Entomological survey for identification of Aedes larval breeding sites and their distribution in Chattogram, Bangladesh

Studying the characteristics of Aedes mosquito habitats is essential to control the mosquito population. The objective of this study was to identify the breeding sites of Aedes larvae and their distribution in Chattogram, Bangladesh. We conducted an entomological survey in 12 different sub-districts (Thana) under Chattogram City, during the late monsoon (August to November) 2019. The presence of different wet containers along with their characteristics and immature mosquitoes was recorded in field survey data form. Larvae and/or pupae were collected and brought to the laboratory for identification. Different indices like house index, container index, and the Breteau index were estimated. The multiple logistic regression analysis was applied to identify habitats that were more likely to be positive for Aedes larvae/pupae. A total of 704 wet containers of 37 different types from 216 properties were examined, where 52 (7.39%) were positive for Aedes larvae or pupae. Tire, plastic buckets, plastic drums, and coconut shells were the most prevalent container types. The plastic group possessed the highest container productivity (50%) whereas the vehicle and machinery group was found as most efficient (1.83) in terms of immature Aedes production. Among the total positive properties, 8% were infested with Aedes aegypti, 2% with Aedes albopictus, and 1% contained both species Ae. aegypti and A. albopictus. The overall house index was 17.35%, the container index was 7%, and the Breteau index was 24.49. Containers in multistoried houses had significantly lower positivity compared to independent houses. Binary logistic regression represented that containers having shade were 6.7 times more likely to be positive than the containers without shade (p< 0.01). These findings might assist the authorities to identify the properties, containers, and geographical areas with different degrees of risk for mosquito control interventions to prevent dengue and other Aedes-borne disease transmissions.


Background
Aedes mosquitoes (Diptera: Culicidae) are a highly efficient vector of various arboviruses, including dengue, chikungunya, and Zika around the globe. Dengue fever is an important mosquito-borne disease (MBD) that poses a risk for more than 128 countries globally [1]. Aedes aegypti is the primary vector of the dengue virus (DENV) found in urban areas due to its anthropophilic blood-feeding behavior. Aedes albopictus belongs to the same subgenus (Stegomyia) considered as a secondary vector of DENV. They are primarily found in rural areas and peridomestic sites of urban areas like parks and green corridors interspersing housing estates. Both of these species were first recorded in Bangladesh in 1952 [2]. Since 2000, dengue fever (DF) cases have been reported every year in all major cities in Bangladesh [3]. The outbreak of chikungunya fever took place in 2008, and the Zika virus was also introduced in the country in 2016 [4,5]. During 2019, Bangladesh recorded its largest dengue virus outbreak with 101,354 cases, more than double the cases in the last 20 years [6]. The Aedes spp. mostly breed in artificial water-holding containers but have been reported in natural containers as well [7]. Vector control remains the center of dengue prevention options, which can be done by limiting the transmission potential by reducing the emergence of adult mosquitoes. This could be achieved by targeting the aquatic habitats of the immature stages of Ae. aegypti through source reduction or biological and/or biocidal treatment [8,9]. The productivity of a container type depends on a range of factors, such as size and shape, the purpose of use, their locations, method of filling (passively/actively), the material of the container (plastic, metal, cement/clay, etc.), temperature, availability of food, and competition among co-species [10,11].
Various entomological indices are used to measure dengue vector infestation in and around infrastructures, such as homes and buildings [12]. Examples of such indices are the different Stegomyia indices-house index (HI, the proportion of Aedes positive houses) and container index (CI, the proportion of Aedes positive containers), and the Breteau index (BI, the number of Aedes positive containers per 100 houses) [13,14]. There is a lack of a sensitive vector surveillance tool to estimate the vector density in the outbreak areas. Therefore, the vector abundances for both Ae. aegypti and Ae. albopictus are still expressed as house index, Breteau index, and container index.
A few studies on Aedes mosquito abundance and breeding preference were conducted in the capital city Dhaka, Bangladesh. However, data on other big cities like Chattogram, the 2nd popular city in the country, are still absent. Therefore, during the late monsoon (August to November) of 2019, we conducted a comprehensive entomological survey in Chattogram. The expansion of residential areas to natural ecosystems, continuous changes in biodiversity, and increasing industrial areas in Chattogram promote the human-vector contact and the transmission of arboviruses to humans [15]. The objective was to identify the breeding sites of Aedes larvae and their distribution in Chattogram metropolitan area.

