malaria is at low risk compared to rural malaria, accounting for 6-28% of the estimated global annual malaria disease incidence, and also warrants special attention in the perspective of global malaria elimination efforts [3]. Regardless of effective control measures by the Urban malaria scheme (UMS) of the local health department in Chennai, malaria still remains to be a threat due to its persistent nature. In Chennai, malaria is transmitted by the Asiatic urban malaria vector, Anopheles stephensi, which predominantly breeds in overhead tanks, besides, wells and other water storage habitats [4]. The adult vectors predominantly are found resting in cattle sheds and thatched structures [5].
In a real urban scenario, it is a strenuous task to nd and collect adult An. stephensi unlike the immature forms due to the complex spatial heterogeneity, the resultant changes, besides, the intricacies in feeding, and resting preferences. The adult mosquitoes rest in places/areas, less disturbed, dark, and damp place with conducive microclimatic conditions [5,6]. Urbanization and better economy have forced the urban dwellers to have modernized, long-lasting roo ng structures such as concrete instead of the older thatched and tiled roof which provided ample places for the vector mosquitoes to hide and rest [7].
Further, the increased use of mosquito repellents has driven the less susceptible An. stephensi out and therefore are forced to nd out suitable resting places in the vicinity of human and cattle populations [5,8,9]. Hence, identi cation of suitable collection methods, potential target sites, and appropriate time of collection is crucial in yielding better estimates of adult vector populations in urban environments.
The present study was done following the resting collections in cattle sheds (dusk) and pyrethrum spray sheet collection in human dwellings (dawn) [5]. The aim of the study was to check the possibility to have more accurate assessments for adult vector density by performing different collection methods (resting in cattle sheds and CDC light trap in human dwellings all night) and timeframes other than the existing routine methods.

Study site
The study was conducted at Besant Nagar (13.0002˚N, 80.2668˚E), a residential area with slums adjacent to the seashore in the south-eastern part of Chennai, characterized by its meso-endemic perennial transmission of malaria, predominantly Plasmodium vivax, by Anopheles stephensi [4,5,10]. The study was conducted in two phases, (i) resting dawn collections were undertaken in cattle sheds from Jan 2015 to Dec 2016 to nd out the possibility of collecting the maximum number of An. stephensi compared to dusk collections done previously and to ascertain the most appropriate collection strategy [5] and (ii) light trap collections from Jan 2016 to Apr 2017 to assess the e ciency and reliability of CDC light trap in sampling the anophelines including the local malaria vector Anopheles stephensi, in human dwellings besides, other mosquito species, Culex and Aedes species.

Resting Collections (dawn) In Cattle Sheds
Cattle sheds were surveyed in the dawn (04:30 − 06:00 am) using a ashlight and an oral/mouth aspirator. The sheds were selected based on the previous longitudinal surveys [5]. In each cattle shed, 15-30 minutes were spent depending on the size/area and presence or absence of mosquitoes at the time of collection. The collected mosquitoes were brought to the laboratory in a cold chain condition and identi ed to the species level following standard identi cation keys. The females were enumerated and graded based on their abdominal conditions. The late-stage fed, gravid, and/or semi-gravid appearance of the abdomen was considered as resting stages, while the unfed guts, and/or freshly fed as feeding stages [11]. The man-hour density (MHD) of An. stephensi was calculated by dividing the total number of female mosquitoes collected by total time spent for 1 hour period i.e., (Total female An. stephensi collected/total time spent) × 60 [5]. The mosquitoes were then processed for circumsporozoite sandwich enzyme-linked immunosorbent assay (ELISA) following the Malaria Research and Reference Reagent Resource Center (MR4) protocol [12].