Methods
This entomological survey was conducted in late monsoon in 2019 between August and November, when dengue incidence was high in the study area and throughout Bangladesh [6]. The study location was Chattogram City of Bangladesh. Chattogram is the second largest city of Bangladesh and an area poorly studied in terms of mosquito surveys. A total of 12 sub-districts (Thana) under the Chattogram City Corporation (CCC) area were chosen for this survey (Fig. 1). A Thana is defined as the small administrative boundaries within the metropolitan area.

Entomological survey
An entomological study for DENV was designed to detect immature stages of the Aedes aegypti (L.) (Diptera: Culicidae). Our survey targeted all groups of juveniles (1st-4th instar larvae and pupae) by inspecting all accessible water-holding containers in public and private areas to identify the most productive and efficient container types of these species. The wet container is defined as any container with stagnant water at the time of the survey. Samples were taken by pipetting, dipping, or netting [16] in small plastic jars with water. Each sample jar was labeled with the unique identification number, date, location, and number of collected larvae/pupae. Breeding habitats of the collected mosquito species were recorded in a pre-defined survey data sheet during the sample collection (Appendix 1). Different indices were calculated to document the primary breeding source and density of the Aedes mosquito. A container was recorded as positive for Aedes if one or more juvenile Ae. aegypti or Ae. albopictus was found in the given type of container and was distinguished from those with no juveniles (negative). During the survey period, the ambient temperature of the area ranged from 21 to 32°C. The rainfall was 435.6 mm in August, and then gradually decreased to 18.9 mm in November 2019 [17].

Selection of properties
A total of 18 different locations were surveyed, including government-owned properties. The properties were purposely selected from each Thana and were categorized into nine classes, depending on the possession of living properties, working stations, and other public gathering places, as shown in Table 1. Moreover, all properties were broadly divided into two classes-residential and  non-residential. All types of households were included under residential areas whereas other properties considered as non-residential areas.

Classification of containers found as larval development sites
The containers were divided into two broad groups based on their purposes: controllable and disposable ones. Controllable containers were household containers that could be manipulated by man to avoid mosquito larval breeding. It included concrete tanks, metal drums, flower pots, aluminum tanks, small buckets, and other plastic containers used to carry or store water. Disposable containers were those that are not used in households, are abandoned or stored in backyards having the potential as breeding sites in the rainy season. Examples of disposable containers include tires, cans, and tubs.

Identification of mosquitoes
After collection, the larvae and/or pupae were brought to the pathology and parasitology laboratory of Chattogram Veterinary and Animal Sciences University, located within the city of Chattogram. Larvae were identified under microscopes in the laboratory. Pupae and the rest of the larvae were reared in rearing trays for the identification of adult mosquitos. Species identification was completed using standard identification keys as described in a literature [18]. The laboratory findings were recorded in the corresponding survey sheet.

Determine the key containers
All wet containers were divided into six categories based on the materials they were composed of or the purposes they served. The categories were plastic receptacles, tin and metal receptacles, cement and clay receptacles, natural receptacles, vehicle and machinery parts, and other receptacles. The role of these container categories regarding the production of Aedes mosquito was estimated by the value of container productivity and container efficiency [19] PC ¼ positive containers CP ¼ container productivity no:of immaturex Â 100=all immature ð Þ CE ¼ Container efficiency productivity=prevalence of container ð Þ Prevalence of container ¼ no:of wet containers Â 100=all containers

Estimate different indices
The number of different wet containers, the percentage of positive containers, and the percentage of Aedes larvae in each container category were calculated to determine the various Stegomyia indices as follows [16]: Container index: the percentage of water-holding containers infested with larvae and/or pupae House index: the percentage of house infested with larvae and/or pupae Breteau index: the number of positive containers per 100 houses inspected.

Statistical analysis
Zero-inflated negative binomial regression analysis was performed to find the factors associated with the number of immature Aedes positive mosquito per container. The binary logistic regression was used to find the odds of the background characteristics causing the containers to be Aedes positive. The exponential of the logistic regression parameter was the odds of the category of the background characteristics being Aedes positive of water-holding containers. The exponential of zero-inflated negative binomial regression was the ratio of the average number of immature Aedes positive mosquitoes among the groups of background characteristics. Data analysis was conducted in R version 3.5.2, and the binary logistic regression model was performed in SPSS version 25.

Results
A total of 704 wet containers of 37 varieties were identified around 216 surveyed properties, with a mean of 3.3 containers per property. Of these, 204 (29%) were controllable, and 500 (71%) were disposable containers. More containers were identified on privately owned properties (N = 531, 75%) than government (N = 173, 25%) owned properties. We collected 573 Aedes juveniles, including 101 pupae. The number of Ae. aegypti (n=371) was higher than the number of Ae. albopictus (202). We calculated the house index as 17.35%, container index 7%, and the Breteau index 24.49. The distribution of wet containers found in the entomological survey is summarized in Table 2.

Distribution of positive containers and juvenile Aedes spp.
Among the total of 704 wet containers, 49 (6.9%) were positive for immature Aedes mosquitoes. Among positive containers, 31% (n=15) were controllable (plastic bucket, plastic drums, plastic bag, and earthen jar), and 69% (n=34) were disposable (containers like tires, coconut shell, cock sheet, mineral water jar, and bamboo hole). A total of 32.65% (n=16) positive containers were found in government-owned properties, and the rest 67.35% (n=33) were in private properties. Among positives, 96% (n=47) of containers was under shades, and 39% (n=19) had vegetation in and around the container. Independent houses possessed maximum numbers of positive containers amounting 33% (n=16). Other than living places, different open places or parks contained 13% (n=6), and educational institutions had 10% (n=5) of the positive containers.
The highest positive containers were recorded in Panchlaish followed by Pahartali and Halishahar Thana. No positive containers were found in Bayezid Thana. Figure 2 shows the distribution of positive containers in different Thanas.

Determining the key containers
According to immature Aedes production in different container categories, the plastic container had the most abundantly (33%) found positive and possessed the highest container productivity (50). Vehicle and machinery group was the most efficient (1.83) and 2nd most productive (30) group. The natural group contained 27% positive containers with container productivity of 11. So, these three container categories could be considered as necessary for Aedes spp. mosquito breeding. The role of all container categories is shown in Table 3.
Zero-inflated negative binomial regression (Table 4) revealed that the number of immature Aedes mosquito produced per container had a significant association with the property category and shade (p< 0.01). Binary logistic regression (Table 5) elucidated that containers having shade had 6.7 times more likely to be positive than the containers without shade (p< 0.05). Containers in multi-storied houses had significantly lower positivity in comparison to independent houses. Besides, Kotowali and Bandar Thana produced significantly lower number of positive containers (p< 0.05) as their odds found below one.