Light Trap Collections In Human Dwellings
A total of 203 trap collections were carried out in 113 houses out of which, 107 (52.7%) were indoor and 96 (47.3%) outdoor. In respect of the light traps placed in 113 houses, 90 houses had both indoor and outdoor placement of traps. However, in the remaining 23 houses, the traps were either placed indoor (17) or outdoor (6) which was mainly based on the permission of the house owner/ head of the family and availability of appropriate locations/places to x them. Among the 203 collection sites, An. stephensi was trapped in 98 collections which include 79 out of 167 collections (47.3%) in thatched houses, 4 out of 11 (36.4%) in concrete houses, 15 out of 21 (71.4%) in asbestos houses, and none in 4 tiled houses. A total of 224 female An. stephensi were collected, 180 in thatched houses, 9 in concrete, and 35 in asbestos roofed houses (Table 2).  (Fig. 2, 4). During the study period, the average temperature ranged from 26.05ºC to 28ºC in winter, although it was 29.65ºC to 32.75ºC and 27.65ºC to 31.95ºC in summer and monsoon seasons. The average relative humidity ranged from 68.5 to 75% in winter, 68 to 75.5% in summer, and 62 to 72.5% in the monsoon season.
Composition of other mosquito species collected along with An. stephensi in human dwellings in both indoor and outdoor is illustrated in Fig. 3b. Interestingly, more species were attracted/trapped indoor than outdoor. Culex quinquefasciatus was the predominant species both indoor and outdoor followed by An. stephensi.
Binary logistic regression analysis revealed that presence of An. stephensi was signi cantly in uenced by

Variables And Data Analysis
The data was analyzed using SPSS version 21. The different parameters obtained and selected for analysis were seasonal variations, roof type of the houses, indoor or outdoor, the number of rooms in the house, number of household members, use of repellents, monthly data of mean temperature, mean relative humidity, total rainfall, and mean wind speed as they were likely to in uence the availability of vectors during the collection period [18][19][20]. Binary logistic regression analysis was performed to nd out the parameters that affect the collection of Anopheles stephensi using a light trap.

Resting collections in cattle sheds
Resting collections undertaken in cattle sheds from 2014 to 2016 showed that collections made at dawn (2015 and 2016) yielded more Anopheles stephensi than at dusk (2014) in the same collection settings (Table 1) Table 1). The reduction in vector incrimination in spite of the high vector density was also re ected in the declining malaria prevalence observed in the study site compared to the previous years. The adult vector density peaked during November 2015 and was maximum in July 2016 (Fig. 2). MHD was observed to be very high (115.7) in the 2016 dawn collection due to heavy rainfall (cyclones) during the end of 2015 and mid of 2016 (Fig. 2). The composition of other mosquito species collected along with An. stephensi in cattle sheds (2016) has been depicted in Fig. 3a. Culex gelidus was perceived to be more abundant (44.63%), followed by An. stephensi (38.14%) and Culex quinquefasciatus (16.23%). Other species such as Armigeres subalbatus, Aedes aegypti, Anopheles subpictus, An. vagus, An. barbirostris, An. annularis and Mansonia annulifera collected were in negligible numbers (1%).

Discussion
The present study analyzed the e ciency in terms of the yield of adult vector collections during dawn and evaluated the CDC-light traps placed inside or outside of residential dwellings in the urban slum settings of Chennai. Thatched houses were surveyed more since a number of vector mosquitoes were collected in these structures besides, the results of the vector incrimination in earlier studies [5,21]. While the aspirator collections in the cattle sheds and pyrethrum spray sheet collections in human dwellings were used as the main collection tools to get the adult vector density, collections with CDC light traps were used as a means to encompass the whole night period to collect the mosquito fauna foraging the different quarters of the night.

Dawn Collection (resting) Of Mosquitoes
The dawn resting collections in 2015 yielded 3 times the number of adult mosquitoes compared to the dusk collections in 2014. Surprisingly, there was an increase in vector density and it was even twice more in 2016 compared to the dawn collections in 2015. Although there was an escalation in the vector density, malaria incidence was declining due to the intensi ed vector surveillance activities targeting two vectorborne diseases (malaria and dengue) and hence the focussed intervention measures. In contrast, Anopheles mosquito surveillance in Madagascar revealed that the mosquito density was least during the early morning collections whilst they peaked at 00:00 hour [22]. Elsewhere in Uganda, the vector densities peaked up during early evenings and mornings [23]. The increased number of mosquitoes collected could be attributed to the collection time, as the early morning time corresponds to minimal disturbance and hence the mosquitoes could rest for a longer period. Furthermore, it may also be due to the possible chance of availability of all the mosquitoes, which had otherwise own in seeking appropriate, cryptic resting places during the later hours of the night besides, fully fed mosquitoes resting and unable to y after a full, heavy blood meal. Studies from Benin have reported a shift towards the early morning biting behaviour of malaria vectors [24]. This is analogous to the present ndings observed in the highly populated urban slum settings. Such possibilities need to be examined further over the years for the successful implementation of effective control measures targeting the vectors.