Discussion
We inspected a broad spectrum of wet containers from different properties in the second most populated city of Bangladesh and identified Aedes larvae and/or pupae. The total number of Ae. aegypti larvae and/or pupae was higher than the number of Ae. Albopictus, and the record was consistent with previous studies [20,21]. Opposite results were found in Malaysia that 77% of breeding containers were positive for Aedes albopictus and 23% for Aedes aegypti [22]. As the survey was conducted in city areas only, the breeding sites of Ae. albopictus were reached partially which might be a reason  for the lower count of Ae. albopictus. The mean number of wet containers per property was 3.3 in our study, which is comparatively lower than the mean numbers of two previous studies 4.20 found in Dhaka [7] and 7.9 found in Southern Mexico [23]. A possibility could be because people were more aware due to the devastating dengue epidemic in 2019. The overall house index (HI), container index (CI), and Breteau index (BI) were calculated high which is supported by a similar study conducted in Dhaka earlier stating the overall HI 14.2%, BI 24.65, and CI 5.9% [7]. All of the indexes showed a high level of risk for dengue transmission. A survey in Nepal found higher CI in the post-monsoon period than monsoon [24]. A recent study in Malaysia showed maximum HI= 13.33%, BI=13, and CI=19.05% [22]. In a literature [13], it has been criticized that these indices are of operational value and are of limited use in assessing the transmission risk. However, these indexes are indicative of the global transmission risk at the community level.
The biggest portion of Aedes larvae containing waterholding containers was under the group of disposable containers. The estimated larval productions in disposable containers in other settings were 70% and 55.4% [23]. This indicates that the proper and regular disposal of unused containers will make a great contribution to vector control programs. The common containers for Aedes breeding were plastic containers (30%) in Kuala Lumpur, Malaysia [22]. As individual container types, we found the contribution of unused tires as the highest for Aedes larval breeding along with other key containersplastic buckets, plastic drums, coconut shells, and flower tubes and trays. Our results are consistent with the findings of two previous studies in Dhaka, Bangladesh, conducted during 2011-2013 [25,26]. A study in Nepal stated discarded tires as the preferable wet containers [24]. Another study revealed flower pots as the principal types of containers with Ae. aegypti larvae [27]. Figure 2 shows that Panchlaish, Pahartali, and Halishahar Thana contained maximum positive containers. These Thanas possess many residential areas residing a huge population of the city. And many parts under these areas become flooded after a rain due to poor sewerage system, whereas Bayezid Thana had no positive containers as this area is an intensive industrial hub with limited access and having minimum residential properties. Within a community, not all houses have the same potential for larval breeding. The identification of those houses that are likely  to have high larval production is important to better direct control activities. Independent households having disposable wet containers in shaded outdoor premises were significantly associated with the infestation of Aedes larvae [7]. Notably, the survey found more juveniles in containers of private properties compared to those on government properties, similar to the findings of a previous study which showed more containers on privately owned properties than in government properties [28]; private properties also have 1.7 times greater odds (95% CI = 1.1-2.5) of being infested compared with containers on government properties [29]. We observed that Ae. aegypti and A. albopictus co-existed in different properties supported by an earlier study in Jinghong City, China [30]. It suggested that both species should be managed during the dengue epidemic season. This study revealed that independent houses were the most infested with containers possessing Aedes larvae, followed by slums and open places/parks. The findings are concordant with the earlier studies conducted in other cities of the country which stated that the household positivity rate was the highest in independent houses (18.6%), followed by slum houses (14.3%), semipermanent houses (12.9%), and multi-storied houses (12.8%) [7]. A study in Japan during a dengue epidemic showed that more than 80% of dengue patients visited a city park where Ae. albopictus was present [31]. It reveals that positive containers in open places/parks might impose risk of the dengue fever for city dwellers. The observations or experimental units were the water-holding containers spread over the different locations in Chittagong city. We analyzed the count data of the number of immature Aedes-positive larvae and/or pupae per container, with access zero accounts for non-Aedes, by a statistical model. A multiple regression model was applied using the predictors of background characteristics (container function, ownership, vegetation, shade, property category, container category, and geographic location) of water-holding containers [28]. To deal with access zero, we considered the zeroinflated Poisson regression model [32] and zero-inflated negative binomial regression model [33] to find the association between the number of immature Aedes-positive mosquitoes per container and the background characteristics of the water-holding containers. Both of the considered models are modifications of the count data regression model. The Akaike Information Criteria (AIC) value was calculated to find the improved model between the two [34,35]. The zero-inflated negative binomial regression model (543.82) has less AIC value than the zero-inflated Poisson regression model (693.16). Hence, the zero-inflated negative binomial regression model better fit our data.
Containers located under any shelter that did not allow rain water to get into are considered as containers with shade. We observed that wet containers under shade were more likely to have Ae. aegypti, unlike previous findings where more larvae and pupae were detected in containers without shade than the containers located in areas with full shades [26]. The opposite findings may have been due to the classification of containers as partially shaded and fully shaded by the researchers. However, a recent study in Dhaka found that containers with partial shades produced 4.6 times more pupae than without shade [25]. Containers located outdoors and proximal to vegetation were significant for producing larvae [29]. We also found odds greater than one for containers possessing vegetation. Vegetation in and around containers might enhance mosquito breeding by facilitating resting sites for gravid females to lay eggs and important sugar-feeding resources for larvae [36]. The larval production of any species of mosquitoes is significantly affected by vegetation near the containers [26].
Probability sampling was not performed in this study which could lead to a selection bias in the result. We examined the surroundings outside of the properties, excluding the inside. The survey covered different residential and non-residential properties like open places/parks, educational institutes, hospitals but devoid of shopping/open market areas, other office buildings, and religious buildings. The total number of containers was also limited to some extent. A few recent studies used different container parameters like size or volume of the container, exposure to sunlight, cover status, and filling methods for evaluating container productivity. We considered exposure to sunlight/shade and vegetation status, function of composing materials of containers. Previous studies suggested that the extensive vector control program by reducing breeding sites could maintain a low Aedes house index [27]. Therefore, the weekly cleaning of the water-holding containers could effectively control larval production [21].

Conclusions
The study detected Ae aegypti in 8%, Ae albopictus in 2% of the properties, and the co-existence of both Ae. aegypti and A. albopictus in 1% of properties, which posed high risks of dengue transmission. The presence of shades on the breeding sites could significantly increase the risk of having Aedes spp. mosquito larvae. Plastic containers were the most abundant and productive containers while the vehicle and machinery category was found as the most efficient container group for Aedes mosquito breeding. We recommend vector control through regular inspection and destruction of potential container types to prevent the dengue outbreak.