Cdc Light Trap Collection
Overall, there was a 24% detected increase in the number of An. stephensi trapped indoors compared to outdoors. This is consistent with the other studies that the CDC light traps are good at trapping indoor mosquitoes, particularly the stage of host-seeking [25]. Further, unfed female An. stephensi were more compared to other stages in both indoor and outdoor. This could be due to the presence of the light trap, which partially diverts the mosquitoes from seeking potential hosts [26]. Unfed and half-fed were also more in outdoor collections compared to indoors. Whereas fully fed, the semi gravid and gravid stage was observed to be twice more indoor.
The presence of more fully fed, semi gravid, and gravid stage vector mosquitoes indoors re ects the resting preference in thatched structures and is accordant with the previous studies undertaken at the same site [5]. Moreover, when the proportion of vectors incriminated with malaria parasites was checked, it was found that the percentage of positivity was considerably higher (2.3%) than the previous study (0.65%). This nding a rms the importance of thatched structures as preferred resting sites of the malaria vectors in an endemic urban setting. Further, it points out that, in a busy urban scenario, identi cation of such a potential resting preference is really worth as it can help to locate the resting hotspots and can prioritize the vector surveillance efforts. This coupled with better allocations of supplies in terms of resources, manpower can signi cantly aid in scaling down the disease prevalence. Nevertheless, the presence of more infected vectors in the outdoor underlines the need for evidencing outdoor biting and implementing effective control strategies.
Strati cation by season revealed that the presence of An. stephensi was signi cantly in uenced by the seasons. An. stephensi count varied throughout the study period, unlike in previous years. Over the years, it has been noticed that the climate of Chennai is unpredictable, similar to the scenario detected elsewhere in the world. Precipitation/rainfall varies (Fig. 2) every year which affects the mosquito density.
Generally, the most abundant species captured in traps were Culex quinquefasciatus, probably because of its preference for human beings [27]. Hence, it is assumed that CDC light traps are bene cial in multiple ways, in terms of mosquito diversity assessments, species prevalence, vector control, and protection from the bites of other mosquito species while targeting Anopheles species. These ndings are in tune with previous studies, which have shown that CDC light traps were attractive to various genera of mosquitoes [15,28]. Overall, the number of Aedes species collected in this study was low and most of the captured ones were Aedes aegypti. Previous studies have also indicated the ine ciency of CDC light trap in capturing Aedes species [15,29]. This could be ascribed to their diurnal biting nature and restricted ability to y high [30]. The collected Anopheles mosquitoes represented all the grades of abdominal conditions. However, with a higher number of feeding stages, it is assumed that feeding status did seem to impact capture e ciency while comparing indoor and outdoor trap locations. This stands in support of a previous study that indicated a preferential capture of fully fed/ unfed by CDC light traps in indoor locations in Zambia [31].
The number of female An. stephensi was signi cantly in uenced by various factors. For instance, the use of repellents was having an adverse impact on the presence of the vectors. This may be due to the deterrent effects of the chemicals that forced the mosquitoes to drive away from the houses and its close proximity [8]. Also, the number of rooms was found to have a positive correlation with the number of An.
stephensi. An increased number of rooms, though not necessarily, may indicate the presence of more inhabitants and thereby providing an escalation in the olfactory cues. Also, it will offer the vectors plenty of hideouts, which prevents them from being caught. Surprisingly, the number of inhabitants was negatively correlated with the number of An. stephensi collected by the light traps. It was assumed that, as the olfactory cues got stronger, the mosquitoes were preferably attracted to humans than to the traps. However, the number of female An. stephensi was not appeared to be in uenced by climatic factors like local mean temperature, relative humidity, rainfall or wind speed, or the location/placement of traps, indoor or outdoor.
It is well known that adult collections of Anopheles stephensi in an urban setting are an onerous task unlike locating the immature breeding habitats. The situation is more acute and complicated when malaria is on a declining trend. This practical di culty may be attributed to the successful and more organized vector control efforts targeting the potential breeding habitats thereby reducing/ eliminating them. Further, the intensive vector-control efforts by the health department over the years against dengue has really resulted in a collateral bene t to malaria vector control as few of the clear breeding habitats are mutually shared by both malaria and dengue vectors. In addition, the attitude and exercise of the community to switch over from the age-old cemented overhead tanks to easily available and economically cheaper bre tanks with mosquito-proof lids provide ample occasions to prevent vector breeding and therefore low adult abundance. Interestingly wells which are also another potential breeding habitat of Anopheles stephensi have been drastically reduced and most of them are closed and converted to tube wells. This custom may be due to repeated monsoon failure over the years resulting in the low water table, increase in household consumption, all of which led to a compounding effect of wells being dried up which were otherwise perennial in nature. Wells due to its damp, dim conditions act as an ideal resting habitat for many mosquito species including malaria vectors. In the absence of such habitats, the vectors are forced to rest elsewhere with less disturbed, appropriate microclimate conditions and hence cannot be collected easily unlike the earlier period. The progressive and healthy urban economy over the years has also resulted in the execution of changing the roof structures of houses that were once thatched or tiled, which belonged to economically weaker, low middle-income group people to asbestos and concrete roof structures which do not serve as an ideal resting habitat for malaria vectors.
Furthermore, the intense and regular indoor repellent usage over the years owing to the mosquito nuisance and also with the rising economy has driven the susceptible An. stephensi irrespective of the density to take shelter or rest outdoors, regardless of any ideal place it could nd.
One of the limitations of the study was that parallel dusk collection (aspirator) data could not be generated due to the lack of manpower for undertaking collections simultaneously and refusal from the community in disturbing the limited privacy in urban settings especially a busy metropolitan city like Chennai. The trap placements were sparse, irregular and the number of traps differed considerably between months and house structures as these contingencies are sometimes bound to happen in eld settings. The study period had to be extended for another 4 months to con rm the unusual peak density of mosquitoes following the devastating oods during Nov-Dec 2015 (rainfall − 752 mm). So in order to understand the unprecedented peak vector density and to nd out the actual scenario, the study was extended to 4 more months (Jan-April 2017) of the following year, the period of which corresponded to the preceding year with normal precipitation (rainfall − 151.1 mm). It is unclear whether the 34% of vector mosquitoes were human blood-fed or bovine blood-fed. The reason for omitting the blood meal analyses was mainly because of a previous study [5] from the same site indicating 95% of the analyzed An. stephensi positive for bovine blood. Also, the unfed mosquitoes were not ascertained if they have newly emerged (nulliparous) or parous females that have not yet taken a blood meal. Further, we couldn't con rm or exclude the possibility that concurrent usage of nets or repellents increased the number of outdoor collections and so decreased indoor biting. Nevertheless, the study construes the various ndings that are intervention driven points translational to the vector control programme.

Conclusion
The different collection (resting and light trap) strategies were assessed to nd out the appropriate collection method and time which could yield the maximum collection of An. stephensi with the available resources. In this regard, the present study indicated dawn collections during the early hours to be the most appropriate time to collect wild An. stephensi in an urban setting. The study also highlighted the need for an in-depth adult mosquito/vector collection/trapping strategies (resting or light trap) in an urban setting to determine the capture/collection e ciency. Besides, an appropriate time interval targeting the precise location/ site (indoor or outdoor) based on the mosquito/vector behavioural changes is a requisite to assess vector abundance for effective control measures. Elaborate and more extensive collection strategies in future depending on the resources may be useful to determine the capture e ciency in an endemic area.

Consent for publication
Not applicable.

Availability of data and materials
The dataset generated during and/or analyzed during the current study shall be available on request mentioning the purpose by contacting the corresponding author.
AE designed the experiment with input from MTM and VJ. AE, JAJ, AA, and SR participated in study design. SR, ST, and AE wrote the manuscript. JAJ and AA conducted the experiment at the study site. VJ, SR, AA, and ST contributed to data analysis. AE, SR, VJ, and MTM edited the manuscript. All authors read and approved the nal manuscript. Monthly man hour density in cattle sheds with corresponding malaria incidence and rainfall Anopheles stephensi collections in cattle sheds and human dwellings

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.